Vous êtes sur la page 1sur 16

Geomorphology 108 (2009) 823

Contents lists available at ScienceDirect

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / g e o m o r p h

Fluvial processes, morphology and sediment budgets in the Coon Creek Basin, WI,
USA, 19751993
Stanley W. Trimble
Department of Geography, 1255 Bunche Hall, UCLA, Los Angeles, CA 90095-1524, United States

a r t i c l e i n f o a b s t r a c t

Article history: Sediment budget processes in Coon Creek, Wisconsin, from 1975 to 1993 changed only moderately, generally
Received 8 June 2006 continuing the trends that were evident in 1975. Perhaps most importantly, sediment yield to the Mississippi
Received in revised form 12 November 2006 River appears to be unabated, with much of this clearly coming from storage loss, especially stream bank
Accepted 12 November 2006
erosion. Vertical accretion continued on the lower (new) oodplains inset within their meander plains in
Available online 12 February 2009
tributaries and the upper main valley. The lower main valley also continued to aggrade but at the rate of
about 6% of that occurring in the 1920s and 1930s. The major revision from earlier studies is that the
Sediment budget upper main valley is a less signicant net source of sediment than previously thought and the earlier
Sediment yield prognostications of it being a signicant net source of sediment in the future may be wrong. Perturbations
Alluviation have been caused by changes of riparian vegetation and, much more importantly, by the introduction of sh
Flood plains shelter structures and protected cut banks along the stream system reaching downstream to the end of the
Stream bank erosion upper main valley. Because these structures do not permit natural stream migration, bank erosion and
Watershed management downstream sediment transfer, the Coon Creek basin has lost much of its suitability as a natural laboratory of
uvial processes.
2009 Published by Elsevier B.V.

1. Introduction paper reports on the results of resurveys of 92 proles conducted

during the 5-year period centering on 1993. Additional data come
Sediment budgets and attending uvial processes have increas- from surveys conducted in 1982. Data from similar streams in the
ingly occupied the attention of geomorphologists. Indeed, Reid and Driftless Area are also used. Detailed descriptions of the region, the
Dunne (2003) consider sediment budgets as the organizing frame- basin and the research design are given in those earlier papers. The
work in quantifying catchment-scale geomorphic processes (see that period 19751993 considered here, was remarkable because it
work for a ne review of work and concepts). The sediment budget of featured two oods of at least 100 year magnitude. The rst of these
Coon Creek, Wisconsin, dating from the mid-19th century has been was a basin-wide ood in 1978 and the second was a more local
reported since 1939 (McKelvey, 1939; USDA, 1942; Haverland, 1944; convectional storm in 1984 which centered on the upper part of the
Trimble, 1975a,b,c,d, 1976a,b,c, 1981, 1983, 1993, 1997a,b, 1999, 2004; basin.
Trimble and Lund, 1982). The Coon Creek study has been termed by
Science as the most comprehensive study of its kind in the world 2. Overall historical sediment budgets of coon creek
(Glanz, 1999). The present paper is an extended version of a short
paper appearing in Science in 1999 and is also a continuation of the The basin-wide sediment budget shows a great decrease in
1983 paper cited above. Unlike some earlier reports (e.g., Trimble and sediment uxes and storage in all parts of the basin over the historical
Lund, 1982) which also considered slope processes, only stream period (Fig. 1). The major point to be gained from this budget is that
processes and rates are considered here. efux, the sediment yield to the Mississippi River, has remained
The data and previously reported analyses for Coon Creek have constant over time despite the great decrease in upland erosion and
come from over 120 cross sections originally surveyed in 1938, all overall sediment ux. Conservation practices on farms successfully
resurveyed in detail during the period 19741979, but centering on reduced soil erosion, but sediment yield continued unabated as
1975. Earlier proles and sedimentary sequences were constructed sediment stored in the bottomland continued to leave, primarily from
from buried soil proles, archeological excavations, and historical stream bank erosion. As explicated in the 1983 Trimble paper, a stream
data, and many new surveyed proles were added 19751978. This with its oodplain has a maximum conveyance capacity and this
capacity is clearly indicated for Coon Creek by the historical sediment
Tel.: +1 310 825 1314; fax: +1 310 206 5976. budget. This disjunction between sediment supply and downstream
E-mail address: trimble@geog.ucla.edu. sediment transport has signicant geomorphic implications: if alluvial

0169-555X/$ see front matter 2009 Published by Elsevier B.V.

