MISSISSIPPI RIVER GULF OUTLET EFFECTS ON

STORM SURGE, WAVES AND FLOODING DURING

HURRICANE KATRINA

EXPERT REPORT

BY

G. PAUL KEMP, Ph.D.

ROBINSON V. UNITED STATES

JULY 11, 2008

MISSISSIPPI RIVER GULF OUTLET EFFECTS ON STORM SURGE, WAVES
AND FLOODING DURING HURRICANE KATRINA

TABLE OF CONTENTS

PREFACE 2

1.0 INTRODUCTION 5

2.0 FUNNEL GEOGRAPHY, TERMINOLOGY AND KATRINA

FLOOD DAMAGES 8

3.0 BRIEF HISTORY 21

4.0 EARLIER SURGE STUDIES 40

5.0 METHODS 69

6.0 FUNNEL EFFECT ON SURGE 96

7.0 EFFECTS OF REACHES 1 AND 2 ON SURGE 148

8.0 EFFECTS ON WAVES 159

9.0 NEED FOR INSTITUTIONAL REFORM 185

10.0 CONCLUSIONS 197

11.0 REFERENCES 199

12.0 APPENDICES 205

A. QUALIFICATIONS

B. MRGO CHRONOLOGY

1
 MISSISSIPPI RIVER GULF OUTLET EFFECTS ON STORM SURGE, WAVES
AND FLOODING DURING HURRICANE KATRINA

G. PAUL KEMP, Ph.D.

July 11, 2008

PREFACE

I, G. Paul Kemp, am a geologist and oceanographer currently employed by the National

Audubon Society, and at the time of Katrina by the Louisiana State University (LSU)
Hurricane Center. My qualifications are attached as Appendix A. The opinions
expressed in this report are solely my own, and not those of the National Audubon
Society or LSU. This is the fourth expert report that I have written as a member of the
Robinson expert team, including ones submitted to the Court on July 28, 2007 (Kemp
Class Certification 2007a), September 14, 2007 (Kemp 702(c) immunity 2007b) and May
1, 2008. The current report supplements and essentially replaces the report of May 1,
2008, with the addition of information developed over the past two months. I incorporate
by reference everything I have written in the three previous reports into the current
report.

The Government takes the position that the Mississippi River Gulf Outlet navigation
project (MRGO) had no effect on the catastrophic flooding of Greater New Orleans
during Hurricane Katrina. Our analyses show that the Government is incorrect. It is my
opinion that the MRGO navigation project:

(1) created a funnel, the dangerous convergence of channels and spoil

disposal areas, later augmented by addition of the Lake Pontchartrain &
Vicinity (LPV) berms east of New Orleans, and by the subtraction of
buffering wetlands, that foreseeably amplified the threat posed by
hurricane surge to the Greater New Orleans area; and
(2) greatly enlarged the original Gulf Intracoastal Waterway (GIWW)
connection between the throat of the funnel and the IHNC (MRGO Reach
1), which foreseeably increased surge transmission into the city earlier,
adding to the height and duration of surge experienced in the IHNC during
Katrina and contributing to the early failure of floodwalls and levees
adjacent to the IHNC; and
(3) created a Reach 2 channel with unstable side slopes that caused it to
predictably expand over time, reducing the natural marsh buffer that
previously separated it from Lake Borgne and from the adjacent LPV
berms, thereby compromising foreshore protection fort the man-made
surge protection elements and hastening the onset of damaging wave
action on these delicate structures (EBSBs) so that they breached earlier in
the storm sequence; and

2
 (4) created a channel in Reach 2 that predictably exposed freshwater swamps
and marshes within the Lake Borgne funnel to increased salinity,
accelerating their conversion to open water and thereby reducing the
wetland surge and wave buffer between Lake Borgne and populated areas;
and
(5) was a substantial contributing factor to the catastrophic flooding of the
Robinson plaintiffs’ homes and communities.

In addition, throughout the history of the MRGO project, up to the time of Hurricane
Katrina, the U.S. Army Corps of Engineers (USACE or Corps), in my opinion, has
continuously adopted a policy of denial and deliberate disregard of well documented
adverse effects of the channel, most of which were known or suspected prior to
construction and others that became apparent later, soon thereafter ensuring that a real
threat for flooding during hurricanes remained unaddressed for five decades. Throughout
this period, the USACE issued increasingly soothing but inaccurate statements to two
generations of New Orleans and St. Bernard residents that they were well protected and
that the MRGO project posed no hazard to their lives or property.

Furthermore, this approach is much the same today, in my opinion, coloring the
conclusions of the Corps sponsored investigations that have followed Katrina, and the
nature of solutions that have been proposed. This, in particular, includes an obdurate
unwillingness to provide wave protection to the hastily rebuilt berms that parallel MRGO
Reach 2 despite overwhelming evidence that they were largely destroyed by waves.
Today, the Corps still refuses to acknowledge, in the face of compelling scientific
evidence, that the MRGO project was a significant cause of the early and catastrophic
flooding of the Upper and Lower 9th Ward, St. Bernard Parish and New Orleans East
during Hurricane Katrina.

Ultimately, the extent of damage and the harm caused by the flooding of St. Bernard and
the Lower 9th Ward through the MRGO was not significantly reduced by the LPV berms
and floodwalls. The structures that Dr. Bea has described as “earthen berm/spoil banks”
(EBSBs) (to differentiate them from properly engineered coastal defense dykes) along
MRGO Reach 2 served merely as “speed bumps” that were swept aside during Katrina by
the surge and waves generated in, and transmitted by the MRGO channel. Most of the
flooding of the Lower 9th Ward and New Orleans East, as well as a significant portion of
the early flooding of the Orleans Metro area, is attributable to the enlarged cross-section
of MRGO Reach 1.

Finally, I believe that the USACE, and certainly the larger oceanographic community,
possessed a knowledge base prior to the construction of the MRGO project, and certainly
after the Betsy disaster, that should have led to actions to tightly integrate the MRGO
economic development project with the LPV public safety project, and thereby reduce or
eliminate the added and substantial hazards uniquely posed by the ship channel. Instead,
with regard to this particular project and, significantly, not in all comparable cases around
the country, the Corps adopted a policy of institutional denial that stymied efforts of local
leaders, engineers and scientists to address critical, long-known deficiencies that were

3
 CHAPTER 1
INTRODUCTION

The Mississippi River Gulf Outlet (MRGO) is a free-flowing, man-made navigation

channel connecting the Gulf of Mexico to the interior of the City of New Orleans that we
will demonstrate played a substantial role in the initiation and enhancement of flooding
during Hurricane Katrina (Figure 1.1). The MRGO project was approved by the U.S.
Congress under the Rivers and Harbor Act of 1956 (PL 84-455). The U.S. Army Corps
of Engineers (USACE) New Orleans District (NOD) began construction of this project in
1958 and completed it in 1968 at a cost of approximately $92 million. In the process,
they dredged more earth than was moved during construction of the Panama Canal and
destroyed thousands of acres of swamps and marshes. Subsequent operations and
maintenance costs have totaled in excess of $500 million (Gunn, 1977; USACE 2007).
The MRGO was authorized as a 76-mile ship channel with a 36 foot controlling depth,
500 feet wide at the bottom and 650 feet wide at the top, that would cut through the
marshes of lower St. Bernard Parish and, with a somewhat larger cross-section (Figure
1.2), across the shallow waters of Breton Sound (USACE, 1999). The MRGO was
promoted by New Orleans port interests as a federally-funded economic development
project that would stimulate a boom in industrial and port development for New Orleans
and St. Bernard Parish. The anticipated economic renaissance, like the ship traffic that
peaked in 1978 and diminished afterward, never materialized (USACE 2007).

Figure 1.1 Graphic created by the New Orleans Time-Picayune newspaper to

explain the role of the Lake Borgne funnel in the flooding of New Orleans during
Hurricane Katrina.
5
Figure 1.2 MRGO Project from the Inner Harbor Navigation Canal (IHNC) lock in
New Orleans to the Gulf of Mexico (USACE 2007). IPET Reach 1 extends from the
IHNC to the turn southeast and Reach 2 runs through the marsh from there to the
jetties. IPET Reach 3 traverses Breton Sound past the Chandeleur Islands.

1.1 Objectives and Organization of Expert Report

The Government takes a position that the 1958 MRGO navigation project had no effect
on the flooding of New Orleans during Hurricane Katrina or on the damage suffered by
adjacent flood protection structures erected as part of the Lake Pontchartrain & Vicinity
(LPV) project following severe flooding from Hurricane Betsy in 1965. We have carried
out a detailed analysis of the oceanographic dynamics stimulated by Hurricane Katrina.
This analysis focuses on the effect of the MRGO project on surge and waves throughout
the Lake Borgne funnel (Figure 1.1) using information derived from observations made
in this area immediately after the storm coupled with results of a comprehensive
modeling program. This modeling program was carried out primarily by colleagues
working under my direction in the Netherlands at Svasek Hydraulics and the Delft
Institute of Technology, as is described in a chapter to follow. We believe the evidence
we have developed clearly shows that the MRGO project:

(1) created a funnel, the dangerous convergence of channels and spoil disposal
areas, later augmented by addition of the Lake Pontchartrain & Vicinity
(LPV) berms east of New Orleans, and by the subtraction of buffering

6
 wetlands, that foreseeably amplified the threat posed by hurricane surge to the
Greater New Orleans area; and;
(2) greatly enlarged the original Gulf Intracoastal Waterway (GIWW) connection
between the throat of the funnel and the IHNC (MRGO Reach 1), which
foreseeably increased surge transmission into the city earlier, adding to the
height and duration of surge experienced in the IHNC during Katrina and
contributing to the early failure of floodwalls and levees adjacent to the
IHNC; and
(3) created a Reach 2 channel with unstable side slopes that caused it to
predictably expand over time, reducing the natural marsh buffer that
previously separated it from Lake Borgne and from the adjacent LPV berms,
thereby compromising foreshore protection fort the man-made surge
protection elements and hastening the onset of damaging wave action on
these delicate structures (EBSBs) so that they breached earlier in the storm
sequence; and
(4) created a channel in Reach 2 that predictably exposed freshwater swamps and
marshes within the Lake Borgne funnel to increased salinity, accelerating
their conversion to open water and thereby reducing the wetland surge and
wave buffer between Lake Borgne and populated areas; and
(5) was a substantial contributing factor to the catastrophic flooding of the
Robinson plaintiffs’ homes and communities.

Each of the conclusions presented here is addressed in separate chapters to follow,

combining results from new analyses carried out by the Robinson expert team with those
contributed by others since Katrina. To facilitate this discussion, however, we first lay
out in Chapter 2 our understanding of the geography, terminology and hydraulic
significance of the various man-made and natural components of the Lake Borgne funnel,
our study area for this project. Then, in Chapter 3, we describe a chronology of actions
and positions taken by the USACE that cover the roughly 50 year period from the
authorization of the MRGO project through its recent de-authorization, and continuing on
into current mitigation efforts. We review earlier studies of surge dynamics in Chapter 4,
and explain the modeling techniques adopted for the current investigation in Chapter 5
showing how that forms the basis of the technical oceanographic discussion. The results
of that analysis are then discussed in separate chapters (Chapters 6 through 8) that
address sequentially the points enumerated above. Chapter 9 reviews high points of an
MRGO chronology that has been created and is included with this report (Appendix B)
and is followed by a short statement of my opinion in Chapter 10. We conclude with a
summation chapter that pulls together all of the elements supporting our overarching
contention that the MRGO project played a significant causative and contributory role in
the catastrophic flooding and damages experienced by the Robinson plaintiffs in New
Orleans East, the Lower 9th Ward and St. Bernard Parish, and a supporting role in the
inundation of the Orleans Metro area west of the Inner Harbor Navigation Canal (IHNC).

7
 CHAPTER 2
FUNNEL GEOGRAPHY, TERMINOLOGY AND KATRINA FLOOD DAMAGES

The east bank of New Orleans and all of St. Bernard Parish flooded on the morning of
August 29, 2005, when two storm surges sequentially struck the city. The first originated
east of the City in the Lake Borgne funnel, while the second struck the south shore of
Lake Pontchartrain (Figure 2.1). The Lake Borgne surge, although it was similar in
many ways to that in Lake Pontchartrain, attained a maximum elevation about 6 feet
higher against the earthen berms protecting St. Bernard Parish along the MRGO channel,
18 compared to 12 feet (Figure 2.2). Decades of erosion of the MRGO channel banks
into the protective wetland buffer south of Lake Borgne enhanced the severity of the
wave attack there in a predictable way. The EBSBs along the MRGO, unlike the coastal
sea dikes constructed along the south shore of Lake Pontchartrain, were not designed to
resist the threat posed by waves reforming and growing over thousands of feet of deep
water in the artificial channel.

Figure 2.1 ADCIRC hind cast of maximum surge elevations around Greater New
Orleans (http://hurricane.lsu.edu/floodprediction/katrina32/images/katrina32.JPG)
created by the LSU Hurricane Center

The Lake Borgne surge, though it reached a higher maximum than in Lake Pontchartrain,
dropped more quickly after Katrina passed as it drained through many routes back to the
Gulf of Mexico (Figure 2.2). Lake Pontchartrain stayed higher than Lake Borgne for
two more days, allowing floodwaters to flow continuously into the Orleans Metro area
through the deeply breached walls on the 17th Street and London Avenue drainage canals
long after water was draining out of St. Bernard.

8
 Com parison of Lake Borgne and Lake Pontchartrain Surge During Hurricane
Katrina

20
Lake Borgne Surge in MRGO Re ach
18 1 Ne ar Paris Road

16
Water Elevation (ft, NAVD88)

14

12

10

2 Lake Pontchartrain Surge Ne ar

Se abrook at the IHNC Entrance
0
8/27/05 8/28/05 0:00 8/28/05 8/29/05 0:00 8/29/05 8/30/05 0:00 8/30/05 8/31/05 0:00
12:00 12:00 12:00 12:00
Date/Tim e (CDT)

Figure 2.2 Katrina surge hydrographs reconstructed from ADCIRC model output
and high water mark data for Lake Borgne surge in the throat of the funnel and in
Lake Pontchartrain near the mouth of the IHNC at Seabrook.

Our focus here is on surge and waves generated in the Lake Borgne funnel (Figure 2.3).
The geography of the funnel includes the pre-existing GIWW canal along the northern
margin, the enlarged portion of the GIWW that IPET has called MRGO Reach 1 that
serves as an outlet to the IHNC, the MRGO Reach 2 channel along the south margin, and
LPV embankments paralleling all of the artificial channels on the inland side. The funnel
also contains more or less natural features including the southern half of Lake Borgne
itself, which is divided into two embayments or lobes, and thousands of acres of wetlands
both on the inboard and outboard sides of the LPV hurricane protection structures. Three
drained and developed polders surround the Lake Borgne funnel, including New Orleans
East to the north, the Orleans Metro area to the west, and the combined Lower 9th Ward
and St. Bernard Parish neighborhoods. All of these developed polders received
floodwaters from the Lake Borgne funnel, specifically through or across channels
constructed or enlarged as a part of the MRGO navigation project (Figure 2.3).

We will show how the flooding of these areas during Hurricane Katrina was initiated and
exacerbated by the presence of the MRGO project in a way that rendered the
subsequently authorized 1965 LPV Hurricane Protection projects largely futile. In the St.
Bernard polder, where the water rose to an elevation of +11 feet (NAVD88) almost
everywhere, it was as if the LPV flood protection works had never been built. Given that
98 percent of the structures within the parish flooded, as they did, it is difficult to imagine
how it could have been worse. Yet to the people living there, the 200 to 300 year level of

9
LPV protection that the USACE claimed it had provided was a crucial factor
underpinning the choice to move or stay there after the Hurricane Betsy disaster. The
same argument applies to a lesser degree for New Orleans East and the Orleans Metro
polders that also received water through the MRGO during Katrina.

Figure 2.3 Relationship of MRGO Reaches 1 and 2, and the GIWW to the Lake
Borgne funnel, showing undeveloped marsh contained within the LPV flood
protection system on the east side of New Orleans East, and north of the developed
area of St. Bernard Parish. Notice breakthrough of eastern lobe of Lake Borgne
into MRGO Reach 2 channel at Bayou Dupre.

2.1 MRGO Reach Classification.

We adopt the terminology used by the IPET to describe four relevant reaches of the
MRGO project (Figure 2.3), beginning, first, with the portion of the Inner Harbor
Navigation Canal (IHNC) that extends north from the Mississippi River lock to the
junction with the 6-mile long portion of the combined Gulf Intracoastal Waterway
(GIWW) and MRGO. The junction with the IHNC is discussed on navigation charts as
MRGO Mile 66 (Figure 3.X). IPET describes the conjoined segment extending due east
and bounded by earthen LPV berms as MRGO Reach 1. East of the Paris Road high-rise
bridge, near MRGO Mile 60, MRGO Reach 1 becomes Reach 2 heading southeast
toward the Gulf, while the much smaller GIWW barge canal continues to the east north
east (See Figure 3.1).

10
Inner Harbor Navigation Canal. The Inner Harbor Navigation Canal (IHNC), also called
the Industrial Canal, is the oldest of the pre-existing artificial channels that were partially
incorporated into the MRGO project. It was originally dredged in the 1920s, and runs
north from the IHNC Lock at the Mississippi River to an opening into Lake Pontchartrain
at Seabrook. It has a complicated geometry, with constrictions at bridge crossings, and
side slips constructed to accommodate various port activities over the years. It has a
controlling depth of about 40 ft, with a number of deeper sections up to 80 feet deep
where scour has occurred near constrictions (Team Louisiana 2007).

Gulf Intracoastal Waterway. A portion of the shallow-draft GIWW was incorporated into
the MRGO project as MRGO Reach 1, but it trends in its original state on a northeasterly
direction along the south side of New Orleans East polder adjacent to earthen berms
erected as part of the LPV. The GIWW is one of the features defining the north side of
the funnel where it passes south of the New Orleans East polder. It was originally
dredged during World War II to facilitate barge traffic with authorized dimensions of 125
feet wide and 12 feet deep. Like all channels dredged through the marsh, it has become
wider over time, but remains relatively shallow. The GIWW provided the first
connection between the funnel and the IHNC.

MRGO Reach 1. The east-west oriented 6 mile section of the GIWW that connects with
the IHNC (Mile 60 to 66) was enlarged by more than a factor of 10 to give it a
controlling depth greater than 36 feet and a bottom width of 500 feet as part of the
MRGO project authorized in 1958. This section is commonly referred to by the Corps as
the GIWW/MRGO, or simply as a part of the GIWW, but the majority of its cross-section
is attributable to dredging carried out under the MRGO project. The top-width of
GIWW/MRGO Reach 1 has enlarged, on average, by nearly 300 feet to nearly a 1,000
feet (Figure 2.4), largely through erosion into the south bank (Morris Expert Report
2008). MRGO Reach 1 is paralleled by LPV levees on both the north and south sides,
located about 1,800 feet apart, with about 800 feet of overbank between the channel and
the toe of the levee along the south side.

MRGO Reach 2. IPET describes the deep-draft reach that turns to the southeast from the
GIWW junction (Figure 2.5) and passes south of Lake Borgne as MRGO Reach 2
(MRGO Mile 60 to 23). MRGO Reach 2 extends nearly 40 miles southeast through the
marsh past the end of the hurricane protection system and the unprotected community of
Shell Beach (Alluvial City) until it reaches the open, and generally salty, waters of Breton
Sound. Between Mile 50 and 55, in the vicinity of the Bayou Dupre water control
structure, the MRGO channel has eroded into the eastern lobe of Lake Borgne through
the lake rim marsh (Figure 2.3). This gap figures prominently into wave model results as
will be seen later. Farther south, beyond the Chalmette LPV EBSBs, dredging for
MRGO breached a ridge constructed by Bayou LaLoutre when it was a distributary for
the Mississippi River. This breach played an important role in allowing salt water to
move into what were once relatively fresh swamps and marshes closer to the populated
areas of St. Bernard. Millions are now being spent to rebuild this ridge as mitigation and
restoration for the destruction associated with MRGO, as will be discussed in the next
chapter.

11
MRGO Reach 2 is bounded on its south bank by a large band of former marsh that was
used as a spoil disposal area (Figure 2.5) and extends all the way to the jetties at the Lake
Borgne shoreline. The marsh within the disposal site has been buried under many feet of
material dredged from the channel. A portion of this spoil disposal area was used as a
platform for construction of the EBSBs of the Chalmette portion of the LPV after that
project was authorized in 1965.

A segment of MRGO Reach 2 extending 10 miles southeast from the Bayou Bienvenue
water control structure (Mile 58) was over-dredged to nearly 60 feet deep in the late
1960s to provide the material that was dewatered and mounded to form the LPV berm
(Fitzgerald et al. Expert Report 2008). Both MRGO Reaches 1 and 2 have been dredged
repeatedly to 42 feet during maintenance operations to remove material that sloughed into
the channel. MRGO Reach 2 has an authorized top width of 650 feet but has widened to
more than 3,000 feet in places, particularly on its north bank where, until recently, no
effort was made to control erosion (Figure 2.4). The retreat of the north bank, which has
diminished the marsh buffer on the Lake Borgne side, averages more than 700 feet
(Figure 2.5), while the retreat on the south bank averages more than 300 feet. Dr. Bea
has found that erosion of the south bank affected the severity of wave attack for the LPV
EBSBs as the scrub vegetation that formerly protected the structures has been reduced to
less than 100 feet in some places (Bea Expert Report 2008). Dredging beyond the
controlling depth of 36 feet hastened the slumping of the unstable banks, but was a
common practice over the years even if it appears never to have been specifically
authorized by Congress.

MRGO Reach 3. The remainder of the MRGO is dredged through open water and has
been called MRGO Reach 3 (See Figure 1.2). It served as a deep route for salt water to
invade and damage inland freshwater swamps and marshes that previously were
protected by the Bayou LaLoutre ridge. This reach shoaled after Katrina, as it did after
even minor storms, so that it could not be used by seagoing vessels. The Corps’ inability
to keep this reach reliably maintained at its authorized controlling depth has been a
constant factor in ship-owner decisions to eschew the MRGO in favor of the longer but
more dependable Mississippi River route. When the USACE sought to spend additional
millions to reopen Reach 3 after Katrina, they encountered a public outcry. Congress
responded by refusing to allow expenditure for this purpose. This provided some
breathing space for public debate that led to the de-authorization of the MRGO project
that was finalized by the November override of the Presidential veto and ultimate passage
of the Water Resources and Development Act of 2007.

12
 4,000
y = -58.242x + 4744.3
2
R = 0.5046

3,500

IHNC B. Bienvenue B. Dupre End HPS Shell Beach B. LaLoutre Jetties

3,000

2,500

Top Width (ft)

2,000

1,500

1,000
DESIGN TOP WIDTH

500

REACH 1 REACH 2 REACH 2 REACH 3

0
70 65 60 55 50 45 40 35 30 25
MRGO Mile

Figure 2.4 Expansion of MRGO channel by mile from the IHNC on left to jetties on
right at Breton Sound shoreline. Top width is variable but tends to increase toward
the coast. Original design top width is also shown (Team Louisiana 2007).

2.2 The Funnel

A triangle that opens east to Lake Borgne and thence to Mississippi Sound and the Gulf
of Mexico is formed by the GIWW on the north and MRGO Reach 2 on the south
(Figure 2.5). This layout has been described as a “funnel” by virtually everyone who has
taken a bird’s-eye look at the plan developed by the USACE in the 1950s to
accommodate navigation and economic development east of New Orleans (see Chapter
3). The funnel is dominated by Lake Borgne, a very large shallow bay similar to Lake
Pontchartrain that is elongated along a southwest to northeast axis, providing more than
40 miles of open water in this direction -- a very long “fetch” over which hurricane winds
can build surge and waves. Any hurricane generating winds from the north (0o), east (90
o
) or southeast (135 o) will cause a buildup of surge against flood protection levees and in
the adjacent artificial channels somewhere on the margins of the funnel, which is then
conducted through the common GIWW/Reach 1 outlet toward the IHNC.

13
 GIWW

Reach 1, Funnel Throat

*Top-width of Reach 2
has eroded to more than
2500 feet (from 650 feet)
in some sections

EBSBs

Reach 2

Figure 2.5 “Golden Triangle” marsh separating Lake Borgne from throat of the
Lake Borgne funnel.

The funnel still includes about 5 miles of degraded tidal wetlands in the “throat” area
between the western shore of Lake Borgne and the MRGO/GIWW junction that is
euphemistically called the “Golden Triangle” by local residents (Figure 2.5). Broken
marsh also fringes the north and south shores of Lake Borgne, except where erosion has
connected the Lake directly with the MRGO channel. Additional, more intact wetlands
occur in the Biloxi marshes to the north and east beyond the ends of the LPV berm
system. As will be seen, these marshes separating Lake Borgne from Breton Sound
played an important role in keeping long-period swell waves from propagating into the
funnel, but they did nothing to stop the shorter-period seas that were generated within
Lake Borgne itself. We discuss the dynamics of surge and wave amplification facilitated
by the geometry of the funnel in Chapter 5.

2.3 Developed Areas Under Pump

Orleans Metro. This is the part of New Orleans that most visitors see. It extends from
the Inner Harbor Navigation Canal (IHNC) at the inland terminus of the MRGO west to
the 17th Street Canal on the Jefferson Parish line, and covers all the land between the
Mississippi River and Lake Pontchartrain (Figure 2.3). This compartment includes the
central business district and some of the highest lands along the River (5 to 15 ft), but
14
also very low areas on the lake and river sides of the Metairie Ridge. The suburbs on the
lake side of the Metairie ridge are reclaimed swamp and marsh that have sunk well below
sea level as a result of forced drainage. A rim of higher land occurs along the lakefront
where it was built up by dredging from the Lake and along the IHNC where dredged
material was also placed. The part of the Orleans Metro polder that flooded during
Katrina averages about 2 ft below sea level (Figure 2.6). Floodwaters that initially
entered the Orleans Metro polder came through the MRGO into the IHNC and from there
initially through a missing railroad flood gate, then through a major breach in an earthen
embankment along the south side of the Port of New Orleans facility, and then finally
over floodwalls and through one section that failed (van Heerden and Bryan 2006, Team
Louisiana 2007).

New Orleans East. This compartment includes the easternmost suburbs of Orleans
Parish. It is located between the IHNC on the west and Lake Pontchartrain to the north.
MRGO Reach 1 and the GIWW form the southern boundary (Figure 2.3). Orleans East
consists almost entirely of drained wetlands and is the “deepest” of the drained marsh
polders, having subsided to 5.8 ft below sea level on average in the area flooded during
Katrina (Figure 2.6). The federal LPV levees that surround Orleans East have
impounded a formerly tidal marsh that covers nearly 22,000 acres now at or below sea
level. This impounded marsh became the Bayou Sauvage National Wildlife Refuge after
it was recognized that further drainage and development would not be permitted. It is
separated from suburban subdivisions by a local levee that allows more intensive
drainage of the developed area. MRGO water first entered New Orleans East from
MRGO Reach 1 over the Citrus Levee, and, to a lesser degree, from overtopping of
floodwalls on the east side of the IHNC. Floodwaters also entered this compartment
later when the federal New Orleans East Back Levee was breached from the GIWW.
This failure allowed water into the Refuge from the GIWW, which then flowed over the
low interior drainage levee (Maxent Levee).

St. Bernard Polder. “The Parish” extends east from the IHNC along the higher land of
the Mississippi River natural levee, and then farther east along the old Bayou LaLoutre
ridge (Figure 2.3). The St. Bernard polder includes the Lower 9th Ward even though
that neighborhood is a political subdivision of Orleans Parish. It also contains a large
swatch of tidal marsh, nearly 32,000 acres, called the Central Wetlands, located between
the main federal LPV berm along the south bank of the MRGO and a lower state-built
levee known as the 40 Arpent or Florida Avenue Levee. Two gated structures were
constructed as part of the LPV project through the federal EBSB alignment where
Bayou Bienvenue crosses on the west end, and at Bayou Dupre farther to the east, to
allow water exchange with the MRGO, and for small vessels to pass between the
wetlands and the ship channel during normal tides.

The developed and drained portion of the St. Bernard polder lies south of the 40 Arpent
Levee, sheltered behind 2.4 to 3.0 miles of former fresh water cypress-tupelo swamp
that have become intermediate to salt marshes since the construction of the MRGO.
Pump stations located along the 40 Arpent levee discharge storm drainage into these

15
wetlands and this limited freshwater introduction has preserved a few stands of the once
more extensive swamp forest that covered this area prior to construction of the MRGO.

St. Bernard has some of the highest land on the East Bank of New Orleans, following as
it does the natural levee of the Mississippi River and some of its abandoned
distributaries (Figure 2.3). Despite being relatively high by local standards, the St.
Bernard polder experienced the most violent, spatially expansive and deepest flooding
in the entire metro area during the Katrina event. Except for a limited contribution
from rainfall, all flooding of the St. Bernard polder was caused by water that passed
through or across one or more reaches of the MRGO. This water entered the
developed area as a result of catastrophic floodwall failures along the IHNC on the
western margin, by overtopping of berms on MRGO Reach 1, and by flow through
breaches in the federally built EBSBs along the MRGO. The interior 40 Arpent Levee
was protected by over two miles of the Central Wetlands and was relatively
undamaged, but it average only 6.5 feet high and was completely overtopped when
floodwaters from the MRGO filled the Central Wetlands beyond this level.

Figure 2.6 Comparison of mean maximum floodwater elevation with average

elevation of flooded land in three New Orleans polders (Team Louisiana 2007).

16
2.4 Connecting the MRGO Project with Damages to People and Property

To show that the extent of damage to property in St. Bernard Parish, the Lower 9th Ward,
Orleans Metro, and New Orleans East was initiated or aggravated by the presence of the
MRGO, we must demonstrate that flooding occurred earlier in the surge sequence than it
would have otherwise. The maximum elevation of flooding experienced in each of the
three Greater New Orleans (GNO) polders most affected by Katrina was quite different
(Figure 2.6), reaching 10 to 12 feet above sea level in populated areas of the Lower 9th
Ward and St. Bernard Parish, but only a little over a foot above sea level in the much
lower New Orleans East polder (Team Louisiana 2007). The extent of damage to
property is related to the depth of flooding, a measure obtained by subtracting the land or
floor elevation of buildings from the maximum elevation of water. The average land
elevations in each of the flooded polders also differed (Figure 2.6), ranging from nearly 6
feet below the NAVD88 datum (-7 feet relative to mean sea level) to +2.5 feet in St.
Bernard (Team Louisiana 2007). In the Lower 9th Ward and St. Bernard, 98 percent of
all structures were seriously damaged by flooding. While the maximum elevation of
water is important to property damage, the maximum rate of rise, estimated at 8 feet per
hour in the Lower 9th Ward (Team Louisiana 2007), was perhaps the most important
factor contributing to greater loss of life there.

The difference among the polders in maximum flood elevation during Katrina was
independent of the relative elevation of the land flooded, attaining its highest level in the
St. Bernard Polder, which has the highest average land elevation of all of the GNO
polders that flooded (Figure 2.6). The maximum elevation of water attained in each of
the GNO polders was more closely related to the time of flooding onset and the length of
time during which water continued to flow in. The St. Bernard polder flooded to such a
high elevation because it flooded earlier in the surge sequence, while water in the MRGO
was still rising. If the time-of-onset of flooding was advanced by the MRGO navigation
project, then one must conclude that the MRGO project contributed substantially to the
severity of flooding wherever this occurred.

All breaching of LPV structures, whether floodwalls or earthen berms, within the Lake
Borgne funnel took place where these structures were inboard of, and in close proximity
to, artificial channels. Except for the New Orleans East Back Levee EBSB, all of the
LPV structures that breached were adjacent to some part of the MRGO project, whether
between the lock and the MRGO junction in the IHNC, along the north bank of MRGO
Reach 1 or along the south side of MRGO Reach 2. Conversely, no LPV structures
separated from artificial channels breached even though many of them experienced
overtopping. We will show that breaching was generally initiated by the excess stress
applied to LPV structures as a result of proximity to deep channels, either by a surge that
reached a higher elevation or lasted longer, or, in some cases, by a higher intensity of
wave attack than would have occurred if the channel was not there or farther away.
Because the LPV and MRGO projects were never explicitly integrated with each other,
we have found no evidence that the MRGO project was ever modified to reduce the
predictable excess surge stresses and wave attack caused by the encroachment of the

17
channels on LPV structures, or, alternatively, that the LPV structures were bolstered in
any way to withstand the obviously increasing threat.

Reduced to fundamentals, the extent of damage posed by the MRGO in each of these
areas was affected, first, by the timing and duration of flooding during the storm, namely,
whether flooding started early in the surge sequence or later. Because storm surge is a
transient phenomenon, flooding that begins later, at or after the peak of the surge
hydrograph (Figure 2.2), will not be as severe as flooding that begins earlier, while the
surge is still rising. While the Katrina storm surge rose first in Lake Borgne and later in
Lake Pontchartrain, it dropped precipitously everywhere at about the same time (Figure
2.2), as the storm completed its due north traverse through the area (Figure 2.7) where
both lakes are at similar latitude. Multiple features of the landscape affect the onset of
surge at any given location, most notably proximity to large lakes and bays where surge
is generated, as well as to large channels which preferentially convey it. The end of the
surge event is, in contrast, determined rather simply by the speed of translation of the
storm, which sets the schedule for reversal of the prevailing winds as it leaves the area.

Figure 2.7 Track of the eye of Hurricane Katrina on August 29, 2005

18
Some LPV floodwalls, levees and EBSBs failed during the passage of Katrina, while
others did not. We will show that segments adjacent to the MRGO project, whether in
the IHNC, or on MRGO Reaches 1 or 2, were exposed to greater Katrina-induced stress, -
- the effect of higher surge and/or more damaging waves – for a longer period than would
have occurred if the MRGO project had been built and maintained in the manner that it
could be called “hurricane neutral,” namely that it did no harm to the public safety
function of the LPV project. We conclude that for the Robinson plaintiffs this spelled the
difference between survivable flooding and catastrophe.

19
 CHAPTER 3
A BRIEF HISTORY OF THE MRGO PROJECT

In 1957, before the MRGO moved into the construction phase, the Police Jury (county
government) of St. Bernard Parish was concerned enough to establish a “Tidewater
Channel Advisory Committee” to review plans then being developed for the MRGO.
The Chairman presented a report to the full Police Jury recommending that St. Bernard
Parish oppose the project, stating that enhanced tidal action caused by the channel “will
have adverse effects on the entire marsh area with consequent crosive (sic) action and the
intrusion of high saline content water into areas normally fresh or only slightly brackish”
(St. Bernard Tidal Channel Advisory Committee 1957). This report goes on to state
presciently:

“During times of hurricane conditions, the existence of the channel will be

an enormous danger to the heavily populated areas of the Parish due to the
rapidity of the rising waters reaching the protected areas in full force
through the avenue of this proposed channel. This danger is one that
cannot be discounted. No matter how small a flood may be, or how small
the area to which it is confined, to the families that have water in their
houses, it is a major catastrophe.”
(St. Bernard Tidal Channel Advisory Committee 1957, p. 3)

Lake Borgne

GIWW MR-GO Reach 2

GIWW/MR-GO
Reach 1

Figure 3.1 View east from above the GIWW/MRGO Reach 1 into the Lake Borgne
Funnel defined by MRGO Reach 2 on right and the GIWW on left.

20
The brief history of the MRGO that I recount here is indeed brief, and should not be
regarded as comprehensive. To facilitate a more in-depth review, the Robinson expert
team has compiled a searchable, hyperlinked chronological inventory of MRGO
documents that begins in 1852 with the first concepts for a tidewater channel east of New
Orleans and extends up to the present. That comprehensive “MRGO Chronology” is
attached as Appendix B.

Once the MRGO was built, local residents quickly came to understand that the project
was far more effective as an agent of environmental and community destruction than as
an engine of economic development. For five decades, they tried unsuccessfully to put
the MRGO genie back in the bottle, while the taxpayers of the Nation continued to pay an
average of $12.5 million every year to keep the MRGO open to the few ships that ever
came to call (USACE 2007).

Two months after the Hurricane Betsy flood in 1965, the New Orleans East based
Citizens Committee for Hurricane Flood Control, recommended revisions to the 1962
USACE hurricane protection plan for Lake Pontchartrain and Vicinity, that was about to
be adopted by Congress in the Flood Control Act of 1965 (PL 89-298). The Citizens
Committee sent a letter and report to the USACE New Orleans District Engineer
containing these recommendations (Citizens Committee for Hurricane Flood Control
1965). Significantly, the recommendations of this committee were ignored then, but have
now been adopted almost in their entirety by the USACE since Hurricane Katrina for the
100 year protection plan (http://www.mvn.usace.army.mil/hps/100yr_design_map.html).
One reason that the 1965 Committee made such recommendations was as follows:

“The US Engineers proposal for a levee along the south shore of the Gulf
Outlet to Bayou Dupre, and along the north shore of the Intracoastal
Waterway would form a funnel channeling surges and wind driven water
into the Intracoastal Waterway and Industrial Canal.”
(Citizens Committee for Hurricane Flood Control 1965, p. 2, emphasis
added).

The second reference made in this 1965 statement to the “Intracoastal Waterway” is to
that segment of the MRGO project that occupies the original GIWW right-of-way
(Figure 3.1) that has become known since Katrina as “MRGO Reach 1,” the six-mile
connection to the Inner Harbor Navigation Canal (IHNC). The IHNC is still commonly
called the Industrial Canal. MRGO Reach 1 was often described in USACE documents
over the years as a part of the GIWW, though it has the far greater dimensions of the
MRGO ship channel rather than those of the original GIWW barge canal. The channels
of the pre-existing GIWW to the north and MRGO Reach 2 to the south delineate the
sides of the “funnel” referred to in the 1965 Committee report (Figure 3.1). The 1965
Committee report and recommendations included a sketch of what later was called the
“Crosby” plan or alternative discussed below (Figure 3.2).

21
By this time, construction of the LPV flood control project was in progress on the east
side of New Orleans. The oceanographic basis for the LPV Barrier Plan had been
developed by a joint USACE/Weather Bureau study team in the mid- to late-1950s prior
to the construction of the MRGO (Team Louisiana 2007). Local engineers at the USACE
New Orleans District (NOD) faced the task of re-interpreting the 1950s hurricane science
and oceanography to incorporate the massive new channel – not included in the original
hurricane analysis -- they had now built into the City. For this reason, and because the
NOD was facing legal proceedings similar to those now before the federal court in
Robinson v. United States, the USACE commissioned the National Science and
Engineering Company (NESCO) to quickly study the potential of the MRGO to enhance
and transmit storm surge to populated areas (Bretschneider and Collins 1966).
Significantly, the NESCO study was limited to hurricane effects on surge in the Lake
Borgne funnel and included no discussion of waves, although one of the authors, Dr.
Bretschneider, was at the time perhaps the worlds foremost authority on wave
forecasting.

The NESCO study released in 1966 would remain the only investigation commissioned
by the USACE on potential impacts of the MRGO project for almost 40 years. A second,
more narrowly focused investigation was conducted by the USACE at the request of the
State of Louisiana beginning in 1999, but it was not completed until September, 2005, a
month after New Orleans was flooded (USACE 2005). Neither of these studies
considered wave regeneration in the MRGO channel.

The NESCO study was a respectable piece of “armchair science” for the time, given the
limitations of modeling technology and the lack of an independent field data collection
program to confirm and augment information provided by the NOD. This important
document will be discussed in more detail later. Suffice it to say here that it raised as
many questions as it answered, and included numerous caveats about how broadly the
findings should be interpreted (See Team Louisiana 2007, p. 244-252). The limitations
of the NESCO study were well understood by scientists at the time, including the eminent
oceanographic meteorologist, S. A. Hsu, a professor at the Louisiana State University
(LSU) Coastal Studies Institute. He produced a report that was finalized in 1973 entitled
“The Impact of the Mississippi River-Gulf Outlet on Hurricane Floods of St. Bernard
Parish and New Orleans Metropolitan Area.” This report was appended to a larger
volume produced by Coastal Environments, Inc. (CEI), under contract to St. Bernard
Parish (Hsu 1973; CEI 1973). In his analysis of the NESCO report, Dr. Hsu states the
following (p. A-11).

“(a) The basic technique used in the [NESCO] report is to first develop a
reasonably sound theoretical model and then compare it to observations.
The comparison is poor as clearly admitted by the authors on page 62,
(here is an excellent negative critique on its own).

(b) The next steps utilize various techniques which essentially readjust the
assumed coefficients or factors to force them to the agreement with
observations….The weakness of the technique is that it utilizes a circular

22
 argument. The same observations are used to verify the model that
provided the basis for calibration. In other words, the authors started with
a model, then readjusted it with data from one observational case, and then
compared the models predicted results with the same observational case
data. The approach virtually guarantees good agreement.”
(Hsu 1973, p. A-11)

Several proposals were developed by the USACE NOD and by local interests to provide
enhanced hurricane surge protection for port facilities and developed areas adjacent to the
combined MRGO Reach 1/GIWW and the IHNC in the years following Hurricane Betsy,
and in the immediate aftermath of flooding caused by Hurricane Camille in 1969. These
proposals included placing floodgates in the IHNC at its junction with the MRGO, a
floating gate in MRGO Reach 1 in the vicinity of Paris Road, and a crescent-shaped levee
to connect the proposed barrier structure at Chef Menteur Pass with the LPV in the
vicinity of Bayou Bienvenue (Figure 3.2). Furthermore, the Corps realized that the
enormous outflow of surge during Hurricanes Betsy and Camille from the IHNC to Lake
Pontchartrain -- what Dr. Bea has called “venting”-- was effective in preventing surge
levels from rising higher in the IHNC and causing more damage to the adjacent dock
areas then more heavily populated by businesses than they are today. This led the
USACE to reassess the controlling elevation of the “Seabrook Lock,” at that time a
proposed component of both the MRGO -- to prevent saltwater from entering Lake
Pontchartrain under normal tides -- and LPV projects. Its role under LPV would be to
prevent storm surge from entering the IHNC from the Lake, and vice versa. It is
important to note that this first tentative step toward integrating the two projects never
made it past the drawing board.

