Vous êtes sur la page 1sur 14

Classifying and Assessing the Geologic Contribution

to Rockfall Hazard
CHRISTOPHER J. VANDEWATER

WILLIAM M. DUNNE
Department of Earth & Planetary Sciences, University of Tennessee–Knoxville,
Knoxville, TN 37996

MATTHEW MAULDON
Department of Civil & Environmental Engineering,
Virginia Tech, 200 Patton Hall, Mail Code 0105, Blacksburg, VA 24061

ERIC C. DRUMM
Department of Civil & Environmental Engineering,
University of Tennessee–Knoxville, Knoxville, TN 37996

VANESSA BATEMAN
Geotechnical Engineering Section, Tennessee Department of Transportation,
6601 Centennial Blvd., Nashville, TN 37243

Key Terms: Rockfall Hazard, Rockfall Modes, Geo- Tennessee’s RHRS is more informative and permits
logic Character, Logistic Regression description of a wider spectrum of geologic conditions
than does the NHI version. Logistic regression analysis
indicates rockfall type correlates to lithologic varia-
tion and the number of discontinuity sets; and block
ABSTRACT size correlates to structurally controlled rockfall,
Rockfalls from roadcuts are a major hazard and lithologic variation, mechanical layer thickness, and
pose problems for transportation agencies across the number of discontinuity sets. Consequently, roadcuts
country. In the context of rockfall hazard manage- containing potential rockfall modes with two or more
ment, however, no consensus exists about the role of discontinuity sets, no lithologic variation, and me-
geology in assessing rockfall hazard. This study chanical thicknesses that exceed 1.0 m are expected to
investigates the geologic contribution to rockfall have greater Geologic Character scores.
hazard through application of rockfall hazard rating
systems to roadcuts in Tennessee and through INTRODUCTION
additional data collection to reveal correlations
between hazard characteristics and geologic attrib- Rockfall occurrences along roadcuts create consider-
utes. The geologic character of 80 roadcuts in central able risk for human injury and property damage, posing
and eastern Tennessee was evaluated using the problems for transportation agencies across the country.
Tennessee Rockfall Hazard Rating System (RHRS), Negative consequences of rockfall include impact damage
which is a revision of the National Highway Institute to pavement from falling rocks, rocks on roads posing
(NHI) RHRS. Scores for both RHRSs were compared hazards to motorists, road closures, and environmental
to evaluate whether the improved reproducibility of impact due to collisions with vehicles transporting toxic
scoring for the Tennessee RHRS yielded unintended substances (Royster, 1978; Moore, 1986; and Wyllie and
losses of scoring accuracy and sensitivity. Additional Norrish, 1996). Consequently, as the demand for rockfall
geologic attribute data beyond those used in the RHRS protection increases (Flatland, 1993), transportation
system were collected to determine with logistic agencies are expected to respond with practices that
regression analysis whether relationships among the minimize damage and increase driver safety.
geologic attributes, rockfall type, and block size exist. Rockfalls occur when rock or debris is shed from
Results indicate the revised geologic component of a roadcut or nearby steep slope by processes such as planar

Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154 141
Vandewater, Dunne, Mauldon, Drumm, and Bateman

sliding, wedge failure, toppling, differential weathering, has adopted the ODOT RHRS (Pierson and Van Vickle,
and raveling onto the catchment and/or road (Norrish and 1993), hereafter referred to as the NHI RHRS.
Wyllie, 1996). Characterization of rockfall hazard along The NHI RHRS employs a two-phase slope categori-
roadcuts is necessary for identifying hazard level and zation process (Pierson and Van Vickle, 1993). The first
prioritizing remediation activities. The characterization phase is a preliminary rating, where slopes are assigned
includes attributes such as vehicular traffic patterns, a rating of A, B, or C based first on the estimated potential
roadway geometry, and rock-slope geometry (Wyllie for rock to reach roadway and second on historical
and Norrish, 1996). However, in the context of rockfall rockfall activity. A-rated slopes are most hazardous and
hazard management, a variety of methods are used to are characterized with a detailed rating that considers the
characterize the role of geology in the rockfall process. following factors:
Developing an effective approach to characterizing
geology is important, because this factor and its relation  Slope height
to the roadway controls whether material is available to be  Roadway width
shed as a rockfall. Approaches to incorporating geology in  Ditch effectiveness
hazard assessment have included defining hazard by  Average vehicle risk
association with rock type (Anonymous, 1996; Hadjin,  Decision sight distance
2002), ascertaining whether geologic discontinuities are  Geologic character
oriented favorably or unfavorably with respect to pro-  Block size/volume of rockfall per event
moting rockfall (Abbott et al., 1998), and describing the  Climate/presence of water
rockfall process for rock and soil slopes (Lowell and  Rockfall history
Morin, 2000).
The goal of this study is to investigate the role of The factors affected by the geologic conditions at
geology through two approaches: (1) building on the a roadcut are Geologic Character and Block Size.
geologic component of the National Highway Institute Geologic Character identifies whether the rockfall type
(NHI) Rockfall Hazard Rating System (RHRS) by is controlled by the geologic structure or differential
modifying it to explicitly evaluate rockfall modes and erosion. Block Size is controlled by rock type, structural
their salient characteristics, which is hereafter referred to conditions such as joint length and spacing, and roadcut
as the Tennessee RHRS; and (2) collecting additional construction methods.
geologic attributes beyond the RHRS system to determine Since the development and implementation of the NHI
if these geologic attributes correlate with rockfall type and RHRS (Pierson et al., 1990; Pierson and Van Vickle,
block size. The correlation in the second approach 1993), more than 17 state and provincial agencies have
compares rockfall modes and attributes developed for adopted the RHRS for rockfall management. Most
the first approach with additional geological data that transportation agencies have approached roadcut geologic
would not typically be collected during roadcut charac- conditions using the RHRS without modification, but
terization for rockfall hazard rating. A purpose of the about seven have modified the RHRS, most notably
second approach is to provide transportation departments Colorado (Stover, 1992), Washington (Lowell and Morin,
and future investigators with correlations that could be 2000), New York (Anonymous, 1996), and Ontario,
used to consider likely rockfall modes and block sizes for Canada (Senior, 1999).
a prospective roadcut if the geologic unit and its properties The Colorado Department of Transportation (DOT)
are already known. incorporated Slope Inclination and Launching Features,
because their experience suggested that these factors
Role of Geology in Existing Rockfall Hazard significantly contributed to rockfall hazard (Stover, 1992).
Rating Systems The Washington State system distinguishes soil and rock
materials but does not explicitly incorporate rockfall
The RHRS is a tool to systematically inventory and modes, because the DOT wished to utilize raters who were
rank hazardous roadcuts. The system was originally neither geologists nor geotechnical engineers (Lowell and
developed by Pierson et al. (1990) for the Oregon Morin, 2000). The New York DOT modified the RHRS by
Department of Transportation (ODOT) in a study funded considering hazard associated with two rock categories,
by Oregon, nine other states, and the Federal Highway crystalline and sedimentary, based on their assumption
Administration. This effort was preceded by earlier work, that crystalline rocks tend to have structurally controlled
dating back to the 1970s, to develop systematic inventory rockfall, whereas sedimentary rocks tend to have rockfall
and ranking procedures for hazardous roadcuts (Fish and controlled by differential erosion (Anonymous, 1996).
Lane, 2002). The ODOT RHRS is based on a rock slope Ontario’s Ministry of Transportation (MTO) modified the
inventory and maintenance program developed by Wyllie RHRS to deal with rockfall modes common in Ontario
(1987). Since 1990, the Federal Highway Administration roadcuts. For example, in northern Ontario, raveling,