S.W. Trimble / Geomorphology 108 (2009) 823 9

Fig. 1. Sediment budgets for Coon Creek Wisconsin, 1853 to 1993. This basin is about 25 km southeast of La Crosse, Wisconsin and has an area of 360 km2. Numbers are annual
averages for the periods in 103 Mg/year. All values are direct measurements except net upland sheet and rill erosion, which is the sum of all sinks and the efux minus the measured
sources. The lower main valley and tributaries are sediment sinks whereas the upper main valley has been a sediment source (Trimble, 1999).

sediment deposits are available and mobile, a stream can reentrain this could be to some people very disturbing. Indeed, in determining
them and they may appear as downstream sediment yield. This the response of a uvial system to environmental changes, sediment
principle was earlier stated by Gilbert (1917) and Happ et al. (1940), storage within a catchment appears to be much more important than
Trimble (1970, 1974, 1975a, 1976a,b,c) and Schumm (1973, who sediment yield, the usual measure (Happ et al., 1940, 1975; Trimble
includes it as part of his complex response concept), but the 1975c, 1976c, 1977).
evidence from Coon Creek is particularly compelling. These ndings
are of both theoretical and practical interest, the latter through policy 3. Distributed sediment budgets within the basin
issues related to controlling non-point sources of pollutants often
accompanying sediments in transport and deposition. There are also The processes portrayed in Fig. 1 are not uniform throughout the
strong environmental and policy concerns relating not only to the basin. Rather, these processes of sediment storage and ux are
sediment but also any chemical adsorbed onto the particles. occurring at differential rates. However, there is continuity to the
Commenting on the implications of Fig. 1 as it appeared in Science differentiation from headwaters to mouth. Since even the extensive
(Trimble, 1999), M. G. Wolman (quoted in Glanz, 1999) stated that: It data for Coon Creek are still too limited to allow quantication of a
means that if I control the materials coming off a eld or group of fully differential model, the situation can be approximated by
elds, it may be some time before I see the results of thatif I do, regionalizing the basin into 3 reaches; tributaries, upper main valley
within decades. Referring to environmental policy and enforcement and lower main valley (Trimble, 1976a,b, 1981, 1983, 1993, 1999;
based on measured sediment yields, Wolman further stated that [] Fig. 2). Processes, morphology and sediment budgets are homogenous
10 S.W. Trimble / Geomorphology 108 (2009) 823

Fig. 2. Differential stream and valley sediment budgets, Coon Creek, WI, 18531993.

enough in each reach to be described as an entity. This regionalization other assumptions. None is denitive and all are limited, but together
is the same as that used in the 1983 paper but the net rates have been they do indicate the general trends of basin storm ow response over
extended and revised for the period 19751993. These appear as the historical period. More recent analyses of accumulated stream
dashed lines because end points are not yet known. Tributaries have ow data show a decided decrease in storm ow response, and as
increased their roles as sediment sinks while the upper main valley would be expected, an increase of base ow and ow duration (Potter,
appears to be moving from a notorious sediment source to a steady 1991; Gebert and Krug, 1996; Krug, 1996).
state or even a mild sediment sink. The lower main valley continues as
a sediment sink but at much lower rates than earlier. These reaches
are discussed further in later sections.
Before discussing the current conditions of the basin, it might be
well to discuss the historical and morphological changes which have
led up to present conditions in the tributaries and downstream. The
basic driver for both the upland and channel erosion was changes of
storm ow response, in turn driven by the changes of land use
described in Trimble and Lund (1982). Estimates of these hydrologic
changes are problematic because the models used are imperfect and
the inputs used are uncertain and sometimes assumed, especially for
the pre-settlement period. The rst study of these changes was by Kay
(1973) who assumed a linear increase from the time of settlement to
the 1970s (Fig. 3). Knox et al. (1975) and Knox (1977) also did similar
calculations, again assuming a linear increase over the total time of
settlement (Fig. 3). Trimble (1975b) was more concerned with the
improvements of land use after the 1930s showing that expected
discharges, like soil erosion, had been greatly reduced (Trimble and
Lund, 1982) and Knox (1977) followed suit. Fraczek (1987) also looked
at the more recent period and included a hydraulic factor for the
gullied landscape which the previous studies had not done. Trimble
revised his 1975 estimate using a rill and gully factor (USDA-SCS,
1985; Trimble, unpublished study) and extended the period back to
1853, the time of settlement in Coon Creek. These studies have Fig. 3. Estimated peak discharges for Driftless Area streams based on land-use change
different time periods, drainage areas, storm size and duration, and using varied models. See text for explication and discussion.
S.W. Trimble / Geomorphology 108 (2009) 823 11

3.1. Tributaries would have maximized stream power and enhanced channel erosion,
thus placing more storm ow in the channel and further increasing
To place the more subtle changes of the recent period into context, channel erosion. Channels in this bare, eroded state were referred to
it is useful to review the changes through which tributaries went up to by Happ et al. (1940) as ume-like because of their form and
about 1975. Considering the earlier peak ow calculations and the efcient conveyance of water. Tributaries in the region had pebble and
change in erosion rates together with historical reconstructions of cobble beds which were only locally mobile so that channel cutting in
stream channels allows a basic understanding of the processes which most places was lateral. Indeed, the coarse material would have
brought about the changes (Fig. 4). Initial headwater streams were enhanced widening of the stream (Knox, 1977; Schumm, 1977). Such
narrow and small (Knox, 1972). Tributaries were typically small rapidly eroding reaches with inset coarse point bars were referred to
enough for a person to leap across (McKelvey, 1939). Under these as meander plains (Melton, 1936) and this term was applied to
conditions, oods would be forced to leave the channel and ow Driftless Area tributaries by Happ et al. (1940) and Happ (1944). But
across the oodplain. Depths would have been small and vegetational the widening had a negative feedback: depth and velocity would have
resistance would have meant low velocities. The primeval landscape been reduced thus reducing stream power. And by the 1950s when
would have permitted little erosion and sediment so that water storm ow began to abate, stream power was reduced even more.
density would have been low and dependant primarily on tempera- Thus, a thin layer of ne material was deposited on the coarser
ture. By about 1900 and after, ooding was increasing with deposits of the meander plain which gave rise to a grass cover which,
commensurate sediment transport so that oodplains were aggrad- in a positive feedback loop, further enhanced deposition from the
ing, effectively creating larger channels (Happ et al., 1940). The ever-ameliorating storm ows. This was the condition in which I
combination of greater ows, greater depths and higher velocities found the tributaries in 19731974 (Trimble, 1975a,b) and I termed