In all instances, the Corps chose the “no-action alternative.” The three structural
proposals to supply enhanced protection were rejected on the rationale that the
incremental cost increase to the overall LPV project was not deemed justified by the
incremental reduction in expected flood damage to businesses along the IHNC. None of
the relevant design memoranda discuss a weighing of public safety concerns, just
property damage. Significantly, one of these plans, known as the Crosby Plan, appeared
as Alternate Plan C in the Citrus Back Levee Design Memorandum 2, (USACE 1967,
App. C, Plate 4). This plan incorporated features similar to those included in the 1965
recommendations of the Citizens Committee (Figure 3.2). Additional reasons that the
Corps rejected this plan were given in a September 23, 1969 letter from Colonel Herbert
Haar, Jr., NOD District Engineer, to Congressman F. Edward Hebert (Haar,1969),
stating:

“Beyond the fact that the plan is not economically justified, it is

undesirable for a number of other reasons. Its adoption would mean that
none of the work already accomplished by local interests subsequent to the
project authorization would be incorporated into the Federal project and
no credit for such work would be allowed. Further, the modifications
involved are so broad in scope as to be beyond the discretionary authority
of the Chief of Engineers to adopt, so that project review and subsequent

23
 Congressional action would be required. During the time that this process
was being accomplished, progress in planning and construction of some of
the most urgently needed project features would be discontinued…In
addition to the above, operation of two features of the plan, namely the
navigation gate in the Mississippi River-Gulf Outlet and the lock in the
Gulf Intracoastal Waterway, would significantly impede seagoing and
inland navigation…The interest expressed by Mr. Crosby relative to the
need for construction of the aforementioned project is greatly
appreciated.”
(USACE Col. Haar, Jr., in 1969 letter to Congressman F. Edward Hebert)

Upon receiving this shot across the bows from the USACE with respect to the Crosby
Plan, the Orleans Levee District and Louisiana Department of Public Works -- the two
state agencies that had originally asked for its consideration -- quickly withdrew their
support for this alternative. Ironically, a variant of this plan has re-emerged since Katrina
as the 100-year protection plan for which the Corps has recently let a $750 million
design-build construction contract, as will be discussed in more detail below.

This brings us to a central theme that I will return to several times in this report, namely,
that the USACE adopted a policy of denial following Hurricane Betsy that effectively
stifled acquisition or analysis of new information on the potential of the MRGO to
increase hurricane surge, waves and flooding. This policy was clearly expressed in 1973
by Mr. P.A. Becnel, Jr., then Chief of the Hydraulics and Hydrologic Branch, NOD, in a
“Memorandum for Record” addressing the “Apparent Funneling Effects of Citrus and
Chalmette Hurricane Protection Levees” in 1973 (Becnel Memorandum 1973).

Figure 3.2 Alternate Plan C from Citrus Back Levee Design Memorandum 2
showing a rejected proposal to incorporate the MRGO into the LPV flood
protection plan through construction of a levee across the funnel and a floating
closure gate across the MRGO (USACE 1967). See USACE post-Katrina “100-Year
Hurricane Protection Plan” for comparison (Figure 3.6)

24
Mr. Becnel was responding to statements made in public hearings by Dr. Sherwood
Gagliano, a respected coastal geologist, LSU professor and president of CEI who often
served as a consultant to the Corps, and had provided early drafts of the St. Bernard
Parish report quoted above (CEI 1973). In one of them, Dr. Gagliano recommended a
“re-evaluation of hurricane storm surge threat in the funnel formed by the MRGO and the
GIWW and associated hurricane protection levees” stating that “the buildup of tides in
naturally constricting estuaries is a well known phenomenon” (CEI 1972).

Mr. Cecil Soileau, then an employee in Mr. Becnel’s branch, has described his role in
drafting this important memorandum for Mr. Becnel’s signature in his deposition
(Deposition of Mr. Cecil Soileau). In the first enumerated paragraph, the Becnel (1973)
memorandum states that:

“The purpose of this memorandum is to provide a unitive statement on the

position of the New Orleans District concerning the accusation by Dr.
Sherwood M. Gagliano or anyone else, that the alinement (sic) of the
protection levees creates a funnel effect in the vicinity of Paris Road at
Lake Borgne…At these meetings Dr. Gagliano et al., presented three
reports which are a matter of record of the public hearings.” (emphasis
added)

It is an added irony that the hearings referenced in Mr. Becnel’s memorandum were
being held not to address public safety concerns, but to receive input on a proposal being
floated by the Corps at that time to enlarge the MRGO from a bottom width of 500 feet to
750 feet, and to increase the authorized depth from 36 to 50 feet. This context may help
explain the frustrated tone of the memo. Mr. Becnel seems aggrieved that those who
were flooded during Hurricane Betsy would continue, seven years later, to “accuse” the
Corps of operating a ship channel that posed a hazard to public safety. The channel
expansion then under discussion was never authorized, at least in part because such
concerns continued to be expressed. But local residents knew that the channel was
expanding anyway, and taking the swamps and marshes with it. In the second paragraph,
Mr. Becnel’s memorandum goes on to describe what is termed the “Bases of Position.”

“The bases of our position are the results of a study performed by the
National Engineering Science Company (NESCO) in 1966 while under
contract with the U.S. Army Corps of Engineers and the heights of water
levels experienced during hurricanes Betsy and Camille.”
(1973 Memo from USACE-NOD Chief Hydraulics and Hydrologic Branch)

The rest of this memorandum provides an apparent technical justification for a lack of
NOD concern. Becnel (1973) ignores the clearly stated caveats expressed by the authors
of the NESCO study that he cites to avoid extending the Betsy statistical model to other
storms. Instead, he bases an official position that all is well – and therefore that any
further analysis is unnecessary and disruptive -- on the observation that a similar peak
surge elevation was observed at one gage on the MRGO (Paris Road) during Camille as

25
was recorded four years earlier during Betsy. Oceanographers who have tried to follow
this logic find any similarity at a single gage station to be purely coincidental, given the
profound differences between the two storms and the unique characteristics of the surge
that each generated (Figure 3.3). Hurricane Betsy, a category 3 storm at landfall,
followed a track to the south and then west of the City, while Hurricane Camille passed
well east of New Orleans in 1969 to strike the Mississippi coast as a compact category 5
storm.

Figure 3.3 Paths of Hurricane Betsy (left) on September 8-9, 1965 (USACE 1965)
and Hurricane Camille (right) on the evening of August 17, 1969 (USACE 1970).

The technical justification stated in the Becnel Memorandum is thin, but what is missing
entirely is any evidence, again, of a weighing or balancing of competing public policy
concerns. Why did a high-level NOD official find it necessary to counter legitimate
technical questions about MRGO with a subjective “unitive statement” (party line)
designed to be imposed hierarchically upon an entire organization responsible for public
safety? This is a recurring feature in the annals of the District. The reluctance to update
the 1959 Standard Project Hurricane surge analyses with new weather service
information is now well known (Team Louisiana 2007, p. 95-114), as is the infamous
1985 benchmark “policy” forcing contractors to use an obsolete survey datum that
resulted in construction of drainage canal floodwall crowns two feet lower than design
grade in the 1990s (IPET 2006, Vol. II, p. 78). The 1973 Becnel memo is woven of the
same fabric.

Similarly, the Corps’ had a long-standing policy against spending operations and
maintenance (O&M) funds on stabilizing the banks of any of its navigation channels
against erosion unless shoreline slumping and retreat began to affect the ability to
maintain the controlling navigation channel dimensions. The standard practice
apparently has been to force local partners to take on this task if at all possible. The
MRGO GDM 1B (USACE 1959) specifically excluded any engineered measures to
maintain the bank lines following construction, even though the side slopes of the
dredged cut were specified at a highly unrealistic 1V:2H slope. It was known prior to
construction that they would not be stable.

26
 “No channel protection is recommended initially; however, erosion due to
wave wash in open areas can be expected in the upper part of the channel
slope where the peat and highly organic clays are exposed. Protection for
this area can be provided if and when the need for it becomes necessary.
No channel protection is included in the overall cost estimate of the
project. It is presumed that sufficient rights-of-way will be furnished by
local interests to preclude use of channel protection, or that additional
rights-of-way will be furnished if the need arises” (GDM 1B 1959, p. 5).

The Corps did purchase a relatively wide right-of-way extending well outside of the
authorized channel footprint so that any issues arising from bank retreat could be put off
for some time, perhaps forever, given that there was no development or government asset
near the MRGO channel ROW at the time the channel was dredged (USACE 1959).

Later, after the Corps constructed the LPV Chalmette structures, it became apparent that
the first asset that would be threatened by the rapid retreat of the channel bank on the
south side was theirs. A useful chronology of the development of this issue appears in a
briefing package dated September 15, 1982, prepared by the New Orleans District and
titled “Data for Testifying Officers on FY 1984 Civil Works Budget, Mississippi River-
Gulf Outlet, Louisiana,” where it is stated on p. 3 (AIN-046-000000862) that in 1967:

“The project was modified under the discretionary authority of the Chief
of Engineers to include as a mitigating measure, the costs of protecting a
portion of the foreshore lying between the Lake Pontchartrain and Vicinity
Hurricane Protection project and the Mississippi River-Gulf Outlet.
Construction of the Mississippi River-Gulf Outlet project exposed levees
and their foreshore along both banks of the navigation channel in the City
of New Orleans to damages from waves generated by seagoing vessels
utilizing the waterway. The navigation project should have included
adequate protection for these levees and their foreshore. The new levees
in the Lake Pontchartrain project located adjacent to the Mississippi
River-Gulf Outlet ship channel will also require protection. The cost of
this work has been deleted from the Lake Pontchartrain project and added
to the Mississippi River-Gulf Outlet project. (There are about 6 miles of
levees along the north bank and 18 miles along the south bank of the Gulf
Outlet navigation project which require protection.)

The chairman of the House and Senate Committees on Appropriations

were informed by the Chief of Engineers of the modification to include, as
a mitigation measure, the wavewash protection for levees and
foreshore…being included in the project. The above change in scope was
first reflected in the project cost estimate presented to Congress in
connection with the appropriation request for FY 1969.” (emphasis added)

Technical memos were exchanged over the years between 1967 and 1990 between the
USACE District and Division offices indicating an increasing level of alarm over the

27
potential for MRGO bank erosion to undermine the foundational stability of the LPV
EBSBs, but no action was taken beyond installation of some test sections of foreshore
armor for the banks of the channel (not EBSBs) with rocks, geotextiles or articulated
concrete mattresses. These tests demonstrated that the channel bank erosion could indeed
be abated by one or more of these standard techniques.

A bureaucratic snafu appears to have arisen about where the money would come from for
the relatively expensive measures required. This partly explains why nothing happened
on the ground in that diminishing “foreshore,” until the Secretary of the Louisiana
Department of Natural Resources (LDNR) provoked a crisis in a May 5, 1991, letter to
the District threatening to find that the next proposed cycle of maintenance dredging for
the MRGO was “inconsistent” with Coastal Use Guidelines approved by the State under
its newly reauthorized, federally approved Coastal Zone Management Plan (CZMP)
because the dredging plan did include measures for stabilizing the banks of the federal
channel. This letter resulted in a May 30, 1991, briefing memo from the District to the
Chief of Engineers that enumerated ten additional “policy” justifications for inaction in
“Tab B: Policy and other considerations” (NOP-007-000000779) that begins:

1. “Long standing Army/Corps policy that bank stabilization projects have

low funding priority (effectively zero funding)

2. Precedent – if Corps fixes mi 50-56 as condition of dredged material

placement, expect continued demands throughout LA coastal zone

3. Corps policy says Sec 1136 will not be used to fix bank erosion

4. Per 33 CFR 336, when at impasse with State, Corps should defer
dredging awaiting additional Congressional appropriations”

And continuing….

7. If Congressional “add,” Cong must tell us in law – forces LA delegation

to get agreement of colleagues

8. Long term viability is questionable. 36 ft channel is not as competitive

as deeper drafts. PONO is considering container port facilities on
Mississippi River to access deep draft channel

9. Local opposition to MRGO is considerable based on perceptions that

MRGO is environmental disaster and its economic promises never
materialized”

Nearly 40 years after the 1967 discretionary modification of the MRGO project, an
MRGO “Fact Sheet” was prepared for the public (in contrast to the internal 1991 memo
above) by the District and dated May 12, 2004, that gives perhaps the clearest

28
explanation of what had happened in the intervening years on p. 5 (NOP-004-
000000984):

“By-in-large, the MRGO banks remain un-lined with protection. This is a

feature of the channel established upon its initial design and construction.
Ship wakes generated by vessels passing along the MRGO have been for
many years and continue to be major contributors to bank erosion.
Measured long-term erosion rates are about 15 and 32 feet per year, on the
south and north banks, respectively. This is an on-going problem, widely
recognized by the Corps, other agencies, state and local governments, as
well as the general public.

Form 1990 to the present, the district has built and maintained about 8.3
miles of foreshore rock bank protection along the MRGO. This armoring
substantially abates wave erosion; reduces channel sedimentation, and
associated cyclic dredging needs and costs; as well as preserves valued
wetlands, land features, and infrastructure along the channel. An
additional 17.8 miles of foreshore rock protection are planned for
construction and maintenance are planned for construction and
maintenance in the future.”

All of the bank armoring completed at that time was along the south shore of
MRGO Reach 2, where it would more effectively address LPV stability concerns
than any for the environment. But clearly the attitude within the district had
changed since 1991, primarily in response to the growing public support and
funding for environmental restoration, but the District felt obliged to justify the
slow progress as follows:

“Bank armoring will curtail erosion and cyclic O&M dredging

requirements along the inland reach of the MRGO, in proportion to the
rate that program is funded for execution. The project’s O&M budget,
indexed for inflation, has been relatively constant over the last decade.
This has limited the rate of execution in bank armoring, in favor of
dredging to sustain the critical function of providing serviceable
dimensions for shipping. There is a capability to accomplish bank
armoring at a higher rate with increased O&M Program funding levels, as
well as through other authorizations and funding streams.” (emphasis
added)

It can be seen that the old “policy considerations” expressed in the 1991 internal memo
continued to exert an influence on Corps behavior with respect to the MRGO project
right up to a year before Hurricane Katrina. Furthermore, the 1973 Becnel “unitive
statement” appeared to be fully in effect, in that public safety considerations and
protection from hurricane surge and waves exacerbated by the MRGO project are never
mentioned.

29
The most serious consequence of this tendency to meet new ideas, information and
technology from external sources with officially imposed “policies” of denial had a
cumulative effect over time of isolating NOD technical personnel from much of the rest
of the technical community. The self-imposed insularity of the District was compounded
by a lack of truly independent review or critical oversight at the Division or Headquarters
levels, as Woolley and Shabman (2008) have pointed out in their Corps-commissioned
LPV decision-making chronology (that ignores the MRGO). This combination ensured
that mistakes made would not be corrected. Regrettably, we believe, the long
conditioned response of the Corps to ignore or deny conflicting information on MRGO
continues to influence some critical post-Katrina decisions on the re-engineering of the
hurricane protection system on the east side of New Orleans, and the remediation of
MRGO-caused impacts.

In 1999, the State of Louisiana and the USACE started a seemingly endless “re-
evaluation study” of the MRGO that was not concluded until immediately after Hurricane
Katrina in 2005 (USACE 2005). This study provided some useful information,
confirming, for example, that the MRGO had more than doubled salinity in Lake Borgne
(Table 3.1), and in the eastern part of Lake Pontchartrain, providing a robust explanation
for much of the conversion and loss of wetlands that had been observed. The study
documented the retreat of the unstable MRGO Reach 2 channel banks. This re-
evaluation study also included an early ADCIRC model study of the effects of a closure
of MRGO Reach 2 at Bayou LaLoutre on hurricane surge (USACE 2005). Despite being
unable to accurately capture wetland effects on surge, the modelers got roughly the same
answer for a closure east of Lake Borgne (where the surge comes from) as have all
subsequent model studies, namely that the surge reduction effects of a flood gate at this
location would be negligible. Establishing a tradition that IPET also adopted (2007), this
USACE study never looked at the effects of such a closure in the more surge “critical”
GIWW/MRGO Reach 1.

More importantly, however, the 2005 MRGO re-evaluation study explored a range of
remedies to these problems and identified those thought to be most effective short of
complete closure. It was found, for example, that narrowing the Reach 2 channel would
more effectively reduce salinity intrusion on normal tides than reducing its depth, and
that a closure at Bayou LaLoutre would be perform even better for this purpose.
Similarly, the study saw no difficulty with stabilizing the MRGO banks using lightweight
articulated concrete mattresses similar to those used on the Mississippi River banks for
decades (USACE 2005).

But, despite showing that it was feasible to address some of the problems known to be
caused by the MRGO Reach 2, the USACE fell back into the old denial pattern and
concluded in the month after Katrina hit that the MRGO – an anachronism before its time
– remained economically viable:

“The analysis of National Economic Development (NED) benefits shows,

that when compared to the alternatives, continued O & M (maintenance
dredging) of the MRGO is justified. The recommended plan is the No

30
 Action Alternative, i.e., to continue current O & M activities to maintain
deep-draft navigation on the MRGO. Interested parties, the most vocal of
which are St. Bernard Parish and the Lake Pontchartrain Basin
Foundation, remain very concerned about the loss of wetlands due to bank
erosion caused by vessel wave action and the environmental consequences
caused by saltwater intrusion. However, ecosystem restoration measures
are outside the authority of the re-evaluation study and are not included in
this report”
(USACE 2005, MRGO General Reevaluation Study Report Executive
Summary, p. 11)

Further, it is significant that even after Katrina, the USACE never thought to mention in
this report the possibility that encroachment of the Reach 2 channel into the toe of the
adjacent EBSBs might have contributed to their catastrophic destruction by waves. This
is another example of an institutional blind spot (Team Louisiana 2007) that will loom
larger when we compare the results of our more comprehensive analysis using the
SWAN wave model (Gautier et al. 2008) with that conducted by IPET (2006, 2007).

It took almost exactly two years after the 2005 report was issued – and some very
insistent Congressional prodding -- for the USACE to confront reality and reverse itself
in November 2007 with a new report recommending MRGO de-authorization and a
closure at the Bayou LaLoutre ridge (USACE 2007). The Corps still refuses to
acknowledge that the waterway poses a threat to public safety, relying on a newly-
discovered deficiency in the cost-benefit ratio. This break with the past on MRGO, the
first chink in the defensive policy armor built by the USACE after Hurricane Betsy, is
discussed in more detail below. But still no weighing of the costs in lives and property.

Table 3.1 Increase in salinity in Lake Borgne funnel caused by USACE after
construction of the MRGO ship channel in 1963 (USACE 2005)_
Shell Beach/South L. Borgne Chef Menteur Pass/North L. Borgne
Salinity Salinity Salinity Salinity
1951-63 1963-77 % 1951-63 1963-77 %
Month (ppt) (ppt) Increase (ppt) (ppt) Increase
January 6.5 9.8 51 3.8 5.7 50
February 6.4 9.7 52 2.9 4.8 66
March 6.3 10.4 65 2.2 4.3 95
April 7.0 10.0 43 2.2 4.0 82
May 9.5 10.2 7 2.6 4.0 54
June 9.0 12.3 37 3.3 4.2 27
July 7.9 16.0 103 3.2 6.3 97
August 8.6 16.1 87 4.8 7.5 56
September 8.2 12.9 57 6.0 8.5 42
October 7.6 13.8 82 5.2 8.4 62
November 8.0 13.1 64 5.2 8.0 54
December 8.0 12.5 56 4.2 7.0 67
Mean 7.8 12.2 59 3.8 6.1 63

31
3.1 Failure to Integrate the MRGO and LPV

The LPV had a public safety mission while the MRGO was strictly an economic
development project. One might expect that public safety would trump economic
development whenever they conflict, particularly after the extreme flooding experienced
during Hurricane Betsy. The Dutch version of the Corps, called the Rijkwaterstaat, is, for
example, guided by the motto “Safety above All.” On the east side of New Orleans,
however, public safety was considered only to the degree that the design and operation of
the MRGO channel might affect vessel maneuverability to avoid groundings and
collisions.

The USACE chose the MRGO channel alignment partly to reduce future costs of
maintaining the authorized channel by routing it as much as possible within the marsh,
and partly to allow for development of potential industrial sites on the relatively high land
created in the spoil disposal area along its south bank (Figure 3.4). Later, when the LPV
project was authorized in 1965, a decision was made to build the Chalmette loop of the
federal flood control structures in the MRGO spoil disposal area even though foundation
conditions were known to be poorer there than, for example, at the more inland 40 Arpent
location where a levee already existed that was supported by sturdier soils. It was further
known in 1965 that founding the levee on this spoil material would delay completion
because of subsidence and the long consolidation periods required between lifts. More
importantly, the Corps knew that a levee closer to Lake Borgne would be exposed to
higher surge elevations and wave attack during hurricanes than a more sheltered line of
defense farther inland.

These good reasons to look for a different right-of-way for the LPV levee were
overwhelmed by considerations that did not include public safety, including the lure of
100 million cubic yards of inexpensive material that could be mined hydraulically from
the MRGO to make earthen berms. The end result, however, was to shift a portion of
MRGO maintenance dredging costs from the navigation project to the under-funded
hurricane protection project.

The opportunity to supply berm material by using hydraulic dredges to over-deepen the
adjacent MRGO channel also trumped quality assurance concerns about whether material
supplied in this way would be suitable to construct a structure that met USACE standards
for a coastal defense dike, or indeed, an urban flood control levee of any kind. In the end,
as Dr. Bob Bea and Mr. Jesse Arnold have pointed out in earlier expert reports (relative
to the 702 (c) immunity question), the LPV structures along MRGO Reach 2 never met
any established USACE levee construction standards, and should not have been
considered engineered “levees,” but merely earthen berms/spoil banks (EBSBs). Again,
it appears that a narrow expedience prevailed over a careful weighing of the public safety
risks.

The MRGO channel supplied material for the LPV Chalmette EBSBs, but the MRGO
project was never incorporated into the LPV plan except as a pre-existing, static
landscape feature. We discussed earlier how the Corps rejected adding MRGO

32
floodgates to the LPV in the late 1960s that would have given the MRGO project a nexus
with the flood control project. Rocks were later added to the south bank to slow the
alarming encroachment of the channel into the EBSB alignment, but not to protect the
EBSBs against waves. Otherwise, the USACE never contemplated significant
modification of the MRGO project to make it compatibile with – or truly integrate it with
-- the LPV public safety project.

The USACE (2005) re-evaluation study lists numerous tried and true remedial steps that
could have been taken at any time during the life of the MRGO. They might have, for
example, considered protecting the Chalmette EBSBs against waves, as the channel
enlarged and diminished the intervening vegetated buffer on one side, and between the
channel and Lake Borgne on the other. The rapid bank line retreat was well documented
before the storm (USACE 1988). Surely, the USACE knew or should have known that
this would increase the potential that storm-driven forces impinging on the EBSBs might
damage these delicate structures, thereby increasing the likelihood that they would be
overtopped and breached and rendered ineffective earlier in a serious storm sequence
than would have been true otherwise (and cause catastrophic flooding).

The Corps never seriously revisited the original 1950s era oceanographic assumptions
about surge and waves in the Lake Borgne funnel although they were quite aware of the
changing conditions on the ground. The MRGO, a failed economic development project,
was a pre-existing hazard that was maintained and operated by the Corps in such a way
that it became a greater threat with each passing day. For reasons that remain unknown
and inexplicable the shocking expansion of the top width of the MRGO channel --
cultivated under USACE care – never was addressed by the chronically under-funded and
incomplete LPV hurricane protection project or with MRGO funds.

3.2 MRGO De-authorization and Remediation

After Hurricane Katrina, the USACE estimated that $60 to $130 million would be
required to remove the sediment that the storm had pushed into the channel in the open
water MRGO Reach 3 segment (USACE 2007). Responding to public outcry, Congress
directed that no public funds be used to reopen the channel to navigation unless and until
the USACE undertook a public evaluation of the desirability of permanently closing, or
de-authorizing, the MRGO project. With much prodding, the USACE reached a
conclusion that would have been equally true at any time in at least the last three decades,
namely that there was no economic justification to keep the MRGO open (USACE 2007).
Congress thereupon de-authorized the MRGO in the Water Resources Development Act
of 2007, a half century after it had set the destructive jugernaut in motion.

The first physical, as contrasted with the legal, closure is scheduled to begin this fall with
construction of the first of two massive dams. One, at the Bayou LaLoutre ridge south of
Shell Beach (Figure 3.4) is promised before the start of the 2009 hurricane season
(Figure 3.5). This structure is an important first step in reversing decades of
environmental destruction caused by the channel. But because this dam is located east of
Lake Borgne where the surge is generated, and not between the City and the Lake, it will

33
not restore any significant increment of protection to New Orleans against the hurricane
surge hazard that continues to be posed by the now-deauthorized MRGO project
(USACE 2007).

In February 2008, the USACE revealed a plan that, if constructed as promised by 2011,
would represent a major surge protection upgrade for the City (Figure 3.6). This plan
includes a second dam across the MRGO to the west of Lake Borgne in the vicinity of
Bayou Bienvenue, eerily similar to that recommended in the Crosby Plan 42 years ago
(Figure 3.2).

Figure 3.4 Inland portion of MRGO project area showing the spoil disposal area
adjacent to the south bank of the MRGO channel, and the location of Bayou
LaLoutre ridge at Mile 36 where first closure dam is to be constructed under the
approved de-authorization plan (USACE 2007).

34
Figure 3.5 Plan view of Bayou LaLoutre closure dam to be completed by spring,
2009 (http://www.mvn.usace.army.mil).

3.3 U.S. District Court Decision in January, 2008, Subsequent May, 2008 Order, and
Corps’ Technical Defense

Much has happened since the Congressional de-authorization of the MRGO project. In
January 2008, U.S. District Court Judge Stanwood Duval, Jr., issued an opinion calling
the “engineering calculation” upon which the LPV hurricane flood protection system was
premised “outdated and lacking,” with deficiencies “known to the Corps but ignored
because of funding concerns.” He concluded, however, that an immunity provision of the
1928 Flood Control Act frees the government from liability for flooding due to the poor
design, construction and maintenance of LPV floodwalls on the 17th Street and London
Avenue Canals. The court has not, however, been willing to grant the same immunity for
increased flood damages resulting from the design, construction, operation and
maintenance of the separately authorized MRGO navigation canal, an economic
development project with no flood control or public safety purpose.

Judge Duval also issued an order governing the further conduct of this case on May 2,
2008. Therein, he noted that the “MRGO was constructed without any consideration of
its causing any ‘surge’’’ (Order, 5/2/08, p. 10). He writes:

“In the case before this Court, there were two projects, with two different
funding methods and two different concerns driving each. The LPV
sought to prevent flooding; the MRGO sought to promote deep draft
shipping.” (Order, 5/2/08, p. 35)

Judge Duval continued:

35
 “There is no proof or any evidence that the Corps in designing the LPV
ever took into consideration any of these changes that MRGO had
wrought on the area. As such…the LPV was not designed to prevent the
storm surge that occurred.” (Order, 5/2/08, p. 36)

The Government continues to assert that the free-flowing surge connection afforded by
the MRGO to the interior of the city, and the destruction that the ever-expanding channel
caused to the pre-existing wetland wave and surge buffer east of the city, played no role
in exacerbating flood damage to New Orleans East, the Lower 9th Ward and St. Bernard
Parish during Hurricane Katrina. Congress has implicitly rejected this benign view of the
MRGO by authorizing construction of three barrier dams across the bed of the former
navigation channel, the one already mentioned at Bayou LaLoutre for environmental
restoration (Figure 3.5), a second closure near Bayou Bienvenue as part of the USACE
100-year protection plan (Figure 3.6), and another structure in Reach 1 near Paris Road.

The Bayou LaLoutre closure is also the first phase of a jointly funded state-federal
restoration program under the Louisiana Coastal Protection and Restoration Authority
(LCPRA) Masterplan that also calls for a major river diversion across the de-authorized
channel alignment (LCPRA 2007). The Masterplan that was approved by the Louisiana
Legislature in 2007 includes provisions to re-establish the former natural surge and wave
protection afforded by swamp forests and marshlands destroyed by the MRGO in the
funnel area east of the city (LCPRA 2007). Any or all of these barriers could have been
constructed at any time within the last 50 years, and, in fact, were in other places around
the country (See Chapter 6).

The Government has adopted a fall-back position that the surge and waves generated by
Hurricane Katrina were so far beyond those anticipated in the LPV design that any
incremental added effects that might be attributed to the MRGO were inconsequential to
the levels of damage sustained within the populated areas of St. Bernard Parish, the
Lower 9th Ward and New Orleans East. These assertions have drawn on technical
analyses conducted by the USACE directly, or by their contractors, that we have
examined in considerable detail. Some of these internal efforts have been publicly
documented in reports issued by the USACE Interagency Performance Evaluation Team
(IPET 2006, 2007), while we have more recently become aware of others through the
litigation discovery process in Robinson v. United States.

Much of the USACE-sponsored post-Katrina work is of very high quality, but we have
found the analyses of surge and waves in the funnel area to be cursory and superficial
when compared to the more comprehensive analyses performed by the Robinson expert
team and discussed in the next four chapters. This is particularly true of USACE
assessments of (1) the amplification of waves caused by loss of vegetated buffers as
determinative factors in the early failures of the Reach 2 EBSBs, and (2) the effect of the
enlarged MRGO Reach 1 cross-section on the elevation and prolonged duration of surge
in Reach 1 and the IHNC.

36
Our own assessment has employed tools and expertise, including some from our
colleagues at Svasek and the Delft University of Technology in The Netherlands. All of
the techniques that we have applied were in widespread use throughout the world prior to

Figure 3.6 Current USACE plan to protect City against Lake Borgne surge expected
to recur once in 100 years (100-year protection level) includes dam across the throat
of the Lake Borgne funnel between the Central Wetlands and New Orleans East, a
navigable flood gate on the GIWW south of New Orleans East, and a closure dam
across the de-authorized MRGO canal south of the current Bayou Bienvenue
structure(http://www.mvn.usace.army.mil/hps/100yr_design_map.html).

37
Hurricane Katrina. In some cases, we have merely carried some of the same analyses
initiated by the USACE to a more logical and instructive endpoint. Based on everything
that I have seen, I conclude that the MRGO project was a substantial contributing factor
to the catastrophic flooding of the Robinson plaintiffs’ homes and communities. Put
another way, but for the adverse effects of the MRGO (funnel configuration, destruction
of wetlands, channel expansion), Hurricane Katrina would not have triggered the
catastrophic degree of flooding that destroyed New Orleans East, the Lower 9th Ward and
St. Bernard Parish (Figure 3.7).
.

Figure 3.7 View of a flooded city looking west from above the Lower 9th Ward
toward the IHNC and the Orleans Metro area four days after Katrina

38
 CHAPTER 4
REVIEW OF EARLIER SURGE STUDIES

Prior to the work described in this report, eight studies have been published that directly
or indirectly touched on the potential for the MRGO to exacerbate flooding during the
passage of hurricanes. The first of these studies, Bretschneider and Collins (1966), also
known as the NESCO investigation, was sponsored by the USACE NOD after Hurricane
Betsy. It was mentioned earlier and will be discussed in more detail here. The NESCO
study is important because it was the only comprehensive oceanographic analysis of
surge published prior to Hurricane Katrina that considered the interaction between levees
and channels in the Lake Borgne funnel. A second Betsy analysis published in the
National Weather Review in 1968 (Goudeau and Conner 1968) was a data source for the
Hsu study that has been quoted previously (Hsu 1973). Hsu (1973) did not provide a
comprehensive surge analysis, but a well-reasoned critique of the NESCO report.
Together, these might be considered the third MRGO study. Fourth, the USACE-NOD
sponsored its first analysis of the MRGO channel employing a modern surge model in
2003, using the S08 ADCIRC version to assess the effects of constricting or closing
MRGO Reach 2 at the La Loutre Ridge with a structure similar to the dam now slated for
construction as a part of the restoration work associated with MRGO de-authorization
(USACE 2003). This study was not focused on storm surge, however, but on reducing
salinity intrusion to Lake Pontchartrain during normal tides.

Three relevant investigations have been completed since Hurricane Katrina, and, as might
be expected, have been centered around understanding what happened during that event.
The sixth study was commissioned by the Louisiana Department of Natural Resources
(LDNR) before the storm, but was changed in mid-course to include Katrina simulations.
It was an ADCIRC based modeling investigation managed by URS Corporation (2006)
that considered the effects on surge of removing both the Reach 2 channel and adjacent
LPV berms. The URS investigation used the S08 version of ADCIRC that the LSU
Hurricane Center had used for forecasting Hurricanes Katrina and Rita, and was the most
comprehensive analysis of all of the elements of the funnel prior to the work of the
Robinson expert team reported here. It has gotten little attention, however, because it
used an unrealistically powerful wind field for Hurricane Katrina, and did not consider
wetland effects. The seventh investigation was undertaken by the USACE IPET and
appears as an appendix in both the 2006 and 2007 versions of that report. It was not as
comprehensive as the URS study, confining itself to an assessment of the effect of
removing only Reach 2 of the MRGO, but it did correct the wind problem and address
wetland effects in a cursory manner (IPET 2006, 2007). The IPET study used the SL15
version of ADCIRC that improved the fidelity of the geometry of the MRGO channel.
Finally, Team Louisiana conducted an ADCIRC modeling program as part of their
forensic investigation that employed the S08 version of ADCIRC, and has been discussed
in the three previous expert reports that I have submitted (Kemp Class Certification 2007,
Kemp 702c Immunity 2007, Kemp 2008).

Since the filing of the current lawsuit, the Robinson expert team has commissioned its
own modeling program, the ninth study, that is more comprehensive than any published

39
so far, considering the effects of both relevant reaches of MRGO -- Reach 1 and Reach 2
– using the ADCIRC S08 and the TUDelft FINEL models for flow, and integrating these
results with an analysis using the SWAN model for waves. We have considered the
MRGO in its entirety as an element of the funnel, something not accomplished in the
other post-Katrina ADCIRC studies which addressed only Reach 2 of the MRGO and
ignored the effects of enlarging Reach 1, a critical shortcoming.

From materials made available in the course of discovery, we are aware that the USACE
has continued to work on the same questions that we found unresolved by the IPET
investigations. Dr. Bruce Ebersole, an excellent oceanographer and coastal engineer at
the USACE Coastal Hydraulic Laboratory in Vicksburg, put together a presentation in
May 2007 titled “Hydrodynamic Loadings and Levee/Wall Responses Along the
Navigation Channels and Influence of MRGO during Katrina.” This provided us with a
great deal of relevant information developed by the Corps that significantly augments
what was published in IPET (Ebersole Presentation 2007). The long presentation (188
slides) is also styled as an effort to address MRGO issues raised by Team Louisiana
(2007). We know from this presentation, for example, that the Corps attempted to
address one shortcoming of the IPET report by conducting at least one ADCIRC run in
which the cross-section of Reach 1 was reduced by about 50 percent, still far short of the
90 percent reduction that would be required to simulate the pre-MRGO condition. Dr.
Ebersole also considers the dynamics of waves breaking on the MRGO EBSBs in a more
comprehensive and sophisticated manner than did IPET.

Below we highlight findings from the NESCO, IPET, Team Louisiana and Ebersole
investigations that are relevant to the conclusions outlined in the previous chapter, while
examining their assumptions, methodologies, conclusions and limitations. Then we will
review the logic and assumptions underlying our own investigation, and in this way
identify the realities of the MRGO role in hastening the onset and severity of flooding.
Because Team Louisiana (2007) reviewed these studies extensively, we incorporate many
of their observations in describing these studies.

4.1 The NESCO Report (Bretschneider and Collins 1966)

Hurricane Betsy flooded a significant portion of Greater New Orleans (GNO) though
generally to a lesser depth than did Hurricane Katrina. The damage caused by
Hurricane Betsy led those harmed to blame the MRGO, and lawsuits were filed against
the USACE. At this point, the USACE felt compelled to commission the first rigorous
analysis of the role of the channel in storm surge conveyance. Consultants Charles
Bretschneider and J. Ian Collins of the National Engineering Science Company
(NESCO) issued their report in September 1966, a year after Betsy struck
(Bretschneider and Collins 1966). They used early one-dimensional computer
modeling techniques to evaluate “the effects, in the vicinity of the Inner Harbor
Navigation Canal, due to rapidly and slowly rising surge… for four cases.” These cases
were:

40
 I. Existing levees with no MRGO
II. Existing levees with MRGO
III. Proposed levees with MRGO
IV. Proposed levees with no MRGO

The LPV was in the first year following authorization, so the hurricane levee on
the MRGO spoil bank was still “proposed,” though the 6 to 8 feet of spoil had
some effect during Betsy. The GIWW and MRGO were in place, but the
“throat” of the funnel was not as constrained as it is today, because the real
levee constructed later (hauled clay not hydraulic fill) along the south bank of
Reach 1 was just a spoil bank (Figure 4.1). Cypress swamps were experiencing
severe salinity stress and trees began dying as soon as the channel was
connected to Breton Sound, but much of the land loss that has been documented
by other members of the Robinson expert team was still in the future.

Figure 4.1 Diagrams showing the geometry used by Bretschneider and

Collins (1966) for simulation of surge in the funnel under the four test
conditions.

41
The geometry was highly schematized, and the model was constrained by a number of
empirical tuning coefficients that limit the physics in ways that should have been
questioned then, and certainly would not be accepted today. Partly because the set-up
was so simple, however, the results, if considered qualitative and representative rather
than truly quantitative, can provide important insights into how storm surge propagates
through this system. The first of these, apparent in the choice of cases to be analyzed,
is that the enlarged channel and the geometry of the levee system will both produce
effects on surge that can be distinguished and treated somewhat separately. The second
was the recognition of the time-dependency of the surge produced by a moving storm.
This means that surge can propagate only a limited distance in a given direction if the
winds that are forcing it continually change in direction and magnitude, so a steady-
state condition cannot be assumed. Finally, Bretschneider and Collins (1966) made the
first serious effort to integrate wetlands into a surge analysis, and provided an early
estimate of the degree of resistance to flow that a marsh might impose, compared to a
deep channel. They concluded that the resistance would be more than three times
greater for flow over the marsh than through the channel.

“It must be expected that a large channel cut through marsh areas will
permit more water to arrive at a faster rate in the interior of the
marshland, at least in the immediate vicinity of the channel. In addition,
the maximum elevation and the steady state peak will be reached at an
earlier time….Without the channel, the water will rise over the
marshlands at a slower rate and it will take a longer duration to reach
maximum elevation and steady state conditions. Similarly, after the
storm has passed it will take longer for the surge to recede since now
there is no channel to act as a drain…The velocity of flow through the
channel will be two to four times as great as that over the marshland, but
the volume of water (velocity times cross-sectional area) determines the
total amount of water which will enter…It is this later factor which tends
to cause an increase in surge because of the Mississippi River-Gulf
Outlet.” (emphasis added)

Bretschneider and Collins (1966) calculated surge along two traverses through the
funnel for Hurricane Betsy (Figure 4.2).

42
Figure 4.2 Test traverses along which surge was computed by
Bretschneider and Collins (1966) for Hurricane Betsy and for the synthetic
critical SPH storms. MRGO 0 and CCL 0 are at the marsh shoreline of
Breton Sound, while MRGO -16 is at Shell Beach and CCL – 30 is at the
Chef Menteur Pass.

Figure 4.3 Predicted surge hydrographs at distances inland from the

marsh edge at Breton Sound computed for Hurricane Betsy along
Christmas Camp Lake Traverse (left) and Mississippi River – Gulf Outlet
Traverse (right) using Bretschneider and Collins (1966) model. MRGO -6
in the right panel should read MRGO -16 and corresponds to the gage
location at Shell Beach (Alluvial City).

43
Figure 4.4 Surge hydrographs captured at water level gages in the funnel
during Betsy (Bretschneider and Collins 1966). Computed curves at
MRGO -16 and CCL -30 can be compared with gage readings at Alluvial
City and Paris Road Bridge, respectively (Figure 5.3). Note the much
lower surge at Lake Pontchartrain in the IHNC (Seabrook).

The model predicted a dramatic increase in moving inland through the funnel along
each of these traverses, from a maximum of 10 feet above mean sea level at the coast to
about 16 feet in the eastern lobe of Lake Borgne (CCL -30, Figure 4.3). The actual
surge was not known at the coast and had to be developed by the model, and may have
been somewhat lower than predicted (See Figure 4.3). Computed maxima correlated
reasonably well with available gage data at Alluvial City (MRGO -16). The computed
values for Lake Borgne and Chef Menteur Pass, CCL -20 and CCL -30, respectively
were, however, significantly higher than what was observed at the Paris Road gage five
miles to the west (Figure 4.4). The model also predicted that the surge peak would
arrive at the gages about 4 hours earlier than it actually did.

Bretschneider and Collins (1966) attributed the error in the timing of the surge maxima
to a lack of consideration of inertia effects in the storm tide equation used. For the
error associated with the level of the maxima, they stated that “no system of equations
can really be expected to predict, with a great degree of accuracy, the complex physical
phenomenon of flooding over marshland, bayous, houses, trees, etc.” They concluded
that “the computation of storm surges over semidry land must be regarded as an art
rather than a science.”

Bretschneider and Collins (1966) failed to mention far more likely sources of error.
Later surge modelers have found that (1) correct parameterization of the time-
dependent forcing wind stresses, and (2) accurate topography and bathymetry are the
most critical factors needed for realistic surge simulation (Westerink et al. 2004). At

44
the time that this work was done, the dynamics of winds for any particular storm like
Betsy were poorly understood and were essentially unknown both offshore and inland
in the area of interest. With respect to the highly schematized geometry employed,
Bretschneider and Collins (1966) did not challenge the belief prevalent at the time that
mean sea level was equivalent to zero NGVD29, the datum at the few gages operating
in the area of interest (Figure 4.4). Further, they state that the “marsh was assumed to
have a water depth of -2 feet, mean low water.” As a result, the intertidal marsh
surface, normally between 1 and 2 feet above mean sea level, was represented as being
well below mean sea level. Marsh at this elevation would not have retarded surge in
the model nearly as much as in the real world.