142 Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154
Geologic Contribution to Rockfall Hazard

Figure 1. Physiographic (A) and geologic maps (B) of central and eastern Tennessee showing location of investigated counties.

toppling, and ice-jacking are the dominant rockfall a statewide rockfall hazard inventory for the Tennessee
behaviors because of well-controlled blasting methods, DOT. The five counties were selected to cover a broad
pronounced physical weathering, and the fact that roadcut range of physiographic and geologic settings. Data from
relief is typically less than 8 m (Senior, 1999). Additional the roadcuts in these five counties were used to evaluate
parameters used by Ontario include height of the water the role of geology in rockfall hazard and to identify
table at the slope face and looseness of the face. specific geologic controls on rockfall type and block size.
Eighty roadcuts in eastern and central Tennessee were
evaluated with the Tennessee RHRS, and 77 of these 80
STUDY AREA cuts were subsequently investigated to examine the
Location and Physiography influence of geologic factors on the geologic attributes
of the rockfall hazard rating (Figure 1). Physiographically,
Rockfall hazard was evaluated in five counties of this region is composed of the Blue Ridge Province;
Tennessee during 2001–2002 in the first phase of Valley and Ridge; Cumberland Plateau; the Highland

Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154 143
Vandewater, Dunne, Mauldon, Drumm, and Bateman

Table 1. NHI Geologic Character rating scheme. Case 1 is for structurally related rockfall, and Case 2 is for erosion-related rockfall.

Values of Hazard Assessment

Geologic Characteristics Low Hazard ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ! High Hazard

Case 1
Structural Condition Discontinuous joints, Discontinuous joints, Discontinuous joints, Continuous joints, adverse
favorable orientation random orientation adverse orientation orientation
Rock Friction Rough, irregular Undulating Planar Clay infilling or slickensided
Case 2
Structural Condition Few differential Occasional differential Many differential Major differential
erosion features erosion features erosion features erosion features
Difference in Erosion Rates Small difference Moderate difference Large difference Extreme difference

Rim; and the Nashville Basin (Bingham and Helton, because lithologies have different weathering and erosion
1999) (Figure 1). characteristics, the presence of rock structures such as
folds and faults changes overall rock geometry, and the
abundance of discontinuities influences rock strength.
Geologic Setting
The Blue Ridge Province is underlain by mostly Early
Cambrian rifted margin sedimentary and volcanic rocks METHODOLOGY OF RHRS REVISION
that were deposited on Grenville basement. The Valley Geological Revisions to NHI RHRS
and Ridge Province consists of a Cambrian and Ordovi-
cian platform and Ordovician to Pennsylvanian synoro- The Geologic Character category in the NHI RHRS
genic sedimentary rocks that were folded and faulted (Pierson and Van Vickle, 1993) evaluates the geologic
during the Late Mississippian–Permian Alleghenian conditions contributing to rockfall hazard at a roadcut.
orogeny (Hatcher et al., 1989). Nearly flat-lying Devonian However, the NHI approach does not explicitly in-
to Pennsylvanian sedimentary rocks form the Cumberland corporate rockfall modes and uses ambiguous terminol-
Plateau, and these are moderately to deeply dissected, ogy, as described later in this section. Consequently, the
creating significant local relief. Adjacent to the Cumber- NHI RHRS produces scores that do not clearly in-
land Plateau to the west, but at lower elevation, is the corporate geologic conditions at a roadcut, and scores may
Highland Rim, containing Ordovician to Mississippian be operator dependent, because the RHRS terminology
sedimentary rocks that are moderately to deeply dissected. can be interpreted differently by different raters, limiting
The Nashville Basin is a topographic low consisting of reproducibility. Given these issues, we revised the NHI
a structural dome of Ordovician to Mississippian sedi- RHRS Geologic Character to explicitly incorporate
mentary rocks that gently dip away from the geologic apex rockfall modes, eliminate ambiguous jargon, and promote
(Hardeman, 1966; Bingham and Helton, 1999). reproducibility among different raters.
The five counties present contrasting geologic con- The Geologic Character category in the NHI RHRS
ditions. Lithologic variations range from crystalline rocks, (Pierson and Van Vickle, 1993) considers two cases
composed primarily of granite, orthogneiss and para- (Table 1). ‘‘Case 1’’ is for structurally controlled rockfall
gneiss, amphibolite, and gabbro, which occur in the Blue where key factors are discontinuity size, discontinuity
Ridge Province (parts of Carter County), to sedimentary orientation, and rock friction. ‘‘Case 2’’ is for differential
rocks in the other provinces, including mudstones, erosion rockfall, where the factors are differential erosion
siltstones, sandstones, and carbonates that occur in parts features and differential erosion rates. The cases are
of Carter, Anderson, Bledsoe, Grainger, and Smith mutually exclusive, because only the score for the rockfall
Counties (Figure 1). Additionally, structural variations condition with the greater hazard is recorded.
occur: horizontally bedded rocks in Smith, Bledsoe, and ‘‘Case 1’’ considers two factors: Structural Condition
Anderson Counties; moderately inclined and folded and Rock Friction (Table 1). Structural Condition at-
sedimentary rocks in Bledsoe, Anderson, Grainger, and tempts to describe the relative orientation and length of
Carter Counties; and igneous and foliated metamorphic joints in the roadcut (Pierson and Van Vickle, 1993).
rocks in Carter County. Accordingly, roadcuts in the study Discontinuous joints are defined as less than 3.3 m in
area contain a variety of lithologies, structural domains, length, whereas continuous joints are defined as greater
and discontinuity characteristics (i.e., spacing and persis- than 3.3 m in length. Rock Friction describes the surface
tence). This variety potentially influences rockfall modes, smoothness of the joints. Clay-filled and slickensided