Fig. 4. (A) Schematic model of changes of historic stream and valley morphology for Coon Creek and other Driftless Area tributaries, 1860 to 1974 (from page 16 of a mimeographed
pamphlet by S.W. Trimble for a eld trip to the Driftless Area, April 1975, sponsored by the Association of American Geographers and led by G. Dury, J. C. Knox, W.C. Johnson and S.W.
Trimble (Trimble, 1975b). Lateral migration of the stream is not shown in this model. (B) Changes of stream power and the transformation of stream and valley morphology. This
model assumes a constant discharge for each stage. With the small stream channel of stage 1, oods spread out over the oodplain, keeping depth, velocity, and stream power low.
With accretion of the oodplain and stream banks with historic sediment in stage 2 (circa 1900), greater ows were restricted to the channel, thus increasing depth, velocity, and
stream power so that the channel erosion shown in the 1900 stage (left) must have been very rapid. In stages 3 and 4, the channel erodes laterally, so that oods are spread, with
decreases of depth, velocity, and stream power. By the latter stage, ne sediment covered the old gravel meander plains and new oodplains are formed as shown to the left (from
presentation to the Association of American Geographers, San Diego, CA, 20 April 1992. Ron Shreve made important suggestions for preparation of this diagram in 1991).
12 S.W. Trimble / Geomorphology 108 (2009) 823

Fig. 5. Numbers of grazing animals 18531970 and grazed forest 19251970 in Vernon County, Wisconsin. Data from U.S. Census of Agriculture.

these transformed meander plains as new oodplains since their destructive to landscapes. Indeed, sheep are known to crop grass
heights appeared to be adjusted to the more moderate stream ow much shorter than cows. A reconstruction of animal populations in
regime (Trimble, 1975a,b, 1976a,b; Knox, 1977). Indeed, since there Vernon County, WI, which covers about 75% of Coon Creek, shows that
was little evidence from hydrological data, I offered them as animal populations peaked in 1900, about the time that tributary
morphological evidence that storm ows had been ameliorated by channels were destabilizing (Fig. 5). While animal populations have
land use (Trimble 1975a,b, 1976a,b). not declined greatly since 1900, management is much better and
But force was not the only driver of stream change. As always, reaches of Coon Creek and its tributaries are no longer grazed (Trimble
geomorphic resistance must be considered and the resistance of and Lund, 1982). Of course, grazing animals also increase the force
stream beds and banks was being reduced by grazing of animals. component by damaging upland vegetation and soil fauna, compres-
Trimble (1994) and Trimble and Mendel (1995) have shown how sing the soil, decreasing inltration increasing overland ow and soil
cows can make stream banks much more vulnerable to the forces of erosion, and peak ows in streams (Trimble and Mendel, 1995). At
erosion. Less is known about the geomorphic effects of sheep and least some, but probably not all, of that force component was included
horses but there is no reason to believe that they are any less in the earlier peak ow calculations.

Fig. 6. The evolution of tributary stream channels as explained in Figs. 3 and 4. North Branch, Whitewater River, MN about 2 km west of Elba, MN. For reference, note barn, bridge, and
road intersection in all three frames. (1905): Note narrow, apparently deep, stream lined with trees and dense riparian vegetation. (1940): By lateral erosion, stream has widened 2 to
5 times in response to increased storm ow and the supply of abundant bedload from hillside gullies and the eroded stream itself. Low-ow sinuosity has increased greatly but
overall, the channel tends towards braided (see also Trimble and Crosson, 2000; Fig. 2A). (1976): Stream has again narrowed to similar dimensions as seen in 1905 with dense
riparian vegetation.
S.W. Trimble / Geomorphology 108 (2009) 823 13

The changes described above are illustrated with a remarkable set of increased ood ows and coarse bedload has created an almost-
of time-lapse photos from a tributary of the Whitewater River, braided conguration. By 1976 the channel has been transformed
Minnesota, a stream generally analogous to Coon Creek (Fig. 6). In the again being narrow and is again bordered by trees, signaling more
rst frame, the stream is shown in 1905, about 40 years after moderate ows and ner sediment. Further upstream, the then-
settlement in that area. Being settled more than a decade after Coon intermittent and ashy streams had trenched into the ume-like
Creek, stream evolution would be expected to lag Coon Creek by a few condition described by Happ et al. (1940; Fig. 7). These eroded further,
years (Trimble, 1993) so the stream is probably moving towards stage especially laterally, but by the early 1970s when I started my work in
2 by 1905 (Fig. 4) but trees still border the narrow channel. In frame 2 the region, the streams had generally stabilized (Fig. 7).
(1940), the stream has been transformed and has clearly progressed to The conditions of tributaries have improved since the mid-1970s
stage 2 or 3 (Fig. 4) with a broad eroded channel and the cobble and, indeed, the old, coarse meander plains are now often overlain
meander plain clearly visible. In this particular reach, the combination by 1530 cm of ne material. These lower grassed areas are now