A modern interpretation of Bretschneider and Collins (1966) results is that the

Hurricane Betsy wind fields used to develop the surge were not accurate either over the
shelf, over the funnel, or both. They do say that “in most practical cases of hurricanes,
the assumption of constant wind speed and direction over the continental shelf is not
justified.” But, instead of challenging the wind and geometry data they were given by
the USACE, and pointing out the need for better information to improve results, the
consultants assumed that the sketchy information provided was entirely accurate, and
then used a statistical curve matching approach, together with empirical “surge
adjustment factors” and “planform factors” proposed by the New Orleans District, to
bring their model predictions into alignment with the observed values at Paris Road
bridge. They do express some dissatisfaction with this approach, which has the effect
of subordinating their elegant physical scheme to empirically derived factors with no
physical meaning. This is the “excellent negative critique” that Dr. Hsu (1973) noted in
his assessment of this study quoted earlier (See Chapter 3). Bretschneider and Collins
(1966) state:

“Confidence in such a prediction scheme could be expected for a

hurricane which has a similar traverse and speed to Betsy. It is apparent
that more elaborate prediction schemes can be developed. Because of
their empirical nature, prediction schemes can only be expected to give
reliable results for conditions which are almost repetitive. In particular,
for example, any hurricane whose eye passes over one of the traverses
will yield too low a storm surge prediction along that particular traverse
because the resulting prediction equation is not applicable in this
case…For conditions of sparse data and several completely different
hurricane tracks the method appears to be impractical, and it was
therefore necessary to return to the “surge adjustment factor” based on
matching peak surge.”

The Bretschneider and Collins (1966) model, when the physics were unconstrained by
empirical tuning factors, showed the expected increase in surge due to the “funnel”
effect. They abandoned this approach, however, when the results did not accord with
the observed gage readings, and then had to come up with a different way to get an
estimate of the relative contribution to the surge of flow through channels and over the
marsh during Betsy. To do this, they retreated to a steady-state assumption that is

45
inherently at odds with actual storm surge dynamics. The steady-state requires an exact
balancing of forces that removes or disregards the spatial and temporal variations in
velocity that govern surge generation. It is not surprising then, that the residual
influence of the MRGO channel “had an effect of increasing the storm surge throughout
the marshland for Hurricane Betsy by about 0.3 to 0.4 feet maximum.” Although the
methodology is suspect, it is important to note that Bretschneider and Collins (1966)
conclude that the MRGO did have an effect on surge, albeit a small one during Betsy,
which would have contributed to flooding if protective structures were overtopped or
breached.

In considering the weight to give this result when it is applied other storms, it should be
pointed out that Bretschneider and Collins (1966) used a diminutive 19,000 square foot
channel cross-section for the MRGO that corresponds to the “box-cut” dimensions of
the maintained navigation channel, rather than the size of the dredged cut (Figure 4.1).
That channel now averages 29,000 square feet in Reach 1 and 41,000 square feet in
Reach 2 when the water is at mean sea level (Figure 4.5). And of course, the marsh
was represented as a large lake with a bed well below the actual elevation, so this
would have worked to increase the flow over the marsh at the expense of the channel.
Given that the channel is twice the size modeled, and that the marsh has a higher
elevation than modeled, it can be expected that the actual effect of the MRGO on surge
during Betsy might have been greater than the consultants reported by a significant
factor, perhaps a four-fold increase. An increase in surge associated with Betsy of 1.2
to 1.6 feet over the elevation without the MRGO should have been of interest to the
levee designers and to those whose homes flooded during Betsy, but the conclusions of
this report tended to allay rather than raise concern.
10

0
Channel Bottom Elevation (ft, MSL)

0 200 400 600 800 1000

GIWW
-10

-20

MRGO Today

-30

MRGO Design Section

-40

-50
Distance (ft)
MRGO Now MRGO Design GIWW Design

Figure 4.5 Comparison of GIWW and MRGO cross-sections, as designed and post-
Katrina, in Reach 1, the connection to the IHNC. The design cross-section is 23,000
square feet, while the current cross-section averages 29,000 square feet in Reach 1,
and 41,000 square feet in Reach 2.

46
The most intriguing analysis carried out by Bretschneider and Collins (1966), and the
part that best withstands modern scrutiny, pertains to the effect of the levee alignments
and channel in the vicinity of Reach 1 on surge in the IHNC. This was a simpler
problem than the others that the consultants tackled, because it essentially reduced to
one of channel and overbank conveyance in Reach 1. They presented results for fast-
moderately fast- and slow-rising hypothetical storm surges that attained an elevation in
Breton Sound of 10 feet for the four cases described earlier (Figure 4.6). Significantly,
Bretschneider and Collins (1966) predicted that (1) adopting the proposed levee
configuration and (2) including or excluding the MRGO channel would each affect
surge transmission to the IHNC for all rates of surge development in the Lake Borgne
estuary.

For the most rapidly rising and falling surge at the coast, the IHNC does not rise at all
without the MRGO (Cases I and IV), because there is insufficient time for the surge to
propagate up the estuary (Figure 4.6). For the moderate rate of rise, the surge that
reaches the IHNC without the MRGO has a lower peak elevation and a narrower peak
signifying a shorter duration of high water conditions. For the slow rising surge, the
peak elevation in Lake Borgne is attained, or nearly attained, in the IHNC for all cases,
after lags of varying duration, but the width of the peak is reduced without the MRGO
project (Cases I and IV). The period above 8 feet, or the 80 percent exceedence
interval, in the slow-rise scenario drops from 3.5 to 2.0 hours for the condition with and
without the MRGO, respectively, with the new levees in place (Cases III and IV). In all
cases, the predicted effect of building the “proposed” levee system that now forms the
throat of the funnel (Cases III and IV) is to hasten the onset of peak surge and to
lengthen the period of highest water.

Given these results, it is disappointing to look at the consultants’ summation of the

effects of the channel because if there was ever an opportunity to influence the Corps to
think about what they were doing, this was it…

“It is seen that the effect of the Mississippi River-Gulf Outlet is almost
negligible for all large hurricanes accompanied by slow rising storm
surges. It may be expected that once in a while a storm may occur which
has a somewhat freakish, more rapidly rising surge in which case the
Gulf Outlet may have a very marked effect. However, such a storm will
not produce tides as high as the more critical hurricane tracks such as
Betsy or the synthetic hurricanes.”

The synthetic hurricanes referred to are those based on the 1959 SPH that were used as
a basis for design of the LPV. We now know that Bretschneider and Collins (1966)
were generally correct with regard to peak surge height, if not duration. The surge of
Betsy rose in 10 hours and Camille in 8. Katrina’s surge rose in two stages that
together lasted 25.5 hours (Figure 4.7). Nearly 17 hours of slow rise from 3 to 6 feet
(NAVD88) occurred on August 28, the day before Katrina made landfall. A rapid rise

47
Figure 4.6 Simulated surge in the IHNC for a given input surge in Lake
Borgne with a peak elevation of 10 feet (MSL). The top panel is for a
rapidly rising surge that peaks in 3.5 hours, the middle for a moderately
fast (7.0 hrs), and the bottom is for a slow rising surge (14 hrs). Cases
tested: I - Existing levees without MRGO, II - Existing levees with MRGO,
III - Proposed levees with MRGO, IV - Proposed levees without MRGO
(Bretschneider and Collins 1966).

48
 Landfall
18
Modeled GIWW Peak@Paris Rd (17.1 ft)
16 Modeled
IHNC
Peak@I-10
14 Observed IHNC Peak @ Lock (14.2 ft) (15.0 ft)
Water Elevation (ft, NAVD88)

12

10

2
Slow Rise = 3 ft in 17 hours
Fast Rise = 8+ ft in 8.5 hours
0
8/25/05 0:00 8/26/05 0:00 8/27/05 0:00 8/28/05 0:00 8/29/05 0:00 8/30/05 0:00 8/31/05 0:00
Time
IHNC Gage@I-10 IHNC ADCIRC@I-10 GIWW Gage@Paris Rd GIWW ADCIRC@Paris Rd IHNC Gage@Lock

Figure 4.7 Katrina surge in the IHNC and on the MRGO/GIWW at Paris Road.
The only gage that recorded the peak was at the IHNC Lock (blue diamonds). The
USGS gages on the IHNC at the I-10 bridge (purple squares) and on the
MRGO/GIWW at Paris Road (red squares) both malfunctioned at about 10 feet,
but all three gages recorded a slow initial rise of 3 feet over 17 hours, followed by
the a more rapid rise over the first 8.5 hours of August 29. ADCIRC model results
give the best estimate for peak elevations in the IHNC at the I-10 bridge (green
line) and for MRGO/GIWW at Paris Road (purple line).

of more than 8 feet occurred over the first 8 to 9 hours of August 29. Surge in the
IHNC peaked at about 15 feet, an estimated 2 feet lower than the peak on the
MRGO/GIWW at Paris Road (Figure 4.7). Peak stage in the IHNC was lowered by the
increment of the flow lost over levees and floodwalls and through breaches, as will be
discussed in the next chapter. Drainage toward Lake Pontchartrain, which was 3 to 6
feet lower at the time, also lowered surge level in the northern reach of the IHNC.

All three storms that flooded New Orleans through the MRGO ‘back door’ were
accompanied by what Bretschneider and Collins (1966) would classify as slow rising
surges. The study could have been interpreted as cautionary with respect to the design
of the hurricane protection system then being built. It showed that surge elevations
experienced at the apex of the funnel near the Paris Road bridge would be transmitted
to the IHNC without diminution, and that levee and floodwall crown elevations
throughout this system would have to be essentially the same throughout. Further, it
showed that the levee proposed along the south bank of MRGO Reach 2 would enhance
the build-up and transmission of surge into the City. What is disturbing is how the

49
Bretschneider and Collins (1966) study, rather than ringing alarm bells, appeared to lull
the USACE back into pre-Betsy complacency. The large, free-flowing MRGO ship
channel into the LPV clearly posed a hazard to the City, yet the Corps never seriously
considered a flood gate that, 40 years later, IPET (2006) and those designing the 100-
year protection plan have since found was so obviously indispensable.

“The Reach 1 GIWW/MRGO section is very important in determining

the magnitude of the storm surge that reaches the IHNC from Lake
Borgne and Breton Sound. If the hydraulic connectivity between Lake
Pontchartrain and Lake Borgne is eliminated at a point within this
section of channel, tide or surge to the west of this point will become
primarily influenced by conditions in Lake Pontchartrain” (IPET 2006,
IV-6-2).

Ironically, the gate that was proposed as part of the original LPV “Barrier Plan” at
Seabrook, where the IHNC meets Lake Pontchartrain (Figure 4.8), would have
restricted drainage toward the Lake and increased flood level in the IHNC for any surge
originating in the Lake Borgne. Hurricane Betsy provided a “wake-up call” that should
have alerted USACE designers developing the LPV that the plan developed in the
1950s prior to MRGO construction was too exclusively oriented toward a Lake
Pontchartrain surge. Hurricane Camille provided yet another demonstration of the way
a Lake Borgne surge would outflank this protection scheme. Currently, designs are
being developed to overcome the shortcomings of this “Maginot Line.” The barriers
originally proposed for two of the Lake Pontchartrain passes (Figure 4.8) are being
augmented with a third to address surge arising in the funnel (See Figure 3.6). Most
concepts include a levee on the west side of Lake Borgne to break up the funnel
geometry, with a gate on the GIWW and possibly a second to accommodate drainage
and shallow-draft traffic on a deauthorized MRGO (Figure 4.9). If such a system had
been in place during Hurricane Katrina, water levels in the IHNC and MRGO Reach 1
would have experienced only Lake Pontchartrain surge, and flooding of the homes and
communities of the Robinson plaintiffs would have been reduced significantly or
avoided altogether (Figure 4.10). One of the scenarios tested in Chapter 6
accomplished much the same thing and the FINEL model showed Lake Pontchartrain
water entering the IHNC at Seabrook for most of the surge sequence.

50
Figure 4.8 Creation of the ‘funnel,’ as shown in 1966 GDM 3, Supp. 1, for the
Chalmette Levee Extension. GNO HPS authorized in 1965 was quite different from
what has actually been built. It included surge barrier gates at the two passes from
Lake Borgne into Lake Pontchartrain (Rigolets and Chef Menteur), and one at the
mouth of the IHNC (Seabrook). Neither was constructed.

Figure 4.9 An improved flood protection scheme developed by H.S. Mashriqui with
the LSU Hurricane Center as part of Team Louisiana (2007) is similar to the plan
recently adopted by the USACE for the 100-year protection system (See Figure 3.6)

51
Figure 4.10 Predicted flooding during Hurricane Katrina if the Mashriqui plan
(Figure 4.9 above) had been built to close off the funnel (Team Louisiana 2007).

4.2 IPET 2006 and URS 2006

A draft report sponsored by the LDNR appeared in late 2005 shortly after Katrina that
used the ADCIRC S08 model to investigate the effect of the MRGO on storm surge
(URS 2006). It was revised to use a newer version of ADCIRC (SL15) and more realistic
winds and then reappeared, in part, as an appendix of the second report issued by the
USACE IPET (IPET 2006) and appeared again in the final report (IPET 2007, IV-6-1).
IPET (2006) cleared up some of the confusion left by the consultant’s report but still
reached the same conclusion, namely, that the influence of the MRGO on regional surge
maxima is very small, generally less than a foot, for a range of storms including Katrina
(Figure 4.11).

52
Figure 4.11 IPET (2006b, IV-6-34) depiction of the difference between the surge
maxima surface developed by ADCIRC for Katrina with and without Reach 2 of the
MRGO. Scale at left shows the differences (‘With Reach 2’ minus ‘Without Reach
2’) ranged from -0.7 to +1.1 feet.

It is important to understand that neither of these ADCIRC studies addressed the entirety
of the MRGO channel, dealing only with the Reach 2 portion that extends through the
marsh southeast of the more “critical” Reach 1 segment that has been discussed as the
throat of the funnel. Both URS and IPET raised the ADCIRC computational nodes that
previously were in Reach 2 of the MRGO channel to marsh elevation to simulate a filling
of that part of the channel. At the time that this work was done, a single bed friction
value (Manning’s ‘n’) was being used universally in ADCIRC to characterize channels,
marsh, bays and the continental shelf. This was reasonable for a regional surge
forecasting model, but has been found to be inappropriate for more detailed work, such as
apportioning flows between channels and marsh within estuaries, along the lines of what
Bretschneider and Collins (1963) attempted four decades earlier.

Bretschneider and Collins (1963) used a Manning’s value for the marsh of 0.08,
compared to 0.02 for the channel, while a single value of 0.02 was applied everywhere in
the ADCIRC version IPET used for assessment of the influence of MRGO Reach 2. So
the channel bathymetry was eliminated but the differential in friction between the channel
and marsh was not addressed. Each study team ran a version of the model with and
without MRGO Reach 2, extracted the maximum surge elevations attained during both
runs, and then subtracted the “Without Reach 2” surface from the “With Reach 2”
surface. IPET created a graphic depicting the magnitude of difference throughout the
area of interest (Figure 4.11). IPET (2006, 2007) explains what they see as the very

53
limited influence of MRGO Reach 2 in the vicinity of New Orleans for strong storm
events as follows.

“First, the MRGO does not influence the important preliminary east-west
movement of water that drives the significant build up of surge in the early
parts of the storm. Second, the northerly propagation of surge during the
later stages of the storm are only minimally influenced by the MRGO
because the increased hydraulic conveyance associated with the channel is
very limited for large storms due to the large surge magnitude and
especially due to the very large lateral extent of the high waters on the
Mississippi-Alabama shelf that build up early on from the east. In
addition, the propagation direction of this surge wave does not typically
align with the MRGO and furthermore the southeasterly winds which
align with the MRGO occur only very briefly” (IPET 2006, IV-6-6).

Some of this rationale is reminiscent of the general conclusions of Bretschneider and

Collins (1963), while other aspects are unique to the dynamics of Katrina. The surge
generated by Katrina, in contrast to that of Betsy, reached a maximum along the south
side of the funnel rather than in the throat. But it is worth taking a closer look at what the
consultants and the IPET surge modelers actually found (Figure 4.11). First, it is
apparent that while the difference in maximum surge elevation anywhere due to MRGO
Reach 2 is small (-0.7 to +1.1 feet), the presence of this channel segment has effects that
are distributed over a large area.

Effects on surge maxima are generally greater outside the funnel than within it,
suggesting that the main effect of the channel is to convey water away from the funnel,
both northwest into Reach 1 and southeast beyond the eastern end of the Chalmette
hurricane protection levee. An elongated blue area is shown on the south side of the
funnel, indicating a reduction in surge elevation of up to -0.5 feet. This blue zone
corresponds to two key features. The first is the Katrina surge maximum that stacked up
on this portion of the levee along the south bank of MRGO Reach 2 (Figure 4.12). The
second is the MRGO Reach 2 channel itself, which was eliminated in the ‘Without’ run.

Modeled flow direction vectors for the MRGO Reach 2 channel when the surge was
highest show flow away from the zone of highest surge at both the northwestern and
southeastern ends of this feature (Figure 4.12). The flow direction is towards the
entrance of MRGO Reach 1 at the apex of the funnel in both cases, but the mean depth of
the MRGO at this stage is more than 50 feet, while the depth over the marsh is about 15
feet. So the channel acted to distribute water stacked up against the levee, thereby
lowering the peak surge in the immediate vicinity of the Reach 2 channel segment. Flow
out of the southeast end moved around the Chalmette levee and contributed to increased
elevations over a considerable area, particularly in the Caernarvon area where the
increase was more than 1 foot. Similarly, water evacuated by the channel to the
northwest raised the surge elevation in the throat of the funnel in the vicinity of the Paris
Road bridge by about +0.5 feet.

54
Figure 4.12 Flow direction in the funnel at peak surge in Katrina ADCIRC
simulation by H. Mashriqui of the LSU Hurricane Center. The surge reached more
than 5 meters (15 ft) on the MRGO hurricane protection levee. Flow converges
from all directions toward the throat of the funnel, but some that cannot be
accommodated moves east as a return flow in the GIWW. The MRGO conveys flow
away from the zone of maximum surge along the LPV EBSB that follows Reach 2
(Team Louisiana 2007).
Perhaps the most striking feature of the IPET Reach 2 analysis is, however, the effect of
the presence of the ship channel on levee overtopping and flooding of developed areas
(Figure 4.13). As is the case for all ADCIRC simulations including the ‘with’ and
‘without’ Reach 2 runs, levees and floodwalls are overtopped when the surge rises above
the crown elevation, set in ADCIRC at idealized elevations, where the levee and
floodwall crowns were supposed to be. The model does not allow breaching, or
accommodate rainfall or pumping. So, water enters the protected areas only for as long
as the flood side water level is high enough to flow over the crown level of the protection
structure, and overtopping ceases when it drops below this level. Water accumulates
inside the protected area and ponds in low spots.

The difference in maximum surge elevation on the flood side for the ‘with’ and ‘without’
Reach 2 simulations is small, and in the case of the IHNC, less than 0.1 foot (Figure
4.13). Yet the Orleans Metro protected area on the west bank of the IHNC is depicted
with extensive areas of light blue and green indicating an increase in floodwater
elevations of 0.1 to 0.4 feet due to the presence of MRGO Reach 2. More extensive areas
of up to 0.6 feet of increased flooding are apparent in the New Orleans East protected
area. Lesser increases are shown for the Lower 9th Ward and the Chalmette area of St.
Bernard. Increased flooding of up to 0.5 feet is shown in the northwestern half of the St.
Bernard wetlands adjacent to MRGO Reach 1, while there is some lowering to the
southeast caused by reduced overtopping of the federal levee in the surge peak zone.

55
Figure 4.13 Close up of ADCIRC difference map showing the effects of MRGO
Reach 2 on flooding of protected areas. Earlier arrival of surge delivered by
MRGO Reach 2 caused overtopping to start earlier and resulted in increased flood
elevations in protected areas in New Orleans East, St. Bernard and Orleans Metro.

The IPET investigators were so intently focused on surge maxima that they appear to
have completely missed another effect of MRGO Reach 2 that is important to flooding
over the levees and floodwalls. Levee designers are interested only in the maximum
water level expected on the flood side of the structure, but when these structures are
overtopped or breached, other aspects of the surge become important. The only plausible
explanation for the increases in flooding predicted by ADCIRC within the protected areas
is that the presence of MRGO Reach 2 increased the period of time that the surge
exceeded levee and floodwall crowns. This would occur if the efficient conveyance of
the ship channel caused the surge to rise more rapidly in MRGO Reach 1 and the IHNC,
so that overtopping began earlier and lasted longer, regardless of the stage ultimately
attained. This could occur without any detectable increase in surge maxima.

IPET (2006b, IV-6-4) notes that their study reached a similar conclusion to that of earlier
studies going back to Bretschneider and Collins (1963) with regard to the effect of the
MRGO channel on relatively low storm surges, or, alternatively, during the early phase of
a large storm surge.

“In these situations, changes induced by MRGO Reach 2 are rather small,
0.5 feet or less, but this amounts to as much as 25% of the peak surge
amplitude. When the long wave amplitude is very low, the surge is more
limited to propagation via the channels. Once the surge amplitude
increases to the point where the wetlands become inundated, this section
of the MRGO plays a diminishing role in influencing the amplitude of
storm surge that reaches the vicinity of metropolitan New Orleans” (IPET
2006b, IV-6-4).

56
The elevation of the surge in the IHNC at any point in the surge sequence is determined
by a balance between flux entering through Reach 1 and that exiting to the Lake. The
IPET analysis suggests that the first few feet of the surge rise at Paris Road and in the
IHNC -- representing a volume of many millions of cubic feet of water -- enters initially
through the channel. Later, as the surge rises in the funnel, more of the water enters from
across the marsh. Given that mean velocity in the channel is at least twice that of the
flow over the marsh, and that the depth of flow in the channel is three to four times that
over the marsh, it is apparent that the 1,000 foot wide channel can deliver as much
volume in a given period as flow across a 6,000 to 8,000 foot wide marsh cross-section.
If the surge never rises high enough to cause flooding over levees and floodwalls, then
the routing of the surge, whether over wetlands or through the channel, is of little interest,
as the accumulated volume drains harmlessly out the north end of the IHNC into the
Lake.

During the early stages of the surge rise, when the contribution of the MRGO Reach 2
channel is proportionally more important, flux into the IHNC is not limited by the
dimensions of Reach 1. Later, as surge continues to rise in the funnel, and the relative
importance of the MRGO Reach 2 channel contribution is diminished, flux into the IHNC
through Reach 1 becomes limited by the conveyance potential of that connecting channel,
and any additional surge buildup at the apex of the funnel has little effect on discharge to
the IHNC.

The “critical” condition, when Reach 1 conveyance has reached its full potential, is
indicated at least qualitatively in ADCIRC simulations by the development of a return
flow in the GIWW that conveys rejected surge volume away from the entrance to Reach
1 (Figure 4.12). All water that enters MRGO Reach 1 contributes to the surge elevation
along that connection. Some of this water at later stages in the surge was lost as it
overtopped the parallel levee system that confines Reach 1, again limiting discharge into
the IHNC.

The IPET focus on late stage surge development appears to suggest that the volume
introduced early when the flux was not limited by the dimensions of Reach 1, and when
the contribution of the Reach 2 channel was proportionally more important, is somehow
less consequential to the flooding dynamics than the volume introduced later in the storm
sequence. Late stage surge in MRGO Reach 1 and the IHNC added to a surge volume
already present. Another way to think about this is to attribute the base of the fully
developed surge in Reach 1 and the IHNC to volume introduced early through the
MRGO Reach 2 channel. The top of the surge peak is added later by flow across the
marsh. Both sources contribute to the maximum elevation attained by the surge, which is
limited by the conveyance of Reach 1. Because the water from MRGO Reach 2 gets to
the critical reach first, however, the surge maximum is attained earlier than if all water
came more slowly across the marsh, and flooding over levees and floodwalls in Reach 1
and the IHNC is also initiated earlier.

All of the ADCIRC runs discussed above applied a single bed friction factor everywhere.
In a separate analysis, IPET (2006, IV-5-94) investigated the effects of applying a

57
spatially variable bed friction factor to surge simulations for Katrina. They found that
merely increasing the Manning’s ‘n’ value over marsh areas from 0.02 to 0.035 had
significant local effects on the shape of the surge maximum envelope (Figure 4.14).
Unfortunately, this analysis was not sufficiently detailed to allow for differentiation
between the MRGO channel and the surrounding marsh. IPET (2006) states that “the
extensive network of connecting channels that occur throughout the marshes, cyprus (sic)
forests and other regions in Southern Louisiana were not considered in these
simulations.”

Generally, applying a spatially varying bottom friction factor tended to increase surge
height in open water and reduce it in wetland areas. This change reduced the predicted
peak surge in the Caernarvon area south of the St. Bernard levee system by up to 2 feet
(Figure 4.14). This estuary is not penetrated by large, deep, dredged channels. On the
other hand, applying a greater friction factor for the marsh in the funnel was found to
only slightly offset the higher levels generated in the open water to the north. It is
possible that much of the surge reduction expected from remaining funnel marshes was
diminished by the presence of the ship channel.

The effect of wetlands on surge reduction is an active area of research at present, in part
spurred by the availability of high-resolution surge models like ADCIRC that make it
possible to accurately trace surge flow-paths and look at varying frictional effects along
the way. Hurricane Rita, also a Category 3 storm, struck the southwest coast of
Louisiana 26 days after Katrina, as has been discussed. Rita pushed a storm surge in
excess of 15 feet across 50 miles of coast fronted by a sparsely settled 20 mile wide band
of coastal wetlands, while Katrina came ashore on a part of the deltaic plain near New
Orleans where wetlands have been greatly reduced over the past century by natural
subsidence and man’s activities. We will discuss the lessons from Rita on surge
reduction by wetlands in more detail in a later chapter.

Before the MRGO and the Chalmette hurricane protection berms were constructed, and
before much of the landloss caused by the MRGO had taken place, the funnel area
between Lake Borgne and the IHNC included a 10 mile wide band of wetlands that
should have reduced storm surge propagating into the IHNC. We have set up the FINEL
surge model to incorporate appropriate wetland friction effects, as well as the very
significant effects that dense forests have on the winds that generate the surge, essentially
separating the wind, which affects the tree canopies from the water. This work is
discussed in Chapter 6.

58
Figure 4.14 Difference in IPET ADCIRC surge maxima with spatially varying bed
friction factor (0.035 for marsh and 0.02 for open water) and the base run (0.02
everywhere). Variable less base is shown in feet. Open water areas generally
experienced higher surge (yellower), while surge in marsh areas without major
channels like the Caernarvon bight south of the St. Bernard HPS was reduced by up
to 2 feet (bluer). A similar reduction might be expected in the MRGO funnel but
does not appear, possibly offset by the presence of the large channel (IPET 2006, IV-
5-105).

While the USACE modelers have claimed that this analysis proves that the MRGO
project had no effect on the Katrina surge, we have pointed out in earlier filings that this
is an overgeneralization and an inaccurate conclusion. MRGO Reach 2 is but one
element of the funnel, and clearly not the most important one from a surge standpoint.
The MRGO project also included other features and impacts that they chose to ignore.
These must each be numerically isolated and tested before such a sweeping conclusion is
justified. Additional effects of the MRGO project, separately addressed in this report,
address the hydraulic impact of the destruction of the wetland buffer caused by MRGO
Reach 2, the interactions resulting from the adjacency of the channel to the flood
protection berms, the large effect on structure overtopping volumes developed from
apparently small differences in peak surge elevation and timing, as well as the effect of
the channel and diminished foreshore protection on wave energy faced by the LPV

59
earthen berms exposed on its south bank. None of these were touched on in the limited
analysis contained in the referenced IPET appendix (IPET 2007).

URS (2006) used the S08 ADCIRC model to simulate surge dynamics with the LPV
berm and MRGO Reach 2 removed, almost as we have done (see Chapter 8) (Figures
4.15 and 4.16). However, URS used much stronger winds than actually occurred during
Katrina, and reduced the LPV berm only to the elevation of the underlying material in the
spoil disposal area. We remove it entirely to marsh elevation in our Scenario 2 (Chapter
8). When URS lowered the LPV berm and removed MRGO Reach 2, this lowered the
maximum surge height along the LPV levee alignment at Bayou Dupre (without the
levee) by 1.5 feet, which is expected, but the URS simulation showed no reduction in
maximum surge elevation across the Central Wetlands. URS (2006) interpreted the
model output to mean that the marshes, and presumably the swamp forest that was there
before it was killed by the MRGO, were of no consequence in reducing surge.

Figure 4.15 ADCIRC S08 levees at idealized levels, including 17.5 feet for the
EBSBs along MRGO Reach 2, and 14.0 feet along MRGO Reach 1, and for the
IHNC floodwalls. The 40 Arpent levee varies from 5 to 8 feet high (URS 2006).

60
Figure 4.16 Modified levees in URS (2006) showing EBSBs along the MRGO
reduced to the 6.5 foot elevation of the underlying spoil disposal site, and the 40
Arpent levee crown raised to 17.5 feet (NAVD88).

4.3 St. Bernard Flooding Analysis by Team Louisiana (2007)

When Orleans Metro began flooding from Lake Pontchartrain through the drainage canal
breaches, the stage was set for a second disastrous inundation of St. Bernard east of Paris
Road. This area had already received some flooding from the west caused by the
breaches and overtopping of the IHNC floodwall in the Lower 9th Ward, and from the
south across the crown of the Caernarvon to Verret levee along the south side of the HPS
(Figure 4.17). The Caernarvon to Verret levee was overtopped briefly for about 1.5
miles where it was less than 13 ft high, and this happened when the water was highest at
that point, at about 0800. This overtopping was relatively brief, and the levee was
protected from waves by a vast wetland area to the south (Figure 4.17). Despite
overtopping, the levee crown experienced only minor damage (Morris expert report
2008), perhaps also due to a construction design that included a substantial clay cap
placed over a core of sand pumped from the Mississippi River (Figure 4.18).

61
Figure 4.17 Different materials and methods used for construction flood control
berms and levees around St. Bernard and the Lower 9th Ward (Ebersole
Presentation 2007)

Figure 4.18 Looking west on the Caernarvon to Verret Levee that protects the
south side of St. Bernard drained area east of Chalmette. This levee was overtopped
for about 1.5 miles where it was between 12.5 to 13.0 ft high, but was protected from
waves by an extensive marsh to the left. It suffered little damage.

62
Team Louisiana (2007) estimated that flooding from overtopping and breaching of the
IHNC into the Lower 9th Ward, together with the more limited overtopping of the
Caernarvon to Verret Levee, accounted for about 10 percent of the total flooding of the
protected area within the St. Bernard HPS. At around 08:30, the most substantial
flooding event in any part of the City began as the 40 Arpent back levee, with a crown
elevation of between 7 and 9 ft, experienced massive overtopping from the wetland
buffer area to the north (Figure 4.17). This was a very destructive overtopping event that
leveled substantial two-story homes in subdivisions in the Meraux area that backed up to
the 40 Arpent Canal (Figure 4.19). The water did not enter uniformly over the whole
levee but formed rivers that started at several low spots, were joined in the 40 Arpent
Canal that was just inside the levee, and raced preferentially through subdivisions and
other cleared areas as they flowed toward the Mississippi River levee. This flooding
could only have started after the vast wetland impounded between that levee and the
MRGO levee, nearly 32,000 acres, had filled to the elevation of the lowest reaches of the
40 Arpent Canal levee (about 6 feet).

Figure 4.19 View northwest toward 40 Arpent Canal and Levee in Meraux showing
homes destroyed down to the slab by flooding from the MRGO across the back
levee. The levee crown is visible in background.

The water that filled the impounded St. Bernard wetland and flowed over the 40 Arpent
Canal levee into the developed part of St. Bernard Parish came initially through breaches
and overtopping of the MRGO berm along MRGO Reach 2. Filling the impounded
wetland to a level of 7 ft required introduction of nearly 9 billion cubic feet of water
across the MRGO EBSB within a relatively narrow time window. That time window is
available from the ADCIRC hindcast of surge on the MRGO that has been corroborated

63
by gage data from the IHNC lock (Figure 4.7) and by the high water mark survey inside
the protected area.

The surge window on the MRGO can be defined by the time the water level exceeded a
specified elevation. If that elevation is defined as 6 feet, then the surge lasted 18 hours,
but if it is defined by the time above 10 ft, which is about the height any previous surge
had attained on the MRGO EBSBs, the period is reduced to 6 hours, starting at about
0500. Since the problem is to explain water levels in excess of 11 feet within the St.
Bernard developed area, we are particularly interested in the exceedence interval for this
value.

Water levels in MRGO and the 10 ft exceedence interval define the period during which
the St. Bernard EBSB breaches and flooding have taken place, as does the maximum
water elevation attained inside the protected communities, about 11 ft (NAVD88). If too
much water entered too early, the interior water level would have gotten higher than what
was observed, while too little inflow would not fill the protected area high enough.
Finally, we know that water did not enter the wetland buffer area uniformly along the
entire 12 mile length of the MRGO EBSBs. This structure, like the 40 Arpent Canal
levee, was not initially of uniform height but had a crown that ranged from 14 to 18 ft
(NAVD88). The maximum surge elevations without the wave component for these
EBSBs ranged from 16 to 18 ft, and so were comparable in most places to the pre-Katrina
crown elevation (Figure 4.21). Post-storm surveys disclosed that about half of the
MRGO EBSBs had experienced significant degradation or breaching. If they had not
failed and had been at the design grade, general overtopping should have lasted only for
about an hour prior to the surge peak at 0830, and this window would not have provided
enough water to explain the filling of the St. Bernard developed area.

Figure 4.20 IPET damage assessment on St. Bernard HPS showing 30 miles of
levees that experienced some overtopping, and 12 miles of breaches along the
MRGO (IPET 2006, V-14-15). LOB means levees overtopped and breached, and
LONB means levees overtopped but not breached. TF designates transition failures.

64
 Lake Pontchartrain and Vicinity Hurricane Protection Project
Chalmette System

Intracoastal
Waterway

Lake Borgne
Mississippi River Gulf
Outlet

Structure at Design
Grade
Structure Zero to 1’ Low
Structure 1’ to 2’ Low
Structure More Than 2’
Low

Pre-Katrina structure elevations shown with respect to design grade based on U.S. Army COE levee elevation data.
Structure elevations taken from COE Shape file entitled “Pre_Katrina_Levee_Floodwall_Max_Elev.shp”
Design Heights based on original authorizations (Corrected for MSL to NAVD difference)

Figure 4.21 Pre Katrina deviation from design elevation relative to mean sea level
for the Chalmette levee loop (source: Morris expert report 2008).

Team Louisiana (2007) adapted a simple flooding model from IPET to satisfy the
constraints discussed, while delivering the requisite water volume across the MRGO
EBSBs in the 6 hours available (IPET 2006, IV-240). The authors assume for this model
that front side wave breaching of EBSBs with poor foreshore protection (EBSBs closer to
channel, little protective vegetation) and overtopping of the lowest areas begins when the
surge attains 10 ft. This is the level at which IPET estimates that waves began breaking
on the levee face. This surge level occurred on the southern leg of the funnel at about
0500, at least an hour before landfall (Figure 4.22).

As the surge rose, the zone of wave-induced overtopping and structure degradation
expanded until the peak was attained at 0830. For the model, Team Louisiana (2007)
assumed, as Dr. Bea has subsequently proved (Bea expert report 2008) that the crown
relatively quickly degraded to its post-Katrina condition once overtopping by waves
began for any given reach. Flow over the levee was modeled using the IPET weir flow
equations, first for wave overtopping and then for overtopping by waves and surge.
Outflow from the inundated areas, both wetlands and developed zones, begins once the

65
MRGO drops below 10 ft at 1030. We also included overtopping from MRGO Reach 1,
the 7 mile reach that connects the funnel with the IHNC.

Best Katrina Surge Hydrographs for MRGO Traverse

20

18

16
Water Level (ft, NAVD88)

14
Shell Beach
12
MRGO B. Dupre
10 MRGO Halfw ay
MRGO B. Bienvenue
8
GIWW@Paris Rd.
6

0
8/25/05 8/26/05 8/27/05 8/28/05 8/29/05 8/30/05 8/31/05
0:00 0:00 0:00 0:00 0:00 0:00 0:00
Tim e

Figure 4.22 Surge hydrographs along MRGO Reach 2 from Shell Beach to Paris
Road, created by adjusting ADCIRC predicted peaks to match high water mark
data.

Team Louisiana (2007) results for overtopping flow were similar to those calculated by
IPET, if flow over the south Reach 1 levee is added (Figure 4.23). The rate of transfer of
water across the MRGO alignment peaks at around 0830 at more than 2.0 million cubic
feet per second (cfs). This is an extraordinary volume comparable to a large flood on the
Lower Mississippi River. But the volume of the 32,000 acre wetland buffer that must be
filled before the 40 Arpent Canal levee is overtopped is also very large, and the deep
MRGO channel provides an efficient conduit for supply of this water. Team Louisiana
(2007) predicted that that filling of the buffer would take nearly 2.5 hours, putting the
onset of overtopping of the back levee at 0830 in agreement with eye-witness accounts
(Figure 4.24).

66
 Comparison of IPET Estimate of MRGO Levee Overtopping Volume

18,000,000,000

Team La. Volume

16,000,000,000

14,000,000,000
Cumulative Volume of Overtopping (cf)

12,000,000,000

10,000,000,000

IPET Volume
8,000,000,000

6,000,000,000

4,000,000,000

2,000,000,000

0
3:36 4:48 6:00 7:12 8:24 9:36 10:48
Time (CDT) 29 August, 2005

Figure 4.23 IPET and Team Louisiana estimates of cumulative water volume
introduced by waves and surge across the degrading MRGO levee crown.

Flooding of St. Bernard from MRGO Required to Attain 10.5 ft in St. Bernard

20.00
MRGO Level
S. Bernard Water Level
18.00
Wetlands
16.00 Filling
Water Level (ft, NAVD88)

14.00

12.00

10.00

8.00

6.00

4.00
Wetlands Area Filled
2.00 Overtopping on 40 Arpent
Levee
0.00
0 1 2 3 4 5 6
Time Since Start of Breaching on MRGO Levee (hr)

Figure 4.24 Team Louisiana (2007) flood model prediction of flood hydrograph
inside St. Bernard polder compared to surge hydrograph at MRGO EBSBs.

67
If the onset of breaching were delayed until the MRGO surge elevation reached 12 ft,
then the spreading of levee degradation and failure would have to advance more quickly
to attain the required length, and make up the necessary discharge. Under no realistic
scenario, however, is the filling of the buffer wetlands an instantaneous event, but almost
certainly required hours. Similarly, given the degradation mechanisms available –
particularly the EBSB “capping” that involves a combination of front and backside wave
damage - it is unrealistic to assume that breaching of many miles of the MRGO berm
occurred at exactly the same time.

Figure 4.25 High-resolution oblique Pictometry image of MRGO levee east of

Bayou Dupre. MRGO is at bottom, and protected side at top. Photo shows several
stages of “levee capping” breaching that initiates with wave bench formation on
unprotected side, followed by wave overtopping and back side erosion, which
enlarges upstream through the levee crown to join the front side bench (Team
Louisiana 2007).

68
 CHAPTER 5
OUR METHODS

Following Hurricane Katrina, investigating engineers and scientists noted that levees and
floodwalls adjacent to the MRGO project had failed catastrophically while performance
was generally better elsewhere except on the New Orleans East Back Levee and the
newer Orleans Metro drainage canals (Team Louisiana 2007, ILIT 2006, IPET 2006,
2007). The USACE has consistently maintained that any linkage between the MRGO
project and failures of floodwalls and levees constructed under the LPV hurricane
protection project is entirely coincidental. To show that it was not, we have been forced
to imagine a landscape in which the 1958 MRGO project never existed.

This is a world in which the IHNC is connected to the Gulf solely by the Gulf
Intracoastal Waterway (GIWW), a 150 foot wide, 12 foot deep barge canal, instead of by
a 2,000 to 3,000 foot wide, 42 foot deep arm of the sea; where the main line of flood
defense for St. Bernard is a real levee built on a stable foundation along the 40 Arpent
Levee alignment two to three miles south of the MRGO channel of today behind
thousands of acres of the surge and wave protection afforded by cypress-tupelo swamp
forest and healthy marsh. Using computer models and historic maps and photographs,
we have brought that world to life in an effort to determine the incremental augmentation
of surge and wave attack that can be reliably attributed to the design, construction,
maintenance and operation of the MRGO navigation project, from the Mississippi River
lock on the IHNC to the Gulf of Mexico.

The only direct measurements of waves and surge during Hurricane Katrina in the
MRGO funnel outside of the flood protection system and even within the various reaches
of the MRGO and IHNC boil down to (1) a few dozen high-water marks surveyed after
the storm, (2) hourly readings made by the USACE lockmaster on a staff gage at the
IHNC lock, (3) partial records from USGS gages at Paris Road and on the IHNC at the I-
10 crossing, (4) video and still camera footage from the Entergy Plant at the north end of
the Paris Road I-510 bridge, and (5) records from two small wave buoys installed near
the south shore of Lake Pontchartrain. This is a meager dataset, even compared with
what was available after Hurricane Betsy in 1965, because the damage to electronic gages
was so complete.

5.1 Direct Observations of Surge and Waves

The high water marks are not unequivocal because they always document surge with
some added unknowable surplus due to waves. The amount of wave surplus depends on
where the mark is located and particularly on how well protected it is from waves.
Through my work with the LSU Hurricane Center, I was personally involved in the
identification of many of these marks and understand their limitations as I detailed in my
second expert report (Kemp 702c Immunity 2007). The lockmaster recorded stage at
hourly intervals, meaning that he may well not have captured the peak, and would
certainly have missed any more subtle changes as the floodwalls and levees on each side
of the IHNC gave way. Furthermore, neither of the partial gage records available from

69
the USGS captures the peak of the surge. Finally, information from pictures at Paris
Road has proven difficult to calibrate because of the single perspective available. The
photos give an idea of conditions at only one point where the levee did not break. As a
consequence, no one – not LSU, USACE, IPET, the URS, Inc. consultants, or any other
entity – actually knows with certainty the surge and wave conditions that occurred at each
critical point where the flood protection system failed.