144 Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154
Geologic Contribution to Rockfall Hazard

joints are assigned the highest hazard, because less shear Table 2. Rockfall modes and characteristics of the Tennessee Geologic
stress is required to exceed the coefficient of friction. Character rating scheme. X indicates inclusion of rated criteria for
a rockfall mode.
‘‘Case 2’’ considers two factors: Structural Condition
and Difference in Erosion Rates (Table 1). Structural Characteristics
Condition describes the surficial weathering features of
a roadcut, whereas Difference in Erosion Rates describes Rockfall Block Block
the formation rate of surficial weathering features (Pierson Modes Abundance Size Inclination Friction Relief Shape
and Van Vickle, 1993), as inferred from the amount of Planar X X X X
weathering relief in a slope face. Wedge X X X X
Examination of the scoring factors for the NHI Topple X X
Geologic Character category during development of the Differential
weathering X X X
Tennessee RHRS, as well as field tests for consistency and Raveling X X X
repeatability among several raters, highlighted the need
for an improved approach to resolve issues of assessment,
field application, and terminology. Issues included: the ratings where more than one rockfall mode is present;
and to eliminate ambiguous terminology.
(1) The NHI RHRS does not apply well-established
geotechnical or geologic terms (Turner and Schus- Tennessee Geologic Character Scoring System
ter, 1996) or rockfall modes (e.g., plane, wedge,
topple, etc.). The Tennessee RHRS considers five rockfall modes:
(2) The NHI RHRS does not assess the abundance or plane, wedge, topple, differential weathering, and ravel-
degree to which a potential rockfall mode is present ing, using appropriate combinations of six characteristics
in a rock slope. This factor often controls the (Tables 2 and 3). Characteristics common to all rockfall
volume of rock that may be shed to the road. modes are the relative Abundance of the rockfall mode
(3) Only the most hazardous condition is considered, and Block Size. The relative Abundance of a rockfall mode
whereas a better assessment of risk is to consider all is expressed as a percentage of the total slope face surface
rockfall modes in a roadcut that could deliver rock area containing the mode (Figure 2 and Table 3). In the
to the road. NHI RHRS, Block Size is treated separately from
(4) The system does not consider raveling explicitly, Geologic Characteristics, but it is incorporated into the
even though it may be a prominent rockfall type, as Tennessee RHRS because block size is primarily a
Ontario’s MTO (Senior, 1999) and New York function of geology.
(Anonymous, 1996; Hadjin, 2002) recognized. Characteristics unique to planar and wedge rockfall are
(5) Use of ‘‘structural condition’’ for both structural and Steepness of the failure plane(s) and wedge intersection,
nonstructural rockfall creates confusion. respectively, and the Roughness Profiles of the failure
(6) Use of ‘‘random’’ for intermediate orientation plane(s). As rockfall hazard increases with steepness,
hazard condition is problematic, because ‘‘random’’ frictional resistance, which is largely a function of
could include potentially very hazardous orienta- roughness, controls whether the rock mass fails. First-
tions, and the term does not encompass the case of order (planar or undulating) and second-order (rough or
parallel discontinuities as an intermediate-risk smooth) friction is evaluated in a profile parallel to the
orientation. likely movement direction of the rock mass (Figure 3).
(7) Use of ‘‘joints’’ is confusing to geologists, because First-order (macroscale) roughness is considered to have
bedding surfaces, faults, and cleavage may provide the greatest effect on shear strength, because substantial
discontinuities for structural cases. shear stress is required to overcome first-order asperities,
(8) Use of ‘‘continuous’’ and ‘‘discontinuous’’ to de- whereas slip over the second-order asperities requires only
scribe discontinuity persistence is confusing. localized shearing (Patton and Deere, 1970; Barton, 1973).
(9) Use of ‘‘favorable’’ and ‘‘adverse’’ for hazard The traditional topple rockfall mode requires disconti-
conditions is ambiguous. ‘‘Favorable’’ means favor- nuities dipping steeply into a slope face (Norrish and
able to stability, but it could be misinterpreted as Wyllie, 1996). However, the degree of discontinuity
favorably disposed to rockfall. steepness usually has little effect on this rotational failure
(10) Differential erosion is more accurately described by mechanism. Therefore, discontinuity steepness is not
the term ‘‘differential weathering.’’ considered in the Tennessee RHRS for the toppling mode.
Additionally, because interlayer slip is a very small
The NHI RHRS was revised for use in Tennessee to component of the topple resistance, friction is not
incorporate rockfall modes explicitly; to include raveling; considered in the Tennessee RHRS for the toppling mode
to use measurable attributes, including abundance; to sum (Table 3).

Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154 145
Vandewater, Dunne, Mauldon, Drumm, and Bateman

Table 3. Scoring schemes for the Tennessee RHRS Geologic Character.

Planar Rockfall Mode


Abundance ,10% 10–20% 20–30% .30%
Score 3 9 27 81
Block size ,1 ft (,0.3 m) 1 to 3 ft (0.3–0.9 m) 3 to 6 ft (0.9–1.8 m) .6 ft (.1.8m)
Score 3 9 27 81
Steepness 0–208 20–408 40–608 .608
Score 2 5 14 41
Friction (micro/macro) Rough/undulating Smooth/undulating Rough/planar Smooth/planar
Score 2 5 14 41
Wedge Rockfall Mode
Abundance ,10% 10–20% 20–30% .30%
Score 3 9 27 81
Block size ,1 ft (,0.3 m) 1 to 3 ft (0.3–0.9 m) 3 to 6 ft (0.9–1.8 m) .6 ft (.1.8 m)
Score 3 9 27 81
Steepness 0–208 20–408 40–608 .608
Score 2 5 14 41
Friction (micro/macro) Rough/undulating Smooth/undulating Rough/planar Smooth/planar
Score 2 5 14 41
Topple Rockfall Mode
Abundance ,10% 10–20% 20–30% .30%
Score 5 4 41 122
Block size ,1 ft (,0.3 m) 1 to 3 ft (0.3–0.9 m) 3 to 6 ft (0.9–1.8 m) .6 ft (.1.8 m)
Score 5 14 41 122
Differential Weathering Rockfall Mode
Abundance ,10% 10–20% 20–30% .30%
Score 3 9 27 81
Block size ,1 ft (,0.3 m) 1 to 3 ft (0.3–0.9 m) 3 to 6 ft (0.9–1.8 m) .6 ft (.1.8 m)
Score 3 9 27 81
Relief ,1 ft (,0.3 m) 1 to 3 ft (0.3–0.9 m) 3 to 6 ft (0.9–1.8 m) .6 ft (.1.8 m)
Score 3 9 27 81
Raveling Rockfall Mode
Abundance ,10% 10–20% 20–30% .30%
Score 3 9 27 81
Block size ,1 ft (,0.3 m) 1 to 3 ft (0.3–0.9 m) 3 to 6 ft (0.9–1.8 m) .6 ft (.1.8 m)
Score 3 9 27 81
Shape Tabular Blocky Round —
Score 3 9 27 —