Fig. 7. Evolution of a small tributary, 19401976, Beaver Creek, MN about 5 km upstream of Beaver, Whitewater River, MN. (1940): Stream trenched into a ume-like channel as
described by Happ et al. (1940). S.C. Happ points to contact of old mollisol overlain by historical sediment. The vertical channel walls attest to the recent quality of the channel erosion
and the large sediment particles suggest the contemporary stream power involved to erode the channel. (1976): The stream continued to erode laterally after 1940 and is now quite
sinuous. There are no large sediment particles visible since the older deposits have been covered by silt, now grassed, and no new particles have been washed down from upstream.
The scene is typical of such tributaries for the period 19741993 and later (see also Fig. 2 in Trimble and Crosson, 2000).
14 S.W. Trimble / Geomorphology 108 (2009) 823

considered to be new oodplains set within the high banks of the old 13 years old when subjected to the rst ood and may not have been
covered oodplain. The smaller channel sizes relative to those of he sufciently hardened. The second point is that Wisconsin sh
1930s are considered to be diagnostic of the present moderated managers concede that these structures would have not been effective
hydrologic regime (Trimble, 1975a). The cross sections up to the level if they had been installed in the early part of the century when ood
of the old oodplain are considered to be a oodway (Trimble, 1993) response of the streams was so great (e.g., David Vetrano, Wisconsin
so that the old oodplains rarely ood. Dept of Natural Resources, oral communication, June 2001). That is,
As previously noted, the period 19751993 featured two oods of while these structures may be able to resist the hydraulic forces of one
at least 100-year magnitude. One of these two events was more local or two big oods at present, the frequent oods of the earlier period
and centered on the headwaters to the east but the greatest stream would have eventually destroyed the structures.
change was from the basin-wide ood of July 1978. Earlier studies of The overall sediment budget for tributaries was shown in Fig. 2,
the region already cited have emphasized the increasingly stable but there was considerable variation among proles. One well-
nature of tributaries as the result of greatly improved soil conservation documented example is Bohemian Valley Range 13A (Fig. 10). From
measures with reduced storm response of streams (e.g., Trimble, 1853 to 1938, the banks were covered with up to a meter of sediment
1975a, 1976a,b; Knox, 1977). Thus, these two large storms were a test by vertical accretion but the channel itself was eroded to several
of sorts. In general, the tributaries held reasonably well and did not times its original size. As noted earlier, the pre-settlement channel
return to the highly disturbed cobble and gravel road look they size is not known exactly but from the description that a man could
always had back in the rst half of the 20th century (see also Fig. 2 in leap across (McKelvey, 1939), dimensions of a meter wide and a
Trimble and Crosson, 2000). meter deep were assumed. From 1938 to 1974, the channel eroded
A second test brought by the storms was that of channel structures. laterally, rst about 20 m to the left, then moved about 35 m back to
Starting in 1974, the Wisconsin Department of Natural Resources the right, demonstrating the extreme stream forces extant during the
(DNR) instituted a program of installing sh shelters along the early part of that 36-year period. The channel bounded by the
tributaries of Coon Creek, the structures also being intended make the eroded high banks of the historic ood plain was, indeed, now so
stream more stable and thus improve trout habitat. These structures large that the old oodplain had experienced no accretion after 1938
comprised a log platform to form the sh shelter, normally set into the and was thus a new terrace. Indeed, this condition obtained in the
cut bank of an outside stream bend. This platform was covered with whole tributary reach so that measurable vertical accretion occurred
boulders, then covered with earth and sowed to grass (Fig. 8). Judging on only one of 72 cross sections between 1939 and 1974 (Trimble,
from the scientic literature, stream structures have generally not 1976a). By the time of the 1974 survey, this reach was perfectly stable
been effective in helping to stabilize streams. But since the later and the cobbles of the bar had been covered by 315 cm of silt with an
resurveys (c. 1993) encountered stream reaches with and without excellent stand of grass (Fig. 11). As early as 1975, such silt-covered
structures, this provided an opportunity to test the effectiveness of and grassed areas were proposed to be new oodplains diagnostic
these structures (Trimble, 1997b). The two diagnostic measures were of the more moderate storm ows of the conservation period
(i) net annual erosion of channel cross sections and (ii) net annual (Trimble 1975a,b, 1976a,b; Knox, 1977) but it will be seen in the
lateral migration. The results, shown here cartographically, demon- next two periods (19741982 and 19821991) that these new
strate little difference between the two conditions (Fig. 9). Statistical oodplains have continued to aggrade by vertical accretion, thus
analysis, however, demonstrated a small advantage of structures in raising the bank, making the channel cross-section larger, and
reducing channel erosion but reduction of channel migration was making the morphological determination of bankfull discharge
seemingly evident but not statistically signicant (Trimble, 1997a,b). increasingly difcult. This cross section (Fig. 10) has experienced
Two points need to be made, however. First, most structures were only more accretion than average during recent decades, perhaps because
the prole is located in the downstream reach of Bohemian creek, just
upstream from the upper main valley so that backwater effects are