5.2 Computer Models Used to Reconstruct Surge and Waves

Like all of the other technical teams that have studied Katrina hydrodynamics, we have
been forced to rely heavily on computer models to reconstruct the surge and wave
dynamics that were in play during the passage of this storm. We have used five in the
course of our work on this case that are widely used by oceanographers, hydrologists and
hydrologic engineers around the world, and have tabulated them with a brief overview of
capabilities, acronyms and references (Table 5.1). We used the S08 version of ADCIRC
that has been fully described in earlier reports to study some features (Kemp 2007a, b).
For this report we used ADCIRC primarily to provide boundary conditions for the FINEL
model developed by SVASEK (Figure 5.1).

The FINEL model was set up with far more detail in the bathymetry and topography than
has been available to date in any ADCIRC version in part to better study the effects of
bottom friction differences between open water and various wetland types (de Wit et al.
2008). Bays and channels tend to provide less resistance to flow than do marshes and
particularly swamp forest wetlands. FINEL gave our team the flexibility necessary to
examine incrementally the effects of adding, removing or modifying channels, wetlands
and flood protection elements within the landscape to gain a greater understanding of
how each contributed to surge dynamics (Figure 5.1). It should be noted that because of
time constraints, the FINEL model domain does not include the New Orleans East and
Orleans Metro polders, and therefore does not directly simulate overtopping into these
regions (de Wit et al. 2008). While it faithfully replicates surveyed overbank, floodwall
and levee elevations as they existed prior to Katrina along the north bank of
GIWW/MRGO Reach 1 and the IHNC, it also does not consider the effects of breaches
that occurred during that storm in the southern portion of the IHNC between the MRGO
junction and the lock. This means that it is expected to over-predict surge elevations
actually observed in the southern part of the IHNC, and, consequently, overtopping into
the Lower 9th Ward and Central Wetlands just to the north. These limitations have been
addressed to some degree by added analyses of overtopping potential using a 1D weir
model for the polders not explicitly included in the model domain (de Wit et al. 2008).

SWAN is the most widely used wave model today. It was developed by our colleagues
in the Netherlands to look at wave generation and dissipation in nearshore, inshore and
inter-tidal areas like the MRGO funnel (Figure 5.2). This is a finite-difference model
built, unlike ADCIRC or FINEL, of equal-sized rectilinear cells, that achieves greater
resolution through the use of nested grids, in which larger grids provide boundary
conditions for smaller ones. It is very similar to the WAM that was used by IPET to
generate offshore wave conditions. It was not necessary to rerun the WAM wave

70
simulator because we had confidence that the WAM output from offshore that was
generated by the USACE IPET could be used to set boundary conditions for the largest of
the nested grids of the SWAN wave model (Figure 5.2).

We applied SWAN to study the Katrina storm scenarios with and without the MRGO
project, and with and without the marshes and swamps that it destroyed (Gautier et al.
2008). As is true for the steady and unsteady flows that contribute to surge, vegetation
type profoundly affects the rate of wave energy dissipation that ultimately determines the
size and character of the waves that attack man-made structures like the EBSBs. The
SWAN model is well set up to investigate such interactions, including the unique
attributes of a swamp forest canopy to separate the winds that build waves from the water
surface, leading to a more rapid loss of wave height and energy than can be attributed to
bottom friction effects (Gautier et al. 2008). The SOBEK flood model was first described
when it was used as part of the class certification process (Kok et al. 2007), and is used
again here to study the severity of flooding that would result from what Dr. Bea has
called an “MRGO Neutral” condition, as will be described shortly.
Table 5.1 Computer Models Used to Reconstruct Hurricane Katrina Surge and
Waves
Name Acronym Purpose Reference
ADvanced CIRCulation ADCIRC 2D Westerink, J.J., Leuttich, R.A., Jr.,
Vertically and Muccino, J. 1994. Modeling
Averaged tides in the Western North
Coastal Atlantic using unstructured graded
Circulation grids. Tellus 46a(2): 178-199.
FINite ELement 2D FINEL2D 2D Flow
and
Transport in
Rivers and
Coastal
Waters
Simulating WAves SWAN Random, Holthuijsen, L.H., Booij, N. and
Nearshore Short- T.H.C. Herbers, 1989: A
Crested prediction model for stationary,
Wind- short-crested waves in shallow
Generated water with ambient currents,
Waves in Coastal Engineering, 13, 23-54.
Coastal
Waters
WAve prediction Model WAM directional WAMDIG 1988: The WAM
spectra as model - A third generation ocean
well as wave prediction model. Journal of
wave Physical Oceanography, 18,
properties 1775-1810.
offshore
SOBEK Flood
forecasting

71
 GIWW

MRGO reach 1 /
IHNC GIWW

Lake
Borgne
MRGO
reach 2

Figure 5.1. FINEL model computational domain included area outlined in red, and
was extended for scenario 2B to include the entire St. Bernard polder from the 40
Arpent levee to the Mississippi River indicated by dashed red line (See Table 5.2).

ADCIRC
Through our work with the ADvanced CIRCulation model (ADCIRC) since 2003 and
particularly during the forecasting period prior to Katrina’s landfall, my colleagues and I
at the LSU Hurricane Center had perhaps the best overall perspective on the surge
oceanography at the time of the storm. We had worked closely with Dr. Joannes
Westerink, one of the inventors of ADCIRC, and his students to improve the fidelity of
the way land and water features were represented in the model, and to improve the speed
with which forecasts could be run on the LSU supercomputer (Kemp 2007). The S08
version of ADCIRC that we used at that time, and that the LSU Hurricane Center

FINEL
The numerical model FINEL generates detailed information about water motion in the
area of study and is similar to ADCIRC in many ways in that it is also a two-dimensional,
depth averaged hydrodynamic model running on an unstructured grid using the finite
element method (de Wit, et al. expert report 2008).

The FINEL numerical model was developed by Svašek Hydraulics and the Fluid
Mechanics section of the Civil Engineering faculty of the Delft University of Technology
(TUDelft) in the Netherlands. The model has been used for hydrodynamic simulations

72
all over the world. The validity and calibration of the model has been extensively tested
on projects in the Western Scheldt estuary, the North Sea (the Netherlands), Dublin
flooding (North Ireland), Hazira (India) and Posorja (Equador). FINEL is based on the
depth averaged equations for 2-dimensional horizontal flow, the equations of Saint
Venant. In FINEL, this system of equations is solved with a robust numerical finite
element method, which can easily handle drying and flooding. The model area in FINEL
is divided into a number of triangles, the so-called elements, instead of a grid of
rectilinear cells. The size and distribution of the triangles can vary over the grid, so
complicated geometries are easily represented in the schematization and the grid can be
more detailed in areas of specific interest. This is very important in the complex area
around New Orleans, with channels that require a lot of detail adjacent to higher ground
and lake bottoms which require less.

Figure 5.2 SWAN Nested Grids A-G positioned over bathymetry and topography.
Boundary conditions for largest grid supplied by WAM output.

Our present study is focused on the funnel area that includes a portion of Lake Borgne,
the GIWW, MRGO Reaches 1 and 2, and the IHNC (Figure 5.1). A series of scenario
runs were executed are described in more detail below that incorporate either the present
topography, bathymetry and degraded vegetation cover, or one that we have
reconstructed for the time before the MRGO was built with the 1956 healthy wetlands
cover no MRGO situation (No reach 2, pre-MRGO GIWW/Reach 1, no levees along
MRGO).

73
SWAN
The SWAN wave model (Version 40.51AB) is used more than any other around the
world to simulate waves, particularly in coastal and estuarine settings. It is known as a
third-generation wave model and represents the current state of the art for wave
simulation. Our colleagues at Svasek and Delft University of Technology (TUDelft)
were involved in its development and were particularly qualified to simulate the wave
climate along MRGO Reach 2 during Katrina. SWAN uses a grid to compute how waves
generate, dissipate and propagate under the influence of wind, bathymetry, vegetation
type, water levels, currents and incoming waves. See Gautier et al. (2008) for more
detailed information on set-up and scenarios. The Robinson expert team provided the
input data that was converted into a SWAN compatible format. Output is given for a
number of pre described output locations in the area of interest. For this study, SWAN
version 40.51AB is used in the stationary mode. Winds and surge vary slowly enough
for the waves to reach equilibrium. We believe that this assumption – one also made by
IPET (IPET 2007, IV-4-31) is reasonable, therefore the steady-state solution was
considered adequate for the simulations. Nested grids are used to increase resolution in
the areas of interest, with the smallest, highest resolution grid having made up of square
cells, about 70 feet on a side (Figure 5.3).

The ADCIRC S08 version that the LSU Hurricane Center used to produce such accurate
predictions of the Katrina surge as Katrina approached, provided an excellent hindcast of
the Katrina event that has been little improved upon since (Team Louisiana 2007). This
well-tested and stable version is adequate for the work described here, namely providing
boundary conditions for FINEL, and exploring the effect of constricting the
GIWW/MRGO Reach 1 channel cross-section on surge height and duration in the IHNC.
ADCIRC has been the primary hurricane surge model used by the USACE during IPET,
and by the Federal Emergency Management Agency (FEMA) to set 100-year surge levels
for the Federal Flood Insurance program since Hurricane Katrina. I provided a detailed
description of the ADCIRC S08 version that we have used for this work in my first expert
report (Kemp Class Certification Report 2007).

The USACE has funded development of several versions of ADCIRC since Katrina that
have increased the fidelity of the geometry and intensified the grid in important areas.
They have also experimented with the use of spatially varying roughness parameters to
try to capture the retarding effects of marshes, swamps and other inland features, which
were not well handled in the S08 version. We did the same thing using the SVASEK-
TUDelft FINEL model, which has essentially the same capabilities within the more
restricted area of interest for this work.

SOBEK
SOBEK-1D2D was developed by WL-Delft Hydraulics and is used worldwide for flood
simulations and specifically is the standard flood model used in the Netherlands.
Flooding is simulated with a combination of one-dimensional (1D) and two-dimensional
(2D) elements. The flooding itself is modeled as a 2D process which allows for a
continuous simulation. This model was used to simulate flooding in the Orleans Metro,

74
New Orleans East, Lower 9th Ward and St. Bernard polders, and is well described in an
earlier report submitted in July, 2007 (Kok et al. 2007). We have since run additional
SOBEK simulations on each of the affected polders for what Dr. Bea has described as the
“neutral MRGO” scenario that is described in more detail below.

5.3 Analytical Sequence

A merger of ADCIRC output and high water mark data was used to provide boundary
conditions for FINEL, as we sought a higher level of accuracy than could be obtained
from any of the ADCIRC versions now in use. The IHNC channel, for example, is very
complex, with numerous constrictions and expansions associated with bridge crossings
and barge slips, as well as a variety of overbank, levee and floodwall configurations that
are not captured in ADCIRC. We continued to use the Planetary Boundary Layer (PBL)
wind generator linked to ADCIRC that had provided such a good basis for Katrina surge
prediction, given that IPET gained little improvement in surge fidelity for the MRGO
funnel area by using other wind reconstructions developed after Katrina. So, the PBL
model provided input that was processed on the LSU supercomputer to yield 10-minute
surge output for Katrina.

Hydrographs were produced at key locations, including three that were necessary to
provide boundary conditions for FINEL (de Wit et al. 2008). The surge hydrographs at
these boundary points were adjusted by the smallest amount necessary to match high
water mark measurements without causing numerical instability. FINEL was then
calibrated to provide benchmark surge hydrographs at specific read-out locations (Figure
5.3). For subsequent scenarios, ADCIRC was run first with the same modifications
proposed for the more detailed FINEL run, following a schema that Dr. Bea and I worked
out with our Dutch colleagues (Table 5.2). Resulting hydrographs were compared to
Katrina base case ADCIRC hydrographs, and those at the FINEL boundaries were
adjusted for elevation and time to incorporate ADCIRC changes predicted at the
boundaries. FINEL was used to generate a surge surface at hourly intervals beginning at
0400 on the morning of August 29, 2005. SWAN was then run over the surge surface
generated by FINEL to provide hourly snapshots of wave characteristics.

Like ADCIRC, FINEL was used only to simulate overtopping, and not breaching, or flow
through breaches. The task of reproducing the flooding, if any, in the developed areas of
New Orleans East, the Lower 9th Ward and St. Bernard Parish with and without MRGO-
induced breaching and overtopping was accomplished with the SOBEK model (Kok et al.
2007, 2008). Such simulations were discussed in my first expert report for the “Katrina
As-Is” scenario. The SOBEK Katrina flood simulation praised by Dr. Dalrymple in his
deposition (Dalrymple Deposition) is compared here with other scenarios in which
breaching and overtopping is started at different times determined by the combination of
the DYNA and Miner’s Law cumulative fatigue models employed by Dr. Bea and Rune
Storesund for the wall structures and turf covered EBSBs that failed during Katrina. The
models used at Berkeley allow estimation of when these structures will fail, whether by
under-seepage, front-side wave erosion prior to overtopping, or overtopping induced
erosion, for what Dr. Bea has called the “MRGO Neutral,” and the “Katrina As-Is”

75
conditions. The FINEL, SWAN and SOBEK models were run in The Netherlands under
my direction by Svasek Hydraulics, a consulting company linked to the Delft University
of Technology (TUDelft).

Figure 5.3 Locations for FINEL time-series output of water level, current direction
and velocity.

We have been asked to restrict ourselves to documenting the effects of MRGO project
elements on the severity of flooding experienced by the Robinson plaintiffs.
Accordingly, we have modified the MRGO funnel landscape within the models to run
scenarios that range from “Katrina-As-Is” (Scenario 1) to “pre-1958 no MRGO”
(Scenario 2D), with a number of additional scenarios that include components of both
end member conditions (Table 5.2). We describe below how various man-made and
natural components of the Lake Borgne funnel were modified for each scenario.

76
Table 5.2 Katrina Simulation Scenarios for characterization of MR-GO Reach 1
and Reach 2 surge, current, and wave characteristics

Scenario MRGO MRGO/GIWW Reach 2 Reach 1 40 Vegetation

Reach 2 Reach 1 EBSBs Levees Arpent
Channel channel Levee
1 Existing Existing 8/2005 Existing Existing Existing Existing
8/2005 8/2005 8/2005 8/2005 8/2005
2A None Pre MRGO None None LPV Pre MRGO
authorized 1958
& no flow
past
2B None Pre MRGO None None Existing Pre MRGO
8/2005 & 1958
flow to
Miss.
River
2C None Pre MRGO Existing Existing Existing Pre
Neutral 8/2005 8/2005 8/2005 MRGO
MR-GO 1958
2D None Pre MRGO Existing None Existing Pre MRGO
8/1965 8/2005 1958
3 As As authorized Existing Existing Existing Pre MRGO
authorized 8/2005 8/2005 8/2005 1958

77
 Reach 1

Borgne Funnel

Reach 2
40
&
Arpent
EBSBs
Levee

Figure 5.4 MRGO Reaches 1 and 2 with important LPV and other flood protection
structures modeled.

MRGO Reach 2 Channel

This portion of the MRGO project was modeled in three configurations (Table 5.2). The
greatly expanded channel 2,000 to 3,000 feet wide and more than 40 feet deep that
existed in front of the Reach 2 EBSBs before Katrina -- and persists today despite all the
talk of remediation – occurs only in Scenario 1, the Katrina As-Is Scenario (Figure 5.4).
This MRGO channel reach is completely absent in all scenario 2 runs (2A, 2B, 2C, 2D).
It reappears in Scenario 3 at its much diminished “authorized” width and depth (650 feet
by 36 feet).

MRGO Reach 1 Channel

This portion of the MRGO project was modeled in three configurations (Table 5.2). The
1,000 foot wide channel that provided the outlet from the funnel to the IHNC during
Katrina occurs only in Scenario 1, the Katrina As-Is Scenario (Figure 5.4). This MRGO
channel reach is reduced to the pre-MRGO dimensions of the antecedent GIWW barge
channel in all Scenario 2 runs (2A, 2B, 2C, 2D), about ten percent of its current cross-
section. It reappears in Scenario 3 at its “authorized” width and depth (650 feet by 36
feet), a still considerable, but smaller channel than exists in this location today.

Reach 2 EBSBs
Bretschneider and Collins (1966) discussed in the NESCO report how placing barriers
parallel to the GIWW and MRGO channels that define the Lake Borgne funnel affects the
time to arrival of surge and its elevation for storms tracking at different speeds. We have
tried to gain a greater understanding of the synergistic relationship between the pile-up of
78
water on the flood defense structures and flows in the adjacent channels. Accordingly,
the EBSB structures that were periodically raised and allowed to sink along the south side
of MRGO Reach 2 prior to Hurricane Katrina are placed in the models just as they were
prior to Katrina in three of the five simulations, namely, scenarios 1, 2C (MRGO Neutral)
and 3 (Table 5.2). They are completely absent and reduced to marsh elevation in
scenarios 2A and 2B. For scenario 2D, only the spoil banks are present as they were in
1965 during Betsy but prior to construction of the LPV EBSBs. This last is true even
though the channel itself is not present in this scenario (Figure 5.5).

Reach 1 Levees
True levees, not EBSBs, were built along most of both sides of the MRGO/GIWW Reach
1 channel as part of the LPV project. Dr. Bea has identified a few notable exceptions to
this statement where floodwalls and transition elements were involved (Bea Report, July
11, 2008). All of these structures are present as they were during Katrina in scenarios 1,
2C (MRGO Neutral) and 3 (Table 5.2). They are not present at all in scenarios 2A, 2B
and 2D, so that the throat area of the funnel was considerably less constricted by levees.

40 Arpent Levee
This is a local levee that predates both the MRGO and LPV projects. It is currently used
to facilitate pumped drainage from the Lower 9th Ward and the St. Bernard polder
(Figure 5.4). The federal LPV flood protection works would have been constructed
along this alignment if the Corps had not chosen to place them in the spoil disposal area
created by the dredging of the MRGO. This levee had a crown elevation at the time of
Katrina averaging +6.5 feet (NAVD88). This levee is included in the models in two
configurations, namely, the pre-Katrina condition (scenarios 1, 2B, 2C, 2D and 3), and as
a hypothetical alternative in which it was raised to 17.5 feet (scenario 2A). For scenario
2B, the FINEL model domain was extended all the way to the Mississippi River (Figure
5.1). This is the only scenario for which this was done.

Vegetation
We are finding that vegetation effects on both surge and waves are not generally
understood as well by American coastal engineers as they are in The Netherlands. It is
well known that storm driven water piles up against elevated flood protection structures,
so that the highest surge for miles around is often experienced on the slope or crest of a
levee. The countervailing effect of even low-lying marsh vegetation to retard and
attenuate storm surge has also been known for as long as the MRGO has existed and was,
in fact, one criterion for the original Chalmette LPV design (USACE 1963; USACE
1967). At the time Katrina struck, however, U.S. engineers had not yet reduced these
empirical observations to reliable physics algorithms for numerical storm surge and wave
models. More progress had been made by researchers in the United Kingdom and the
Netherlands, as will be discussed later (Moller and Spencer 2002).

One of the major adverse impacts of the MRGO project was to change wetlands to open
water and swamp forests to marsh (Day 2008; Fitzgerald 2008). Accordingly, it has been
very important to incorporate the effects of these MRGO-induced changes on surge and
wave dynamics experienced during Hurricane Katrina, and then to assess whether the

79
presence or absence of swamp and marsh affected the timing and severity of flooding of
populated areas. We modeled vegetation in two ways, namely in the existing pre-Katrina
configuration (scenario 1), and in the 1958 pre-MRGO condition (scenarios 2A, 2B, 2B,
2C, 2D and 3). After careful investigation of the stem densities of cypress forests, for
example, Gautier et al. (Appendix, 2008) selected the range of Manning’s coefficients
given below (Table 5.3) to parameterize the frictional effects of vegetation types that
occur today in the Lake Borgne funnel (Figure 5.5a), and that were found in this area
prior to the construction of the MRGO project (Figure 5.5b).

Table 5.3. Manning’s coefficients selected to characterize bottom friction in the

FINEL model for the range of vegetation types found in the Lake Borgne funnel
Area Manning coefficient n
Open water 0.02
Marsh along Lake Borgne 0.05
Normal Cypress forest 0.1
Dense Cypress forest 0.12
Commercial area next to IHNC 0.07
Low residential area between 40Arpent and Mississippi 0.07

Other members of the Robinson expert team have confirmed that cypress-tupelo swamp
forests occurred throughout the Central Wetlands area between the current MRGO
alignment and the 40 Arpent Levee prior to construction of the MRGO project in 1958,
and, in particular, formed a substantial barrier to surge intrusion from Lake Borgne into
the IHNC (Figure 5.5b). The conversion of the GIWW barge canal into the MRGO
Reach 1 ship channel greatly increased the potential for surge conveyance into the IHNC.

Bottom friction effects are addressed somewhat differently in the SWAN wave model as
a result of research conducted in salt marshes primarily at the University of Cambridge in
collaboration with TUDelft (Moller and Spencer 2002). This becomes particularly
important as waves approach a structure like the EBSBs along MRGO Reach 2. The
approach used is a variant of that developed by Collins (1972), therefore the resultant
value is the “Collins Friction Coefficient”, Cfveg.

Cfveg = Cfr * d * n * hveg

Where Cfr is a measure of stem rigidity, d is the plant diameter in meters, n is the
number of plants per m2, and hveg is the stem height of the vegetation in meters. Gautier
et al. (2008) developed three different values for the friction coefficient
- Cypress forest Cfr= 0.9 * 0.5 * 0.9 * 5 = 2.0
- Marsh Cfr= 0.5 * 0.01 * 49 * 2 = 0.49
- Water (Lake Borgne, MRGO channel) Cfr= 0.015 (SWAN default value)

80
Figure 5.5a. Spatial distribution of Manning’s coefficients in the pre-Katrina
FINEL model domain

Figure 5.5b. Spatial distribution of Manning’s coefficients in the pre-MRGO

FINEL model domain

81
Because the vegetation that would have occurred on the unprotected, outboard side of the
EBSBs between the MRGO channel and the EBSBs was neither marsh nor cypress forest,
Dr. Bea has selected an intermediate Cfveg value of 0.98 to be used in SWAN for this zone
(Bea Declaration, July 11, 2008). Gautier et al. (2008) conducted a sensitivity analysis of
the error that might be introduced by choosing an inappropriate value for this coefficient
that should be considered in assessing the SWAN output at the EBSB toe that served as
input to Dr. Bea’s erosion model (Bea Declaration, July 11, 2008). Furthermore, the
heavily vegetated swath separating the MRGO Reach 2 channel bank and the toe of the
Chalmette Loop EBSBs decreased by an average of 300 feet between the time the
channel was completed and arrival of Hurricane Katrina (Morris 2008). The loss of these
shrubs and small trees proved to quite significantly affect wave energy transmission to
the LPV structures (Bea expert report 2008). An average loss of more than 700 feet of
the marsh buffer between Lake Borgne and the north bank of MRGO Reach 2 had a
similar, but somewhat lower impact on incident wave energy, as will be discussed
(Morris 2008)
The SWAN model allows friction to be defined on the basis of the Collins’ friction coefficient
5.3 Calibration and validation of FINEL and SWAN

Output from FINEL for the Katrina calibration dataset along along the MRGO Reach 1
and 2 and the GIWW mirror those derived from adjusting ADCIRC hydrographs to
match available peak high water mark data (Figures 5.6a, 5.6b). I have described the
collaborative acquisition of high water mark data from the area affected by the Lake
Borgne funnel in an earlier report (Kemp 2007, Immunity Report). The Root Mean
Square (RMS) difference between adjusted ADCIRC and FINEL for the entire time-
series is on the order of 0.6 feet in the Lake Borgne funnel, while the difference in peak
water level is less than 1 foot. The difference between FINEL and adjusted ADCIRC is a
little larger (RMS average 0.8 ft) in the IHNC, but this was expected for reasons that have
been described above. In addition, we created the FINEL model to develop more fidelity
in the bathymetry and roughness of the IHNC channel. The ADCIRC S08 model does not
account for the narrowing of the flow profile under the bridges in the IHNC, while in
FINEL the surveyed depths of the IHNC are used which show narrowing at the bridges.
In FINEL the harbor areas next to the IHNC are also given a higher roughness to account
for the obstruction of the flow, while in ADCIRC a uniform roughness was used.
Second, neither ADCIRC nor FINEL, as they are presently configured can address local
reductions in the surge peak caused by the failures of more than a 1,000 feet of floodwall
near the southern terminus of the IHNC. So, both of these models over-predict peak
surge height by 2 to 3 feet in this area. Given these complications, the agreement in peak
water level in the IHNC is reasonable. While a more satisfactory validation is to be
desired, the calibration results give confidence in the FINEL flow results.

Calibration and validation of the surge model is problematic given the lack of reliable
gage records, and the variability in observed high water mark data. But evaluating the
output of wave models like SWAN or the STWAVE model used by the USACE is even
more difficult given that no direct measurements of waves exist in the MRGO funnel
area. The only direct observations of waves generated during Katrina come from two
small wave buoys that were deployed in 12 feet of water near the south shore of Lake
Pontchartrain (Figure 5.7). These buoys recorded peak significant wave heights (Hs) of

82
about 9 feet with a peak wave period (Tp) of 6.5 seconds, for maximum wind speeds
estimated at between 60 and 70 miles per hour (IPET 2007). The USACE was able to
reproduce these results using their STWAVE model run in a full-plane mode to capture
waves coming from all directions (Figure 5.8).

The situation in southern Lake Borgne is expected to be similar to that in Lake

Pontchartrain, as both lakes have about the same depth and length along the critical
direction from which the hurricane winds blew, typically called “fetch,” when the storm
was at its most powerful stages in this area. Lake Borgne is just slightly offset to the
south relative to Lake Pontchartrain. Because it is situated closer to the track of the storm
center, wind speeds in Lake Borgne are presumed to have been somewhat higher than in
Lake Pontchartrain and were modeled as such using the PBL model used with ADCIRC.
The critical fetch for Lake Borgne was from the northeast (Figure 5.9), rather than from
the north and northwest as in Pontchartrain (Figure 5.8). Nonetheless, SWAN results for
Katrina at peak surge and wave development predict a maximum significant wave height
at 0800 CDT that is between 9 and 10 feet in southern Lake Borgne (Figure 5.9), quite
comparable to what was recorded by the buoys in Lake Pontchartrain (Figure 5.7).

SWAN predicts that not only were the maximum significant wave heights at peak surge
in southern Lake Borgne comparable to those measured in Lake Pontchartrain, but the
peak wave period was also almost identical, between 6 and 7 seconds (Figure 5.10).
Such waves are local seas that have been generated within Lake Borgne, rather than long-
period swell waves propagated inland from beyond the islands offshore. As with FINEL,
the SWAN results compare favorably to the only direct measurements obtained during
Katrina, even if they are from an adjacent lake. This leads us to believe that the SWAN
results should be reliable as a basis for further analysis of the effect of the GIWW and
MRGO Reach 2 channels on the waves striking the EBSBs along both sides of the Lake
Borgne funnel.

It should be pointed out here that the USACE STWAVE simulations for the MRGO
funnel area produced results that are quite different from those generated by SWAN
(Figure 5.11). The USACE STWAVE output indicates that the peak significant wave
height in southern Lake Borgne was about 5 feet, with a peak period throughout the
funnel of as long as 16 seconds. These would be classified as residual offshore swells
originally generated by the storm while far offshore, rather than locally within Lake
Borgne. These smaller amplitude, long-period waves are virtually absent in the SWAN
simulations. One possible explanation is that the STWAVE results were produced using
a “half-plane” version of the model which, depending on grid orientation, could have
missed the larger amplitude, shorter-period local seas developed within the grid in Lake
Borgne (IPET, 2007, IV-4-5).

We are convinced that the waves that developed in southern Lake Borgne were, in fact,
similar to those that we know developed in Lake Pontchartrain. They were large, up to a
significant wave height of 10 feet, but they were locally generated seas, rather than
residual offshore swell, with a maximum period that is half of the 12 to 16 second
dominant period that the USACE postulates in IPET (2007), based on STWAVE (Figure

83
Figure 5.6a. FINEL model results (blue) from Lake Borgne funnel compared to
reconstruction of actual hydrographs during Hurricane Katrina (red). See Figure
5.3 for locations.

84
Figure 5.6b. FINEL model results (blue) from GIWW/MRGO Reach 1 and IHNC
compared to reconstruction of actual hydrographs during Hurricane Katrina (red).
See Figure 5.3 for locations.

85
Figure 5.7 Significant wave height (top) and peak wave period (bottom) recorded
by two wave buoys near the south shore of Lake Pontchartrain during Hurricane
Katrina (IPET 2007, IV-4-13, 14).

86
Figure 5.8 Maximum significant wave height (top) and peak wave period (bottom)
during Katrina as modeled by the USACE for Lake Pontchartrain using STWAVE
in full-plane mode (IPET 2007, IV-4-11, 12) closely matches SWAN results for
southern Lake Borgne.

87
Figure 5.9 SWAN prediction of significant wave height distribution for Katrina in
southern Lake Borgne at peak development, showing 9 to 10 foot seas, similar to
waves measured in southern Lake Pontchartrain.

5.11). An erroneous doubling of the incident wave period is not trivial because long-
period swell waves tend to drive run-up to higher elevations on beaches and sloped levee
faces for a given energy or prevailing wave condition than do shorter period seas. In
contrast, shorter-period chop or seas of the same amplitude transport less water up slope,
therefore producing less potential for overtopping, but do more erosive work on the
outboard, unprotected face of a sloping dyke because the number of breaking wave cycles
is doubled. We believe that this led to a biasing of subsequent IPET analyses of the type
of erosion that occurred on the EBSBs along MRGO Reach 2 and New Orleans East
Back Levees toward a conclusion favoring almost exclusively back side overtopping
erosion

88
Figure 5.10 SWAN generated snapshots showing the distribution of peak wave
period in southern Lake Borgne during Katrina. The 0800 CDT panel in the
bottom left shows a peak wave period in southern Lake Borgne of between 6 and 7
seconds (Gautier et al. 2008).

89
Figure 5.11 USACE estimates of maximum significant wave height (top) and period
(bottom) obtained using STWAVE model in half-plane mode show smaller waves
with longer periods at peak development than predicted by SWAN or measured in
Lake Pontchartrain (IPET 2007, IV-4-16, 17).

90
The STWAVE results suggest less potential for front side damage to structures like the
MRGO Reach 2 EBSBs than do the SWAN results. As Dr. Bea has shown using the
more plausible SWAN results, significant reaches of these structures were damaged early
by front-side wave attack (Bea Expert Report 2008).

5.3 Summary of modeling program

Our approach was to run ADCIRC in the U.S. followed by FINEL and finally SWAN in
the Netherlands for the Katrina event to ensure calibration, and then to modify the
geometry in steps to separate out the role of different elements of the MRGO project in
Katrina surge and wave dynamics. The sequence of landscape modifications was
designed to allow us to differentiate the effects on surge elevation, duration and waves of
separable portions of the Lake Borgne funnel created by the MRGO project.

Construction of the IHNC and GIWW barge canal predate the USACE actions being
investigated here, as did the initial construction of the 40 Arpent Levee separating the
Central Wetlands from the portion of St. Bernard Parish then coming under forced
drainage. A more substantial privately constructed levee already followed the north bank
of the GIWW, providing some protection to New Orleans East before Hurricane Betsy.

The pre-existing channels and levees, as well as the natural levees of Bayou LaLoutre
created a large embayment including large expanses of marsh. Much of the western
portion of this embayment in the area of what became GIWW/MRGO Reach 1, however,
was a heavily forested swamp, and the outlet GIWW channel through the trees was small
and little deeper than existing natural bayous in the adjacent wetlands (Figure 5.12). A
densely forested swamp also occurred in the southern part of the Central Wetlands
forming a buffer for the 40 Arpent Levee, though it appears from other information that a
less dense cypress forest existed north of the boundaries shown by Penland et al. Expert
Report (2008).

The amplification of hurricane-driven oceanographic forces that occurs in the Lake

Borgne funnel results not from a simple addition of influences attributable to individual
MRGO channel reaches and LPV structures, but also from synergistic interactions
between them that are a direct result of the MRGO project. As we have discussed, this
synergy was first predicted by Bretschneider and Collins (1966) in the NESCO report
prior to construction of the LPV floodwalls, levees and EBSBs for the imprudent
juxtaposition of channels with the flood protection alignment ultimately selected. One of
the best illustrations of this interaction from south of Lake Borgne in the stretch of the
MRGO Reach 2 channel between Bayous Dupre and Bienvenue. There, expansion of the
top-width of the MRGO channel from 650 feet to more than 2,000 feet allows for
increased regeneration and growth of the waves crossing from Lake Borgne to break on
the adjacent EBSBs. Water piles up on the EBSBs, in turn, locally driving up surge
elevation in the adjacent channel (See Figure 4.12). This super-elevation of the water
surface above the Reach 2 channel creates a gradient that increases the velocity of flow in
that channel toward the throat of the funnel. Finally, a greater discharge earlier from

91
MRGO Reach 1 into the IHNC leads to an earlier initiation of overtopping of floodwalls
along both sides of the IHNC.

Our modeling program was developed to allow quantification of both the individual and
interactive effects of (1) the expanded channels to conduct surge and waves, (2) the
effects of the LPV flood protection structures to drive up surge and break waves, and (3)
the effects of various types of marshes and swamps to attenuate surge and dissipate wave
energy. This was accomplished with both FINEL and SWAN. Only then could we
identify the interactive effect of, for example, allowing the MRGO Reach 2 channel to
expand to the toe of the Chalmette EBSBs, or the loss of the swamp.

We also conducted a series of experiments using ADCIRC to quantify incrementally the

effect of constricting the GIWW/MRGO Reach 1 from its pre-Katrina cross-section down
to its pre-MRGO condition for the Katrina base case. The pre-MRGO end-member of
this analysis was later confirmed with a series of FINEL runs (2a, 2b, 2c, 2d). Finally,
using SOBEK, we were able to examine the relative severity of Katrina flooding if the
MRGO project had been constructed and operated in a more prudent manner that did not
expose the LPV flood control structures to elevated surge and wave forces.

We have used hydrodynamic modeling packages that are, we believe, the best of those
commonly used in modern engineering practice. But some of the analyses by
Bretschneider and Collins (1966) should have led an alert oceanographer or coastal
engineer to many of the same conclusions over 40 years ago following Hurricane Betsy.
Evaluating the effect of enlarging the GIWW/MRGO Reach 1 cross-section that IPET
has called “critical” to surge transmission could have been accomplished with HEC-2, a
ancient 1D model developed by the USACE that engineers have used since the late
1960s. Similarly, wave regeneration in deep channels is by no means a new concept, and
should have particularly aroused concern as the MRGO channel expanded into Lake
Borgne and toward the toe of the Chalmette Loop EBSBs. Analytical models for wave
hindcasting were a particular specialty of Dr. Bretschneider, the lead author of the 1966
NESCO report, but he was never asked to look at waves.

Neutral MRGO analyses

Special attention is directed to scenario 2C in FINEL (de Wit et al., appendices 2008) and
2Ci in SWAN (Gautier et al Appendix 2008), a configuration that Dr. Bea calls the
“Neutral MRGO” condition. It differs from the “2005 Katrina-As-Is” situation in one
way. It has no trace of the MRGO project, or of the wetland destruction that it caused,
but retains all of the LPV structures that were added over the years in the pre-Katrina
condition. The “Neutral MRGO” analyses conducted by Dr. Bea are based on the
rationale that the USACE had a Congressionally directed responsibility to manage the
MRGO navigation project so that it caused no added unmitigated impact on the ability of
the LPV hurricane protection project to fulfill its mission -- also Congressionally
mandated – to protect the City of New Orleans and St. Bernard Parish from hurricane
induced flooding. Further, it is accepted under this hypothesis that the levees, EBSBs,
floodwalls and other structures built by the Corps as part of the LPV project are beyond
reproach in the condition in which they were tested by Hurricane Katrina.

92
Figure 5.12 Habitat vegetation map – 1950s (Penland et al 2008, FitzGerald et al
2008).

93
So, we analyze the surge and wave conditions that would have pertained during Katrina
without the MRGO project. Then, we assess the incremental differences in the
oceanographic forces that would have been applied to the LPV structures throughout the
funnel with and without the MRGO project to determine whether and where these
differences would have changed the performance of each of the LPV structures that failed
during Katrina. Dr. Bea has carried out this analysis to determine where breaching was
avoided entirely, or occurred later in the surge sequence, thereby limiting flood damage
within the developed area (Bea Expert Report July 11, 2008).

94
 CHAPTER 6
THE MRGO FUNNEL AND ITS EFFECT ON SURGE

Point (1). The MRGO created a funnel, the dangerous convergence of channels and
spoil disposal areas, later augmented by the LPV berms east of New Orleans, that
conducted surge toward the IHNC and contributed significantly to overtopping
potential for a surge like Katrina that reached a surge maximum comparable to the
crown elevations of most protection structures.

We have previously discussed information derived from model generated time-histories

of surge elevation at a single point, known as a surge hydrograph (See Figures 5.6a, b).
The surge during Katrina dropped after the peak at roughly the same time everywhere as
the storm completed its transit past New Orleans heading for the Mississippi coast. The
pattern of surge rise was more complicated than its fall. For any approaching storm,
wind speeds are higher closer to the eye, but also the pattern of land and water features
that the surge encounters differs from one place to another. Surge will build in large
water bodies like Lake Borgne and Lake Pontchartrain and be transmitted inland
preferentially through channels, rather than over land. It will pile up on elevated features
like levees, particularly if they are oriented close to perpendicular to the direction of the
winds driving the surge. This is readily apparent in a snapshot from ADCIRC
simulations of Hurricane Betsy (Figure 6.1), as well as Katrina (Figure 6.2).

Figure 6.1 Surge build-up in the funnel and against the east bank levees of the
Mississippi River in a snapshot from an ADCIRC simulation of Hurricane Betsy by
Westerink at the University of Notre Dame. Arrows indicate wind direction. The
eye of the storm is just making landfall at Grand Isle. Warmer colors indicate
higher surge elevation in meters on scale at right
(http://www.nd.edu/~adcirc/betsy.htm).
95
Figure 6.2 ADCIRC hindcast of surge sequence from August 29 0700 (top left), 0800
(top right), 0900 (bottom left), 1000 (bottom right). Surge is in meters with
warmer colors higher. The vectors indicate wind direction with the center of
circulation (eye) visible in the top right and bottom left panels (Team Louisiana
2007).

Wave action generally increases with surge height (Figure 6.3), as both are forced by the
same properties of the wind, namely speed and direction (Gautier et al. 2008). So surge
and wave action built up at varying rates in different places during the passage of Katrina,
but diminished rapidly as soon as the storm left the area. The severity of flood damage in
protected areas is related to when flooding begins in the surge sequence, more if earlier
and less if later, which is also a function of when overtopping or breaching of the LPV
structures takes place in the surge sequence. Surge and wave energy (proportional to
significant wave height, Hs) builds up at varying rates and to varying levels at different
places depending on details of the geometry of the landscape impacted. Accordingly, it is
appropriate to consider whether the construction of the MRGO, which dictated the
selection of the subsequent alignments of the LPV flood protection structures built later,
played any role in hastening the onset of surge or waves that could have damaged LPV
flood protection elements sooner rather than later.

The way that this could -- and did -- happen is if the geometry created by the combination
of channels and paralleling man-made protective structures act together to drive up surge
through creation of what local residents have called since Hurricane Betsy the Lake
Borgne “funnel” (Figure 6.4). As we will show, it was well known within the
oceanographic and coastal engineering community that this was a likely outcome of the
geometry selected by the USACE at the time that it was chosen.

96
Figure 6.3 Evolution of significant wave heights (feet) in eastern lobe of Lake
Borgne predicted by SWAN for the southeastern lobe of Lake Borgne from 0600
(top left), 0700 (top right), 0800 (bottom left) and 0900 (bottom right) corresponding
to surge elevations on the MRGO EBSBs of 13, 15, 18 and 14 feet, respectively
(Gautier et al. expert report 2008).

97
Figure 6.4 Lake Borgne is bounded by channels that converge to the west with
parallel flood protection structures on the landward side that has been called a
“funnel.” Most breaches of the Greater New Orleans flood protection system
occurred within this funnel and in the IHNC which was connected directly to it
(Morris expert report 2008).

6.1 Historical USACE understanding of funnel effects on storm surge

Fitzgerald (expert report 2008) has documented an extensive scientific knowledge base
that existed on factors affecting storm surge elevation prior to MRGO construction.
Devastating hurricanes struck heavily populated coastal communities on the east coast in
1938 and 1954, as well as those living around Lake Okeechobee in south Florida in the
1920s and 1949. The 1938 hurricane produced storm tides of 14 to 18 feet across most of
the Long Island and Connecticut coast, with 18 to 25-foot tides from New London east to
Cape Cod (Jarvinen http://www.aoml.noaa.gov/hrd/Landsea/12Tides.pdf). As the surge
drove northward through Narragansett Bay, it was restricted by the Bay's funnel shape,
and rose to a level of nearly 16 feet (15.8) feet above normal spring tides, resulting in
more than 13 feet of water in some areas of downtown Providence (Figure 6.5).

Narragansett Bay is a coastal estuary connected by relatively deep outlets to Long Island
Sound, but Lake Okeechobee is a completely landlocked freshwater lake about 35
percent larger than Lake Pontchartrain in the center of the Florida peninsula. There, the
August 26, 1949 hurricane sequentially generated surges of more than 20 feet against the
south and north shores as the wind directions changed (Figure 6.6)

98
In 1954 the U.S. Army Corps of Engineers published their first manual entitled: “Shore
Protection, Planning, and Design”, which was the forerunner to the well-known “Shore
Protection Manual” that continues to this day to be the coastal engineer’s “Bible.” A
second edition was published in 1961 that used the Lake Okeechobee example shown in
Figure 6.6 to illustrate the problems associated with converging, funnel shaped shoreline.
Dr. Fitzgerald (expert report 2008) discusses this manual in more detail to illustrate the
state of USACE knowledge of hurricane surge in the 1950s prior to construction of the
MRGO.