In many parts of Tennessee, especially on the potential for spherical blocks than tabular or blocky
Cumberland Plateau, adjacent sedimentary beds of the blocks. Therefore, block shapes are described in order of
same lithology dip into the roadcut. This geometry increasing hazard as tabular, blocky, or spherical.
produces rockfall because the blocks fail in tension across The NHI RHRS uses point values equal to powers of 3
bedding, releasing from the bed above. The behavior to score characteristics (Pearson and Van Vickle, 1993). A
occurs without a difference in lithology between beds, and score of 31 ¼ 3 points is assigned for the lowest hazard
thus it is not the result of differential weathering, so a new category and 34 ¼ 81 points for the highest hazard
descriptive term is required. We refer to this failure mode category, with intermediate values of 32 ¼ 9 and 33 ¼ 27.
as ‘‘bedding plane release.’’ Although the mechanism is The premise behind exponential scoring is to give added
not classic topple failure, the failure is not governed by weight to high-hazard roadcuts so they will be readily
sliding, such as wedge or planar failure, but does involve identifiable in the database.
discontinuities dipping into the roadcut, and thus it is Following the NHI approach, each characteristic in the
included in the topple category in the Tennessee RHRS. Tennessee system was scored using four categories (Table
Amount of Relief is an attribute unique to differential 3). For Raveling, only three categories were used for the
weathering, and it is defined in this article as referring to block shape score, because it was felt that the use of
the extent that a rock mass overhangs the material directly a fourth category overemphasized the hazard for this
underneath (Tables 2 and 3). Block Shape is an important mode, which typically sheds only small blocks in
attribute for raveling, because of the greater runout Tennessee. With this reduction for Raveling, each of the

146 Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154
Geologic Contribution to Rockfall Hazard

Figure 2. Roadcuts showing different failure modes and example Abundance values for the failure modes. Abundance values reflect only the percentages
visible in the photograph. (A) Example of subhorizontal sedimentary rocks: differential weathering at the sandstone/coal contact with abundance of 10 to
20 percent, and bedding release at the cross-bed surfaces in the sandstones as toppling at .30 percent. (Percentages neglect portion of slope above bench,
which is not fully visible in the photograph. Vehicles and people for scale.) (B) Example of inclined sedimentary rocks: planar mode along inclined
bedding surfaces in the sandstone at .30 percent and raveling on the right in mudstone at ,10 percent. (C) Example of metamorphic rocks: raveling with
abundance .30 percent related to discontinuities created from cleavage surfaces, bedding surfaces, joints, and blasting fractures.

Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154 147
Vandewater, Dunne, Mauldon, Drumm, and Bateman

Figure 3. Visual scoring aid for friction. Terms indicate micro- and macrofriction profiles, respectively (modified from Barton, 1973).

other four rockfall modes can yield scores of up to 244 attributes. Logistic regression requires categorical values
(Table 3). For Toppling, which has two scoring character- (Table 4) for many geologic attributes (Hosmer and
istics, each characteristic has a maximum score of 122. Lemeshow, 1989). Slope orientation data were collected,
For Differential Weathering, with three scoring character- because it was expected that north-facing roadcuts could
istics, the maximum characteristic score is 81. For planar have high abundance of differential weathering rockfall.
and wedge rockfall, with four scoring characteristics, Rock type, fissility/cleavage, and geomechanical layer
Abundance and Block Size have maximum scores of 81, thickness could affect block size, and lithologic variation
but to maintain a total possible score of 244, Friction and could affect rockfall mode and abundance. Discontinuity
Steepness are limited to 41 points each. This difference geometric data were collected to investigate association
in maximum score between characteristics is justified on with rockfall mode and block size.
the basis that the Friction term is the most difficult to Most data values in Table 4 are self-explanatory, but
judge in the field (Barton, 1973), and so it was combined a few require comment. Layer thickness as defined in this
with Steepness to share 82 points. study is not actual bedding thickness in sedimentary rocks,
but rather geomechanical thickness between upper and
lower bounding discontinuities for a set of beds with the
METHODS FOR RELATING GEOLOGIC same geomechanical and weathering characteristics. In
ATTRIBUTES TO RHRS CHARACTERISTICS metamorphic rocks, well-defined foliation surfaces and
rock layer boundaries were used to define the geo-
Geologic Attributes mechanical thickness. A discontinuity set, as defined in
Slope geometry, lithology, and discontinuity data for this study, is typically (but is not limited to) systematic
134 potential rockfall modes at 77 roadcuts were fractures or geomechanical layer surfaces that have
collected. A statistical analysis using logistic regression parallel to subparallel orientation and measurable spacing.
was conducted to relate RHRS characteristics to geologic For analytical purposes, all data, including the number of
discontinuity sets, were collected with respect to each
Table 4. Geologic attributes collected for logistic regression.
identified rockfall mode and not necessarily the entire
roadcut.
Attribute Values

Slope orientation and failure mode information Logistic Regression: Method Overview
Slope aspect Slope trend þ 90
We chose to investigate the presence or absence of
Rockfall type Structural, nonstructural
Rockfall mode Planar, wedge, topple, differential correlations between RHRS characteristics and geologic
weathering, raveling attributes using logistical regression, because Tennessee
Lythology of potential rockfall RHRS data are categorical, and the method is well suited
Rock type Based on field description; clastic for analyzing categorical data (Hosmer and Lemeshow,
sedimentary, carbonate 1989; Allison, 1999; Stokes et al., 2000; and Ott and
sedimentary, crystalline Longnecker, 2001). Logistic regression is often used in
Lithologic variation Yes, no the social sciences (Cleary and Angel, 1984; Wang and
Fissility/cleavage Yes, no
Fitzhugh, 1995; and Studenmund, 1997) and is now being
Geomechanical layer thickness ,0.2 m, 0.2 m–0.5 m,
0.5 m–1.0 m, .1.0 m applied to fields such as landslide hazard assessment (Dai
and Lee, 2001; Apt et al., 2002), hydrology (Zain, 2001;
Discontinuity data
Bent and Archfield, 2002), and resource exploration
Number of discontinuity sets 0, 1, 2, .2
(Harris and Pan, 1999; Sahoo and Pandalai, 1999).