3.2. Upper main valley

The upper main valley has been perhaps the reach of greatest
interest but also of greatest concern. It went through changes much
like those described for the tributaries, the differences being that
sediment sizes rarely exceed sand, stream banks were much higher,
and the changes came later. As late as 1940, this reach resembled a
ume with the stream owing between high banks which had been
raised by the vertical accretion of by up to 3 m of historical alluvium.
With the curtailment of upland erosion and the resulting decrease in
sediment load, the stream's capacity was increased, allowing
increased lateral erosion and retreat of the high banks. But because
the storm ow regime had been ameliorated, the new point bar and
oodplain built behind the advancing stream was only about half as
high as the old historical oodplain built in the rst half of the 20th
century (Figs. 1214). As in the tributaries, the result of this process
was the massive downstream export of sediment and the morphologic
result was a oodway down the stream and over the new oodplains
which could convey most ood ows (Trimble 1975a, 1976a,b, 1993).
Thus, deposition on the old historical oodplain was greatly reduced
(Trimble, 1975a, 1976a,b). Indeed, the old oodplain was now a
Fig. 8. A typical sh shelter built in some reaches of Coon Creek tributaries and
elsewhere in the basin. A heavy oak bench is set by the bank, covered with limestone
terrace. The oodway reduced local ooding but the efcient down-
boulders and then soil. The bank is then beveled and planted to grass. Such structures stream routing of water may have increased ooding downstream
had a mild but statistically signicant effect in stabilizing streams. (Woltemade, 1994; Woltemade and Potter, 1994).
S.W. Trimble / Geomorphology 108 (2009) 823 15

Fig. 9. Changes in tributary channels, 19751993, structures versus no structures. (A) Net annual change in channel cross-sectional area. Channels with structures had statistically less
erosion; and (B) net annual lateral channel migration. Reduction of migration seemed evident but was not statistically signicant.

The processes just described for the upper main valley were Indeed, I forecast that this erosion might not only furnish sediment for
essentially the same as occurred earlier in the tributaries (see Trimble centuries but might actually result in unit sediment yield increasing
1975a,b, 1976a,b, 1983; and Figs. 9.24 and 9.26 in Ward and Trimble, downstream, giving a sediment delivery ratio greater than 1.0
2004). The same processes are moving downstream but appear to be (Trimble, 1981, 1983). But as the cut banks retreated and the new
slowing relative to earlier rates, a process discussed in the next oodplains expanded, the area for vertical accretion was expanded
section. As one moves downstream towards the lower main valley thus promoting deposition and creating a morphology-based negative
where still-aggrading oodplains are lower, the new oodplain feedback. I also did not realize that the new oodplains were not yet at
appears higher relative to the old oodplain so that the escarpment their true channel-forming or effective-ow height. Thus, they have
between the two becomes lower. The downstream end of the upper continued to aggrade by vertical accretion over the period since then
main valley is where the new oodplain, formed as a bench behind and, indeed, still appear to be accreting. Thus, local accretion was
the meander, merges at the same level as the still-aggrading tending to offset, at least to some degree, the loss of sediment by
oodplain of the lower main valley (Fig. 15). erosion of the high banks.
Like the tributaries in the earlier period c. 19201950, the upper By the early 1990s, I had enough data to realize what was
main valley was a prolic net source of downstream sediment for the happening and devised a model to show the feedback function of
period c. 19501980. In the 1970s, this reach was still losing sediment expanded depositional surface of the new oodplains (Fig. 14). Only
at a very high rate and it looked as though it would continue for after the 19911995 surveys were complete was it clear that the upper
centuries until the historical alluvium (high banks) was removed. middle valley reach was past stage 3 and nearing stage 4 as depicted in
16 S.W. Trimble / Geomorphology 108 (2009) 823

Fig. 14. In the last report on Coon Creek (Trimble, 1999), I thought it
prudent to retract my earlier statements about future centuries of
erosion from this reach and state that the reach may no longer be a
great net exporter of sediment.
During a eld excursion in 2005, it was suggested to me that the
planform of this reach with cut banks often at the outside radius of
the curve would bias the samples I was getting from my surveys
(Fig. 12). That is, the ratio of eroding land to accreting land might be
greater than my calculations would indicate (Wes Lauer, University
of Minnesota, oral communication, Oct. 2005). Early in the Coon
Creek study I had realized the hydrologic importance of the new
oodways and the sediment source potential of the upper main
valley, so starting in 1978 and continuing to the present, I have
mapped long sections of this reach in detail (e.g., Fig. 12). Using
these large-scale maps, I was able to draw isopachs of sediment
depths and remeasure changes in this reach. However, they were not
signicantly different from my earlier published values so no revision
is possible. That is, my earlier predictions of future net sediment loss
over centuries from this reach still appear to be wrong, at least based
on the present evidence.