“The publication discussed the physical forces acting on coasts and shore
protection structures. In a heading called: Wave Set-up, storm surges are
explained as a function of wind stresses and low pressure accompanying
storms. They also state: “Surge may also be increased, particularly in
coastal areas, by a funneling effect in converging open mouth bays." In
this same section they state: “Engineers must know expected water levels
to design harbor improvements and shore protective works.” They use
Lake Erie to illustrate the concept of “wind set-up” or as they also called
the process “wind tide” … The Corps’ Shore Protection, Planning, and
Design (1961) also demonstrates the effect of funnel-shaped coastlines by
showing water elevation changes in Lake Okeechobee during the passage
of the 26-27 August 1949 hurricane. This was a Category 4 hurricane that
weakened after coming ashore at Palm Beach, Florida. Wind velocities
and directions and water surface elevation data were collected throughout
Lake Okeechobee as storm winds first blew from the northerly quadrant
and then several hours later blew from the southerly quadrant. Maximum
wind velocities occurred at 2100 hours (26 August) coinciding with
maximum storm surge elevations (23.7 feet above a still-water surface of
13.9 feet) in the funnel-shaped bay at the southern end of the lake. As the
storm tracked northwestward, winds shifted 180 degrees and came out the
south. After the lake adjusted to the new wind field, a maximum storm
surge elevation of 23.1 feet was measured at the very northern funnel-
shaped lake shoreline. Using data sets from Lake Erie and Lake
Okeechobee, the Corps showed clearly that funnel-shaped shorelines
cause a significant increase in storm surge elevation.”

Fitzgerald (expert report 2008) has clearly shown how the science of hurricanes and their
impact to coasts grew from detailed studies made of major storms that affected American
coasts and lakes, but that it was quite mature prior to construction of the MRGO or the
LPV projects. Redfield and Miller (1957) published a very detailed survey of storm
surge levels due to hurricanes in New England in a paper entitled “Water Levels
Accompanying Atlantic Coast Hurricanes” in the journal Meteorological Monographs.
What is even more intriguing is the USACE response to the 1938 and 1954 east coast
storms. They built flood-gates and barriers cutting off bays in Providence, Rhode Island
and New Bedford, Massachusetts (Figure 6.7) to protect these regions from future storm

99
surges precisely because effective protection could not be assured without addressing the
natural funneling geometry of the coast that amplified storm surge.

Figure 6.5 Geometry of Narragansett Bay on left and observed surge hydrograph
for September 21, 1938 storm surge in Providence, Rhode Island at head of bay
(http://www.aoml.noaa.gov/hrd/Landsea/12Tides.pdf)

100
Figure 6.6 Convergence of the north and south shorelines of Lake Okeechobee into
narrow bays amplified surge in these areas during the 1949 hurricane as the wind
direction changed. Diagram reproduced from the Army Corps of Engineers’ (1961)
Shore Protection, Planning, and Design (Fitzgerald expert report 2008).

101
Figure 6.7 Rock-armored surge barriers covering more than 3 miles built by the
USACE in the mid-1960s at the mouth of New Bedford Bay equipped with navigable
flood gates span nearly 3 miles (Andrew Morang, USACE ERDC 2007 presentation
titled “Hurricane Barriers in New England and New Jersey – A Four-Decade
History”

6.2 A “bowling” analogy for storm surge effects in the Lake Borgne funnel

The Lake Borgne funnel is not a feature arising from a natural convergence of coastlines,
but a creation of man that took a natural situation that was problematic, a large coastal
lake partly enclosed by low-lying developed areas similar to Lake Okeechobee, and made
it worse. Unlike the situations considered above along the New England coast and in
Lake Okeechobee, the USACE first constructed the MRGO channel, much deeper than
any natural bayou in the area, along the axis of an estuary in a way that penetrated natural
ridges and marsh land bridges. They then compounded the problem by building flood
protection works adjacent and inboard of these channels that had much higher crown
elevations than any natural feature found in the area. These structures cut off wetland
areas with significant storm surge storage capacity while deflecting surge toward
westward into an ever narrower cross-section. We will show that the interaction between
the channel coupled with the artificial berms greatly exacerbated flow toward the City.
The question they never considered is “Where is the water pushed by a storm to go, if not
up?”

“Funnel” is an apt term to describe the geometry of the system of channels and flood
protection berms the USACE developed for this area (Figure 6.4). But a physicist might

102
prefer an analogy that gives a better idea of what happens to surge during a hurricane.
Consider a bowling alley in which the lanes are not rectangular, but are much wider at the
baseline behind which the bowlers must release, and converge to a point at the other end.
At the far end, where the pins might be placed in a normal alley, the two gutters (GIWW
and Reach 2) on each side of the lane come together to form a single trough that serves as
an outlet (GIWW/MRGO Reach 1).

Participants in this curious game take on the role of hurricanes, with names like Betsy
and Katrina, or any storm from the past or future that passes nearby on any track from the
south or east. The goal is to throw as many balls as possible in a period that corresponds
to hurricane passage, and see who gets the most balls through the trough on the other end
before time is called.

In addition to its shape, this alley has another unusual feature. The gutters are the
smoothest features, and frequently waxed (dredged), while the rest of the lane is never
even cleaned. So, in contrast to what happens in a real bowling alley, balls move faster in
the gutters than in the lanes. In fact, areas that are marsh in the real funnel are covered
with deep-pile carpet in the hypothetical hurricane bowling alley. Balls that hit these
patches are deflected and slow down, and some never make it to the end of the alley.

A skilled bowler knows that the best way to win is to send all balls sequentially down the
largest and smoothest gutter (MRGO Reach 2), because these balls will take a straight
path, will end up in the right place with no aiming required, and will experience minimal
interference from other balls. No matter how a ball arrives in a gutter, as long as it does
not stop, it ends up heading toward the outlet trough.

But the players who enjoy this game most are youngsters who love to throw balls in from
every point on the baseline as fast as they can. These balls bump and jostle and
sometimes even bounce toward the next lane, where they carom off the divider, drop into
the gutter, and continue on their way. The divider is like a flood protection berm, at least
like one that works, in that it keeps the balls in the right lane and redirects errant balls
back toward the target.

The kids’ approach described above gives a better idea of the hydrologic chaos that
pertains during a surge than does the method of the more skilled bowler. While water,
like the balls, is always coming across the baseline into the mouth of the funnel from the
east, flows within the funnel are complex and can interfere with each other. Current
speed and direction at points outside of the gutters changes often and is influenced by the
occurrence of lakes, ridges, swamp, marsh and other landscape features, so that water
particles, like colliding bowling balls, follow erratic paths. Erratic paths are longer,
meaning that the balls may not make it through the trough at the end before winds shift to
the opposite direction as the storm passes.

The USACE designed this particular bowling alley, built and maintained every part of it.
Even if we are constrained to assume that the lane dividers, i.e. flood protection berms,
are beyond reproach, there is a much larger system here, with many other features that, in

103
combination, enhance surge into New Orleans. When the system is at full capacity and
full of water, this water links all of the parts of the funnel together so it is clear that what
happens one place, like a breach, can have significant effects elsewhere in the system,
like lowering peak surge or delaying the arrival of it.

6.3 Storm Surge Dynamics and the Funnel

Damaging storm surges propagated into the IHNC from Lake Borgne through the MRGO
channel reach west of the junction with the GIWW in 1965, 1969 and 2005, during
hurricanes Betsy (Figure 6.8), Camille (Figure 6.9) and Katrina (Figure 6.2),
respectively. These were all storms that generated higher surges in Lake Borgne than in
Lake Pontchartrain, and caused flooding in New Orleans through the MRGO “back
door.” The convergence of the two navigation channels, and the berms built along their
banks, with the mouth open to Lake Borgne on the east, favors the amplification of
surges generated by hurricanes tracking over the city or to the east. Once the adjacent
Central Wetlands were isolated by the LPV berms, the only outlet for surge is the 6 mile
portion of the MRGO channel that was constructed inside the original GIWW right-of-
way that IPET discusses as MRGO Reach 1.

Hurricane-generated surge arises from the interaction between wind-forced water flow
and the geometry of the receiving coastal landscape. Like a tsunami, the elevation of the
surge is trivial in deep water but rises up against any barrier, whether an island, beach,
floodwall or levee, particularly if there is no ready avenue for escape. In coastal
Louisiana, the hurricane protection and river levees are the highest features in the
landscape, so this is where the surge rises to its greatest heights. At the same time,
dredged channels, far deeper than any natural waterway, convey the surge inland
particularly if, like the MRGO or the Calcasieu Ship Channel, they run perpendicular to
the coast. Adjacent wetlands also convey surge, but provide far greater resistance and a
shallower depth of flow.

Figure 6.8 Peak water levels in funnel during Hurricane Betsy, September 8-9, 1965.

104
Figure 6.9 High water caused by storm surge of Hurricane Camille on August 17,
1969. Peak elevation was 9.7 ft (NGVD29) at Paris Road, 10.2 ft on the north side of
the IHNC lock, and 6.5 ft at the IHNC opening into Lake Pontchartrain. New
Orleans flooding was limited to a small area of the Upper 9th Ward on the west side
of the IHNC. New Orleans experienced a Lake Borgne/MRGO surge during
Camille, while Lake Pontchartrain levels remained relatively low.

The throat of the funnel, MRGO Reach 1, is a large navigation channel (44 by 1,000
feet), but it is small relative to the gape at the end of the flood protection structures (7.5
miles, Figure 6.10) so that water pushed by wind into the jaws tends to pile up as can be
seen in an ADCIRC simulation of the surge from Hurricane Betsy (Figure 6.1). IPET
describes MRGO Reach 1 with respect to storm surge as follows.

“The critical section of the MRGO is Reach 1, the combined

GIWW/MRGO. It is through this section of channel that Lake
Pontchartrain and Lake Borgne are hydraulically connected to one
another via the IHNC. Reach 1 existed as the GIWW prior to the
construction of the MRGO, although the maintained depth was lower.
Because of this connectivity, the local storm surge and astronomical tide
in the IHNC and in the section designated GIWW/MRGO is influenced by
the tide and storm surge in both Lake Pontchartrain and Lake
Borgne…The Reach 1 GIWW/MRGO section is very important in
determining the magnitude of the storm surge that reaches the IHNC from
Lake Borgne and Breton Sound. If the hydraulic connectivity between
Lake Pontchartrain and Lake Borgne is eliminated at a point within this
section of channel, tide or surge to the west of this point will become
primarily influenced by conditions in Lake Pontchartrain” (IPET 2006,
IV-6-2). (emphasis added)

105
The term “critical section” as used by the IPET surge modeling team refers to that portion
of the channel in which hydraulic gradients are steepest and the velocity increases as a
result of constriction (Figure 6.11). The highest mean velocities observed in deep rivers
and dredged channels like the MRGO are generally constrained by turbulent and inertial
forces to 10 feet per second (fps) or less, which is far lower than the classic transition
from critical to supercritical flow (Froude number > 1) for a channel of the depth of the
MRGO. But conveyance or discharge through a given channel cross-section is still
limited by this constraint, as we will discuss in the next chapter.

1,000,000
15 ft
900,000
REACH REACH
2 L. BORGNE
800,000
1
THROAT MARSH 10 ft
Funnel Cross-Section (sq. ft.)

700,000

600,000
5 ft

500,000

400,000
0 ft

300,000

200,000

100,000 GIWW +
MRGO
Channels
0
0 10,000 20,000 30,000 40,000 50,000 60,000
Distance East of IHNC (ft)
0 5 10 15 C0 C5 C10 C15

Figure 6.10 Cross-section of the funnel between the New Orleans East and
Chalmette (St. Bernard) HPS levees at surge levels from 0 to 15 ft MSL, showing
constriction in MRGO Reach 1 (Team Louisiana 2007).

In the 1,800 foot narrows between the New Orleans East and St. Bernard hurricane levee
systems in MRGO Reach 1, the cross-section available rises more slowly with the
hurricane surge than the rate at which water enters the eastern end, so velocity rises
(Figure 6.11). If we look at the maximum depth-averaged velocities predicted by
ADCIRC throughout the funnel area during Katrina, the highest values occurred at the
eastern end of MRGO/GIWW Reach 1 and in the northern reach of the IHNC at inlet
constrictions (Figure 6.12). At the time that the surge is highest against the hurricane
protection levee along the south bank of the MRGO in Reach 2, flow throughout the
funnel converges on the throat, but cannot be fully accommodated, so that a return flow is
predicted to the northeast in the GIWW (Figure 6.12). Flow also moves away from the
peak surge zone out both ends of MRGO Reach 2. It should be noted that IPET has
quibbled with use of the term ‘funnel’ to describe Lake Borgne surge dynamics because

106
surge merely built up against the Chalmette EBSBs south of the Lake rather than in the
‘throat as it did during Betsy (Figure 6.1), but this interpretation is complicated by the
close proximity of MRGO Reach 2 to the LPV EBSBs which acted to efficiently conduct
surge essentially undiminished into the throat from wherever it originally built up (IPET
2007)

Figure 6.11 Peak flow velocities in the funnel in Katrina ADCIRC simulation by H.
Mashriqui of the LSU Hurricane Center. The red regions are zones of highest
velocity (>8 fps). MRGO Reach 1 and the northern reach of the IHNC are ‘critical,’
where a reduction in cross-section has the most effect on velocity (Team Louisiana
2007).

Figure 6.12 Currents in the funnel at peak surge in Katrina ADCIRC simulation by
H. Mashriqui of the LSU Hurricane Center. The surge reached approximately 18
feet on the MRGO EBSBs south of Lake Borgne (orange). Even though the surge
maximum was along the south side of the funnel, flow converges from all directions

107
toward the throat, but some that cannot be accommodated moves east as a return
flow in the GIWW (Team Louisiana 2007).

6.4 FINEL analysis of funnel effects

Three scenarios modeled with FINEL by our colleagues at Svasek Hydraulics are
considered (de Wit et al. expert report 2008). These have allowed us to learn far more
about Katrina surge dynamics than any previous analyses. First, we take an in-depth look
at Hurricane Katrina driven flow patterns in the funnel as it was during the storm,
Scenario 1 or Katrina-As-Is. Surge hydrographs for this scenario were given in the last
chapter (Figures 5.6a, b). Then we compare this to the condition that would have existed
in the absence of the man-made features constructed under either the MRGO or LPV
authorizations (Scenarios 2a, 2b, 2d). Then, we put back first the flood protection
structures absent the MRGO project (Scenario 2c). Finally, we put the MRGO back in,
but at its authorized dimensions, rather than what it has since become (Scenario 3).

Katrina-As-Is Scenario 1
The progression of surge in the funnel is shown hourly from 0400 through the peak at
around 0800 and as it is dropping at 0900 (Figure 6.13a, b). The surge maximum occurs
on the MRGO Reach 2 EBSBs south of the east lobe of Lake Borgne as has been true for
all ADCIRC simulations by the LSU Hurricane Center and IPET. But FINEL predicts
that this maximum is transmitted through MRGO Reach 1 with little decrease to the south
end of the IHNC (Figure 6.13b). Overtopping of flood protection structures occurs
where the surge is highest south of Lake Borgne, and where levee and floodwall crowns
are lowest, along MRGO Reach 1 and at the southern end of the IHNC, but because the
structures do not breach in the model, the Central Wetlands storage area is not filled by
the end of the sequence (Figure 6.13b). Under the modeled conditions for the flood
protection structures, flooding of developed areas of the St. Bernard polder through
overtopping of the 6.5 foot 40 Arpent Levee would not have occurred except at the
western margin into the Lower 9th Ward. There, overtopping would have lasted less than
3 hours, and would have been less than 2 hours on the MRGO EBSBs south of Lake
Borgne.

The role of both reaches of the MRGO in conveying surge into the IHNC is quite
apparent in maps of the spatial distribution of current velocity (Figure 6.14). Velocities
are generally less than 2 feet per second (fps) even at the peak of the surge everywhere
but in the channels and where flow is overtopping flood protection structures. It is much
lower in the GIWW along the north side of the funnel than in the two reaches of the
MRGO and the IHNC. The highest velocities predicted are in the constricted reach of the
IHNC north of the MRGO junction that was not enlarged as part of the MRGO project.
There, velocities of more than 10 fps are shown for water draining to Lake Pontchartrain
(Figure 6.15). Current direction in the IHNC is always north toward Lake Pontchartrain
through the whole sequence that ends after the peak at 0900.

108
Figure 6.13a FINEL surge simulation for Katrina-As-Is Scenario 1 at 0400 (top),
0500 (middle) and 0600 (bottom). Warmer colors indicate higher surge on scale (ft).
Inset indicates surge progression on B. Bienvenue hydrograph. Arrows show wind
direction. No overtopping of intact MRGO EBSBs into Central Wetlands (de Wit et
al. expert report 2008)

109
Figure 6.13b FINEL surge simulation for Katrina-As-Is Scenario 1 at 0700 (top),
0800 (middle) and 0900 (bottom). Overtopping of intact MRGO EBSBs into
Central Wetlands begins close to peak surge at 0800 is most significant at MRGO-
IHNC junction and in vicinity of Bayou Dupre (de Wit et al. expert report 2008).

110
Figure 6.14 Flow velocities predicted by FINEL (top) at 0800 with detail of
overtopping between Bayous Dupre and Bienvenue for Katrina-As-Is Scenario 1 (de
Wit et al. expert report 2008).
111
Figure 6.15 Flow velocities and direction in MRGO Reaches 1, 2 and IHNC at 0800
for Katrina-As-Is scenario, showing overtopping across south levee of MRGO
Reach 1. Maximum velocities in Reach 1 and IHNC are 8 and 11 feet per second
(fps on scale), respectively (de Wit et al. expert report 2008).

112
What is most apparent in the velocity plots is that the larger channel dimensions of the
channels constructed or enlarged as part of the MRGO project are carrying a
disproportional share of flow. The channel from about Bayou Bienvenue all the way into
the IHNC looks like a single river in the velocity maps (Figure 6.15). We will compare
discharges calculated by FINEL for each of the reaches once the scenario results have
been introduced, but it can be seen that velocities in MRGO Reach 1 are only slightly
higher than in Reach 2 between the Bayou Bienvenue control structure and the turn at the
junction with the GIWW. This suggests that the vast majority of flow entering Reach 1 is
coming from Reach 2 (Figure 6.15), and that this flow is being efficiently conveyed by
the Reach 1 channel all the way into the IHNC. From this point on in the surge analysis
we consider the various “what if” surge scenarios provided by our esteemed Dutch
colleagues (de Wit et al. expert report 2008).

No MRGO, No LPV structures

Scenarios 2A and 2B were designed to simulate a Katrina surge in the absence of any
man-made channels or flood protection works constructed under either the MRGO or
LPV projects (de Wit et al. expert report 2008). The two scenarios differ in only one
way. For scenario 2B, the 40 Arpent Levee is left at the average 6.5 foot (NAVD88)
elevation that it had during Katrina. For scenario 2A, this levee has been raised to the
17.5 foot (NAVD88) design elevation that it could have had under an LPV protection
scheme in which the pre-existence of the MRGO did not lead the USACE to choose a
more exposed alignment.

We consider scenario 2B first. For this, we have to envision a situation similar to what
existed in the late 1950s prior to construction of the MRGO and LPV projects.
Accordingly, we simulate surge along the shores of Lake Borgne in a model in which we
change the Scenario 1 model input in the following ways:

(1) 90% reduction in cross-section area of MRGO Reach 1 to the

dimensions of the GIWW (pre-MRGO);
(2) MRGO Reach 2 is removed and replaced with land at the elevation of
the surrounding marsh and swamp areas;
(3) The spoil disposal area and LPV EBSBs are degraded to natural
swamp or marsh elevation depending on the 1958 habitat map;
(4) Bayou LaLoutre ridge is restored to fill-in the hole dredged for
MRGO Reach 2;
(5) The developed area between the 40 Arpent levee and the Mississippi
is added to the model grid. The pre-Katrina crest level of 6.5 ft NAVD
is used for the 40 Arpent levee.
(6) The cypress forests originally present between the MRGO alignment
and the 40 Arpent levees, as well as between Lake Borgne and the
IHNC are restored and exert an influence on the water current
velocities and the effective wind speed on the water surface (Figure
6.16), as was discussed in Chapter 4 (de Wit et al. expert report 2008).

113
Figure 6.16 Healthy stand of second-growth cypress-tupelo swamp in the Maurepas
basin will effectively retard flow both by bottom friction and lateral friction on
trunks throughout the water column even at high surge levels, and by largely
separating the water surface from hurricane velocity winds, which will affect the
forest canopy instead (de Wit et al. expert report 2008, Day and Shaffer expert
report 2008, S. A. Hsu, personal communication).

Figure 6.17 Output locations for FINEL simulated surge hydrographs (de Wit et al.
expert report 2008).

114
Figure 6.18a FINEL surge simulation for Scenario 2b at 0400 (top), 0500 (middle)
and 0600 (bottom). Warmer colors indicate higher surge on scale (ft). Inset
indicates surge progression on B. Bienvenue hydrograph. Arrows show wind
direction. Overtopping of intact 40 Arpent Levee into developed area has already
begun by 0400 (de Wit et al. expert report 2008).

115
Figure 6.18b FINEL surge simulation for Scenario 2b at 0700 (top), 0800 (middle)
and 0900 (bottom). Warmer colors indicate higher surge on scale (ft). Inset
indicates surge progression on B. Bienvenue hydrograph. Arrows show wind
direction. Six foot difference between surge at L. Borgne shore and 40 Arpent
Levee at 0800. Developed area flooded with surge at average 12 foot elevation by
0900 (de Wit et al. expert report 2008)

116
Figure 6.19a FINEL generated Katrina hydrographs for Scenarios 1, 2a, 2b and 3
(de Wit et al. expert report 2008). The red line for Scenario 3 is almost identical to
the Scenario 1 hydrograph and lies beneath the blue line. Scenarios 2a and 2b
produce similar hydrographs, lower than Scenario 1 at these locations (Figure 6.17)

117
Figure 6.19b FINEL generated Katrina hydrographs for Scenarios 1, 2a, 2b and 3
(de Wit et al. expert report 2008). The red line for Scenario 3 is almost identical to
the Scenario 1 hydrograph and lies beneath the blue line. The Scenario 2a peak is
higher than that at 2b, but the peak is delayed for both relative to Scenario 1 at
these locations (Figure 6.17).

118
Figure 6.19c FINEL generated Katrina hydrographs for Scenarios 1, 2a, 2b and 3
(de Wit et al. expert report 2008). The red line for Scenario 3 is almost identical to
the Scenario 1 hydrograph and lies beneath the blue line. The Scenario 2a peak is
higher than that at 2b, but the peak is delayed for both relative to Scenario 1 at
these locations (Figure 6.17).

119
Figure 6.19d FINEL generated Katrina hydrographs for Scenarios 1, 2a, 2b and 3
(de Wit et al. expert report 2008). The red line for Scenario 3 is almost identical to
the Scenario 1 hydrograph and lies beneath the blue line. The Scenario 2a peak is
higher than that at 2b, but the peak is delayed for both relative to Scenario 1 at
these locations (Figure 6.17).

120
Figure 6.19e FINEL generated Katrina hydrographs for Scenarios 1, 2a, 2b and 3
(de Wit et al. expert report 2008). The red line for Scenario 3 is almost identical to
the Scenario 1 hydrograph and lies beneath the blue line. The Scenario 2a peak is
higher than that at 2b, but the peak is delayed for both relative to Scenario 1 at
these locations (Figure 6.17).

121
Figure 6.19f FINEL generated Katrina hydrographs for Scenarios 1, 2a, 2b and 3
(de Wit et al. expert report 2008). All of the scenarios are similar at these locations
(Figure 6.17) controlled by the elevation of Lake Pontchartrain, a boundary
condition.

122
The surge for scenario 2b approximates that which would have occurred under more
natural conditions (Figures 6.18a, b). The area between the developed polders of New
Orleans East and St. Bernard Parish would have been prone to high surge originating in
Lake Borgne even without the MRGO and LPV projects. The wetlands in the Central
Wetlands and closer to the IHNC did reduce the maximum surge elevation at the 40
Arpent Levee by about 3 feet, from 16 feet at the Lake Borgne shoreline to 13 feet
(Figure 6.18b), but this still allowed significant overtopping of the 6.5 foot levee, and
flooding of the developed area so that St. Bernard served as an outlet that reduced surge
elsewhere in the funnel. The reduction in the size of the MRGO Reach 1 channel to the
dimensions of the pre-existing GIWW, coupled with the swamp buffer through which it
was cut, reduced maximum surge elevation at the lock on the south end of the IHNC by 5
feet and delayed the onset of surge greater than 10 feet by more than 2 hours, as
Bretschneider and Collins (1966) predicted (Figure 6.19e).

The funnel would have been larger with more wetland storage capacity under natural
conditions, and, of course, the St. Bernard Polder would have provided additional storage
comparable to another outlet as it flooded to essentially the same level as actually
occurred during Hurricane Katrina, although it would have been a more gradual process
(Figure 6.18b). This simulation is the basis for the observation made in Chapter 1 that
with regard to the flooding of the Lower 9th Ward and St. Bernard Parish it was as if the
LPV flood protection works had never been built.

I will not present all of the output from the scenario 2A simulation in which the 40
Arpent Levee was raised to 17.5 feet. The scenario 2A hydrographs are included with the
2B output (Figure 6.19a-f), and show that loss of the St. Bernard storage would have
raised maximum surge level in the Central Wetlands to 16 feet, about 2 feet lower than
the level experienced by the MRGO EBSBs during Katrina, though for a much shorter
period, and later in the surge sequence. But an earthen levee at this location would be
shielded from waves by miles of swamp and marsh, and would not have been damaged
by overtopping. Therefore, it would not have been exposed to the destructive
combination of front and back side erosion that destroyed the MRGO EBSBs, and would
not have breached almost regardless of what materials were used to build it. As surge
levels rose in the Central Wetlands, water that would have drained over the much lower
40 Arpent Levee in scenario 2b made its way into the IHNC under scenario 2a, raising
peak surge level at the lock by about 2 feet relative to scenario 2B, but still 3 feet lower
than during Hurricane Katrina (scenario 1).

The surge peak in the IHNC during scenario 2B was controlled more by inflow from
Lake Pontchartrain than from Lake Borgne (Figure 6.20). Water level in the IHNC
during scenario 2B was essentially what it would have been if a surge barrier similar to
that now being planned by the USACE for the 100-year protection plan had been in place
during Katrina (See Figure 3.6), except that the barrier effect would have been achieved
by a 90 percent reduction in the cross-section of MRGO Reach 1 to that of the GIWW,
and by saving the cypress forest that once blocked the remainder of the hydraulic
pathway for surge into the IHNC.

123
Figure 6.20 Flow velocities and direction in the GIWW and IHNC at 0900 for
Scenario 2b, showing overtopping across south levee of MRGO Reach 1. Maximum
velocities in GIWW and IHNC are 7 and 6 feet per second (fps on scale),
respectively, and flow in the IHNC is south, from Lake Pontchartrain (de Wit et al.
expert report 2008).

124
Neutral MRGO Scenario 2c
The final FINEL scenario that I will describe, and I have not addressed all that have been
run, is that which Dr. Bea refers to as the “Neutral MRGO” simulation (Bea expert report
2008). For this, we added the pre-Katrina levees to the 1950s landscape that was used for
scenario 2B. Accordingly, we simulate surge along the shores of Lake Borgne in a model
in which we change the Scenario 1 model input in the following ways:

(1) 90% reduction in cross-section area of MRGO Reach 1 to the dimensions

of the GIWW (pre-MRGO);
(2) MRGO Reach 2 is removed and replaced with land at the elevation of the
surrounding marsh and swamp areas;
(3) The levees along MRGO Reach 1 and 2 are at their pre-Katrina crest
elevations (same as in Scenario 1);
(4) Bayou LaLoutre ridge is restored to fill-in the hole dredged for MRGO
Reach 2;
(5) The area between the 40 Arpent levee and the Mississippi is added to the
model grid. The pre-Katrina crest level of 6.5 ft NAVD is used for the 40
Arpent levee.
(6) The cypress forests originally present between the MRGO alignment and
the 40 Arpent levees, as well as between Lake Borgne and the IHNC are
restored and exert an influence on the water current velocities and the
effective wind speed on the water surface (Figure 6.16), as was discussed
in Chapter 4 (de Wit et al. expert report 2008).

The “Neutral MRGO” analyses conducted by Dr. Bea are based on the rationale that the
USACE had a Congressionally directed responsibility to manage the MRGO navigation
project so that it caused no added, unmitigated impact on the ability of the LPV hurricane
protection project to also fulfill its mission -- also Congressionally mandated – to protect
the City of New Orleans and St. Bernard Parish from hurricane induced flooding.
Further, it is accepted under this hypothesis that the levees, EBSBs, floodwalls and other
structures built by the Corps as part of the LPV project are beyond reproach in the
condition in which they were tested by Hurricane Katrina.

Surge hydrographs were prepared at the same output locations as shown for the earlier
scenarios, but also at additional points along the 40 Arpent Levee (Figure 6.21). The
replacement of the MRGO channel by marsh had a minor effect on the surge hydrographs
at locations along the MRGO EBSB alignment toward the east side of the funnel (Figure
6.22a), but a progressive lowering of the peak surge and a time-lag becomes more
prominent at locations farther west (Figures 6.22b-d). This is particularly true for the
locations inside the GIWW and IHNC west of Paris Road, so that at the lock on the south
side of the IHNC, the surge is about 3.0 feet lower than for the Katrina-As-Is condition,
and is similar to the hydrograph for scenario 2a, where the only LPV structure was a 17.5
foot levee on the 40 Arpent alignment (Figure 6.22d).

125
Figure 6.21 Output locations for FINEL simulations of the Neutral MRGO scenario
2c (de Wit et al. expert report 2008).

The most interesting result of this analysis is that it produces the least amount of flooding
of the Central Wetlands area of any of the simulations run (Figures 6.22e, f). The closest
competitor for this distinction is scenario 3, one that we have not discussed yet, in which
the MRGO Reach 1 and Reach 2 channels are modeled as at the authorized dimensions
with a 500 foot bottom-width and a 650 foot top-width. Scenario 3 produced reductions
in peak surge elevations of less than a foot only within MRGO Reach 1 and the IHNC
(Figures 6.22c, d) when compared with the Katrina-As-Is scenario (scenario 1). The
reason for this has eluded all earlier researchers but is quite obvious, and explains the
IPET ADCIRC results discussed in Chapter 4 that showed decreased flooding of
protected areas from what the USACE researchers considered negligible reductions in
maximum surge elevation when they eliminated only MRGO Reach 2 and did nothing to
Reach 1 (See Figure 4.11).

The answer is that the apparently minor increases in peak surge elevation and duration
that the ADCIRC and FINEL models allow us to attribute to the MRGO project have a
disproportionate impact on overtopping caused flooding across the adjacent LPV
hurricane protection structures because in many places the crown elevations are not much
lower than the surge elevation. Because

126
Figure 6.22a FINEL generated Katrina hydrographs for Scenario 1, 2a, 2b, 2c, 2d
and 3 (de Wit et al. expert report 2008). Scenario 2c, the grey dashed line is of
interest but is only slightly lower than the Katrina-As-Is run, scenario 1 at these
locations (Figure 6.21), with no lag.

127
Figure 6.22b FINEL generated Katrina hydrographs for Scenario 1, 2a, 2b, 2c, 2d
and 3 (de Wit et al. expert report 2008). Scenario 2c, the grey dashed line is of
interest but is only slightly lower than the Katrina-As-Is run, scenario 1 at these
locations (Figure 6.21), with a slight lag.

128
Figure 6.22c FINEL generated Katrina hydrographs for Scenario 1, 2a, 2b, 2c, 2d
and 3 (de Wit et al. expert report 2008). Scenario 2c, the grey dashed line is of
interest and is now quite a bit lower than the Katrina-As-Is run, scenario 1, at these
locations (Figure 6.21), with a pronounced lag similar to the scenario 2a and 2b
hydrographs.

129
Figure 6.22d FINEL generated Katrina hydrographs for Scenario 1, 2a, 2b, 2c, 2d
and 3 (de Wit et al. expert report 2008). Scenario 2c, the grey dashed line is of
interest and is now quite a bit lower than the Katrina-As-Is run, scenario 1, at these
locations (Figure 6.21), with a pronounced lag similar to the scenario 2a and 2b
hydrographs.

130
Figure 6.22e FINEL generated Katrina hydrographs for Scenario 1, 2a, 2b, 2c, 2d
and 3 (de Wit et al. expert report 2008). Scenario 2c, the grey dashed line is of
interest and is now lower than the Katrina-As-Is run, scenario 1, or any other at
these 40 Arpent locations (Figure 6.21) indicating less flooding of the Central
Wetlands.

131
Figure 6.22f FINEL generated Katrina hydrographs for Scenario 1, 2a, 2b, 2c, 2d
and 3 (de Wit et al. expert report 2008). Scenario 2c, the grey dashed line is of
interest and is now lower than all other simulations at these 40 Arpent locations
(Figure 6.21) indicating no flooding of the Central Wetlands.

132
Figure 6.23 Output locations for overtopping volume analysis (de Wit et al,
appendices, expert report 2008)

the length of these structures extend for several miles, even a minor increase in the
overtopping rate per linear foot results in much larger volumes of water getting over the
structure. This can be seen in an analysis of structure overtopping for a number of the
FINEL scenarios (de Wit, et al. expert report 2008).

Overtopping Analysis Using Weir Equation

Locations were selected for calculation of structure overtopping using the FINEL surge
output as the upstream input and considering the surveyed elevation of the structure
crown in the pre-Katrina condition (Figure 6.23). de Wit et al.(expert report 2008) state
that during overtopping the levees can be considered as a perfect weir. The perfect weir
discharge can be calculated by:
q = C d 2 / 3E 2 / 3gE .
With:
q = discharge per m levee (m3/s/m),
Cd = discharge coefficient (-),
E = upstream energy level with reference to the levee crest (m),
g = gravity constant (9.81 m/s2).

The upstream energy level can be considered equal to the upstream water level. The
Dutch ministry of public works executed a study to determine the discharge coefficient
for different levee heights/slopes and for perfect and imperfect weir flow (de Wit et al.

133
expert report 2008). They found that the discharge coefficient Cd≈1 for perfect weir flow
for all investigated levee slopes/heights. Therefore, we use a Cd=1 is to determine the
overtopping rates.

Knowing the elevation difference between the water level in FINEL and the levee crest, a
time line of the overtopping rate can be calculated for each selected location (Figures
6.24a-f). The levee crest elevations used to determine the overtopping rates are shown in
Table 6.1. The overtopping rates and total amount of levee overtopping water volume for
all scenarios are shown in Figures 2.40a to 2.40g.

Table 6.1: LPV structure crest elevation for overtopping analysis

Levee crest elevation for all runs
Symbol Structure Name Except where noted otherwise
Fig. 6.22 (ft+NAVD88)
1 Citrus Back Levee 13.0
2 New Orleans East Back EBSB 16.5
3 Lower Ninth Ward Floodwall 12.5
4 IHNC West Floodwall 11.5
5 IHNC East Floodwall 11.6
6 Chalmette West of Paris Road 11.6
I Bayou Bienvenue EBSB 15.5
Ii MRGO halfway EBSB 17.0
iii Bayou Dupre EBSB 14.5
II Loc II 40 Arpent 17.5 (Sc 2A) 6.5 (Sc 2B, 2C, 2D)
A Loc A 40 Arpent 17.5 (Sc 2A) 6.5 (Sc 2B, 2C, 2D)
B Loc B 40 Arpent 17.5 (Sc 2A) 6.5 (Sc 2B, 2C, 2D)
C Loc C 40 Arpent 17.5 (Sc 2A) 6.5 (Sc 2B, 2C, 2D)

The Orleans Metro and New Orleans East polders are not included in the model grid so
overtopping into those bowls, as well as breaching, do not occur in the FINEL model. So
the FINEL hydrographs in the IHNC and MRGO Reach 1 will be a little higher than what
actually would occur in each scenario, as was discussed in the calibration section.
Overtopping rates in MRGO Reach 1 and to a greater degree in the IHNC must therefore
be considered an upper bound. The surge peak on structures within the Lake Borgne will
be more realistic estimates when no breaching occurs because the surge in Lake Borgne
is unaffected by overtopping that is not included (de Wit et al. expert report 2008).

For the four locations along the 40 Arpent Levee in Scenario 2B where the levee crown is
low (Table 6.1) and the developed St. Bernard polder fills up completely, the perfect
weir discharge assumption is not valid. The overtopping discharge for those four
locations in Scenario 2B is determined from the FINEL flow velocities and water depths
at the levee crest. In FINEL the elements containing the 40 Arpent Levee have a higher
bottom friction value (equal to the bottom friction in the area between the 40 Arpent
Levee and the Mississippi River), therefore the current velocities are relatively low and
the overtopping discharge in FINEL over the 40 Arpent Levee may lead to an
underestimate (de Wit et al. expert report appendices 2008).
134
Figure 24a Overtopping rates calculated for Scenarios 1, 2a, 2b, 2c and 3 at location
1 on the Citrus Back Levee (top) and location 2 on the New Orleans East Back
Levee (de Wit et al. expert report 2008)

135
Figure 24b Overtopping rates calculated for Scenarios 1, 2a, 2b, 2c and 3 at location
3 on the Lower 9th Ward floodwall (top) and location 2 on the IHNC West floodwall
(de Wit et al. expert report appendices 2008).

136
Figure 24c Overtopping rates calculated for Scenarios 1, 2a, 2b, 2c and 3 at location
5 on the IHNC east floodwall (top) and location 6 on the Chalmette EBSB west of
Paris Road (de Wit et al. expert report appendices 2008).

137
Figure 24d Overtopping rates calculated for Scenarios 1, 2a, 2b, 2c and 3 at location
I on the B. Bienvenue EBSB (top) and location Ii on the Chalmette EBSB halfway
between Bayous Bienvenue and B. Dupre (de Wit et al. expert report appendices
2008).

138
Figure 24e Overtopping rates calculated for Scenarios 1, 2a, 2b, 2c and 3 at location
iii on the B. Dupre EBSB (top) and location II on the 40 Arpent Levee near
Chalmette (de Wit et al. expert report appendices 2008).
139
Figure 24f Overtopping rates calculated for Scenarios 1, 2a, 2b, 2c and 3 at location
B on the 40 Arpent Levee (top) and location C on the 40 Arpent Levee near Verret
(de Wit et al. expert report appendices 2008).

140
Results of the overtopping analysis are summarized for each area that received the
overtopping volume, namely, the New Orleans East polder, St. Bernard/Lower 9th Ward
polder, Orleans Metro polder and the Central Wetlands (Table 6.2). These are only
cumulative volumes at a single location and do not take into account the length of the
flood protection structures that are being characterized by a few points for each receiving
area. This is just one point in the case of overtopping of the IHNC west floodwall into
the Orleans Metro polder, but as many as four points for the St. Bernard/Lower 9th Ward
polder. For scenarios 2a and 2b, where all flood control structures were eliminated
except at the 40 Arpent Levee alignment, it was not possible to calculate flow into the
Central Wetlands. Some of the water that entered the Central Wetlands later overtopped
the 40 Arpent Levee into St. Bernard under the 2b scenario, while it was merely stored
harmlessly in the Central Wetlands for 2a with the raised 40 Arpent levee.

Even so, it is possible to derive much information from this analysis. While we know
that FINEL is over predicting surge in the IHNC and to a lesser degree in MRGO Reach
1, and that this results in higher overtopping rates over the structures along these
channels, this is true across the board for all scenarios. The point by point comparison
acts to normalize these effects (Table 6.2), rendering the comparisons meaningful.

The first observation is that the Katrina-As-Is scenario caused more overtopping at all
locations where common points could be compared except along the 40 Arpent Levee
alignment (Table 6.2). Conversely, scenario 2b, which simulated the most natural, pre-
MRGO, pre-LPV landscape performed better in terms of causing less overtopping into
developed areas everywhere but into the St. Bernard polder which acted as a sacrificial
outlet in this scenario. The added room for storage in the Central Wetlands and in the St.
Bernard polder itself reduced overtopping to an almost negligible level in the other two
developed areas that flooded during the actual Katrina event. This was, however, the
only scenario that included significant overtopping of the 6.5 foot 40 Arpent Levee. But
raising the 40 Arpent Levee in scenario 2a sent the surge elsewhere in the funnel where it
contributed to increased overtopping compared to scenario 2b, although the overall effect
was a significant 70 to 80 percent improvement over Katrina-As-Is (Table 6.2).

For scenario 3, the ravages of time and ship traffic were reversed to return the dimensions
of both channel reaches of the MRGO project to the authorized values. This reduced
overtopping at points along MRGO Reach 1 and the IHNC by 20 to 25 percent, but did
little to reduce it along the MRGO Reach 2 EBSBs (Table 6.2). The best overall
performance was achieved in scenario 2c by returning the landscape to its pre-MRGO
condition, but with the LPV levees in place as they were before Katrina. This resulted in
a fairly uniform reduction in overtopping volume of 80 to 85 percent at points
discharging into all three of the developed polders. Overtopping of the Reach 2 EBSBs
was reduced by 60 percent (Table 6.2).

141
142
MRGO channel reach discharge analysis
Our Svasek Hydraulics modeling team did one other analysis that sheds a great deal of
light on surge behavior in a funnel situation. de Wit et al. (2008 expert report appendix)
developed discharge estimates at cross-sections laid across the MRGO Reach 1 and
Reach 2 channels, or in the case where they were modeling a no MRGO channel
condition, across the total area that could convey surge inland in the vicinity of Reach 1.
This long traverse extends from the LPV berm protecting New Orleans East all the way
to 40 Arpent alignment and included conveyance through both wetlands and the pre-
existing GIWW channel (Figure 6.24).

Figure 6.25 Cross-sections for which surge discharge was calculated using FINEL
(de Wit et al. 2008 expert report)

The cross-sectional area for each of these traverses with an assumed +16 foot (NAVD88)
water elevation is given in Table 6.3, and the discharges for each cross-section and each
scenario is provided at hourly intervals in Table 6.4.