148 Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154
Geologic Contribution to Rockfall Hazard

In logistic regression, the dependent ‘‘outcome’’ vari-


able is restricted to two values, 1 and 0, which represent
the occurrence or nonoccurrence, respectively, of an
outcome event. The analysis determines if the outcome is
predicted by one or more independent ‘‘explanatory’’
variables and produces the coefficients of a prediction
model formula, with their standard errors of estimate, and
significance levels and odds ratios, with their 95 percent
confidence intervals (Allison, 1999; Stokes et al., 2000).
Odds are defined as the probability of the event divided by
the probability of the nonevent. An odds ratio is the ratio
of the odds of an outcome occurring, given a value for the
explanatory variable, compared to the odds of the same
outcome occurring for a given reference value of the
explanatory variable.
The logistic regression equation has the log-linear
form (Allison, 1999):

lnðp=½1  pÞ ¼ a þ BX þ errorB , ð1Þ


Figure 4. Graphical relationship between the log odds of Y, where py is
the probability that event Y occurs, and the independent variable, X, in
where p is the probability that ‘‘outcome’’ event Y occurs the logistic regression equation.
(yes ¼ 1), ln( p/[1  p]) is the natural-log odds, a is the
intercept coefficient, B is the coefficient of the in-
dependent ‘‘explanatory’’ variable X, and errorB is the
standard error of B (Figure 4). The method uses the lnðodds j X ¼ 1Þ  lnðodds j X ¼ 0Þ
 
independent variable, X, values and the probabilities of odds j X ¼ 1
the dependent variable, p, to calculate values for the ¼ ln ¼B ð2Þ
odds j X ¼ 0
coefficients by maximum-likelihood estimates, and co-
efficient significance is indicated by the significance where the vertical bar means ‘‘given that.’’ Taking both
probability value ( pw value) determined from the Wald sides of Eq. 2 as exponents of e yields the odds ratio,
statistic with significance level of .05 (Hosmer and
Lemeshow, 1989).
Evaluation of model parameters is accomplished by odds j X ¼ 1
considering: (1) the likelihood ratio (chi-square statistic) odds ratio ¼ ¼ eB ð3Þ
odds j X ¼ 0
to determine if the overall model is statistically significant
by testing the global null hypothesis that the coefficients
of all independent variables are equal to 0; (2) the
For example, if coefficient B ¼ 1.6, then the odds ratio
Hosmer-Lemeshow goodness-of-fit statistic to test the
(e1.6) equals 5, meaning that when the independent
global null hypothesis that the model fits the data,
variable, X, increases 1 unit, the odds that Y will equal 1
rejecting the hypothesis if the model does not fit the data;
increase by a factor of 5. Odds ratios of 1 indicate a 50/50
and (3) the c-statistic, which measures the logistic
chance the event will occur with a change in X. If the 95
equation discriminatory power. The c-statistic varies from
percent confidence interval for eB does not include the
0.5 (predictions no better than chance) to 1.0 (predictions
value 1.0, a change in X from 0 to 1 produces a statistically
are always correct) (Hosmer and Lemeshow, 1989).
significant change in the odds for Y.

Odds Ratios Data Models


The slope coefficient in logistic regression is inter- For this study, the software package SAS (Stokes et al.,
preted as the rate of change in the log-odds of Y as X 2000) was used to perform logistic regression, with the
changes (Figure 4). However, a more intuitive interpre- dependent variables being the two major factors in both
tation of the coefficient utilizes the odds ratio (Hosmer the Tennessee RHRS and the NHI RHRS: Rockfall Types
and Lemeshow, 1989; Allison, 1999; and Stokes et al., and Block Size (Table 5). The dependent variable
2000). The graph in Figure 4 shows that the coefficient, B, ROCKFALL is equal to 1 if rockfall type is nonstructural
is equal to the change in log odds as follows: and 0 if structural. Nonstructural rockfall types are

Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154 149
Vandewater, Dunne, Mauldon, Drumm, and Bateman

Table 5. Variables and interactions for logistic regression models.

Logistic Regression Rockfall Block Size


Model Model Model

Dependent ROCKFALL BLOCKSIZE


variable, Y
Modeled category Nonstructural Block size . 0.3 m
( p j Y ¼ 1)
Independent Slope aspect Rockfall type
variables, X
(geologic
attributes)
Rock type Geomechanical
layer thickness
Lithologic variation Lithologic variation
Number of Fissility/cleavage
discontinuity sets
Rock type
Number of
discontinuity sets

differential weathering and raveling modes, and structural


rockfall types are plane, wedge, and topple modes. The
dependent variable BLOCKSIZE is equal to 1 if block
length is greater than 0.3 m and 0 otherwise.
Independent geologic or physical variables were
selected to test for correlation with the dependent variable
of ROCKFALL type and BLOCKSIZE (Tables 4 and 5).

Figure 6. Number of rockfall modes at a roadcut (A) and rockfall


mode occurrence at roadcuts (B), showing that multiple modes
may be present at a roadcut.

Independent variables are categorical or ordinal, and one


value for a variable becomes the reference value (Hosmer
and Lemeshow, 1989; Kleinbaum et al., 1998; and Stokes
et al., 2000).

RESULTS
Comparison of Tennessee RHRS Scores to
NHI Scores for Geologic Character
Roadcuts were scored using both the Tennessee RHRS
and the NHI Geologic Character schemes (Figure 5).
Both systems yielded the same score at roadcuts where
Differential Weathering is the only mode. Yet, the
average Geologic Character score of 84 for the Tennessee
system is higher than the average NHI Geologic
Character/Block Size score of 66 (Figure 5). One reason
that the Tennessee system scores are higher is because
scores are cumulative for all potential rockfall modes at
Figure 5. Comparison of Tennessee Geologic Character scores with a roadcut, whereas the NHI RHRS records only the
NHI Geologic Character and Block Size scores. Note: Points without
lettering indicate only a single rockfall mode at that roadcut. P ¼ plane; highest-case score. Of the 80 roadcuts rated, 62 percent
W ¼ wedge; T ¼ topple; D ¼ differential weathering; R ¼ raveling. have multiple rockfall modes, with Differential Weather-
Some points represent identical scores for multiple roadcuts. ing and Raveling being most common (Figures 5 and 6).

150 Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154
Geologic Contribution to Rockfall Hazard

Table 6. Logistic regression results.