3.3. Lower main valley

The lower main valley remains primarily a net sediment sink.

While the average vertical accretion for this reach in the 1920s and 30s
was extremely high, being about 15 cm/year (Trimble and Lund, 1982;
Trimble, 1983) the mean for 19751993 was only 0.53 cm/year, but
this gure was reduced somewhat by channel erosion induced, at least
in part, by afforestation of stream banks (Trimble, 1997a). The present
rate from surveys is slightly higher than that suggested by Fig. 7C in
Trimble (1983) but that estimate was based on dendrochronology and
other short-term measures. Additionally, the present measured rates
include the oods of 1978 and 1987 and the very wet spring of 1993. A
typical prole is Coon Valley Range 27 (Fig. 16). Note that the bed,
banks, and oodplain have all aggraded.
This study has depended on fully surveyed transverse proles.
Fig. 10. Evolution of tributary cross-section, Bohemian Valley Range 13R., approximately
The utility of these full proles as compared to coring and some
4 km northeast of Coon Valley, WI. While part of the 1853 prole is based on the old
mollisol, part is informed conjecture. The 1938 prole shows about 1 m of historical stratigraphic dating methods (e.g., 137Cs) is demonstrated in Fig. 16.
sediment on the old oodplain and the channel had eroded to the point that the old Five locations along the prole were randomly selected and a
covered oodplain had become a terrace. By 1974, the stream had moved laterally, simulated chronology was constructed for each based on the data
taking out a large (20 m) section of the old oodplain (see Fig. 11 for photo). After 1974, from the surveyed prole as though a core had been extracted and
the primary process was the vertical accretion of the new oodplain with ne sediment.

Fig. 11. Photo made along the axis of Bohemian Range 13R, 1974. The x marks where the toe of the cut bank stood in 1938.
S.W. Trimble / Geomorphology 108 (2009) 823 17

Fig. 12. An example of detailed plane-table mapping of an eroding reach (the Marshal Reach) of the upper main valley, approximately 4 km northwest of Coon Valley, WI, showing
the network of interconnected proles all set on MSL datum. Historical sediment covered the old oodplain to a depth of about 23 m. According to air photos and S.C. Happ (oral
communication), this reach was ume-like in 1940 and had only moderate sinuosity. Lateral erosion removed large sections of the old historical oodplain and replaced it with the
new, lower oodplain. Inset prole is Range C. See Fig. 13 for photo.

sequential dates had been established. These are then compared to the Given that this reach is aggrading, suggesting a surplus of sediment
mean chronology as established by the full prole itself. What the and lack of excess energy, it might be surprising to know that many
simulations suggest is that there is much more variance inherent in short reaches are migrating laterally quite rapidly. While the stream at
corings compared with full proles as used in this study. They also Range 27 moved laterally only about 12 m in 65 years, a meander loop
suggest that the variation is greater for shorter time periods and/or for about 80 m downstream has migrated about 30 m in 30 years (Fig. 17).
lesser depths. Dating of oodplain accretion sequences from limited Unlike the morphology with lateral migration upstream in tributaries
borings can lead to questionable notions such as that of Knox who and the upper main valley, the point bar here is built up to about the
suggests that each of a few individual large oods between 1950 and same level as the cut bank and both banks continue to slowly aggrade.
1954 deposited blankets of 1525 cm of sediment across the valley Indeed, this is the major reason that the meander belt can intrude
oor (Knox, 2001). The period 19741978 includes the 100-year ood only slowly into the Lower Main Valley. That is, the bank height is by
of 1978 and the variance of deposition depth and rates, mainly from the denition the aggrading oodplain (here, the natural levee). So
1978 event, are clearly shown, the rates varying from 0 to 3 cm/year. although some reaches here are actively meandering, there can be no
The thicker deposits were primarily sand splays, so commonly seen inset surfaces as seen in the upstream reaches. Citing Knox (1987),
with accelerated sedimentation (Happ et al., 1940). Individual Woltemade (1994) ascribes the lack of a meander plain in the Lower
measures like cores, whether related to erosion or accretion, tend to Main Valley to a lack of stream power but that clearly cannot be the
demonstrate high variance and should be used only with a large case since the stream is actively meandering. The point upstream
sample and error analysis (Trimble, 1975d; Rommens et al., 2006). where the existing eroding banks are higher than the presently-
18 S.W. Trimble / Geomorphology 108 (2009) 823

Fig. 13. Photo across the Marshall Reach showing Range C in inset, Fig. 12. The man on the right stands on remnant of old Pleistocene terrace at the end of Range C. Man on the left
stands at water's edge on Range C.

forming oodplain is the lower end of the meander plain, and, by that during the 19911995 surveys the backswamps were much
denition, the upper main valley. Or expressed another way, the lower wetter than in the c. 1975 surveys so that survey passage on foot was
end of the meander plain is where the new oodplain is built up to at much more difcult and almost impossible in some cases. In 1993, this
least the level of the old historic oodplain. condition came in part from excess rain but the more general problem
It will be observed that for the period 19751993, the natural was that higher levees blocked drainage of the backswamps allowing
levees close to the banks appear to be accreting more rapidly than them to become deeper. That is, it is more difcult for the water from
distal backswamp areas of the oodplain. This trend appears springs and small streams to reach the main stream so that it must
throughout the reach. The perception was strengthened by the fact ow parallel to Coon Creek as a Yazoo stream until it can make its way