Table 6.3 Cross-sectional area for each traverse in Figure 6.24 at a uniform
elevation of +16 feet (NAVD88) from de Wit et al. 2008 expert report appendix
Cross-sectional area (ft2)
Scenario 1 MRGO Reach 1 67.000
MRGO Reach 2 75.000
Scenario 2A MRGO Reach 1 incl. wetlands 225.000

Scenario 2B MRGO Reach 1 incl. wetlands 221.000

Scenario 2C MRGO Reach 1 38.000

MRGO Reach 2 36.000
Scenario 3 MRGO Reach 1 55.000
MRGO Reach 2 58.000

The largest cross-sections by more than a factor of three are those applicable to scenarios
2a and 2b. In these 1950s-like scenarios neither the MRGO Reach 1 ship channel, nor

143
the LPV flood protection structures that flank it today are present, while many acres of
wetlands that have since been destroyed are thriving. The cross-section of Reach 1 is
roughly half that of the Katrina-As-Is section for scenario 2c because about half of the
total cross-sectional area of Reach 1 for scenario 1 is above the overbank when the water
level is at + 16 feet (Table 6.3). The “as-authorized” Reach 1 channel used for scenario
3 has 82 percent of the Katrina-As-Is cross-section, indicating that erosion and
enlargement is contributing an additional 18 percent to the section authorized for this
reach. For the Reach 2 traverse of similar length in the vicinity of the Bayou Bienvenue
control structure, the contribution of post-construction cross-section enlargement is
larger, 23 percent (Table 6.3). These differences would loom larger at lower water
elevations. The contraction in section from Reach 2 to Reach 1 is 6 and 11 percent for
the ‘as authorized’ and ‘as is’ conditions, respectively.

Table 6.4 Discharge to the west across the test cross-sections in cubic feet per
second (cfs) predicted by FINEL for scenarios 1, 2a, 2b, 2c and 3 at hourly intervals
for Katrina surge (de Wit et al. expert report 2008)

4:00 am 5:00 am 6:00 am 7:00 am 8:00 am 9:00 am

Scenario MRGO Reach 173,000 202,000 236,000 293,000 430,000 234,000
1 1
MRGO Reach 149,000 161,000 179,000 211,000 264,000 72,000
2
Scenario MRGO Reach 124,000 156,000 205,000 280,000 363,000 246,000
2A 1 incl.
wetlands

Scenario MRGO Reach 50,000 71,000 103,000 154,000 225,000 245,000

2B 1 incl.
wetlands

Scenario MRGO Reach 44,000 63,000 86,000 117,000 157,000 114,000

2C 1
MRGO Reach 14,000 20,000 28,000 39,000 48,000 -4,000
2
Scenario MRGO Reach 141,000 174,000 209,000 255,000 354,000 213,000
3 1
MRGO Reach 107,000 117,000 129,000 145,000 171,000 41,000
2

These discharges are very large, ranging for MRGO Reach 1 in the Katrina-As-Is
scenario from 173,000 to 430,000 cfs over the course of five hours (Table 6.4). The
Reach 1 channel entrains water in addition to that contributed by Reach 2 from the
GIWW channel and from across the remaining wetlands. This added water becomes
more significant at the 0800 surge peak. At peak surge, the volume contributed from
outside of Reach 2 rises from about 25 to 40 percent and continues to increase to 70
percent at 0900 as the discharge in Reach 2 drops precipitously when the wind shifts.

144
The reduced dimensions of the ‘as-authorized’ channel in scenario 3 also reduce
discharge proportional to cross-section compared to scenario 1 only at the surge peak
when flow velocities reach a maximum when it appears that conveyance is restricted.

As might be expected, discharge through the Reach 1 section is greatly diminished when
the ship channel is removed as is the case in scenarios 2a, 2b and 2c. Flow then is largely
overland even if it is constrained by LPV berms, as under scenario 2c. Peak discharge for
this scenario in Reach 1 is only slightly more than a third of the ‘Katrina-As-Is’ discharge
(Table 6.4). The Reach 2 discharge across the unbroken marsh at the Bayou Bienvenue
cross-section is less than 20 percent of its value when the fully enlarged channel section
is available. The direction of flow reverses to the east at this location by 0900 as the
winds shift. When a channel is absent in Reach 2, there is little reason to think that this
cross-section should contribute much more flow to Reach 1 than any other marsh traverse
of similar length, but it does, roughly 30 percent, at least until the winds shift. This
occurs in all likelihood because as the surge piles up on the LPV EBSBs along the south
margin of the funnel, flow is deflected preferentially along the levee face toward the
entrance to Reach 1. To return to the ‘hurricane bowling’ analogy, balls that hit the lane
divider will tend to follow it toward the outlet even if the MRGO ‘gutter’ is not there.
Such is the nature of a ‘funnel.’ Because the water is incompressible and cannot escape,
changes made to the geometry in one place propagate throughout the system either as
surge elevation increases or decreases, or as delays in surge rise or fall.

The cross-section of the 2a and 2b scenario that crosses the Reach 1 location occupied by
the GIWW conveys far more flow, essentially the same as Reach 1 with the ‘as-
authorized’ channel (scenario 3). This water is moving much slower through the marshes
and swamps toward the IHNC than it does in the smaller cross-section available in
scenario 2c, which is also missing the ship channel but is constrained by the LPV berms
spaced about 2,000 feet apart. The discharge for scenario 2b climbs slowly throughout
the sequence reaching its peak discharge at 0900 when the surge is already dropping in
most parts of the funnel. The increased ability for the surge to spread out and flow
through the Central Wetlands and over the 40 Arpent Levee delays the arrival of peak
surge at the entrance to Reach 1 almost until the storm has already left for Mississippi.
This is not the case for scenario 2a where the higher levee on the 40 Arpent alignment is
not overtopped and the surge backs up toward the entrance to Reach 1.

6.5 Funnel Summary

We established first that converging shorelines and channels can increase peak surge
elevation and hasten the onset of overtopping. Further, the USACE had a mature
corporate knowledge of these effects from analyses conducted in many places around the
U.S. prior to construction of either the MRGO or LPV projects and from experience
designing surge barriers and gates developed specifically to address them. Then, we
reported on a series of experiments using the FINEL model to detect changes in surge
dynamics that could be attributed to the man-made geometry created by the USACE
around Lake Borgne. We found that indeed surge would be a serious problem in this
embayment even without the presence of MRGO or LPV project elements because of the

145
proximity of the margins of Lake Borgne to low-lying developed areas. This already
precarious situation was made worse, however, by the construction of the MRGO ship
channel that destroyed the cypress forest and much of the coastal marsh which had
provided a very significant natural protection against most storms, both in the Central
Wetlands and at the head of the estuary on its western margin near the IHNC.
Comparison of scenarios 1, 3 and 2c showed that enlargement of the GIWW to create
MRGO Reach 1 provided a much greater potential for conveyance of Lake Borgne surge
into New Orleans particularly after the forests were destroyed.

The Reach 2 channel was the main cause of the deterioration of the wetland buffer
because of its effect on salinity (See Table 3.1). It posed less of a problem, however, for
enhancing storm surge than Reach 1 until the surge funnel geometry was worsened by
construction of the LPV EBSBs along the its south bank. First, the LPV EBSBs took ‘off-
line’ the significant amount of storage afforded by the 32,000 acre Central Wetlands.
Second, the levee project introduced a new and very dangerous dynamic in the hydraulic
feedback between the ship channel and the adjacent berms. After the LPV construction
surge could then build up on the slopes of the EBSBs and drive added flow through the
adjacent deep channel toward Reach 1, an interaction that would not have occurred if one
or the other of these projects had not been built in the close conjunction that they were.
These interactions were studied in the Reach 1 and Reach 2 discharge analyses just
summarized above (See Table 6.4).

It has taken some time, but we have finally discovered the reason why when ADCIRC
removed only MRGO Reach 2 from their ADCIRC model, it reduced flooding of the
developed polders even though the drop in peak surge elevation that this produced
appeared trivial to the IPET researchers at the time. We tested a scenario in which the
MRGO was missing, but the LPV berms were present and allowed to survive throughout
the Katrina simulation. This is the ‘Neutral MRGO’ condition. We found that surge
elevation was reduced slightly and the rise was delayed almost everywhere around the
margins of the funnel, but more so toward the west. We made use of a rigorous approach
for predicting overtopping of structures developed by our colleagues in the Netherlands
that was normalized to prevent aliasing due to model inaccuracies. This analysis showed
that even slight reductions in peak surge elevation or delays in peak onset could combine
to significantly reduce overtopping of the LPV flood protection structures throughout the
funnel. When both reaches of the MRGO were removed, overtopping was reduced by
about 80 percent for all of the three developed polders that experienced catastrophic flood
damage on August 29, 2005.

The USACE has quibbled with the use of the term ‘funnel’ to describe the geometry and
dynamics that developed during the Katrina event, as they have with the term ‘hurricane
highway’ to characterize MRGO Reach 1. We believe these are quite accurate
evocations of what happened. Because the greatest build-up of surge was observed along
the Reach 2 EBSBs south of Lake Borgne rather than in the throat at the entrance to
Reach 1 as it did during Hurricane Betsy (Compare Figures 6.1 and 6.2), the USACE
IPET team has stated that there was no ‘funnel’ effect during Katrina. We believe that
this argument is rendered untenable by the presence of the Reach 2 channel at the toe of

146
the EBSB alignment which effectively delivered surplus surge to the throat almost
undiminished during Katrina. The Lake Borgne funnel, with all its natural and man-
made features and the geometry that has been created, becomes an integrated system
connected hydraulically when it is filled to the brim with hurricane delivered surge as it
was during Katrina. We believe we have rigorously demonstrated that because the water
is incompressible and cannot escape, changes made to the geometry in one place
propagate throughout the system either as surge elevation increases or decreases, or as
delays in surge rise or fall. Even small changes in the onset or elevation of surge when
the funnel is at capacity translate into very significant differences in the likelihood of
overtopping or damage to protective structures. If we do not want to call it a surge
funnel, then what better name should we use to alert engineers designing integrated
defensive systems about the integrated nature of the hazard they are confronting, and the
duty they have to at least do no harm when they are considering modifications to the
geometry that appear unconnected outside of the critical surge condition.

147
 CHAPTER 7
EFFECTS ON FLOODING OF MRGO REACHES 1 AND 2

Point (2). The MRGO project greatly enlarged the original GIWW connection
between the throat of the funnel and the IHNC (MRGO Reach 1), that increased
surge transmission into the city earlier adding to the height and duration of surge
experienced in the IHNC during Katrina

Construction of the MRGO project to accommodate deep-draft navigation required that

the cross-section of the original six mile GIWW connection between the throat of the
funnel and the IHNC, what IPET has called MRGO Reach 1 (Figure 7.1), be increased
more than ten-fold, from about 2,000 to nearly 30,000 square feet. Here, we address the
effect of this enlargement on the onset and rate of surge transmission into the city through
the IHNC during Katrina. Again, the purpose is to investigate whether earlier
transmission of more water could have been a factor affecting when flooding of protected
areas began, and, thereby, the severity of that flooding.

10

0
Channel Bottom Elevation (ft, MSL)

0 200 400 600 800 1000

GIWW
-10

-20

MRGO Today

-30

MRGO Design Section

-40

-50
Distance (ft)
MRGO Now MRGO Design GIWW Design

Figure 7.1 Comparison of GIWW and MRGO cross-sections, as designed and post-
Katrina, in Reach 1, the connection to the IHNC. The design cross-section is 23,000
square feet, while the current cross-section averages 29,000 square feet in Reach 1,
and 41,000 square feet in Reach 2.

This part of the investigation was accomplished by a relatively simple course of model
runs, initially using ADCIRC, and confirmed with FINEL. In this analysis we modeled
the Katrina surge using the Advisory 31 hind cast that had given such good results in the
IHNC immediately after the storm to provide the base case (Team Louisiana 2007).
Then we sequentially made the following modifications to the ADCIRC S08 grid by (1)

148
raising nodes that were previously in the channel to either the elevation of the
surrounding marsh in the case of Reach 2, or (2) by an increment to reduce the cross-
section of the channel in Reach 1 by reducing its depth. Then we documented the effect
of these changes on the maximum surge attained at a number of points on Reach 2
(Figure 7.2) and within Reach 1 and the IHNC (Figure 7.3).

Figure 7.2. Reporting points for ADCIRC modeling studies within the funnel.

Figure 7.3. Enlargement showing ADCIRC reporting points within MRGO Reach 1
and the IHNC.

149
To date, the USACE has pursued modeling using ADCIRC which addresses only Reach
2 though a consultant contracted by the Louisiana Department of Natural Resources
(LDNR) has taken the analysis further as will be discussed (URS 2006). Briefly, the
USACE analysis was contained in Appendix 6 of IPET Volume IV (IPET 2006, 2007).
ADCIRC nodes within the MRGO Reach 2 footprint were raised to the level of the
surrounding marsh in the SL15 version of the model. All other features of the funnel,
including the LPV berms along the south bank of the Reach 2 channel alignment and the
current configuration of MRGO Reach 1, were retained without modification. A Katrina
simulation was run, and it was reported that maximum surge elevations at a number of
read-out points throughout the funnel and within the IHNC were virtually unaffected.

We repeated this analysis using the S08 version of ADCIRC (Table 7.1) and found
similarly that while the removal of Reach 2 did reduce surge levels over a large area that
included Lake Pontchartrain, the reductions were less than 0.5 foot (Figure 7.4). More
significant effects were observed when this modification was extended to include
constricting the cross-section of MRGO Reach 1 by varying percentages (Figure 7.4).
Reducing the cross-section progressively to 60, 40, 30, 20 and 10 percent of its current
cross-section resulted in reductions everywhere within the connecting channels but
increased surge in the throat of the funnel at Bayou Bienvenue and Paris Road where the
benefits of removing Reach 2 were counteracted by the slower drainage to Lake
Pontchartrain (Table 7.1). The reductions ranged from 1 to 1.5 feet at the southern end
of the IHNC near where MRGO Reach 1 enters it (Figure 7.4).

The time of onset and duration of surge at levels where it can damage levees and
floodwalls is as important as the peak elevation in influencing when flooding will begin
and its ultimate severity. The potential for damage that can lead to floodwall or levee
failure is, in part, a function of the time that the structure is exposed to high surge.
Floodwalls and levees in the IHNC and MRGO Reach 1 had crown elevations generally
between 11 and 13 feet (NAVD88) prior to Katrina. One important finding from this set
of modeling experiments was that the length of time during which the surge was above
10 feet, a condition that would impose stress, was sensitive to the cross-section of MRGO
Reach 1 and varied with location (Table 7.1).

The longest period of surge greater than 10 feet was experienced at the southern end of
the IHNC at the Lock, where such conditions were predicted for 5 hours during Katrina
(Figure 7.5). But 4 hours of surge above 10 feet apparently prevailed throughout the
Reach 1/IHNC channel system. Removal of Reach 2 had a limited effect on surge
duration, reducing it by up to 20 minutes at the southern end of the IHNC (Table 7.1).
Reducing the cross-section of Reach 1 progressively to closer to its pre-MRGO project
dimensions had far more effect, and this was observed everywhere throughout the
connecting channels. The reduction in surge duration above 10 feet ranged from 1 to
nearly 2 hours. As has been discussed earlier, the rise of surge is affected by local
geometry while the initial fall of the hydrograph is set by the speed of translation of the
storm and is not subject to much local variation. Reducing Reach 1 back to close to its
original pre-MRGO project dimensions slows the rate of rise of surge, delaying the time
at which it reaches whatever elevation is critical for any given floodwall or levee. The

150
Table 7.1. ADCIRC Results Giving Change in Surge Peak Elevation and Duration
above 10 Feet at Locations in Reach 2, Reach 1 and the IHNC
MRGO Reach 2 B. Bienvenue
Change
Peak Ht T>10 ft
Treatment (ft) (hr)
Katrina MRGO 0.000 3.5
No R2, 100%R1 -0.177 3.5
No R2, 60%R1 -0.008 3.5
No R2, 40%R1 0.089 3.5
No R2, 30%R1 0.155 3.5
No R2, 20%R1 0.251 3.5
No R2, 10%R1 0.257 3.5

MRGO Reach 1 Paris Rd Bulk Loading

Change Change
Peak Ht T>10 ft Peak Ht T>10 ft
Treatment (ft) (hr) (ft) (hr)
Katrina MRGO 0.000 4.0 0.000 4.0
No R2, 100%R1 -0.311 4.0 -0.093 4.0
No R2, 60%R1 -0.160 4.0 -0.049
No R2, 40%R1 -0.076 4.0 -0.014
No R2, 30%R1 -0.011 4.0 -0.113
No R2, 20%R1 0.084 4.0 -0.228
No R2, 10%R1 0.093 4.0 -0.254 3.0

IHNC PONO
IHNC Junction IHNC Lock IHNC I-10
Change Change Change
Peak Ht T>10 ft Peak Ht T>10 ft Peak Ht T>10 ft
Treatment (ft) (hr) (ft) (hr) (ft) (hr)
Katrina MRGO 0.000 4.8 0.000 5.8 0.000 3.5
No R2, 100%R1 -0.023 4.7 -0.039 5.5 -0.006 3.5
No R2, 60%R1 -0.095 4.3 -0.071 5.2 -0.040
No R2, 40%R1 -0.471 -0.423 -0.383
No R2, 30%R1 -0.741 3.7 -0.674 4.7 -0.628
No R2, 20%R1 -1.143 -1.040 -0.979
No R2, 10%R1 -1.531 3.2 -1.420 4.0 -1.306 2.0

151
surge will fall at the same time regardless of how high it has gotten by that time (Figure
7.5).
Effects of Removal of R2 and Progressive Restriction of R1

0.4
B.
Bienvenue IHN Mouth
0.2 Bulk 17th St.
Loading L. Pont. Canal
Offshore
0

-0.2

Paris
Change in Peak Surge (ft)

-0.4
Rd. Katrina MRGO
IHNC at No R2, 100%R1
-0.6 Seabrook No R2, 60%R1
MRGO No R2, 40%R1
Reach 2 MRGO Reach
-0.8 No R2, 30%R1
No R2, 20%R1
No R2, 10%R1
-1

-1.2

-1.4 IHNC
at I-10
IHNC
-1.6 Lock
IHNC PONO
Junction
-1.8
Location

Figure 7.4. Changes in peak Katrina surge elevation at read out points in Reach 2,
Reach 1, the IHNC and Lake Pontchartrain predicted by ADCIRC for (1) removal
of Reach 2, and differing degrees of closure of Reach 1, ranging from 40 to 90
percent.

The degree to which maximum surge elevation and critical surge duration are reduced is
not a linear function of the extent of constriction of Reach 1 (Figure 7.6). The
polynomial that appears to predict it best is a cubic equation. This means that
constricting the channel from 30 to 20 percent of its current cross-section will have more
effect than from 60 to 50 percent, for example, and this can be seen in the surge
hydrographs (Figure 7.5). As is usually the case in channels, the first response to the
kind of gradient imposed by a hurricane surge is for the velocity of the flow down
gradient to increase, in this case toward Lake Pontchartrain, so that the rate of
conveyance or discharge is initially little affected. The water is just moving faster. As
the gradient increases, however, turbulence develops within the flow that limits the
ability of the current to increase in average velocity. At this point, the rate of increase in
discharge for each added foot of driving head becomes limited by the dimensions of the
channel and whatever overbank area is available. This is a bit simplistic, of course,
because in this case the water is out of its normal channel and flowing overbank and
perhaps over levees as well, but the same hydraulic principle applies.

152
 ADCIRC Katrina Surge in IHNC at Junction With (1) Current Condition, (2)
Without Reach 2 and (3) Without Reach 2 and With 40 to 90% Constriction of
Reach 1
SIMULATES FLOODWALL OVERTOPPING WITHOUT BREACHING

18

16 DURATION ABOVE CRITICAL

ELEVATION
14 CRITICAL
ELEVATION
ABOVE 10 FT
Stage (ft, NAVD88)

12

10

0
8/28/05 0:00 8/28/05 6:00 8/28/05 12:00 8/28/05 18:00 8/29/05 0:00 8/29/05 6:00 8/29/05 12:00 8/29/05 18:00 8/30/05 0:00
Time (CDT)

KatrinaMRGO+ NoR2100R1+ NoR260R1+ NoR230R1+ NoR210R1+

ADCIRC Katrina Surge in IHNC at LOCK With (1) Current Condition, (2) Without
Reach 2 and (3) Without Reach 2 and With 40 to 90% Constriction of Reach 1
SIMULATES FLOODWALL OVERTOPPING WITHOUT BREACHING

18
DURATION ABOVE CRITICAL
16 ELEVATION

14 CRITICAL
ELEVATION
12 ABOVE 10 FT
Stage (ft, NAVD88)

10

0
8/28/05 0:00 8/28/05 6:00 8/28/05 12:00 8/28/05 18:00 8/29/05 0:00 8/29/05 6:00 8/29/05 12:00 8/29/05 18:00 8/30/05 0:00
Time (CDT)

KatrinaMRGO+ NoR2100R1+ NoR260R1+ NoR230R1+ NoR210R1+

Figure 7.5. Change in IHNC Katrina surge hydrographs predicted by ADCIRC at

the Junction with MRGO Reach 1 (top), and at the IHNC Lock (bottom) from
modifying two reaches of the MRGO project.

153
 Effect of R1 Reduction at IHNC/MRGO Junction

0.5

0
0 10 20 30 40 50 60 70 80 90 100

-0.5
Surge Reduction (ft)

-1

-1.5
3 2
y = -1E-06x - 0.0001x + 0.0048x - 0.04
2
R = 0.9992

-2

-2.5
Percent Restriction

Figure 7.6. Reduction in ADCIRC predicted surge peak in the IHNC at the MRGO
Reach 1 junction as a function of the degree to which the MRGO Reach 1 cross-
section is constricted.

The dynamics of how Reach 2 connects to Reach 1 in the funnel is perhaps best
illustrated by a sequence of FINEL generated maps of water velocity and direction at
different times in the Katrina surge sequence (Figure 7.7a-c). A detail of flow through
and around the point where Lake Borgne and Reach 2 have merged also shows
overtopping flow on the especially low levee crowns near Bayou Dupre (Figure 7.8). It
should be recalled that the flow model does not incorporate velocities generated by wave
runup on the front and back sides of the EBSBs, which would have greatly augmented
overtopping flow velocity.

154
Figure 7.7a. Depth-average water velocities in the two reaches of the MRGO and in
adjacent area within the funnel (de Wit et al. expert report 2008). Velocity scale on
right ranges from 0-13 feet per second (fps). Line on inset hydrograph shows
position in Katrina surge sequence at specified hour on August 29, 2005: 0400 and
0500.

155
Figure 7.7b. Depth-average water velocities in the two reaches of the MRGO and in
adjacent area within the funnel (de Wit et al. expert report 2008). Velocity scale on
right ranges from 0-13 feet per second (fps). Line on inset hydrograph shows
position in Katrina surge sequence at specified hour on August 29, 2005: 0600 and
0700.

156
Figure 7.7c. Depth-average water velocities in the two reaches of the MRGO and in
adjacent area within the funnel (de Wit et al. expert report 2008). Velocity scale on
right ranges from 0-13 feet per second (fps). Line on inset hydrograph shows
position in Katrina surge sequence at specified hour on August 29, 2005: 0800 and
0900.

157
Figure 7.8. Close-up of FINEL simulation of EBSB overtopping on low spots near
Bayou Dupre, adjacent to Lake Borgne on MRGO Reach 2 (de Wit et al. expert
report 2008). Velocities over the crowns of the structures, which are indestructible
in the model (unlike reality), exceed 11 feet per second. If you look closely, you will
notice a thin red line just before the velocities diminish on the back toe of the
EBSBs. This does not include velocity contributed by wave run up, which would
have been added.

158
 CHAPTER 8
EFFECTS OF MRGO REACH 2 ON WAVES AND FLOODING

Point (3). The MRGO Project created a Reach 2 channel with unstable side slopes
that caused it to predictably expand over time, reducing the natural marsh buffer
that previously separated it from Lake Borgne, and from the EBSB alignment,
thereby hastening the onset of damaging wave action on adjacent LPV berms that
breached early in the storm sequence.

The Robinson expert team and the USACE agree to a significant degree on the role of
waves as a primary agent of destruction of the EBSBs along the south bank of MRGO
Reach 2 (Figure 8.1). Most of the EBSBs were constructed of uncompacted hydraulic
fill material dredged from the MRGO or GIWW (Figure 8.2). Those that faced a large,
open water body like Lake Borgne or the MRGO channel experienced significant damage
and were generally classified by IPET after the storm as Levee Overtopped with
Breaching (LOB), and in some cases with Wave Scour.

Figure 8.1. Hydraulic fill EBSBs adjacent to the MRGO and GIWW navigation
channels on the margins of the Lake Borgne funnel are the only reaches that
experienced damage designated by the USACE as LOB, standing for Levees
Overtopped with Breaches (IPET 2006).

Other similarly constructed EBSB reaches that had marsh instead of a channel at the toe
on the unprotected side, like the reach of the Chalmette levee loop that turns south from
the MRGO channel toward the LaLoutre Ridge and road at Verret (Figure 8.3),
experienced virtually no damage, and were designated LONB by IPET, for Levee
Overtopped but No Breaches (Figure 8.1). This east-facing EBSB segment experienced
nearly the same overtopping as that facing the MRGO Reach 2 channel to the west
(Figure 8.3), and was no better maintained with respect to crown elevation (Figure 8.4),
but experienced virtually no damage. This is a tale of two levees, if the structures could
qualify, one with and one without large waves. Here, we are focusing strictly on the
oceanography of the surge and storm wave sequence. It is up to other members of the
Robinson expert team to say what the presence or absence of large waves means in

159
Figure 8.2 EBSBs were constructed of silt/sand rich hydraulic fill in the MRGO
spoil disposal area south of Reach 2 that performed poorly under wave attack
(IPET 2006).

Figure 8.3. A tale of two equally fragile embankments. The one facing north into
MRGO Reach 2 was destroyed, while the other facing east with marsh at the toe was
hardly damaged (modified from IPET).

160
Figure 8.4. Pre Katrina crown deviation from design elevation relative to mean sea
level for the Chalmette levee loop (white = at design el., green=1 ft low, yellow=2 ft
low, red>2 ft low) (Morris expert report 2008).

explaining the necessarily early onset of flooding of the Robinson plaintiffs homes and
communities that was described in Chapter 4 (See Figure 4.23). The critical point here is
that the size of the waves is related to the size of the channel out front. We will show that
MRGO Reach 2, now up to 3,000 feet across in places, and more than 40 feet deep at low
tide, is a powerful wave magnifier (Figure 8.5). Anything that magnifies waves breaking
on an unprotected earthen structure increases the likelihood of early failure. This battle
between grass-covered earth structures and crashing waves was enacted foot by foot, and
second by second, along more than 10 miles of MRGO EBSBs, and on more than four
miles adjacent to the GIWW (Figure 8.1). The earth lost. We will return to this question
using the SWAN model as a way of quantifying the wave energy that the channel, in its
expanded form, contributed to the destruction process.

One question appears to have captivated the IPET investigators (Ebersole USACE
Presentation 2007). Really, the big question to all of us has been “Were the EBSBs
destroyed by water eroding the back side after overtopping or by front side wave attack
prior to overtopping?” While this may seem like quibbling to those not associated with
this case, it is actually important to know the answer because it will have implications for

161
how these structures are designed, built and maintained into the future. The USACE
investigators (IPET 2007) have preferred to explain the observed destruction of these
poorly built structures primarily as a result of overtopping, rather than by erosion of the
front side (Figure 8.6). Presumably, one could attribute a virtue to these structures if
they waited for the water to go over the top, no matter how far below design grade the
crown might be. They derive this opinion primarily from post-storm observations of
damaged but still surviving structure segments (IPET 2007, Ebersole USACE
presentation 2007). We do not find these post-storm observations to be a good basis for
such an opinion.

As Dr. Bea points out, the potential for preserving unequivocal signs of breaching from
front side wave attack all the way through the surge sequence are essentially nil (Bea
expert report 2008). EBSB segments breached early by front side wave attack would
have attracted intense inflow through those breaches that would remove all signs of the
precipitating cause. Many deeply scoured EBSB breaches also later served as preferred
routes for drainage of impounded flood waters after the surge dropped. Dr. Bea (expert
report 2008), using far more rigorous analytic techniques has shown that this is not an
‘either/or’ question. Different segments of the Reach 2 EBSBs breached at different
times as a result of front side wave attack, back side scour, and a combination of both.
Dr. Bea (expert report 2008) has estimated that “approximately 35 percent of the EBSBs
along the Reach 2 alignment were breached due to attack by waves, 47 percent were
breached by a combination of wave attack and surge overtopping – backside erosion
(including sheet pile repair breaches), and 18 percent were overtopped but did not breach
(intact sheet pile repair sections, Dupre and Bievenue navigation structures).”

Both waves and the “still water” surge can destroy an unprotected earthen embankment
by backside erosion and head-cutting, though the combination of the two is wicked. On
the other hand, only waves can destroy a levee by destroying the turf cover, eroding the
front, unprotected side and then chewing through the structure. As a post-storm
investigator standing in a scour hole where a levee was, I often found it difficult to tell
exactly what had happened. So, we have to look at portions of the structure that almost
failed, and I, like most other post-storm investigators saw signs of front side wave scour
and erosion as well as back side erosion (Figure 8.7).

162
Figure 8.5. URS (2006) concept of change in significant wave height along a Lake
Borgne traverse that crosses MRGO Reach 2 and the EBSBs near Bayou Dupre,
showing wave regeneration in the channel.

163
Figure 8.6. Slide from Ebersole USACE presentation (2007) explaining regions on
front and back of embankment, where waves work on the front face and flows
caused by still water surge and wave run-up flow over the crown and cause back
side erosion.

I have found in the fairly cursory IPET (2007) analysis of the waves that destroyed the
MRGO Reach 2 EBSBs, that the waves they predict based on the STWAVE model,
operated in a half-plane mode, are just too small (See Figure 5.11). IPET (2007) used
STWAVE in the full plane version in Lake Pontchartrain, a Lake comparable to Borgne
in depth and fetch, affected by a weaker wind field, and found waves of 7 to 8 feet, while
they estimated waves on the MRGO EBSBs that were no more than 5 feet. Their
analysis is also biased toward long-period ocean swell (13-15 seconds), which seems
unlikely given the shallow water and marshes that such waves would have to cross, and
what was found in Lake Pontchartrain where 6 to 7 second waves contributed most of the
energy to the spectra (See Figure 5.7). Long-period swell tends to cause more run-up and
overtopping on levees, but shorter period local waves with periods of 3 to 7 seconds can
cause more damage to the front side. But if the total energy contained within the wave
field indicated by the significant wave height is grossly underestimated, then the potential
for damage will also be minimized.

164
 Flood side scour due to
wave breaking in varying
degrees; none in places
Headcutting due to
overtopping and overflow
evident with or without flood
side erosion

Figure 8.7. Slide from Ebersole presentation (2007) showing erosion on levees east
of Bayou Dupre structure. MRGO is to left in all photographs, showing removal of
grass by waves on the unprotected MRGO side, erosion of a wave bench in some
places, and back side erosion and headcutting with no front side erosion in others.

STWAVE is an older program that works much like SWAN, using nested grids. If it is
operated in a half-plane mode, however, and the grids are not oriented properly, in this
case, to accurately capture waves moving in the direction of the wind, which was
essentially perpendicular to the MRGO EBSBs, directly down the 40 mile long axis of
Lake Borgne, then those locally generated waves will not make it into the results. In
addition, it is possible that the USACE may have used excessive depths in Chandeleur
Sound and Lake Borgne, and this could have also allowed longer-period waves to show
up more prominently in the inshore spectra. For whatever reasons, it should be noted that
there is a fundamental difference between the SWAN simulation of the Katrina wave
field developed by Gautier et al. (2008 expert report) and that used by the USACE IPET
(Figure 8.8).

165
 From approx 6:00 to 8:30 am
(2.5 hour duration) SWL was between 15 and 17.5 ft

Levee Cross Section and Wave/Water Level Conditions

Levee Cross Section - Flood Side Scour - Cross Section 2

2000 - Pre-Storm LiDAR

19
2005 - Post-Storm LiDAR

17
Elevation (ft) N A VD 88 2004.65

15

13 2 ft to 5 ft
wave heights
11

5
0 20 40 60 80 100 120 140

Figure 8.8. Wave conditions predicted by the USACE on the MRGO Reach 2
EBSBs at the peak surge elevation during Katrina, still only 2 to 5 feet, albeit with a
15 second period (Ebersole Presentation 2007).

8.1 SWAN modeling of effects of the MRGO Reach 2 channel on waves

Gautier et al. (2008 expert report) modeled waves impacting the MRGO EBSBs from
well offshore. Waves offshore were huge and reflected the whole evolution of the
massive storm, as it intensified and then relaxed somewhat before coming ashore in
Louisiana and then Mississippi. Wind speeds over the Lake Borgne-Chandeleur Sound
complex ranged up to a maximum of about 90 knots (Figure 8.9). The prevailing wind
direction over Lake Borgne was from the northeast, almost perfectly aligned with the
long axis of the Lake, for the entire surge sequence modeled, from 04:00 to 09:00 on the
morning of August 29, 2005 (Figure 8.9).

Very long-period ocean swells were indeed present at that time, but were well offshore
(Figure 8.10). In looking at SWAN results for the distribution of dominant wave
periods, Tp throughout the domain, it appears that the waves that dominate inshore of the
Chandeleur Islands are locally generated wind waves with periods less than 6 seconds.

166
Figure 8.9 Snapshots of wind speed and direction for Hurricane Katrina at hourly
intervals from 04:00 (top left) to 09:00 (bottom right), covering the surge sequence.
Arrows indicate wind direction and colors indicate speed (knots) on legend (Gautier
et al. expert report 2008).

167
Figure 8.10 SWAN simulation of the distribution of waves of different periods in
the study area at 07:00. While waves with periods of more than 16 seconds are
offshore of the Chandeleur Island chain, waves inland of this feature are locally
generated wind waves with less than 7 second period, and less over the marshes
(Gautier et al. expert report 2008).

Waves with a significant wave height, Hs, greater than 40 feet occur offshore of the
Chandeleur Islands, but were generally less than 12 feet inland of this feature (Figure
8.11). The MRGO channel is visible in the marsh as a thin line of higher waves, up to
about 9 feet, similar to what is seen in the southerly lobes of Lake Borgne.

Looking more closely at what is occurring in MRGO Reach 2 from 06:00 to 07:00, we
see some large waves propagating up the channel from Breton Sound. They diminish,
however, before they reach the area of most interest south of the eastern lobe of Lake
Borgne, suggesting that the large waves there have regenerated while crossing the
channel (Figure 8.12).

168
Finally, looking even closer specifically at transitions near the surge peak in the channel
immediately south of the eastern lobe of Lake Borgne, it is apparent that significant wave
heights build in the channel up to a maximum of about 9 feet along the whole of Reach 2
in front of the EBSBs (Figure 8.13a, 8.13b). Significant wave heights drop significantly
as the surge maximum passes, as can be seen in the 09:00 image (Figure 8.13b). The
growth of the waves as they cross the 2,000 to 3,000 foot wide channel is much more
apparent in these close-up views. Similar things are also happening in front of the New
Orleans East Back Levee on the GIWW, but the significant wave heights there range
from 6 to 8 feet.

These waves are all locally generated within Lake Borgne, as can be seen in the
distribution of predominant wave period (Figure 8.14). At the surge peak, the 8 to 9 foot
waves in the MRGO channel have periods of between 4 and 6 seconds, and periods are
even shorter in the GIWW, from 3 to 5 seconds. As an oceanographer prone to take
risks, I imagine myself in a tower at Bayou Dupre watching these very large, short-
crested seas crashing on to the north face of the EBSBs. While I can just make out a very
chaotic situation in Lake Borgne, the seas are filtered, reformed and refracted as a more
organized sequence of closely-spaced crest rolling toward the shoreline within the
MRGO Reach 2 channel.

Figure 8.11 SWAN simulation of the distribution of significant wave height, Hs, in
the study area at 07:00 (Gautier et al. expert report 2008). While waves with Hs of
more than 40 feet are offshore of the Chandeleur Island chain, waves inland of this
feature are locally generated wind waves with significant heights of 6 to 12 feet, and
less over the marshes. The deeper MRGO channel is visible as a line with waves up
to about 9 feet.

169
Figure 8.12 Significant wave heights in southern Lake Borgne and the MRGO at
06:00 and 07:00, showing 9 foot waves in the channel at 07:00 (Gautier et al. expert
report 2008) Figure 8.13a Close-up of significant wave heights in the throat of the
funnel and western lobe of Lake Borgne and the MRGO channel at 06:00 and 07:00,
showing waves between 8 and 9 feet in that channel, and 6 to 7 feet in the GIWW.

170
Figure 8.13b Close-up of significant wave heights in the western lobe of Lake
Borgne and the MRGO channel at 08:00 and 09:00, showing waves between 8 and 9
feet in that channel, and 6 to 8 feet in the GIWW at 08:00. A significant reduction is
apparent everywhere by 09:00 as the surge drops.

171
Figure 8.14 Close-up of the distribution of predominant wave period, Tp, in the
funnel and MRGO and GIWW channels at the peak 08:00, and as the surge falls
(Gautier et al. expert report 2008).

172
We asked the SWAN modelers to provide us with some transects, what they call “rays,”
across the channel at seven key locations, six crossing the MRGO, and one crossing the
GIWW (Figure 8.15). Here we will look in more detail at two, Rays 2 and 6 that
correspond to the locations of the Bayou Bienvenue and Bayou Dupre water control
structures, respectively, restricting ourselves only to the 08:00 snapshot, approaching the
surge peak (Figure 8.16a,b).

Figure 8.15 Locations of wave transects, including six crossing the MRGO, and one
crossing the GIWW, with details of bathymetry Ray 2, Bayou Bienvenue, and Ray 6,
Bayou Dupre.

173
Figure 8.16a SWAN generated wave parameters for 08:00 along rays including 2
(Bienvenue) and 6 (Dupre), ending on the EBSBs at right, showing color-coded
points described in Figure 8.15. The top blue line is significant wave height, Hs, in
feet, dashed red line is wave period, Tp, in seconds; black line is wave direction
multiplied by 10 in degrees north; and the bottom green line is water depth
multiplied by 10 in feet (Gautier et al. expert report 2008).
174
Figure 8.16b SWAN generated wave parameters for 08:00 along rays including 2
(Bienvenue) and 6 (Dupre), ending on the EBSBs at right, showing color-coded
points described in Figure 8.15. The top blue line is significant wave height, Hs, in
feet, dashed red line is wave period, Tp, in seconds; black line is wave direction
multiplied by 10 in degrees north; and the bottom green line is water depth
multiplied by 10 in feet (Gautier et al. expert report 2008).

175
Ray 2 starts in the marsh where the water depth is about 15 feet, and significant waves
are 7 feet high (Figure 8.16a). The period is about 5 seconds and the waves direction is
55 degrees, from the north east. That general orientation holds true for all waves
impinging on the MRGO and reflects the prevailing wind direction. As the waves
approach the channel, they cross a slightly lower shelf at about 5,000 feet from the origin
where the marsh has been lost, as it has in many places along the MRGO north bank
(Figure 8.16a). Significant wave height begins to rise, reaching 9 feet before getting to
the channel proper at 6,000 feet. Hs rises again to peak at close to 10 feet as the waves
cross the 60 foot deep channel and do not diminish until they break as 6 foot seas against
the submerged levee face.

The pattern is similar for Ray 6 at Bayou Dupre, except that the ray begins on the margin
of Lake Borgne, so it starts at a depth of about 20 feet (Figure 8.16b). The period is
longer, about 6 seconds, and the wave direction is somewhat more easterly than at Bayou
Bienvenue. The significant wave height in Lake Borgne is higher than over the marsh,
about 10 feet and diminishes gradually as it encounters marsh at 6,000 feet from the
origin, at the Lake edge. Significant wave height drops dramatically to 7 feet as the
waves cross 1,500 feet of marsh, but begins to rise again as the waves begin to cross the
channel. The waves grow to 9 feet before breaking abruptly on the more complex
shoreline at this location.

SWAN can provide directional wave spectra, showing distribution of wave energy across
the frequency domain, as well as its distribution on the compass. We will again focus on
the ends of the two rays we have been discussing (Figure 8.17). The energy contained
within the spectra for both show the expected loss between the toe and higher up the face
of the EBSBs. Slightly more energy appears to be in lower frequencies at Bayou Dupre
than at Bienvenue, as would also be expected given the proximity to Lake Borgne and the
more limited filtering by the narrower intervening marsh. But these lower frequencies do
not include any waves with periods greater than 10 seconds (0.1 Hz). The waves are all
coming from the northeast, as would be expected for waves locally generated in Lake
Borgne by the prevailing winds blowing down the long axis of that waterbody.

These illustrations by Gautier et al (expert report 2008) clearly show the effects of the
marsh, even a relatively narrow swath, to greatly reduce the significant height of
significant waves both on the Lake Borgne side of MRGO Reach 2 and on the side facing
the EBSB toe. Morris et al (expert report 2008) has shown the effect of wave erosion on
both sides of the channel since it was constructed (Figure 8.18)

In summary, it is my opinion that the USACE and IPET got the wave climate on the
MRGO and GIWW wrong in two respects. First, they underestimated the total wave
energy and the size of the waves that impacted the EBSBs flanking the MRGO and
GIWW navigation channels. Second, they missed entirely the nature of the waves,
mischaracterizing them as residual offshore swells rather than higher frequency, locally
generated waves. Third, we believe that all analyses to date have missed the significance
of channel widening on the energy of waves impacting the Reach 2 EBSBs because they
have underestimated both the potential for waves to reform in the channel and of the

176
effects of marsh and foreshore erosion to increase the wave energy to which they were
exposed during Katrina

Returning to our original “tale of two levees,” it is interesting to note that the waves
along the Verret return EBSB reach, in contrast to those fronting the MRGO, never
experienced significant waves over 4 feet high, less than half the significant height of the
waves that grew as they crossed the MRGO (Figure 8.19). This is what made the
difference for these fragile structures, and for the early onset of the flooding or the
Robinson plaintiffs in St. Bernard.

Figure 8.17 Wave energy spectra at the toe (green) and face (blue) of the EBSBs at
Bayou Bienvenue (top) and Bayou Dupre (bottom), showing the relatively narrow
range of locally generated wind wave periods, with no energy in waves with periods
longer than 10 seconds (0.1 Hz).

177
Figure 8.18 Aerial view of MRGO Reach 2 Channel at the Bayou Bienvenue control
structure. Authorized channel location and 650 foot top-width is outlined in the
orange showing extent of erosion on both north and south banks that has expanded
top with to about 2,500 feet (Morris expert report 2008)

178
Figure 8.19 A tale of two levees, told in SWAN model output (Gautier et al expert
report 2008) and post-Katrina LiDAR (Ebersole USACE 2007 presentation).