Reference Wald Wald Odds 95% Confidence


Variable Value B ErrorB Statistic pw Value Ratio Interval

ROCKFALL dependent variable


Lithologic variation: yes No 1.8 0.7 6.7 .01 5.8 1.5, 22.0
Number of discontinuity sets: 0 .2 sets 2.3 0.7 11.8 ,.01 9.7 2.6, 35.4
Number of discontinuity sets: 1 .2 sets 1.8 0.8 5.6 .02 6.0 1.4, 26.4
Number of discontinuity sets: 2 .2 sets 1.4 0.6 5.3 .02 4.0 1.2, 13.3
BLOCKSIZE dependent variable
Failure type: structural Weathering 2.6 0.6 15.4 ,.01 19.3 4.4, 84.5
Lithologic variation: yes No 1.72 0.52 11.0 ,.01 5.6 2.0, 15.4
Number of discontinuity sets: 2 .2 sets 1.5 0.7 5.2 .02 0.22 0.06, 0.8
Mechanical thickness: ,0.2 m .1.0 m 2.0 0.7 8.0 ,.01 0.13 0.03, 0.5

For dependent variable ROCKFALL, modeled category ¼ weathering. The likelihood chi-square value was 25.7 (df ¼ 4, p , .01). The Hosmer-
Lemeshow goodness-of-fit value was 0.91 (df ¼ 5, p ¼ .97). The c-statistic value was 0.788. For dependent variable BLOCKSIZE, modeled cate-
gory ¼ block size . 0.3 m. The likelihood chi-square value was 48.8 (df ¼ 8, p , .01). The Hosmer-Lemeshow goodness-of-fit value was 6.7
(df ¼ 8, p ¼ .57). The c-statistic value was 0.824.

Another reason for the higher scores is the large percent of the responses correctly according to the c-
percentage of roadcuts that have Raveling, which would statistic. The goodness-of-fit statistic indicates that the
not typically be scored for the 80 roadcuts with the NHI logistic regression model fits the data.
RHRS, because Raveling is not typically the greatest
scoring rockfall mode.
Abundance is another factor that is not evaluated in the Results of Logistic Regression: BLOCKSIZE
NHI RHRS. However, Abundance causes a relative Dependent Variable
decrease for some Tennessee scores versus NHI scores The results from the BLOCKSIZE logistic regression
for two reasons. First, where multiple rockfall modes are analysis (Table 6) indicate that block size is influenced by
present, the scores for each individual mode tend to be rockfall type, lithologic variation, geomechanical thick-
lower for the Tennessee system than for the NHI system, ness, and number of discontinuity sets. The odds for block
because mode abundances tend to be smaller where sizes larger than 0.3 m in the longest dimension are 19
multiple rockfall modes are present. Second, NHI scores times greater for structural than for nonstructural rockfall
tend to be higher than the Tennessee scores where only and 6 times greater where a lithologic variation is present
structural modes are present and in Low Abundance, than where absent. However, the odds for block sizes
which occurs at five roadcuts. larger than 0.3 m in longest dimension are decreased by
4.5 times (eB ¼ 0.22) where there are two discontinuity
Results of Logistic Regression: ROCKFALL sets as compared to the presence of three or more sets, and
Dependent Variable odds are decreased by 8 times (eB ¼ 0.13) when the
geomechanical thickness is less than 0.2 m than where it is
The results from the ROCKFALL logistic regression greater than 1.0 m. The remaining variables are not
analysis (Table 6) indicate that rockfall type is influenced statistically significant predictors of block size and were
by the occurrence of lithologic variation and by the not included in the model. Overall, the model is significant
number of discontinuity sets. Based on the odds ratio, at the .05 level according to the likelihood chi-square
lithologic variation increases the odds of nonstructural statistic, and the goodness-of-fit statistic indicates that the
rockfall by a factor of 6. Odds for nonstructural rockfall logistic regression model fits the data and predicts 82
are: 10 times greater where no discontinuity sets are percent of responses as indicated by the c-statistic.
present, as compared with slopes containing three or more
discontinuity sets; 6 times greater where there is one
discontinuity set as compared with slopes containing three DISCUSSION
or more; and 4 times greater when there are two Tennessee RHRS
discontinuity sets as opposed to three or more. The
remaining variables are not statistically significant pre- As demonstrated by the greater geologic character
dictors of rockfall type and were not included in the model. scores where raveling is present (Figure 5), the Tennessee
The overall model is significant at the .05 level according RHRS captures the significance of raveling with respect to
to the likelihood chi-square statistic and predicts 79 the production of rockfall material. Though raveling

Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154 151
Vandewater, Dunne, Mauldon, Drumm, and Bateman