Fig. 14. Transformation of the upper main valley morphology and accompanying changes of the sediment budget (redrawn from Trimble, 1993).
S.W. Trimble / Geomorphology 108 (2009) 823 19

Fig. 15. Near the downstream end of the upper main valley dened by the new oodplain, inset within the meander plain, merging into the old aggrading oodplain. In this scene,
about 5 km downstream from Coon Valley, Wisconsin, the toe of the escarpment between the new and old oodplain is shown by the dashed line. Only about 30 cm separate the two
surfaces here whereas the elevation difference several km upstream at the Marshall Reach (Fig. 12) is 23 m.

into the main stream. Of course, streambeds and water levels are also directly after the 100-year ood of 4 July 1978 showed splays of sand
raised so that streams are almost superposed above the distal on levees and even on distal oodplains.
oodplain in some places. The source of the channel sand is also of interest. While the system
Differential vertical accretion rates between banks and distal ood appears to be receiving much less sediment from the uplands than in the
plains for 15 proles in this reach are shown in Table 1. For the banks, 1930s, Coon Creek and its tributaries have been reworking earlier
the measurement was made at the highest point on each bank. For the channel and bank deposits as described earlier in this paper. This appears
distal oodplain, the measurement was made at the lowest point. The to entail a sorting process whereby ne material is transported further,
overall mean rate for banks 19751993 was 0.63 cm/year but for distal some perhaps out of the system, but the sand is left as lag deposits.
oodplains, it was only 0.42 cm/year. It would have been useful to The sand is moved downstream at a slower rate and some of it may
measure the rates of streambeds and water levels but both have so be deposited on banks as just described, but much is transported
much variance in space and time that they are very difcult to downstream further aggrading channels. Many examples could be
measure with any dependability. The changes in water levels shown in given in the region but the situation at Arcadia Wisconsin, on the
Table 1 suggest that they were rising faster than the banks but the Trempeleau River about 100 km north of Coon Creek is illustrative
values are largely artifacts of surveying both during high and low ow (Trimble, 1993). There, migrating sediment from local tributaries
events with the wet, very high base ow year of 1993 predominating. similar to Coon Creek has moved into the Trempeleau River, but the
Actually, stream water levels have probably risen in general accord ner material has moved on leaving the sand as lag deposits.
with the bank heights but that relationship is not certain. Aggradation of the Trempeleau River and its local tributaries has
A test for statistical signicance was conducted by grouping all raised ground-water levels in the city so that, by the 1980s, basements
bank and all distal ood plain values and subjecting them to a were becoming wet or ooded. The only solution for amelioration was
difference of means test. They were different at the 0.05 signicance to pump sand from the streambed downstream of Arcadia thus
level, so I conclude that the banks of the lower main valley are indeed lowering the stream, at least temporarily.
aggrading more rapidly than the distal oodplains. Observation suggested that forested reaches in the lower main
The explanation for these differential accretion rates appears to be valley had a larger cross-sectional area than grass-covered ones. Four
quite straightforward. The stream bed is composed largely of sand, reaches, each with long-term grassed and forested subreaches were
and during ood events, this sand is transported overbank where examined in great detail (Trimble, 1997a, 2004; see also Lyons et al.,
hydraulic resistance is high because of dense vegetation so that much 2000 for benthological implications). Grassed reaches were narrower
of the sand is immediately deposited. Evidence for this is the and had smaller channels than forested reaches, suggesting that
percentage of the sand often found along the natural levees. As grassed channel reaches stored about 2100 to 8800 m3 more sediment
suggested earlier, sand tends to be deposited in irregular splays, per kilometer than forested reaches. Since many reaches of the lower
accounting in part for the variance among the accretion measure- main valley have reverted from grass to forest over recent decades, it
ments. Because upland erosion is so greatly curtailed, there is less silt seems reasonable to assume that riparian afforestation has caused the
in the main channel to be transported in suspension to the distal loss of sediment from the reach and thereby decreased the overall
oodplains. Moreover, the small tributaries entering directly onto the accretion rate in the lower main valley.
oodplain are also transporting little sediment so that there is less
material available for accretion. One can compare this with the 1930s 4. Spatial and chronological distribution of alluviated area
when the main stream and the tributaries were transporting much
ne sediment and the distal oodplains were aggrading almost as fast Earlier studies of Coon Creek have shown the spatial distributions
as the banks (Fig. 16). However, reconnaissance and detailed surveys of sediment volumes over time (Trimble, 1976a; Trimble, 1993). With
20 S.W. Trimble / Geomorphology 108 (2009) 823

Fig. 16. Floodplain accretion and accretion rates at Range CV 27, approximately 3 km downstream from Chaseburg, WI. The actual proles for 1853, 1938, 1974, 1978, and 1992 are
shown at the bottom. Five locations along the prole were selected and a simulated chronology was constructed for each based on the data from the surveyed proles as though a
core had been extracted and the sedimentary unit dates had been established. For comparison, the mean shows the rates as determined from the surveyed proles.