179
 CHAPTER 8
EFFECTS OF MRGO CAUSED WETLAND LOSS ON SURGE, WAVES AND
FLOODING

Point (4). The MRGO Project created a channel in Reach 2 that exposed freshwater
swamps and marshes to increased salinity, accelerating their conversion to open
water and thereby reducing the wetland surge and wave buffer between Lake
Borgne and populated areas.

We have discussed the proximal effect of the MRGO channel widening and wetland loss
along the north and south banks of the MRGO on diminishment of effective foreshore
protection for the LPV EBSBs along Reach 2. But there is also the issue of the tens of
thousands of buffering wetlands, particularly cypress forests as a result of hydrological
modification and salinity intrusion associated with the ship channel.

It should be noted that we use ADCIRC only to simulate the surge at the shoreline of
Lake Borgne, where it reaches its greatest development in a natural system. ADCIRC
should be reliable until we cross the shoreline boundary and leave the open water. We
found this problem after collecting high water marks in the coastal marshes inundated by
Hurricane Rita a month after Katrina (September 24, 2005). Hurricane Rita was a
Category 3 storm that produced surge elevations (15 to 16 feet NAVD88) along the
Cameron Parish shoreline in southwest Louisiana comparable to the Katrina surge in the
funnel east of New Orleans a month earlier (Figure 9.1). For Rita, the same ADCIRC
model used to forecast Katrina predicted surge elevations accurately at the coast but
showed the simulated surge continuing to rise across the shoreline, and dropping only 10
to 15 percent over nearly 20 miles of unbroken marsh in the Grand Chenier area. This
turned out to be wrong.

The ADCIRC S08 model and its successors do not yet include verified physics to
simulate the reduction in peak water level that occurs as a surge moves inland through
swamps and wetlands, even though this phenomenon was parameterized empirically from
7 storms prior to 1960 by the USACE (1965) for levee design purposes (1 foot reduction
every 2.75 miles). This lack was apparent not only in our comparisons of field data with
model results after Rita, but also in a consultant’s report URS (2006) about hurricane
surge effects of the MRGO commissioned by the Louisiana Department of Natural
Resources (LDNR) prior to both storms, and amended after Katrina.

ADCIRC (with the proper wind speeds) did a good job of simulating the Katrina surge in
the funnel only because the LPV berms in the funnel were constructed adjacent to open
water and were not fronted by significant areas of intact wetlands. So to test the effect of
the entire MRGO project, we simulate surge along the shores of Lake Borgne without the
LPV levees, and then project that surge inland across the swamps and marshes of the
Central Wetlands to a 40 Arpent Levee that, as in URS (2006), has been raised to the
design elevation of the LPV levee that was supposed to be built along the MRGO.
Because ADCIRC is incapable of predicting the attenuation that occurs when the surge
crosses the shoreline into coastal wetlands, a second more empirical method is required to

180
Figure 9.1 Locations of surge gages installed before Hurricane Rita landfall
(September 24. 2005) by the USGS with maximum surge elevations in parentheses
(McKee et al. 2006). Maximum surge at the coast for about 50 miles east of Sabine
Lake was 15+ feet and diminished inland.

Figure 9.2 Location of USGS gages and FEMA high water marks in Rita marsh
swath east of Calcasieu Lake in Cameron Parish.

181
project surge across the MRGO right of way and Central Wetlands to the 40 Arpent
Levee. We base this projection of surge across the wetlands on study of the Hurricane
Rita surge of September 24, 2008. Rita was a Category 3 hurricane like Katrina that
struck a marsh fronted coast with no levees in southwest Louisiana (Figure 9.1). Rita
generated a surge that attained an elevation of 14.5 to 16 feet along about 50 miles of
coast extending east from the Texas line. This coastal surge was quite similar in
magnitude to that at the southern Lake Borgne shoreline during Katrina.

I mention the problem with ADCIRC primarily because of what I perceive to be an

undervaluing of the importance of swamp and marsh buffering to reduce storm surge in
places where vegetated storage areas like the Central Wetlands are available. There is a
perception that if a factor is not incorporated in an engineering model that it may not
exist. Our Dutch colleagues have gone to considerable effort to better quantify wetland
effects both on surge (de Wit et al. expert report 2008) and waves (Gautier et al expert
report 2008). Through sensitivity analysis, we were able to understand better when and
where these effects will be most significant as is indicated by discussions in Chapter 7
and 8. We saw both surge reductions of up to 3 feet across the Central Wetlands area as
delays in the onset of surge through cypress forest in FINEL scenarios 2a and 2b, which
modeled 1950s conditions in the Lake Borgne funnel.

For Katrina, unfortunately, no gage data exists either in open water or in the marsh that
allows precise reconstruction of surge propagation, and we must rely almost entirely on
model output and sparse high water mark information from the edges of the funnel. Prior
to the landfall of Rita, however, the U.S. Geological Survey (USGS) exercised great
initiative to emplace 47 recording gages in the path of the surge (Figure 9.1), and these
gages recorded a remarkable sequence of surge propagation across the marsh (McKee et
al. 2007). This data set was checked against hundreds of high water marks surveyed by
the Federal Emergency Management Agency (FEMA) that compared well with marks
previously obtained by the LSU Hurricane Center as the surge receded (Figure 9.2).

We focus here on a swath of marsh about 15 miles wide extending about 25 miles inland
(Figure 9.2) isolated from large water bodies like Calcasieu Lake to the west and Grand
Lake to the east (Figure 9.1). Each of these large lakes experienced its own surge that
was disconnected in time and space from the coastal surge. Calcasieu Lake, for example,
is nearly as long on a north-south axis as Lake Pontchartrain. Downtown Lake Charles,
which is nearly 30 miles inland at the northern end of Calcasieu Lake, was inundated by
9 foot surge that occurred more than 8 hours after the surge at the coast (Figure 9.3).

182
Figure 8.3. Surge at the head of Calcasieu Lake in Lake Charles during Hurricane
Rita lagged the surge at the coast by more than 8 hours (McKee et al. 2007).

Figure 8.4. Surge propagation through the marsh at Grand Chenier during
Hurricane Rita (McKee et al. 2007)

The situation was quite different for surge through the marsh swath to the east (Figure
9.2) also recorded by USGS gages (Figure 9.4). There, the surge that peaked at 14.8 feet

183
2 miles from the coast was reduced to 7.4 feet at 8 miles from the shoreline, and to less
than 4 feet at a gage 27 miles inland, where river flooding was also a factor. It can be
seen that the surge 8 miles inland lagged that at the coast by more than 4 hours. While
the surge rose rapidly at both locations, the falling leg of the more inland hydrograph was
elongated, showing that drainage of the marsh was relatively slow. Water level at the
most inland station 27 miles from the coast did not peak until more than two days after
the storm passed.

We compared maximum surge elevation for all of the USGS hydrographs and FEMA
high water marks in the Calcasieu Lake and Grand Chenier marsh transects to develop an
understanding of the variation in inland response to the same surge at the coast (Figure
9.5). Data from the Grand Chenier marsh transect yielded a statistically significant linear
relationship between the surge maxima and distance from the coast. This rate was 1 foot
for every 1.4 miles (0.71 ft/mile), or approximately twice the reduction traditionally
applied by the USACE (1965) for levee design on the Louisiana coast (0.36 ft/mile).

Figure 9.5. Relationship between maximum Rita surge and distance from the coast
documented by USGS gages (squares) and FEMA high water marks from the
Calcasieu Lake transect (triangles) and from the Grand Chenier marsh transect
(diamonds). Surge was between 15 and 16 feet at the coast for both transects
(McKee et al. 2007). The marsh transect yielded a statistically significant linear
correlation indicating surge reduction at the rate of 1 foot for every 1.4 miles of
marsh. The original USACE (1965) estimate of 1 foot for every 2.75 miles is also
shown for comparison.

184
 CHAPTER 9
NEED FOR INSTITUTIONAL REFORM

Over the course of the past year, the USACE has produced in the Robinson litigation
several million pages of documents relating to the history of the MRGO. These
documents—most of which have never before been made public—include internal and
external USACE communications, draft and final studies, and other writings
memorializing the agency’s planning, investigation, design, construction, operation, and
maintenance of the MRGO navigable waterway as well as the record of the Corps’s
mounting knowledge over time of the risk of flooding or EBSB failure caused by the
adverse effects created by the MRGO. From this repository, over 130,000 documents
were culled and reviewed, and a subset has been further selected because of the light that
they shed on the USACE organizational treatment of these issues.

I have read more than my share of these documents to gain an understanding of how the
USACE “managed” the threat to life, property and the environment posed by the MRGO
and have made use of an invaluable MRGO chronology that I attach as an appendix to
this report (Appendix B). This timeline summarizes information contained in the
referenced documents as it relates to milestone events and acknowledgement of, and
response to knowledge about, adverse effects created by the MRGO, including the
funnel, loss of wetlands, and erosion and widening of the channel. While the listing of
documents is not exhaustive, the contents provide remarkable insights into how and why
the MRGO disaster happened, how it could have been avoided, and, perhaps most
important, how we can prevent another repeat. The Corps got so entrenched in the habit
of fending off reasonable queries about the MRGO after Hurricane that they fooled even
themselves.

Long Time Awareness of Risks Posed by the MRGO

The history of the MRGO is characterized by a stubborn unwillingness on the part of
Corps—to acknowledge, much less seek to ameliorate, the growing body of data
indicating that the channel posed a serious threat of flooding the neighboring
communities along Reaches 1 and 2 and the IHNC.

• Beginning in the mid-1950s, the Corps was well aware that the construction of the
MRGO—as designed to cut through 43 miles of marshland and break the natural
barrier at Bayou LaLoutre separating fresh/brackish water from salt water—would
result in the intrusion of salt water from the Gulf of Mexico and the death of the
Central Wetlands Unit in the upper region of Reach 2. Nevertheless, the USACE
ignored the entreaties of the U.S. Fish and Wildlife Service and Louisiana Wild Life
and Fisheries to delay construction and allow further environmental studies. The
predictable, tragic result—as documented in scores of USACE documents over a half
century—was the destruction of tens of thousands of acres of storm-surge buffering
wetlands. If these wetlands had not been destroyed, Greater New Orleans would not
have been catastrophically flooded.

185
• Similarly, as early as 1966 immediately after Hurricane Betsy, the Corps was told that
the MRGO created a storm surge enhancing “funnel” effect by joining Reach 2 at the
GIWW/Reach 1 and a surge barrier was recommended. The USACE dismissed the
concern—as it did again in 1969 and 1973. And as noted earlier, the NESCO study
contained significant information about Hurricane Betsy and storm surge that should
have further alerted the Corps about the risk of funneling. The solutions
recommended to the Corps in 1966, 1969, and 1973 were feasible and would have
prevented the propagation of enhanced surge through Reach1/GIWW into the IHNC
and the failure of the floodwalls along the IHNC. Indeed, the Corps is now proposing
a surge barrier across the mouth of the funnel—the very recommendation rejected by
the USACE over 40 years earlier.

• Even before the first dredging began in the process that ultimately decimated so many
acres of swamp and marsh in the Lake Borgne funnel, the USACE acknowledged that
the channel’s banks would erode and widen due to the lack of any armoring or other
protection from ship wakes. Over time, the Corps chronicled the ever-widening
channel, documenting the further killing of tens of thousands of acres of wetlands and
lamenting the increased cost of continuous dredging. Meanwhile, as the channel’s
width grew from the Congressionally-authorized 650 feet to 2,000 feet and in some
places even wider, the dredged material was not used until recently to effectively
restore or protect the adjacent marshes, further accelerating the MRGO’s decimation
of the protective cypress swamps and other vegetation.

The Team Louisiana (2007) forensics team, of which I was a part, acknowledged the
USACE’s consistent refusal to take any remedial action with respect to these well known
dangerous conditions caused by the MRGO.

“The MRGO and GIWW channels provide efficient conduits to funnel

surge into the heart of New Orleans. As a result, surge elevations
peaked in Lake Borgne and the IHNC at nearly the same time, and at
higher levels relative to levee and floodwall crowns than would have
been true if the MRGO did not exist and had not caused so much
wetland loss. The effect of these federally constructed and operated
channels on surge and waves has consistently been underestimated by
the USACE from before Hurricane Betsy right through to the recent
IPET report, as has the effect of accelerated wetland loss in the funnel
area. One consequence of this institutional “blind spot” was that a
hurricane barrier of the type proposed in the original pre-1980s HPS
for the other two main passes into Lake Pontchartrain was never even
considered for the MRGO. (Team Louisiana, p. 223) (emphasis
added).

This “institutional blind spot” is starkly evident as one reads the USACE documents
referenced in the MRGO Chronology. What is most noteworthy is that the Corps’s
inveterate response was to undertake study after study that updated its knowledge about

186
the problems, studied possible solutions, and weighed alternatives. But the agency never
took any decisive action to recommend to Congress necessary changes in the MRGO and
remediation of its adverse effects. The Corps simply did nothing.

By no later than 1988 (and likely much earlier), the Corps had identified the
interrelationship between the loss of wetlands and bank erosion and the threat to life and
property. As the 1988 Corps Erosion Report detailed, there was a need to study potential
benefits of completely closing MRGO. Indeed, the Army Corps’ Lower Mississippi
Valley Division (“LMVD”), to which the New Orleans District reported, suggested
prophetically:

[A complete closure]…will control all future channel maintenance

problems by controlling bank erosion, prevent[] the associated
biological resources problem…[and] reduce the possibility of
catastrophic damage to urban areas by a hurricane surge coming
up [MRGO] . . . .

LMVD Comments to the Corps Erosion Report, at 1 (emphasis added).

Tragically, the USACE ignored its own predictions that the MRGO would in fact serve as
a storm surge delivery system aimed at the heart of Greater New Orleans.

Twelve years later, the Corps participated in a multi-agency review of the MRGO and
endorsed the unanimous recommendation that the MRGO be closed. (Mississippi River-
Gulf Outlet (MRGO) Task Force, Report of Findings (2000)). Rather than submit a
recommendation to Congress, however, the Corps insisted that it had to study the issue
further and that it was the only entity that could make a formal proposal for closure to
Congress. Like it had always had done in the past, the agency delayed and delayed its
study and never reached a definitive conclusion. When Hurricane Katrina hit on August
29, 2005, the USACE had still not completed its evaluation—and the draft still did not
recommend closure. Indeed, in its reevaluation study issued shortly after the hurricane
destroyed the nation’s 35th largest city, the Corps still maintained that the MRGO should
remain open.

The MRGO Chronology reveals that at no time prior to Hurricane Katrina did the
USACE ever recommend to Congress a plan to mitigate or eliminate the risk of flooding
posed by the MRGO, whether the recommendation was complete closure or some
prophylactic measures to retard bank erosion and protect banks from wave action of ships
and storms, prevent wetlands destruction, restore wetlands to their pre-1958 condition,
and/or to prevent enhanced surge, waves, and current from the funnel effect. For a half
century, the consistent course of action was no action—despite detailed institutional
knowledge decades before Hurricane Katrina that the MRGO, in the Corps’s own words
in 1988, posed a possible threat “of catastrophic damage to urban areas by a hurricane
surge coming up [MRGO] . . . .”

187
Failure to Heed Specific and Mounting Warnings About Flood Risks
As much as anything else, this is a record of disregard of public safety that is
unimaginable given the wealth of knowledge and expertise possessed by the USACE and
the known potential hazard of introducing a major body of water susceptible of
transporting storm surge into a major metropolitan area from the Gulf of Mexico. As
previously noted, the Corps adopted early in the life of the MRGO (in 1973) a party
line—the “unitive statement”—to reject any suggestions (even from respected experts
like Dr. Sherwood Gagliano who had been a USACE consultant on the MRGO) that the
MRGO’s design created a funnel for transmitting deadly storm surge down Reach
1/GIWW and into the IHNC, just as had occurred during Hurricane Betsy. (CEI 1973)
The Corps would not be swayed by objective scientific facts or empirical data on this or
any other issue relating to the MRGO.

This was only one of many warnings that the Corps received—from its own employees
and outsiders—over the life of the project about the substantial risk of flooding caused by
the MRGO. The following is a sampling of other alarm bells that went unheeded:

*It is well documented that before MRGO construction began, the USACE was
warned in detail by the U.S. Fish and Wildlife Service that the MRGO would destroy the
sensitive ecosystem along its channel by the construction itself as well as the intrusion of
salt water from the Gulf of Mexico. The document—entitled An Interim Report on Fish
and Wildlife Resources as Related to Mississippi River-Gulf Outlet Project, Louisiana
and an Outline of Proposed Fish and Wildlife Studies (April 1958) (the “Fish & Wildlife
Report”)—predicted the very environmental disaster that the MRGO would inflict on the
region of the Central Wetlands Unit.

*In the same vein, the Louisiana Department of Wildlife and Fisheries, or its
predecessor also predicted before groundbreaking that the MRGO would adversely affect
the local wetlands and urged further study.

* The St. Bernard Police Jury passed a resolution in April 1958 opposing the
MRGO channel, stating that “[d]uring the times of hurricane conditions the existence of
the channel will be an enormous danger to the heavily populated areas of the parish.”
This resolution was preceded by the 1957 Tidewater Advisory Committee report
(referenced earlier) predicting that the MRGO would destroy the wetlands and cause
catastrophic flooding.

*In September 1965, Hurricane Betsy flooded New Orleans East, St. Bernard
Parish, and the Lower Ninth Ward. As noted earlier in my report, valuable lessons about
the effect of the MRGO on propagating storm surge were not factored into the design of
the LPV flood protection system.

*In 1970, Louisiana coastal scientist, Dr. Sherwood Gagliano, warned the Army
Corps that the disappearing wetlands just east of MRGO would intensify storm surges
and create “deathtraps.”

188
 *In its 1976 Environmental Impact Statement for the MRGO, the USACE
acknowledged the rapid decimation of the wetlands and bank erosion.

*In October 1988, National Hurricane Center Director Jerry Jarrell depicted a
scenario concerning a Category 4 or 5 hurricane that, after destroying islands in the
Caribbean and parts of Florida, intensifies over the Gulf of Mexico and “assail[s] New
Orleans with a storm surge that overwhelms the city’s levee system.”

*In the fall of 1988, the St. Bernard Parish Council unanimously adopted another
resolution urging closure of the MRGO because it constituted a threat to public health
and safety.

*In 1988, in its MRGO Bank Erosion Reconnaissance Report which as noted
earlier recognized the threat “of catastrophic damage to urban areas by a hurricane
surge coming up [the MRGO,” the Corps admitted that the MRGO had accelerated the
“loss of marshes, ridges, bayous, ponds, aquatic grass beds and shorelines needed for the
Lake Borgne, Lake Pontchartrain and Breton Sound statuaries.” The same Corps study
characterized MRGO as one of the eight areas in south Louisiana where “erosion
stabilization measures are urgently needed.”

*1n 1994, the Corps completed its MRGO Bank Erosion Reconnaissance Report
that detailed the already well documented phenomenon of the widening of the Reach 2
channel and loss of tens of thousands of acres of wetlands.

*Six years before Katrina, the Times-Picayune summarized a consensus of expert

and lay opinion about the MRGO:

But the [MRGO] has not fulfilled its earlier promise of bringing
port-related development along it in St. Bernard parish. It has
instead become an environmental and economic disaster. Its
original width was to be 500 feet, but it has eroded in some places
to 2,000 feet. It has changed the salinity of the marsh, leading to
further erosion and ruining oyster beds. But worse it has
threatened lives, acting as a pathway for hurricane storm surges
and even surges from storms short of hurricane force.

Editorial, “Light at the End of the Channel,” Times-Picayune, June 1, 1999, at B-6
(emphasis added).

*In 2000, the Corps acknowledged MRGO’s shortcomings and endorsed a coastal
restoration plan calling for the canal’s eventual closure. Bending under pressure,
however, the Army Corps did not move forward to recommend to Congress closing the
MRGO, instead initiating a reevaluation study to assess (once again) the environmental
and economic implications of closing the channel or keeping it open. Safety concerns
were once again conspicuously ignored.

189
 *In his book Holding Back The Sea: The Struggle for American’s Natural Legacy
on the Gulf Coast (Harper Collins, 2001), Christopher Hallowell warned (and the Corps
already knew) that “erosion from ships and storms has gouged it [the MRGO] 2,000 feet
wide and made it a freeway to New Orleans for any hurricane that happens to come from
the right direction . . . . The surrounding marsh, now vulnerable to storms and salt water,
has all but died as a result, along with 40,000 acres of mature cypress trees. Now, storm
surges can invade the marsh through the straight arrow channel and smash into New
Orleans.”

*In October 2001, Scientific American magazine published “Drowning New

Orleans” that predicted “[a] major hurricane could swamp New Orleans under 20 feet of
water, killing thousands. Human activities along the Mississippi River have dramatically
increased the risk, and now only massive reengineering of southeastern Louisiana can
save the city.”

*Another forewarning of the MRGO’s imminent dangers came from proponents

of the outlet’s closing. In June 2002, the critics described MRGO as “a shotgun pointed
straight at New Orleans should a major hurricane approach from that angle.”

*In June 2002, Corps acknowledged that its own safety estimates of MRGO—
created in the 1960s with tools that are low-tech and unsophisticated by modern
standards—are antiquated and unreliable. Local engineer Lee Butler estimated the risk of
MRGO overtopping in St. Bernard Parish may be double the Army Corps’ original
estimates. In addition, the Army Corps agreed that MRGO is a “weak spot” and more
likely to be affected by a hurricane storm surge than other areas because of its proximity
to the Gulf of Mexico.

*In July 2004, Hurricane Pam—a federally-funded hurricane simulation drill

conducted by an LSU team of which I was a member—prophetically concluded that
surges from a Category 3 hurricane would be “funneled” by the MRGO, flooding
surrounding areas and killing tens of thousands of people in Greater New Orleans.

*In October 2004, the Louisiana Legislature passed a resolution urging closure of
the MRGO and immediate implementation of remedial measures to address the risk
posed by the MRGO and “more drastic tidal surges and more prolonged flooding as a
result of tropical storms and hurricane.”

*In May 2005, a hydrodynamic model from LSU’s Hurricane Center

further exposed MRGO as a “superhighway for storm surge” and predicted that a
“funnel” created by MRGO could amplify storm surges by 20 to 40 percent.

All of this evidence—and much more in the record—shows that Hurricane Katrina and
its devastating aftermath were clearly foreseeable and foreseen events to the Corps. It
cannot be seriously disputed that the USACE before 1958 had actual notice that the
construction of the MRGO would destroy the storm surge buffering wetlands and sustain

190
significant bank erosion from lack of shore protection. Likewise, the Corps should have
known (because coastal engineers around the world knew) that the design of the
MRGO—the geometry of creating a funnel at the Reach 2 and Reach 1/GIWW
confluence—would exacerbate a hurricane’s power and destructive effect on the local
population and property. Similarly, before construction, the Corps—based on its
experience in designing and operating other waterways—knew or should have known
that the unarmored banks of the MRGO need to be armored to avoid inevitable erosion
from ship waves.

Even if the Corps did not appreciate the potential for these adverse effects in the planning
and design stage, it acquired information throughout the operation and maintenance of
the MRGO—both from its own personnel and civilian engineers and coastal scientists—
that the waterway posed a significant risk of accentuating storm surge during a hurricane.
Unfortunately, such notice was ignored, and no corrective action was ever recommended
to Congress before Hurricane Katrina. Nor was the LPV—40 years in the making and
still incomplete—ever designed or redesigned to take into account the need for added
safety margins because of the MRGO.

Mission Myopia
The question that I continue to ask is “How could this situation have persisted for so
long?”

I think the answer lies in the USACE’s shortsightedness about its role and responsibility
for correcting the serious hazards that it created in placing a deep channel with direct
access to the Gulf of Mexico into a major metropolitan area. As discussed in the next
section, the failures here also stem from a stunning failure of leadership and vision.

The record shows that the USACE had a chronic myopia about its mission with regard to
the waterway. Time and time again, the Corps indicated that its sole responsibility was to
maintain the channel so that it could be used by deep draft vessels. That entailed a
continuous dredging and maintenance program because of the Corps’s decision not to
armor the banks and the resulting soil erosion from ship wakes and periodic storms.

Actions taken (or more accurately, not taken) by the Corps reflect this tunnel vision.
Despite knowing from the outset that the MRGO would destroy the storm surge buffering
wetlands, no remedial program was ever undertaken. The agency perceived that its sole
responsibility was keeping the channel open.

Corps officials did not view protection of the environment as part of its primary mission
at the outset, but this should have changed a long time ago. While the loss of the
wetlands from the MRGO construction and ongoing operation and maintenance was well
documented and maybe even regreted, the agency did not consider itself obligated to fix
the problem. In other words, the USACE knew that it created a bad situation and by
1988 had acknowledged the nexus between loss of wetlands and potentially disastrous
hurricane surge flooding Greater New Orleans. Nevertheless, the Corps never exhibited
any sense of urgency to develop a comprehensive set of policy recommendations for

191
Congress to authorize funding for ameliorating these defects. The Corps’s left hand was
regularly reporting the burgeoning ecological disaster, while its right hand slavishly
persisted in the fiction that making the channel safe for shipping was its only priority.

The Corps also had a narrow view of its constituency. The agency showed concern only
for what it termed its “customers”—the Port of New Orleans, shipping, dredging, and
pilot boat interests. These were the commercial sectors that had successfully lobbied for
the MRGO’s creation, and they were the lobbies that kept the Corps in the navigable
waterway business.

At the same time that they sought to accommodate shipping interests in a channel in
which that group gradually lost interest, the Corps turned a deaf ear to the crescendo of
public, local government, and expert voices sounding the alarm about the potential for
catastrophic flooding and calling for closure or remediation of the MRGO. The St.
Bernard local government—from the Tidewater Advisory Committee’s prescient report
that I quote in Chapter 2 in 1957 to its numerous commissioned expert reports by Dr.
Sherwood Gagliano and others to the resolution urging closure in 2004—continuously
monitored the MRGO and documented for the Corps the escalating risk of catastrophic
flooding from the loss of wetlands, bank erosion, and the funnel effect. The concerns of
other federal and state agencies and respected scientists and scholars were likewise
ignored. Public safety was simply not a factor in the Corps’s decision-making calculus.

A Failure of Leadership
The roots of this catastrophe can also be directly traced to a monumental failure of
leadership. As noted earlier, at the same time that the USACE was planning the MRGO,
it was planning the LPV in an effort to give Greater New Orleans an added measure of
hurricane flood protection. In other words, the same agency—with the same leadership in
New Orleans, Vicksburg, and Washington, D.C.—was making critical decisions about
fundamental engineering and scientific issues affecting the welfare of the residents. Over
the intervening half century between the authorization of the MRGO in 1956 and
Hurricane Katrina, the same Corps senior management would be responsible for dealing
with (or more accurately, ignoring) the known flood risks posed by the MRGO.

Much post-Katrina attention has been focused on the deficiencies in the USACE
decision-making process with respect to the planning, design, construction, and operation
of the LPV and the reasons for the catastrophic failure of the levees. Douglas Woolley
and Leonard Shabman recently prepared a Decision-Making Chronology for the Lake
Pontchartrain & Vicinity Hurricane Protection Project: Final Report for the Headquarters,
U.S. Army Corps of Engineers Submitted to the Institute for Water Resources of the U.S.
Army Corps of Engineers (March 2008) (“LPV Decision-Making Chronology”) that
sheds some light, though it studiously avoids mentioning the MRGO. The authors gained
valuable insights into the processes by which the Corps gains and shares knowledge,
evaluates alternatives, and implements its decisions. The authors found that the Corps
proceeded with the LPV as originally planned despite knowing that it was using
inappropriate technical data and assumptions such as continuing to plan on the basis of
outdated datum benchmarks for levee crown elevations and outmoded Standard Project

192
Hurricane parameters. Many of the same patterns emerged for me in my study of the
MRGO record Woolley and Shabman write:

“Project construction was not yet underway when significant project design
changes were requested and approved following Hurricane Betsy. At that time,
the increase in project costs associated with the design change might have
appeared readily affordable to project sponsors, and the change involved virtually
no delay in project implementation.

In later years, however, the accommodation of new information in project design

and construction would have required adjustments to ongoing construction
activities as well as retrofitting project features that had already been constructed.
Such changes would have significantly increased project costs and
implementation delays at a time when local concerns about project costs and
urgency for project protection were paramount, and a stagnant Corps construction
general budget had to be spread among competing priorities. It was in the context
of a history of local sponsors’ frustrations over project delays and costs, federal
and local budget limits, and increasing scrutiny of water project investment
proposals at the Washington, DC level, that new information suggesting the need
for project reevaluation and redesign that might take years to analyze and get
approved was either put aside for later consideration (e.g., the 1985 datum
benchmark decision), or subjected to further study (e.g., the decision to refine the
ADCIRC surge model before applying it for project reevaluation) (see Chapter
6).”

This pattern of deferring hard choices by engaging in “further study” is a recurring

institutional response to problems not only with regard to the LPV but also the MRGO.
The Corps conducted numerous reconnaissance, evaluation, and reevaluation studies with
regard to the MRGO’s glaring defects. Like the LPV, until the final report on MRGO
closure issued last year, however, there was never a resolution and a recommendation to
Congress to deviate from the originally authorized plan.

Woolley and Shabman also found that the USACE lacked any organizational systems to
monitor and reassess whether program objectives were being realized in light of changed
circumstances or new information.

“The District was not expected to routinely track and as needed revisit—using
whatever tools were available at the time—the ability of the project to provide the
authorized degree of protection as new information became available. The
absence of a standing, agency-wide process for continuing assessment and
reporting of project performance capability left the District to make its own
determination as to whether the analytical foundation was adequate for requesting
changes to project designs, and for satisfying higher federal authorities and local
sponsors that additional project funding was warranted” (see Chapter 6).

193
The same was true with regard to the MRGO. There does not appear to have been any
centralized responsibility within the USACE for determining whether the agency should
seek Congressional authorization and funding for remedying known defects with regard
to the MRGO. While the Corps was able to monitor its dredging costs and make annual
appropriations requests for routine and emergency funding for maintaining the channel,
no similar institutional focus ever materialized as to recommending a comprehensive plan
to Congress before Hurricane Katrina.

The LPV Decision-Making Chronology also criticizes the Corps for how it
communicated flooding risk and the LPV’s level of protection to the stakeholders. Id. at
ES-19. This failing is magnified many times with respect to the MRGO and the agency’s
persistent claim—based on its “unitive statement”—that the channel had not been during
Hurricane Betsy, and would not be in the future, a source of enhances hurricane storm
surge. Certainly, by 1988 if not much earlier, the Corps knew that the destruction of the
wetlands and widening of the channel posed a serious threat of catastrophic flooding.
This ostrich-like approach to its duty to warn and remediate was a critical factor in the
tragedy of Hurricane Katrina.

There is a broad consensus about the failure of leadership by the Corps in protecting New
Orleans and St. Bernard Parish. In his earlier-referenced decision dismissing the lawsuits
against the USACE for faulty levees along the 17th Street and London Avenue Canals,
Judge Stanwood Duval, Jr., in summarizing the evidence, criticized the Corps for its
“monumental miscalculations,” “gross incompetence,” “egregious myopia,” and
“catastrophic failure . . . to fulfill its mission” by constructing seriously flawed levees that
set the stage for the calamity that engulfed 80% of the region. “This story—fifty years in
the making—is heart-wrenching.”

The same can be said with respect to the MRGO. Tragically, the same Corps officials
who egregiously mismanaged the LPV were also derelict in their duty to protect the
public and property from the long-known defects of its waterway. There can be no
justification for the USACE failing to take corrective action to prevent the MRGO from
becoming a major cause of the catastrophic flooding of St. Bernard Parish, New Orleans
East, and the Lower Ninth Ward.

Final Thoughts
For over five decades, the USACE’s stewardship over the MRGO—amidst the mounting
evidence of its adverse effects on the environment and the potential for greatly enhanced
storm surge leading to catastrophic flooding—can best be described as one of studied
indifference to long-recognized threats to life, property and the environment.
Consistently, the Corps preferred the interests of its commercial constituency—its
“customers”—over those of the hundreds of thousands of local residents put in harm’s
way by the MRGO. This “keep the channel open at all costs” mindset never calculated
(or appreciated) the costs in human life, property, and the environment from cataclysmic
flooding that the agency long knew could happen because of the MRGO.

194
Throughout this half century, the Corps conducted numerous studies of the channel’s
adverse effects. However, it never reached a definitive conclusion or recommendation
for any plan of action until more than two years after Hurricane Katrina devastated
Greater New Orleans. Indeed, at no time before the catastrophic flooding did the Corps
weigh alternatives, balance the social, financial, and political considerations in favor of
and against each potential course of action, evaluate the risk to people and property, and
make a recommendation to Congress in the exercise of its policy-making judgment about
the preferable course of action with respect to the MRGO’s defects.

Only in late 2007—more than two years after Hurricane Katrina—did the Corps actually
exercise its judgment and recommend to Congress the MRGO’s closure and one
remediation measure to construct a barrier at Bayou La Loutre to prevent further
intrusion of salt water from the Gulf of Mexico. Even then, the Corps did not
acknowledge the MRGO’s contribution to the severe flooding during Hurricane Katrina
and its continuing danger; nor did the USACE recommend as elements of its plan
essential remedial steps for ecosystem restoration and prevention of storm surge along
Reach 1, Reach 2, and the IHNC. In typical fashion, the agency continues to study these
issues to the present day.

195
 CHAPTER 10
CONCLUSIONS

The Government takes the position that the Mississippi River Gulf Outlet navigation
project (MRGO) had no effect on the catastrophic flooding of Greater New Orleans
during Hurricane Katrina. Our analyses show that the Government is incorrect. It is my
opinion that the MRGO navigation project:

(1) created a funnel, the dangerous convergence of channels and spoil

disposal areas, later augmented by addition of the Lake Pontchartrain &
Vicinity (LPV) berms east of New Orleans, and by the subtraction of
buffering wetlands, that foreseeably amplified the threat posed by
hurricane surge to the Greater New Orleans area; and
(2) greatly enlarged the original Gulf Intracoastal Waterway (GIWW)
connection between the throat of the funnel and the IHNC (MRGO
Reach 1), which foreseeably increased surge transmission into the city
earlier, adding to the height and duration of surge experienced in the
IHNC during Katrina and contributing to the early failure of floodwalls
and levees adjacent to the IHNC; and
(3) created a Reach 2 channel with unstable side slopes that caused it to
predictably expand over time, reducing the natural marsh buffer that
previously separated it from Lake Borgne and from the adjacent LPV
berms, thereby compromising foreshore protection fort the man-made
surge protection elements and hastening the onset of damaging wave
action on these delicate structures (EBSBs) so that they breached earlier
in the storm sequence; and
(4) created a channel in Reach 2 that predictably exposed freshwater
swamps and marshes within the Lake Borgne funnel to increased
salinity, accelerating their conversion to open water and thereby
reducing the wetland surge and wave buffer between Lake Borgne and
populated areas; and
(5) was a substantial contributing factor to the catastrophic flooding of the
Robinson plaintiffs’ homes and communities.

In addition, throughout the history of the MRGO project, up to the time of Hurricane
Katrina, the U.S. Army Corps of Engineers (USACE or Corps), in my opinion, has
continuously adopted a policy of denial and deliberate disregard of well documented
adverse effects of the channel, most of which were known or suspected prior to
construction and others that became apparent later, soon thereafter ensuring that a real
threat for flooding during hurricanes remained unaddressed for five decades. Throughout
this period, the USACE issued increasingly soothing but inaccurate statements to two
generations of New Orleans and St. Bernard residents that they were well protected and
that the MRGO project posed no hazard to their lives or property.

Furthermore, this approach is much the same today, in my opinion, coloring the
conclusions of the Corps sponsored investigations that have followed Katrina, and the

196
nature of solutions that have been proposed. This, in particular, includes an obdurate
unwillingness to provide wave protection to the hastily rebuilt berms that parallel MRGO
Reach 2 despite overwhelming evidence that they were largely destroyed by waves.
Today, the Corps still refuses to acknowledge, in the face of compelling scientific
evidence, that the MRGO project was a significant cause of the early and catastrophic
flooding of the Upper and Lower 9th Ward, St. Bernard Parish and New Orleans East
during Hurricane Katrina.

Ultimately, the extent of damage and the harm caused by the flooding of St. Bernard and
the Lower 9th Ward through the MRGO was not significantly reduced by the LPV berms
and floodwalls. The structures that Dr. Bea has described as “earthen berm/spoil banks”
(EBSBs) (to differentiate them from properly engineered coastal defense dykes) along
MRGO Reach 2 served merely as “speed bumps” that were swept aside during Katrina by
the surge and waves generated in, and transmitted by the MRGO channel. Most of the
flooding of the Lower 9th Ward and New Orleans East, as well as a significant portion of
the early flooding of the Orleans Metro area, is attributable to the enlarged cross-section
of MRGO Reach 1.

Finally, I believe that the USACE, and certainly the larger oceanographic community,
possessed a knowledge base prior to the construction of the MRGO project, and certainly
after the Betsy disaster, that should have led to actions to tightly integrate the MRGO
economic development project with the LPV public safety project, and thereby reduce or
eliminate the added and substantial hazards uniquely posed by the ship channel. Instead,
with regard to this particular project and, significantly, not in all comparable cases around
the country, the Corps adopted a policy of institutional denial that stymied efforts of local
leaders, engineers and scientists to address critical, long-known deficiencies that were
fully manifested during the Katrina event. My reading of the record indicates that
certainly by 1988, if not earlier, the Corps knew about the nexus between many of the
deficiencies noted above and the likelihood of catastrophic flooding. Nevertheless,
before Hurricane Katrina, the Corps never completed any study recommending a plan to
Congress to mitigate known hazards. The storm has provided an opportunity for new
leadership at the Corps to mobilize all of the authority, resources and expertise that it has
been given by Congress to mitigate the damage that the MRGO has caused. We are still
waiting.

197
 REFERENCES

American Society of Civil Engineers (2007). The New Orleans Hurricane flood
protection System: What Went Wrong and Why. ASCE, Reston VA.

Asano, T., Deguuchi, H, Kobayashi, N. (1993). “Interaction between water waves and
vegetation,” Proceedings 23rd Coastal Engineering Conference, ASCE, New
York, 2710-2723.

Bea, R. G. (2008a). Expert report concerning the performance during Hurricane Katrina
of the man-made features bordering the Mississippi River – Gulf Outlet.
Declaration submitted on behalf of the Plaintiffs in Robinson v. United States in
connection with the Flood Control Act immunity issues, July 11, 2008.

Bea, R. G. (2008b). Expert report concerning the performance during Hurricane Katrina
of the man-made features bordering the Inner Harbor Navigation Canal at the
Lower 9th Ward. Declaration submitted on behalf of the Plaintiffs in Robinson v.
United States in connection with the Flood Control Act immunity issues, July 11,
2008.

Bea, R.G. and Arnold, J. L. (2008). Technical Guidance Available for Design,
Construction, and Maintenance of the MR-GO and LPV, Technical Report to
MR-GO Litigation Group Pertaining to MR-GO, Robinson [No. 06-2268], Risk
Assessment and Management Services, Moraga, CA, July 11, 2008.

Bea, R. G. and Cobos-Roa, D (2008b). Analyses of Breaching of the Effects of the

USACE IHNC Lock Expansion Project East Bank Industrial Area Site Clearing
Excavations on Development of the Breaches at the Lower 9th Ward During
Hurricane Katrina. Technical Report to MR-GO Litigation Group Pertaining to
MR-GO, Robinson [No. 06-2268], Risk Assessment and Management Services,
Moraga, CA, July 11, 2008.

Bea, R. G. and Cobos-Roa, D. (2008a). Analyses of Breaching of MR-GO Bayou Dupre

& Bayou Bienvenue Navigation Structures During Hurricane Katrina. Technical
Report to MR-GO Litigation Group Pertaining to MR-GO, Robinson [No. 06-
2268], Risk Assessment and Management Services, Moraga, CA, July 11, 2008.

Bea, R. G. and Storesund, R. (2008a). Analyses of Breaching of MR-GO Reach 2 EBSBs

During Hurricane Katrina. Technical Report to MR-GO Litigation Group
Pertaining to MR-GO, Robinson [No. 06-2268], Risk Assessment and
Management Services, Moraga, CA, July 11, 2008.

Bea, R. G. and Storesund, R. (2008b).Review of Excavation and Backfill Practices Near

Flood Control Systems. Technical Report to MR-GO Litigation Group Pertaining
to MR-GO, Robinson [No. 06-2268], Risk Assessment and Management
Services, Moraga, CA, July 11, 2008.

198
Bretschneider, C. L. (1954). Field investigation of wave energy loss of shallow water
ocean waves, USACE, Beach Erosion Board, Technical Memo. No. 46,
September.

Bretschneider, C. L. and Collins, J. I. (1966). Storm Surge Effects of the Mississippi

River-Gulf Outlet, Study A, Report to USACE (New Orleans), NESCO Report
SN-326-A, September, Pasadena, CA.

Camfield, F.E. (1977). Wind-Wave Propagation Over Flooded, Vegetated Land,

Technical Report Paper No. 77-12, Coastal Engineering Research Center, USACE
Waterways Experimental Station, Vicksburg, MS.

Camfield, F.E. (1983). “Wind-wave growth with high friction.” Journal Waterway, Port,
Coastal, and Ocean Engineering, ASCE, 109, 1, 115-117.

Coastal Environments, Inc. 1972. Environmental Impact Study, Ship Channel Project.
Report Prepared for St. Bernard Parish Police Jury, October, 1972

Collins, J. L. (1972). “Prediction of shallow-water spectra,” Journal of Geophysical

Research, 77, 2693-2707.

Day, J. W. Jr., and Shaffer, G. P. (2008a). Effects of the Mississippi River Gulf Outlet on
Coastal Wetlands and Other Ecosystems in Southeastern Louisiana, Report to
MR-GO Litigation Group, May.

Day, J. W. Jr. (2008b). Effects of the Mississippi River Gulf Outlet on Coastal Wetlands
and Other Ecosystems in Southeastern Louisiana, Report to MR-GO Litigation
Group, May.