occurs at 70 percent of the roadcuts in the study area comes a potential problem for the NHI RHRS, which is
(Figure 6), the mode typically involves a small block size. that the rockfall type with the greatest score for Geologic
However, raveling can be a traffic threat or a source of Character, as identified by the NHI RHRS, may occupy
roadway damage if these small blocks are shed from large only a small portion of the roadcut and may be less
heights; if large masses of raveled blocks are shed abundant than another mode with a smaller score.
simultaneously; or if the blocks roll or launch onto the However, second mode, which is not recorded for the
roadway, necessitating raveling characterization for NHI system, increases the cumulative risk because of
effective rockfall management. greater abundance along the roadcut. Therefore, the use of
The cumulative scoring of potential rockfall modes in Abundance in the Tennessee RHRS acts as a sensitivity
the Tennessee RHRS yields greater total Geologic indicator to the overall role of potential rockfall mode.
Character scores compared to the NHI RHRS. One Finally, the use of rockfall modes and distinctive
possible concern is that the score increase would over- characteristics eliminates some of the terminology prob-
emphasize the role of Geologic Character in total hazard lems of the NHI RHRS.
score, but the following points counterbalance this. First,
although the importance of an accurate geologic evalua- Predictors of Rockfall Type and Block Size
tion is the focus here, the Tennessee system also evaluates
traffic volume and roadway conditions with the same The logistic regression results (Table 6) indicate that
proportion of possible score as the NHI RHRS. For both lithologic variation and number of discontinuities are
example, a roadcut could score very high on Geologic significant predictors of rockfall type. Intuitively, litho-
Character, but if catchment is sufficient to minimize logic variation should affect rockfall type through its
rockfall affecting the roadway, the overall difference in effect on differential weathering. Similarly, as the number
total scores will not be large, despite the Geologic of discontinuity sets increases, structural conditions are
Character score. created that promote planar, wedge, and topple rockfall,
Second, considering all potential rockfall modes, not because planar and topple modes require at least one set
just the rockfall type with the greatest Geologic Character of systematic discontinuities, and wedges require at least
score, as occurs using the NHI RHRS, gives insight to the two sets.
structural and weathering condition of a roadcut. This The logistic regression results (Table 6) indicate that
assessment is critical because, for example, another rockfall type, number of discontinuities, geomechanical
rockfall mode with a lower Geologic Character score thickness, and lithologic variation are significant predic-
may cause greater roadway damage after failure because tors of block sizes. Structural rockfall strongly favors block
of block size. Similarly, the possibility also exists that sizes larger than 0.3 m, and conversely, nonstructural
rockfall by one mode can trigger rockfall by another rockfall favors block sizes smaller than 0.3 m. Lithologic
mode. Therefore, recognition of all modes more com- variation also favors larger block size, because roadcuts
pletely defines the portion of a roadcut that is prone to with such variation have nearly 5.5 times greater odds to
failure, and scoring all modes characterizes the cumulative produce large blocks, suggesting that lithologic contrast
contribution of Geologic Character to the RHRS rating. favors the creation of overhangs greater than 0.3 m.
Third, scoring all potential rockfall modes at a roadcut Rock units with geomechanical thicknesses smaller
provides an indication of likely successful remediation than 0.2 m favor the formation of blocks with length less
techniques (Wyllie and Norrish, 1996). Furthermore, by than 0.3 m. As intuitively expected, rock units with
evaluating all potential rockfall modes at a roadcut, the smaller geomechanical thickness have greater odds for
need to separate the roadcut into segments where different shedding small blocks, whereas units with larger geo-
modes are present is obviated, thereby reducing data mechanical thickness have greater odds for shedding large
collection complexity. Overall, the methodology of blocks.
scoring all rockfall modes captures useful information Interestingly, rockfall modes with more than two
omitted by the single-case methodology in the NHI RHRS discontinuity sets are more likely to produce block sizes
and does not overemphasize the role of geology in the greater than 0.3 m than those with just two discontinuity
hazard assessment. sets. One might expect that as the number of discontinuity
Compared to the NHI RHRS, the contribution of sets increases, a multiplicity of intersecting surfaces is
individual rockfall modes to the Geologic Character score created, and therefore, smaller block sizes should be
is appropriately lower in the Tennessee system when the observed. However, this relationship is not observed.
mode is less abundant. The advantage of this result is that Rather, as the number of discontinuity sets increases,
the use of Abundance as a geologic characteristic in the a multiplicity of intersecting surfaces is created, generat-
Tennessee RHRS differentiates roadcuts that would score ing discontinuity geometries that define structural rockfall
identically in the NHI RHRS, which does not evaluate modes (Norrish and Wyllie, 1996), typically yielding
Abundance. Additionally, the use of Abundance over- larger blocks.

152 Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154
Geologic Contribution to Rockfall Hazard

Implications  Block sizes greater than 0.3 m are predicted by the


occurrence of structural rockfall modes, the presence
The logistic regression results highlight geological of more than two discontinuity sets, lithologic
attributes of roadcuts that correlate to greater or lesser variation, and geomechanical thicknesses greater than
rockfall block size or with particular rockfall modes. 1.0 m.
Lithologic variation increases the odds for nonstructural
rockfall, because roadcuts without lithologic variation With respect to the Tennessee RHRS, roadcuts without
include a large majority of the structural rockfall modes. lithologic variation but with two or more discontinuity
Furthermore, the Tennessee RHRS average Geologic sets are expected to have relatively high geologic
Character score for structural modes is 84, as compared character scores because of the increased odds of large
with 43 for nonstructural modes. intact rock blocks shed by structural rockfall modes. This
Additionally, as the number of discontinuity sets expectation is even greater where geomechanical thick-
increases, the odds of structural rockfall increase. nesses exceed 1.0 m because of the increased odds of
Similarly, the odds of block sizes greater than 0.3 m are block sizes greater than 0.3 m.
greatest with the existence of more than two discontinuity
sets. In the Tennessee RHRS, structural rockfall modes
have an average Block Size score of 26, which translates ACKNOWLEDGMENTS
approximately to the third hazard category (1–2 m).
Therefore, if a roadcut contains rocks without lithologic The authors would like to acknowledge Bill Trolinger,
variation and two or more discontinuity sets, a greater Len Oliver, Harry Moore, and the Tennessee Department
Geologic Character score is expected because of the of Transportation for their contributions and funding for
increased odds of large intact rock blocks shed by this research. The manuscript was improved by reviews
structural rockfall modes. This expectation is even greater by Tom Badger, Martin Woodward, and Robert Watters.
where geomechanical layer thicknesses exceed 1.0 m,
because of the increased odds of block sizes greater than REFERENCES
0.3 m.
ABBOTT, B.; BRUCE, I.; SAVIGNY, W.; KEEGAN, T.; AND OBONI, F., 1998,
Application of a new methodology for the management of
CONCLUSIONS rockfall risk along a railway. In Moore, D. P., Hungr, O.
(Editors), 8th Congress of the International Association of
The Geologic Character category utilized in the Engineering Geologists: Vancouver, BC, A. A. Balkema
Tennessee RHRS gives consistent descriptions of a wider Publishers, Netherlands, pp. 1201–1208.
spectrum of geologic conditions than the NHI version ALLISON, P. D., 1999, Logistic Regression Using the SAS System:
Theory and Application: BBU Press, Cary, NC, 288 p.
because it: ANONYMOUS, 1996, GEM-15: Rock Slope Rating Procedure: State of
New York Department of Transportation Geotechnical Engineer-
 Explicitly identifies potential rockfall modes, including ing Bureau, 41 p.
raveling APT, J.; SKAUGSET, A. V.; AND PYLES, M., 2002, Discriminating
 Accumulates hazard scores for all potential modes at between landslide sites and adjacent terrain using topographic
variables: Geological Society America Abstracts with Programs,
a roadcut Vol. 34, No. 6, p. A40.
 Considers the abundance of occurrence for particular BARTON, N. R., 1973, Review of a new shear strength criterion for rock
modes joints: Engineering Geology, Vol. 7, pp. 287–332.
 Avoids ambiguous terminology BENT, G. C. AND ARCHFIELD, S. A., 2002, A Logistic Regression
Equation for Estimating the Probability of a Stream Flowing
Perennially in Massachusetts: U.S. Geological Survey Water-
The Tennessee RHRS does yield greater Geologic Resources Investigations Report 02-4043, 51 p.
Character scores, in general, than the NHI RHRS, but BINGHAM, E. AND HELTON, W. L., 1999, Physiographic Map of
this difference reflects the importance of scoring all Tennessee: State of Tennessee Division of Geology, scale 1:
rockfall modes, including raveling, that have the potential 500,000.
for delivering rock to the roadway. CLEARY, P. D. AND ANGEL, R., 1984, The analysis of relationships in-
volving dichotomous dependent variables: Journal Health Social
An analysis of relationships between geologic attrib- Behavior, Vol. 25, No. 3, pp. 334–348.
utes and RHRS characteristics shows: DAI, F. C. AND LEE, C. F., 2001, Terrain-based mapping of landslide
susceptibility using a geographical information system: A case
 Lithologic variation and number of discontinuity sets study: Canadian Geotechnical Journal, Vol. 38, No. 5, pp.
are significant predictors of rockfall type, because 911–923.
FISH, M. AND LANE, R., 2002, Linking New Hampshire’s rock cut
lithologic variation coincides with differential weath- management system with a geographic information system:
ering, and a greater number of discontinuity sets Transportation Research Record, No. 1786, pp. 51–59.
increases the odds of structural rockfall modes. FLATLAND, R., 1993, Application of the Rockfall Hazard Rating System

Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154 153
Vandewater, Dunne, Mauldon, Drumm, and Bateman

to the Rock Slopes Adjacent to US50 and State Route 28 in the PIERSON, L. A. AND VAN VICKLE, R., 1993, Rockfall Hazard Rating
East Side of Lake Tahoe, Nevada: Unpublished M.S. Thesis, System: Participant’s Manual: Federal Highway Administration
University of Nevada, Mackay School of Mines, Reno, 316 p. Publication SA-93-057, 104 p.
HADJIN, D. J., 2002, New York State Department of Transportation ROYSTER, D. L., 1978, Landslide Remedial Measures: Tennessee
rock slope rating procedure and rockfall assessment: Trans- Department of Transportation, Prepared for the 37th Annual
portation Research Record, No. 1786, pp. 60–68. Southeastern Association of State Highway and Transportation
HARDEMAN, W. D., 1966, Geologic Map of Tennessee: State of Officials Convention, Nashville, TN, 62 p.
Tennessee Department of Conservation, Division of Geology, SAHOO, N. R. AND PANDALAI, H. S., 1999, Integration of sparse
4 sheets, scale 1:250,000. geologic information in gold targeting using logistic regression
HARRIS, D. AND PAN, G., 1999, Mineral favorability mapping: A analysis in the Hutti-Maski schist belt, Raichur, Karnataka, India:
comparison of artificial neural networks, logistic regression, A case study: Natural Resources Research (International
and discriminant analysis: Natural Resources Research (In- Association Mathematical Geology), Vol. 8, No. 3, pp. 233–250.
ternational Association Mathematical Geology), Vol. 8, No. 2, SENIOR, S. A., 1999, Rockfall hazard remediation along Ontario
pp. 93–109. highways: Proceedings, 50th Highway Geology Symposium,
HATCHER, R. D., JR.; THOMAS, W. A.; AND VIELE, G. W., 1989, The Roanoke, VA, May 20–23, p. 276–286.
Appalachian-Ouachita Orogen in the United States: The Decade STOKES, M. E.; DAVIS, C. S.; AND KOCH, G. G., 2000, Categorical Data
of North American Geology Project Series, Vol. F-2, Geological Analysis Using the SAS System, 2nd ed.: BBU Press, Cary, NC,
Society of America, Boulder, CO, 767 p., 12 plates. 626 p.
HOSMER, D. W. AND LEMESHOW, S., 1989, Applied Logistic Regression: STOVER, B. K., 1992, Highway Rockfall Research Report: Colorado
John Wiley and Sons, New York, 307 p. Geological Survey Special Publication, Denver, CO, 27 p.
KLEINBAUM, D. G.; MULLER, K.; KUPPER, L.; AND NIZATI, A., 1998, STUDENMUND, A. H., 1997, Using Econometrics: A Practical Guide,
Applied Regression Analysis & Multivariable Methods, 3rd ed.: 3rd ed.: Addison-Wesley, Reading, MA, 670 p.
Brooks/Cole Publishing Company, Pacific Grove, CA, 736 p. TURNER, A. K. AND SCHUSTER, R. L., 1996, Landslides: Investigation
LOWELL, S. AND MORIN, P., 2000, Unstable slope management: and Mitigation: Transportation Research Board Special Report
Washington State: TR News, No. 207, pp. 11–15. 247, National Research Council, Washington, DC, 673 p.
MOORE, H. L., 1986, Wedge rockfalls along Tennessee highways WANG, E. J. M. AND FITZHUGH, E. C., 1995, Application of odds ratio
in the Appalachian region: Their occurrence and correction: and logistic models in epidemiology and health research: Health
Bulletin Association Engineering Geologists, Vol. 23, No. 4, pp. Values, Vol. 19, No. 1, pp. 59–62.
441–460. WYLLIE, D. C., 1987, Rock slope inventory: Proceedings, Federal
NORRISH, N. I. AND WYLLIE, D. C., 1996, Rock slope stability analysis. Highway Administration Rockfall Mitigation Seminar, FHWA
In Turner, A. K. and Schuster, R. L. (Editors), Landslides: Region 10, Portland, OR. 25 p.
Investigation and Mitigation: Transportation Research Board WYLLIE, D. C. AND NORRISH, N. I., 1996, Stabilization of rock slopes.
Special Report 247, National Research Council, Washington, DC, In Turner, A. K. and Schuster, R. L. (Editors), Landslides:
pp. 391–425. Investigation and Mitigation: Transportation Research Board
OTT, R. L. AND LONGNECKER, M., 2001, An Introduction to Statistical Special Report 247, National Research Council, Washington, DC,
Methods and Data Analysis, 5th ed.: Duxbury, Pacific Grove, pp. 474–504.
CA, 1152 p. ZAIN, H. H. M., 2001, Assessing ground-water vulnerability using
PATTON, F. D. AND DEERE, D. U., 1970, Significant geologic factors in logistic regression. In Hayes, D. F. and McKee, M. (Editors),
rock slope stability. In Planning Open Pit Mines: A. A. Balkema, Proceedings, American Water Resources Association (AWRA)
Capetown, South Africa, pp. 143–151. Summer Specialty Conference and Universities Council on Water
PIERSON, L. A.; DAVIS, S. A.; AND VAN VICKLE, R., 1990, Rockfall Resources (UCOWR) Annual Conference. Decision Support
Hazard Rating System: Implementation Manual: Federal High- Systems for Water Resources Management, American Water
way Administration Report OR-EG-90-01, 172 p. Resources Association, Middleburg, VA, pp. 265–271.

154 Environmental & Engineering Geoscience, Vol. XI, No. 2, May 2005, pp. 141–154

Vous aimerez peut-être aussi