Dean, R.G. (1978). “Effects of Vegetation on Shoreline Erosional Processes,” Wetlands

Functions and Values: The State of Our Understanding, American Water
Resources Association, 415-426.

Ebersole, B. (2008). Roles of MR-GO, GIWW, and IHNC During Hurricane Katrina,
Hydrodynamic Loadings and Levee/Wall Responses Along the Navigation
Channels and Influence of MR-GO During Katrina. USACE, Engineering
Research and Development Center, Coastal and Hydraulics Laboratory,
Vicksburg, MS.

Ebersole, B (2007) Hydrodynamic loadings and levee/floodwall responses along the

navigation channel and influence of MRGO during Katrina. Powerpoint
presentation 188 slides.

FitzGerald, D, et al (2008). Impact of the Mississippi River Gulf Outlet (MR-GO):

Geomorphology & Geology, Technical Report to Katrina Canal Breaches
Consolidated Litigation Pertaining to MR-GO, Robinson [No. 06-2268],
Environmental Science Services, Inc., Denham Springs, LA.

199
Freeman, G.E., Rahmeyer, W.J., and Copeland, R.R. (2000). Determination of Resistance
Due to Shrubs and Woody Vegetation. USACE, Engineer Research and
Development Center, Coastal and Hydraulics Laboratory, ERDC/CHL TR-00-25,
Vicksburg, MS.

Gautier, C., Kok, M., and Vrijling, J.K. (2008). Wave Modeling New Orleans –
Mississippi River Gulf Outlet, Hurricane Katrina August 2005, Technical Report
to Katrina Canal Breaches Consolidated Litigation Pertaining to MR-GO,
Robinson [No. 06-2268], Delft University of Technology and SVASEK
Hydraulics, Delft, The Netherlands, July.

Gunn, R.L. (1997) Project fact Sheet, Mississippi River Gulf Outlet, Louisiana, US Corps
of Engineers, New Orleans District,
(www.mvn.usace.army.mil/ops/fact_sht/mrgo.htm)

Independent Levee Investigation Team (ILIT) (2006). Investigation of the Performance

of the New Orleans Flood Protection Systems in Hurricane Katrina on August 29,
2005. Final Report to National Science Foundation, University of California
Berkeley, Berkeley Ca.

Interagency Performance Evaluation Task Force (IPET) (2007). Performance Evaluation

of the New Orleans and Southeast Louisiana Hurricane flood protection System.
USACE, Washington DC.

Jarvela, J. (2004). Flow Resistance in Environmental Channels: Focus on Vegetation,

Helsinki University of Technology Water Resources Publication TKK-VTR-10,
Helsinki, Finland.

Kahan, J.P.,Wu, M., Hajiamiri, S., and Knopman, D. (2006). From Flood Control to
Integrated Water Resource Management, Lessons for the Gulf Coast from
Flooding in Other Places in the Last Sixty Years, Gulf States Policy Institute,
Rand Corporation.

Kemp, G.P. (2007). Declaration, Case 1:05-cv-01119-SGB, Document 35, Filed

01/27/2007.

Koelewijn, A.R., and Sellmeijer, J.B. (2006). “Beyond the Prediction of Critical Flood
Levels: Using Artificial Neural Networks for Failure Mechanisms Below Crest
Level,” 7th International Conference on Hydroinfomatics, Nice, France.

Knutson, P. L., Brochu, R.A, Seelig, W.N and Inskeep, N. (1982),. “Wave damping in
Spartina alternifloria marshes.” Wetlands, Journal of the Society of Wetlands
Science, (2), 87-104.

Knutson, P.L (1988). “Role of Coastal Marshes in Energy Dissipation and Shore
Protection, Chapter 13, The Ecology and Management of Wetlands, Vol 1, Timber
Press, Portland, OR.

200
Kobayashi, N., Raichle, A.W., and Asano, T. (1993). “Wave Attenuation by Vegetation,”
Journal of Waterway, Port, Coastal, and Ocean Engineering, ASCE, 119, 1, 320-
48.

Kok, M., Aalberts, M., Maaskant, B., and Wit, L. de (2007). Polder Flood Simulations
for Greater New Orleans Hurricane Katrina August 2005, Report by Delft
University of Technology and SVASEK Hydraulics, Delft, The Netherlands, July.

Kok, M., Aalberts, M., Kanning, W., Maaskant, B, and Wit, L. de (2008). Polder Flood
Simulations for Greater New Orleans: the neutral MR-GO scenario, Report to
MR-GO Litigation Group, Report by Delft University of Technology and
SVASEK Hydraulics, Delft, The Netherlands, July.

Mashriqui, H.S., Kemp, G.P., van Heerden, I., Ropers-Huilman, B.D., Hyfield, E. Young,
Y., Streva, K. and Binselam (2006). “Experimental Storm Surge Simulations for
Hurricane Katrina, Coastal Hydrology and Water Quality, Xu, Y.J. and Singh,
V.J. (Eds), Proceedings of the AIH 25th Anniversary Meeting & International
Conference, American Institute of Hydrology, May, 481-489, Baton Rouge,
Louisiana.

Massel, S.R., Furukawa, K., and Brinkman, R.M. (1999). “Surface wave propagation in
mangrove forests,” Journal of Fluid Dynamics Research, Elsevier Science B.V.,
24, 219-249.

Meer, J.W. van der and Breteler, M. K. (1990). “Measurement and Computation of Wave
Induced Velocities on a Smooth Slope,” 22nd International Conference on Coastal
Engineering, Delft, The Netherlands.

Meer, J.W. van der and Segffert, J.W. (1994). “Erosion and Overtopping of a Grass
Dike,” Proceedings of the International Conference on Coastal Engineering,
ASCE, 2639-2652.

Moller, I., et al (1999). “Wave Transformation Over Salt Marshes: A Field and
Numerical Modelling Study from North Norfolk, England.” Estuarine, Coastal
and Shelf Science , 49, 411-426.

Morris, C. (2008). Survey and Spatial Data in the Vicinity of the Mississippi River Gulf
Outlet (MRGO), Expert Report Prepared for Plaintiffs in Robinson v. United
States, Cmor Consulting LLC, Baton Rouge, LA.

MRGO Litigation Group (2008). MR-GO Bibliogaphy – Chronology, Report to MR-GO

Litigation Group, July.

Penland, S., et al (2008). Impact of the Mississippi River Gulf Outlet (MR-GO):
Geomorphology & Geology, Technical Report to Katrina Canal Breaches
Consolidated Litigation Pertaining to MR-GO, Robinson [No. 06-2268],
Environmental Science Services, Inc., Denham Springs, LA.

201
Reid, R.O. and Whitaker, R.E. (1976). “Wind-Driven Flow of Water Influenced by a
Canopy,” Journal of Waterways, Harbors, and Coastal Engineering Division,
ASCE, WW1, 61-77.

Schuttrumpf, H. (2007). Wave Overtopping Flow on Sea dikes – Experimental and

Theoretical Investigations, Delft Technical University, Delft, The Netherlands.

Seijffert, J.W. (2007). Grass Covers and Reiforcement Measures, Rijswaterstaat,

Hydraulic Engineering Division, Delft, The Netherlands.

Select Bipartisan Committee to Investigate the Preparation for and Response to Hurricane
Katrina (2006). A Failure of Initiative. U.S. Government Printing Office,
Washington DC.

Shaffer, G (2008). Geometric characteristics of marsh and swamp vegetation in south

Louisiana based on field measurements, personal communications, Southeastern
Louisiana University.

Sills, G.L, et al (2008). Overview of New Orleans Levee Failures: Lessons Learned and
Their Impact on National Levee Design and Assessment,” Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, 134, 5, 556-565.

Smith, J. M. (2007). Modeling Nearshore Waves for Hurricane Katrina, USACE, ERDC
TN-SWWRP-07-6, ERDC/CHL CHETN-I-76.

U.S. Army Corps of Engineers (USACE) (1988). Coastal Engineering Manual, EM

1110-2-1100, Vicksburg, Mississippi.

USACE (2007) Interagency Performance Evaluation Task Force (IPET) (2007).

Performance Evaluation of the New Orleans and Southeast Louisiana Hurricane
flood protection System. USACE, Washington DC.

USACE (2008a). Integrated Final Report to Congress and Legislative Environmental

Impact Statement for the Mississippi River – Gulf Outlet Deep-Draft De-
authorization Study, USACE New Orleans District, June.

USACE (2008b). Louisiana Coastal Protection and Restoration Technical Report,

UsACE New Orleans District, Mississippi Valley Division, February.

URS Corporation (2007). Interaction of Hurricanes and Coastal Landscape Features: A

Literature Review, Final Report, USACE Vicksburg District, Engineer Research
Development Center, Coastal and Hydraulics Laboratory, URS Corp, Baton
Rouge, LA.

van Heerden, I.L., Kemp, G.P., and Mashriqui, H. (2007). “Hurricane realities, models,
levees and wetlands,” Cities of the Future Towards Integrated Sustainable Water
and Landscape Management, Novotny and P. Brown (Eds), IWA Publishing,
London, UK.

202
Van Heerden, I.L. and G.P. Kemp. 2006. The Failure of the New Orleans Levee
System During Hurricane Katrina. Loyola L. Rev. 52:1225-45.

Van Heerden, I.L. and M. Bryan, 2006. The Storm—What Went Wrong and Why
during Hurricane Katrina—The Inside Story from One Louisiana Scientist. New
York, NY: Penguin/Viking.

Verheij, H.J., et al (1997). Erosion Resistance of Grassland as Dike Covering, Technical

Report, TAW – Technical Advisory Committee for Flood Defence in the
Netherlands.

Vriend, H.J. de and Barends, F.B.J. (2006). “Scour and erosion: common ground between
hydraulics and geotechnics,” Delft Hydraulics and Delft University of
Technology, Netherlands.

Wit, L. de, Maaskant, B, Kok, M., and Vrijling, J.K. (2008). Flow Modeling New Orleans
– Mississippi River Gulf Outlet, Hurricane Katrina August 2005, Technical
Report to Katrina Canal Breaches Consolidated Litigation Pertaining to MR-GO,
Robinson [No. 06-2268], Delft University of Technology and SVASEK
Hydraulics, Delft, The Netherlands, July.

Woolley D. and Shabman, L. (2008). Decision-Making Chronology For the Lake

Pontchartrain & Vicinity Hurricane flood protection Project, Final Report for the
Headquarters, U.S.Army Corps of Engineers, Institute for Water Resources of the
U.S. Army Corps of Engineers, Washington, DC.

Wu, F-C, Shen, H.W., and Chou, Y-J (1999). “Variation of Roughness Coefficients for
Unsubmerged and Submerged Vegetation,” Journal of Hydraulic Engineering,
ASCE, 125, 9, 934-940.

US Corps of Engineers, New Orleans District (1999) Habitat Impacts of the Construction
of theMRGO, prepared for the Environmental Subcommittee of the Technical
Committee Convened byEPA in Response to St. Bernard Parish Council
Resolution 12-98.

203
 APPENDIX A

QUALIFICATIONS

204
 APPENDIX A

QUALIFICATIONS OF DR. G. PAUL KEMP

I, G. Paul Kemp, declare as follows:

1. I am a coastal geologist and oceanographer living in Baton Rouge,

Louisiana. I have been employed since February 2007 as Vice-President of the Gulf

Coast Initiative of the National Audubon Society, a 100 year old conservation

organization headquartered in New York City. Prior to holding this position, I had been

an Associate Professor, Research, and Director of the Natural Systems Modeling Group

(NSMG) at the Louisiana State University (LSU) School of the Coast and Environment

for twelve years. I received a B.S. degree in 1975 from Cornell University in Ithaca,

New York, and I was awarded M.S. (1978) and Ph.D. (1986) degrees from the LSU

Department of Marine Sciences, now the Department of Oceanography and Coastal

Sciences, and from the Coastal Studies Institute at Louisiana State University. The

opinions given here are my own and should not be construed as positions held by LSU or

the National Audubon Society.

2. I submit this declaration on behalf of Plaintiffs Norman Robinson, Kent

Lattimore, Lattimore & Associates, Tanya Smith, Anthony Franz, Jr., and Lucille Franz

in Robinson v. United States, Case No. Civ. 06-2268 (SRD). I reserve the right to amend

and supplement my Declaration when and if additional information is acquired as a result

of the production of documents by the U.S. Army Corps of Engineers (“USACE”).

3. This Declaration is submitted to address: (a) the effect of the design,

construction, and maintenance of the Mississippi River Gulf Outlet (“MR-GO”) on the

storm surge and waves that occurred during Hurricane Katrina (“Katrina”); and (b) the

205
effect of the MR-GO on the ability of the surrounding wetlands and marshes to serve as

protection against storm surges during Katrina. I have devoted much of my professional

career in the past several years to considering issues of storm surges and environmental

impact caused and/or exacerbated by the MR-GO. In particular, I have conducted

extensive research with respect to the effects of the MR-GO on the storm surges and

devastation during Hurricanes Katrina and Rita.

4. This declaration, my fourth provided as a member of the Robinson expert

team is divided into several chapters as discussed in the main text. This Appendix is

devoted solely to establishing my qualifications to serve as an expert witness with respect

to oceanography, hydrology and, more specifically, the interpretation of information

developed by a numerical modeling program conducted under my direction. I also

gained substantial insight into the chronology of decisions made by the U.S. Army Corps

of Engineers (USACE) regarding the MR-GO project in the course of my previous

engagement by the State of Louisiana as a member of the forensics investigation

commissioned by the Secretary of the Louisiana Department of Transportation and

Development (Team Louisiana 2007).

I. QUALIFICATIONS

5. My areas of expertise include hydrodynamic modeling of high-intensity,

short-term events, such as hurricanes, and low-intensity long-term phenomena, such as

delta formation, river sediment transport and deposition, coastal wetland formation and

deterioration, coastal erosion processes, ecosystem modeling and natural

resources/science policy. While I was still an academic, I provided consulting services in

the past year, as I had in previous years, to the Morgan City Port Harbor and Terminal

206
District on navigation channel dredging issues and to the Lafourche Parish Government

on mitigating flood threats. I also served over the past year as an expert in hydrology and

sedimentology on behalf of the Louisiana Department of Transportation and

Development in a lawsuit in state court in St. Charles Parish, and in Federal Court in

Lafayette, Louisiana, in a case involving the navigability of Rycade Lake in the

Atchafalaya Basin. I have previously worked as an expert for the Louisiana State Land

Office on several cases dealing with ownership of riparian lands along the Mississippi

River and smaller streams in Louisiana, and with the proper location of boundaries

between states and parishes where a lake or an alluvial stream is involved. Attached

hereto as Exhibit A is a true and correct copy of my current curriculum vitae.

6. Before and after Hurricane Katrina, I conducted research on the storm

surge impacts of the funnel formed by the convergence of the earthen structures that

follow the south bank of the MR-GO and the north bank of the Gulf Intracoastal

Waterway (GIWW) for various state agencies including the United States Department of

Homeland Security (FEMA Hurricane Pam exercise), the Louisiana Department of

Transportation and Development and the Louisiana Attorney General’s Office. Some of

this work has been published in the peer-reviewed literature listed as Exhibit B.

7. Since Katrina, I have reviewed all of the primary investigations and

studies that have documented this failure (USACE Interagency Performance Evaluation

Task Force, Performance Evaluation of the New Orleans and Southeast Louisiana

Hurricane Protection System, 2007; Independent Levee Investigation Team,

Investigation of the Performance of the New Orleans Flood Protection Systems in

Hurricane Katrina on August 29, 2005 (http://www.ce.berkeley.edu/~new_orleans/,

207
November 2005 Preliminary Report, May 2006 Draft Final Report, July 2006 Final

Report); D. Woolley and L. Shabman, Decision-Making Chronology For the Lake

Pontchartrain & Vicinity Hurricane Protection Project, Report to Institute for Water

Resources of the U.S. Army Corps of Engineers, June 2007; American Society of Civil

Engineers, The New Orleans Hurricane Protection System: What Went Wrong and Why,

2007; National Academy of Engineering and National Research Council, Committee on

New Orleans Regional Hurricane Protection Projects, Third Report, 2006; National

Institute of Standards and Technology, Performance of Physical Structures in Hurricane

Katrina and Hurricane Rita: A Reconnaissance Report, 2006; Federal Emergency

Management Agency, Hurricane Katrina in the Gulf Coast, 2006; Committee on

Homeland Security and Governmental Affairs, U.S. Senate, Hurricane Katrina A

National Still Unprepared, 2006; Committee to Investigation the Preparation for and

Response to Hurricane Katrina, U.S. House of Representatives, A Failure of Initiative,

Feb. 2006).

8. I was a member of the Louisiana Department of Transportation and

Development forensics team (Team Louisiana) that conducted a study of the failure of

the Greater New Orleans Hurricane Protection System (HPS) associated with passage of

Hurricane Katrina. During 2005 and 2006, I made numerous inspections of the MR-GO

and associated features (from the land, water, and air).

9. I co-authored and served as primary draftsman through the LSU Hurricane

Center of the report titled The Failure of the New Orleans Levee System during

Hurricane Katrina, that was submitted to the Louisiana Department of Transportation

and Development (LDOTD) in late 2006, and was approved by the LDOTD Secretary for

208
distribution on March 16, 2007, after extensive review by the state agencies that comprise

the Louisiana Coastal Protection and Restoration Authority, by the Louisiana Attorney

General’s Office and by the Governor’s Office

(http://www.dotd.louisiana.gov/administration/teamlouisiana/). IPET addressed some

aspects of the storm hydraulics associated with a single portion of the MR-GO (Reach 2),

and ILIT has provided important information on the materials found to have been

incorporated in the earthen berm/spoil banks (EBSBs) that washed away adjacent to this

reach. The Team Louisiana report is unique, however, among all of the forensic studies

that have been published to date, in its comprehensive treatment of the concentrating

effect of the funnel, of the differing roles played by each reach of the MR-GO in

contributing to the catastrophe, and, importantly, of the decision-making that led to the

unfortunate series of flooding events that accompanied Hurricane Katrina. I have

authored with others nine professional conference and journal papers that have

documented this forensic work. I have, when requested, provided staff members of

Congressional and Senate committees and a Department of Defense Task Force

investigating the Katrina related failures in New Orleans with information about

Mississippi River-Gulf Outlet (MR-GO) contributions to that failure.

Storm Surge Modeling

10. At the time Hurricane Katrina struck, I had been an LSU research faculty

member for eleven years, achieving the academic rank of Associate Professor, and had

spent much of that time developing capacity within the State of Louisiana for adapting

standard hydrodynamic models used by the engineering community (1) to meet the needs

of ecologists and government agencies developing plans and projects to offset land loss

209
in the wetlands of coastal Louisiana, and (2) to pilot innovative design approaches for

wetland restoration projects using Mississippi River diversions. At that time, relatively

few numerical computer models had a reliable capability for handling the wetting and

drying that is a major feature of low-lying intertidal wetland landscapes. So I was excited

by the possibilities for both hurricane simulation and restoration project design offered by

the wetting and drying capabilities of the Advanced Circulation (ADCIRC) computer

model that had been developed over more than a decade by Joannes Westerink

(University of Notre Dame) and Rick Luettich (University of North Carolina). One of

my responsibilities since 2003 was to research the potential of applying the ADCIRC

storm surge model as a disaster management tool by providing near real-time early

warning forecasting of the storm surge threat associated with hurricanes approaching the

coast of the United States from the direction of the Gulf of Mexico.

11. I directed a small team, including Dr. Hassan Mashriqui, P.E., a

hydrologic engineer, and various graduate students and research associates, both at the

LSU Hurricane Center and at the LSU Center for Computational Technology (CCT),

which housed the supercomputer assigned for the modeling work. Our team worked very

closely with Dr. Joannes Westerink, one of the ADCIRC developers, and his graduate

students at Notre Dame. At this time, both Dr. Westerink’s team and mine worked under

contracts with the LSU Center for Study of the Public Health Impacts of Hurricanes

(CSPHIH), a sister unit of the LSU Hurricane Center, to carry out hurricane modeling

tasks assigned by Dr. Ivor Ll. van Heerden, Deputy Director of the Hurricane Center and

Director of CSPHIH. After several years of ADCIRC testing that had started before I

became involved, the LSU Hurricane Center was called upon by the Louisiana Office of

210
Emergency Preparedness to provide experimental real-time surge forecasting support

during the 2005 hurricane season. This request came after ADCIRC had gained

significant exposure and credibility with both local and national emergency managers as

a result of our modeling and scripting of the New Orleans flooding sequence caused by

hypothetical ‘Hurricane Pam’ in an exercise in July, 2004. The Hurricane Pam exercise,

coordinated by the United States Federal Emergency Management Agency (FEMA), was

intended to test hurricane forecasting, preparedness and response, and was the subject of

testimony before the Senate Homeland Security and Governmental Affairs Committee in

January 2006. The physical attributes of the fictional Hurricane Pam were developed by

the National Weather Service. Our Hurricane Pam surge simulations can be viewed at

http://hurricane.lsu.edu/floodprediction/PAM_Exercise04.

12. Hurricane Pam was characterized as a Category 3 storm created to follow

a track similar to that taken by Hurricane Betsy in 1965. ADCIRC simulated major

overtopping of levees and floodwalls on both the east and west banks of the City and

starkly demonstrated to me and many other observers the threat posed by the Lake

Borgne funnel geometry created by the convergence of the Mississippi River-Gulf Outlet

(“MR-GO”) and the Gulf Intracoastal Waterway (GIWW) channels and the artificial

embankments that paralleled these channels on the south and north sides, respectively

(Figure 1). As a result of my participation in the Hurricane Pam exercise, I also became

aware of the lack of practical plans for quickly dewatering the City in the event of

overtopping or breaching.

Storm Surge Forecasting for Hurricane Katrina

211
 13. 2004 and 2005 were active hurricane seasons. In 2004, my group

prepared computer models for four storms that threatened the Greater New Orleans area,

in addition to the Hurricane Pam exercise. In 2005, we modeled six storms, beginning

with Tropical Storm Arlene in early June, and including three hurricanes that ended up

making landfall in Louisiana, starting with Cindy in July and ending with Rita in

September. All of these computer model runs were posted on the LSU Hurricane Center

internet website as we completed post-processing, and they remain available there today

without subsequent modification (www.hurricane.lsu.edu/floodprediction).

Throat
The Funnel

Figure A-1. Modified from Interagency Performance Evaluation Task Force, Army Corps
of Engineers, Performance Evaluation of the New Orleans and Southeast Louisiana
Hurricane Protection System, at IV-134, Figure 93 (June 2006) to show the Lake Borgne
‘funnel’ and the ‘throat’ constituted by MR-GO Reach 1.

212
 14. The computer model runs of most interest for this report were the six

made for Hurricane Katrina that were first posted to the internet on Saturday, August 27,

2005 and continued to be posted up to landfall, and concluded with a “hind cast” posted

on September 13, 2005. A hind cast is computer model output based on empirical data

collected by the National Hurricane Center (NHC) after passage of Hurricane Katrina.

During all of this time, Dr. Westerink provided essential advice and encouragement as

our small team endured several sleepless nights. Our modeling team left LSU at 4 AM

Monday morning, about two hours before landfall, and I was startled to have to drive

around several downed trees as the winds picked up in Baton Rouge. The power was

already out in many parts of Baton Rouge at that time, including at my home. So it

would remain for several days as we struggled to get credible information about what had

happened in New Orleans. Our model runs had predicted serious overtopping two days

before the storm hit, and one of these forecast maps had been published on the front page

of the Times-Picayune newspaper on Sunday morning, even though the latest forecast

213
that we modeled had suggested some weakening (Figure 2).

Figure A-2. Graphic that appeared in the Sunday morning (August 28, 2005) edition of
the New Orleans Times-Picayune newspaper, showing LSU predictions of overtopping
on the east side of the city (courtesy of the Times-Picayune).

Post Landfall Investigations

15. Initial reports were heartening and indicated that the winds and rainfall

over the City were not extreme, but it later became clear that the overtopping flooding

that had given us chills as we watched our modeling results unfold on our computer

screens on Saturday and Sunday had been compounded by water flowing through

breaches.

16. It was not until Friday, September 2, that I had a chance to see the

aftermath first-hand during an over flight with the Civil Air Patrol. I was unable to enter

214
the City to ascertain volume of flooding until the following Wednesday when police

escorts could be arranged. Rescue efforts were still in progress, but the breaches had

been shored up to some degree while Lake Pontchartrain had returned to a normal stage

and some pumps in the Orleans Metro area had been restarted. Work crews had also

deliberately breached the levees1 around St. Bernard Parish to let water out. After that

we began to use LSU boats and surey equipment to more systematically probe outside the

perimeters of the alleged “Greater New Orleans Hurricane Protection System” (“HPS”)

to find high water marks that were sheltered from wave action. We were seeking to

validate our model results. Initially using a laser level and height above the still water

elevation, we were able to compare surge elevations to the crowns of the protection

structures. It was clear to us within a few days that our model simulations were

extremely accurate, particularly along the Lake Pontchartrain shoreline. We determined

that the 17th Street and London Avenue outfall canal floodwalls had not been overtopped,

but had failed at surge elevations lower than the crest of those structures. At that time,

the official position of U.S. Army Corps of Engineers was that these floodwalls had been

overtopped (van Heerden and Bryan 2006). Reporters were interested in what we were

finding as so little authoritative information was available, and began to accompany us as

we conducted our surveys, often making it easier for us to gain access to critical

locations. On September 19 and 20, Michael Grunwald of the Washington Post went

with us and we showed him what we had found on the 17th Street, London Avenue and

Inner Harbor Navigation Canals (IHNC). This led to a major story on September 21 that

contrasted our findings with the official USACE position, and the apparent discrepancy

1
By use of the word “levee” or “levees,” I am not expressing an opinion about whether the man-made
structures that were overtopped or breached in Greater New Orleans were technically a levee or levees as
defined by the USACE or generally-accepted engineering principles.

215
led to a frenzy of interviews and stories that quickly changed public perceptions of the

causes of the disaster. This story also stimulated the formation of other investigative

teams organized by the American Society of Civil Engineers (ASCE), by the National

Science Foundation (Independent Levee Investigation Team) and by the Corps itself

through what later became known as the Interagency Performance Evaluation Team

(IPET). These reports also prompted the State of Louisiana to fund its own forensics

team that became know later as Team Louisiana (van Heerden et al. 2007).

17. Hurricane Rita struck southwestern Louisiana on September 24, 2005 and

after we had finished our last computer run, I rushed to New Orleans to see Gentilly and

the Lower 9th Ward flood again through many of the same breaches along the Inner

Harbor Navigation Canal (IHNC) that were opened by Hurricane Katrina. When I took a

survey team to New Iberia the next day, water was still high and many roads remained

impassable. I left Mr. Dane Dartez, a research associate, to continue documenting high

water for the Rita surge over the next week, leading to our understanding that while

ADCIRC had done well for Katrina, the model had over predicted surge, in general, in

inland areas protected by miles of intact wetlands. It was also possible to discern the

effect south of Lake Charles of the preferential conveyance of surge by the Calcasieu

Ship Channel, like the MR-GO. These observations gave me a personal understanding of

the well-established, but largely unquantified, ameliorative effects of healthy wetlands as

buffers against storm surge, and its inverse, the negative impacts of large artificial

channels running inland from the coast.

18. I returned to New Orleans with the intent to more accurately and

permanently document surge levels, and to arrange for confirmation by oceanographers

216
from the USACE Engineering Research and Development Center (ERDC), formerly the

Waterways Experiment Station in Vicksburg, Mississippi, before the high water marks

were lost. This began about a month of extensive work with Dr. Steven Maynord, and a

number of his colleagues from ERDC, in which we painstakingly located high water

marks and debris lines throughout Greater New Orleans, and marked them for survey by

teams of U.S. Army Corps of Engineers (USACE) contracted surveyors. We took great

pains to try to discriminate wave-free surge marks (pure surge) that were susceptible of

prediction using ADCIRC from those that were inflated by the effects of wave breaking

and run-up, which was not predicted by the S08 ADCIRC model then in use. Generally,

we used LSU boats and trucks because the USACE had little equipment available for this

purpose.

19. On October 1, I believe, while this work was in progress, I was contacted

by a colleague at the USACE New Orleans District to attend a meeting at the District

headquarters. This was where I met the advance contingent of the ASCE and

NSF/Berkeley (later ILIT) teams that had been invited to investigate engineering causes

of the catastrophic flooding. The composition of these groups was heavily weighted

toward geotechnical engineers. I welcomed them to New Orleans and arranged for Drs.

Mashriqui and Sharma from LSU to join them in their field investigations and to provide

them with ADCIRC model output while I continued the high water mark work. Our

paths crossed on several occasions during the next week or so, and a number of these

team members told me how valuable Dr. Mashriqui’s presentation of ADCIRC output

had been in helping them to reconstruct the details of the surge sequence.

217
 20. By the time the other teams left New Orleans in mid-October, they had

confirmed most of our initial findings with respect to the outfall canal failures, and had

added a great deal of understanding with respect to the precise mechanisms. By this

time, however, I was more intently focused on oceanography associated with the

destruction of floodwalls along the IHNC and the EBSBs that had been erected along the

MR-GO and the GIWW on the eastern outskirts of St. Bernard and New Orleans East.

These areas had received less attention than the Orleans Metro floodwalls. Many of these

areas could be approached only with considerable difficulty by boat as we surveyed scour

holes and other indicators of the magnitude and direction of flow, noting the orientations

and apparent trajectories of stranded barges and other vessels (Figure 3). In the spoil

disposal area that had been created along the south bank of Reach 2 of the MR-GO as a

part of the original construction, dredged material that had been mounded into an

embankment was completely or partially eroded or washed away, leaving little evidence

other than vast aprons of sand and shell material from which to reconstruct what had

happened (Figures 4). Similar destruction was also observed on part of the northern

boundary of the funnel (Figure 1), along the GIWW (Figure 5). These structures are

henceforth referred to as “earthen berm/spoil banks” (EBSB) to contrast them with

engineered coastal protection levees or sea dykes found in other parts of the HPS,

particularly the Lake Pontchartrain frontage, and more generally elsewhere in the

developed world (See Dr. Bea’s declarations). Few structures were found in these

reaches from which reliable high water marks could be recovered, though wind and

wave-tossed debris was visible in trees and on surviving navigation towers.

218
Figure 3: Large eroded breach at the contact between the south end of the concrete
navigation lock structure at Bayou Bienvenue and the adjacent EBSB section (ILIT 2006)

Figure 4. Oblique view looking northeast across Chalmette EBSB toward MRGO
channel between the Bayou Bienvenue and Bayou Dupre floodgates that was
constructed of hydraulic fill from the channel (IPET 2006, V-13-60).

219
Figure 5. View looking west along New Orleans East EBSB along the north bank of the
GIWW that was constructed of hydraulic fill from the channel in stages over a 3 year
period (IPET 2006, V-13-57). The GIWW channel is to the left in this photograph.

21. We drove every part of the HPS that could be driven, including the non-

federal 40 Arpent Canal Levee that defines the northern edge of the populated area in St.

Bernard Parish.

22. We saw a tremendous range in the degree of overtopping that various

levees sustained, some with and others without damage, and noted ameliorative effects

wherever substantial wetland acreage protected structures from wave impacts (Figure 6).

220
Figure 6. Looking west on the Caernarvon to Verret Levee that protects the south side of
St. Bernard drained area east of Chalmette. This levee was overtopped for about 1.5
miles where the crown was between 12.5 to 13.0 ft high, but was protected from waves
by an extensive marsh to the left. It suffered little damage.

23. By mid November, despite the very dry weather, most high water marks

and debris lines were no longer easily discernible and the post-Katrina surge validation

effort was ended. Through this collaborative effort, however, we were able to gain a

reasonable understanding of actual surge levels outside the HPS. This information was

placed into a common database maintained by the USACE that became an invaluable

common source of information. As a result, few disputes have subsequently arisen

among the various investigative groups about surge elevations. In contrast, to this day,

no one actually knows the contribution of storm-driven waves anywhere, as the only

direct measurements were made in Lake Pontchartrain. IPET has tried to fill in this blank

221
with inferences derived from after-the-fact wave modeling, and some of this work

appears quite good, but it cannot be regarded as a full substitute for contemporaneous

measurements.

Formation of Team Louisiana

24. There were few disagreements among investigators working together to

recover evidence on the ground, but a problem arose during this work with obtaining

survey results from the USACE. It must be remembered that evidence of the storm

effects, particularly in the vicinity of breaches, were being obliterated every day by

efforts to repair the HPS. On October 14, my colleague from ERDC, Dr. Maynord, was

embarrassed to tell me that he had been informed by a senior New Orleans District

USACE employee that I needed to submit a formal request under the federal Freedom of

Information Act (FOIA) before he could send me the survey data from high water marks

that I had earlier identified for the USACE! These issues were resolved by Dr. Paul

Mlakar who was leading what later became the IPET team, but not before we were forced

for a short period of time to field a separate team of surveyors from the Louisiana

Department of Transportation and Development (LDOTD) to develop a back-up set of

values from around the 17th Street Canal that then gave us more confidence in what the

USACE contract surveyors were producing.

25. It was also around this time in mid-October that Secretary Bradberry of

the Louisiana Department of Transportation and Development (LDOTD) and the

Louisiana Attorney General’s Office (AG) approached Dr. Ivor van Heerden about

creating the state forensic team that was later known as “Team Louisiana.”

222
 26. In October, I had been invited to submit a proposal to formally begin to

work with the USACE IPET. This was attractive at the time because we had no

dedicated funding at this point for our post-storm work which was consuming vast

amounts of staff, truck and boat time. I was advised by the AG’s Office, however, not to

pursue this option as it could create a potential conflict of interest if the state later decided

to pursue legal action against the USACE. As a result, once Team Louisiana was

established, our cooperation with other groups was limited to the ILIT team, the only

other forensics investigators who were not funded by the USACE either as employees or

consultants. Dr. van Heerden told me also that efforts he had made to initiate a trade of

data between IPET and Team Louisiana were not successful. In November, 2005, the

LSU Hurricane Center was fortunate to receive a generous grant from the McKnight

Foundation that allowed us to begin to recoup expenses and purchase a new truck to

supplement the limited funding available from LDOTD. I sometimes accompanied

members of Team Louisiana and the ILIT as they conducted various investigations and

ended up being tasked by Dr. van Heerden to serve as the primary drafter of the Team

Louisiana report. I completed most writing in August, 2006, but the final version was not

officially released until March 16, 2007, after extensive review by personnel from a

number of state agencies, the AG’s Office, and the Governor’s Office

(http://www.dotd.louisiana.gov/administration/TeamLouisiana/BradberryLetter.pdf).

27. This was very much a team effort. Team Louisiana was comprised of a

diverse group of distinguished experts, including oceanographers, geologists,

geotechnical engineers, hydrologic engineers, and others. Drs. Mashriqui, van Heerden

and I were primarily responsible for the surge analysis and discussion of MR-GO impacts

223
and wetland effects that are contained in that report, some of which have also been

published elsewhere (See van Heerden and Bryan 2006, Day et al. 2007, for example).

Because of my responsibility to reduce the input of various specialists into a single

narrative, I had a unique vantage point from which to understand all of the many aspects

of engineering and policy that led to the many failures of flood protection structures.

28. After completing work with Team Louisiana, I was asked to join the

MRGO litigation group expert team and have continued to participate in investigations of

the failures as a paid consultant ($300/hr). This has given me the opportunity to pursue

questions that remain unanswered with a very high-caliber team of experts and modelers

from this country and The Netherlands during hours that I can spare from my new job. I

have now completed four expert reports or declarations in this case.

224
 G. PAUL KEMP
633 Magnolia Wood Ave.
Baton Rouge, Louisiana 70808
Office: 225-288-9876; Cell: 225-772-1426; Fax: 225-769-8732

EDUCATION
Ph.D. (Coastal Studies/Marine Sciences) Louisiana State University (LSU), 1986
M.S. (Marine Sciences) Louisiana State University, 1978
B.S. (Natural Resources) Cornell University, 1975

EMPLOYMENT
2007-Present, Vice President, Gulf Coast Initiative, National Audubon Society
Adj. Assoc. Prof., LSU Hurricane Center
1998-2006, Director-Natural Systems Modeling Laboratory & Hurricane Center,
LSU
1994-1998, Associate Professor, Research, School of the Coast and Environment,
LSU
1992-1994, Assistant Professor, Research, School of the Coast and Environment,
LSU
1989-1994, Executive Director/ Science and Technology Director, Coalition to
Restore Coastal Louisiana, Baton Rouge
1987-1991, Project Scientist and Director, Coastal Sciences Unit - Woodward-
Clyde Consultants, Baton Rouge
1987, Post-Doctoral Research Associate-Coastal Ecology Institute, LSU
1985-1987, Hydrogeologist/Project Manager, Groundwater Technology, Inc.,
Mandeville
1984, NOAA/Knauss Congressional Fellow-Office of Senator Edward
M.Kennedy, Washingron D.C.
1983, Geologist, Navarin Basin Project- Woodward-Clyde Oceaneering Inc.,
Dutch Harbor, AK and Houston, TX
1980-1983, Consulting geologist/Graduate Research Assistant- Department of
Marine Sciences, LSU

SELECTED PUBLICATIONS
Day, J.W., Jr., Boesch, D.F., Clairain, E.J., Kemp, G.P., Laska, S.B., Mitsch, W.J., Orth,
K., Mashriqui, H., Reed, D.J., Shabman, L., Simenstad, C.A., Streever, B.J.,
Twilley, R.R., Watson, C.C., Wells, J.T., and D.F. Whigham. 2007. Restoration
of the Mississippi Delta: Lessons from Hurricanes Katrina and Rita. Science
315(5819): 1679-1684.
Van Heerden, I., Kemp, G.P., and H. Mashriqui. 2007. Use of the ADCIRC storm
surge Model for Hurricane Katrina surge predictions and forensic studies IN
Olsen, H.W. (ed) Proc. Geo-Denver 2007, GeoInstitute of Am. Soc. Civ. Engrs,
February 18-21, 2007, Denver. GSP 161 Embankments, Dams and Slopes. 9 pp.
Van Heerden, I., Kemp, G.P. 2007. A conceptual hurricane surge protection plan for
coastal Louisiana IN Olsen, H.W. (ed) Proc. Geo-Denver 2007, GeoInstitute of

225
 Am. Soc. Civ. Engrs, February 18-21, 2007, Denver. GSP 161 Embankments,
Dams and Slopes. 11 pp.
G. Paul Kemp
Curriculum Vitae – Selected Publications
Page Two

Van Heerden, I.Ll., Kemp, G.P., Mashriqui, H., Sharma, R., Prochaska, W., Capozzoli,
L., Theis, A., Binsalem, A., Streva, K., and E. Boyd. 2006. The Failure of the
New Orleans Levee System during Hurricane Katrina. Final Team Louisiana
Forensics Report to Louisiana Department of Transportation and Development,
December, 2006. Baton Rouge, LA. 476 pp.
Day, J.W., Jr., Barras, J., Clairain, E., Johnston, J., Justic, D., Kemp, G.P., Ko, J.-Y.,
Lane, R., Mitsch, W.J., Steyer, G., Templet, P., and A. Yanez-Aranciba. 2005.
Implications of global climatic change and energy cost and availability for the
restoration of the Mississippi delta. Ecological Engineering 24: 253-265.
Reyes E., J.F. Martin, J.W. Day, G.P. Kemp, H. Mashriqui. 2004. River Forcing at
Work: Watershed modeling of prograding and regressive deltas. Wetlands
Ecology and Management 12(2) 103-114.
Lane, R., Kemp, G.P., Mashriqui, H.S., Day, J.W., and Hamilton, A., 2003. Potential
nitrate removal from a river diversion into a Mississippi delta forested wetland.
Ecological Engineering 20:237-249.
Martin, J., M. White, E. Reyes, P. Kemp, H. Mashriqui, and J. Day. 2000. Evaluation
of coastal management plans with a spatial model: Mississippi delta, Louisiana,
USA. Environmental Management. 25: 117-129.
Reyes, E., M. White, J. Martin, P. Kemp, J. Day, and V. Aravamuthan. 2000.
Landscape modeling of coastal habitat change in the Mississippi delta. Ecology.
81: 2331-2349.
Martin, J., E. Reyes, P. Kemp, H. Mashriqui, and J. Day. 2002. Landscape modeling of
the Mississippi delta. BioScience. 52: 357-365.
Lane, R., J. Day, G. Kemp, and B. Marx. 2002. Seasonal and spatial water quality
changes in the outflow plume of the Atchafalaya River, Louisiana, USA. 2001.
Estuaries. 25 (1): 30-42.
Are, D., Kemp, G.P., Guistina, G., Day, J.W., and F. Scarton. 2002. A portable,
electically-driven Dutch cone penetromenter for geotechnical measurements in
soft estuarine sediments. Journal of Coastal Research. 18:372-378.
Kemp, G.P., Day, J.W., Reed, D.J., Cahoon, D.R., and M.Wang. 1999. Sedimentation,
consolidation and surface elevation change in two salt marshes of the Mississippi
River deltaic plain: Geotechnical aspects of wetland loss. In Rozas, L.P., J.A.
Nyman, C.E. Proffitt, N.N. Rabalais, D.J. Reed, and R.E. Turner (eds). Proc.
Conf. Recent Research in Coastal Louisiana, February 2-5, 1998, Lafayette, LA.
P 15-34.
Suhayda, J.N., Kemp, G.P., Jones, R.S., and J. Peckman. 1991. Restoration of
wetlands using pipeline transported sediments, IN Coastal Depositional Systems
in the Gulf of Mexico, Proc. GCS/SEPM Found. 12th Ann. Res. Conf., Dec. 8-11,
Houston, TX.

226
G. Paul Kemp
Curriculum Vitae – Selected Publications
Page Three

Kemp, G.P., and J.T. Wells. 1987. Observations of shallow-water waves over a fluid
mud bottom: implications to sediment transport, In (N.C. Kraus, ed.) Coastal
Sediments ’87 Proc. Spec. Conf. Adv. In Understanding Coastal Sediment
Processes, New Orleans, LA., 12-14 May, 1987. Am. Soc. Civil Engrs., Vol 1:
363-378.
Kemp, G.P., Conner, W.H., and J.W. Day. 1985. Effects of flooding on decomposition
and nutrient cycling in a Louisiana swamp forest. Wetlands 5: 35-51.

227