the foregoing processes, morphologies, and sediment budgets in became more vulnerable to ooding and sedimentation so that the
mind, it is now possible to show the areal distribution of alluviation in area of alluviation increased as a step function as successive terraces
Coon Creek over the historical period (Fig. 18). As agriculture began were covered. Since this cannot be shown by maps at the small scale
and expanded, stream response and sediment loads increased thus permitted here, area itself is graphically shown in Fig. 18B. By the 1938
increasing alluviation of the functional oodplains of that period surveys, all potentially oodable terraces were being reached by
(Fig. 18). That is, all active oodplains in the basin were being covered oods although not all parts of terraces showed measurable accretion.
by vertical accretion. As aggradation continued, stream terraces then An example is Coon Valley Range 28, about 0.5 km downstream of
S.W. Trimble / Geomorphology 108 (2009) 823 21

Fig. 17. Meander loop looking downstream approximately 80 m downstream of Range CV 27 which has migrated about 30 m laterally since 1974. Note that advancing point bar is
accreted to the same elevation as the retreating cut bank/oodplain. This entire reach of several hundred meters is rapidly migrating, and a meander neck just to near side of the trees
in the foreground is only a few meters wide. There exists the possibility of a meander cut-off here in the near future.

Range 27 (Fig. 19). Here, the original oodplain was relatively narrow also because stream storm ow peaks were ameliorated. As explained
(~100 m) and was anked by a 4 m high stream terrace on which a earlier, expansion of the new oodway has been moving downstream
railroad embankment had been constructed in 1904. At CV 30, 1.6 km so that the upstream-most point of historical oodplain alluviation
further downstream, the oodplain had aggraded only 0.75 m by 1904 moved downstream about 67 km during the period 19751993
so it would appear unlikely that this high terrace would ever have (Fig. 18B). The downstream movement can be expected to continue its
been ooded at this time. By 1938, however, the oodplain had downstream movement but at a much lower rate. Whether this
aggraded to the level of the old high terrace and despite the partial slowing is a function of decreasing slope, or decreasing storm ow
diking effect of the railroad embankment, the lower part of the terrace response, or both, is uncertain. What will happen upstream is also
was already being covered by historical sediment. uncertain, largely because of human intervention. If natural processes
While the old historical oodplains were being aggraded, channel had been allowed to continue, it seems likely that oodway expansion
and bank erosion was beginning in the tributaries (Figs. 4, 6, 7, and would have continued for a long time into the future, but as explained
10), but most of the old tributary oodplains were still being ooded earlier, bank protection structures were installed in the tributaries by
and aggraded as late as the 1938 surveys. But with the rapid bank governmental agencies starting in the mid-1970s. Largely after the
erosion of that time, historical oodplains were quickly becoming 1993 surveys, bank structures were also increasingly installed in the
terraces so that vertical accretion was halted (Fig. 18, 19381975) so upper main valley. With the eventual cessation of bank erosion there,
that the old meander plains became the new, lower oodplains. Again, and given the rate of vertical accretion on the low oodplains, it is fair
this occurred not only with the expansion of the new oodway, but to wonder if the old historical oodplains may again be ooded in the
future. Moreover, the building of stream stabilization structures by
governmental agencies progresses downstream so that the down-
stream expansion of the oodway may be totally halted. And with
Table 1
that, the whole Coon Creek study as originally envisioned will be
Differential accretion, natural levees and distal oodplains, lower main valley, Coon effectively ended also because the full gamut of natural stream
Creek, Wisconsin, (19751993, cm). processes is no longer allowed to operate.
Range Left f.p. Left bk Water level Right bk Right f.p.
25A 30 24 42
25R 15 15 72 9 6 5. Conclusions and prognosis
26 15 15 18 9
27 18 21 12 27 18 Sediment budget processes in Coon Creek from 1975 to 1993
28 3 9 78 9 6 changed only moderately, generally continuing the trends that were
29 15 15 15 15 6
evident in 1975. The most important of these continued trends is that
30R 6 3 33 6 6
30(0) 9 0 0 6 0 the modest but constant sediment yield to the Mississippi River
31 18 15 9 12 6 appears to be unabated, the result of storage removal primarily by
33 0 15 9 9 bank erosion. The major revision from earlier studies is that the upper
33b 0 6 33 0 0
main valley is a less signicant source of sediment and my earlier
36 9 9 60
36R 0 9 6 9 12 prognostications of it being a future sediment source may be wrong.
38 15 15 15 Perturbations have been caused by changes of riparian vegetation, but
Mean 10.6 12.0 15.9 10.6 7.5 more signicantly, by the introduction of sh shelter structures and
Mean rate of all natural levees (banks): 0.63 cm/year. protected cut banks along the stream system reaching downstream to
Mean rate of all oodplains: 0.42 cm/year. the end of the upper main valley. Because these structures do not
22 S.W. Trimble / Geomorphology 108 (2009) 823

Fig. 18. Allocation of alluvial land in time and space. (A) Aggrading historical oodplains, 18531993. (B) Change of alluvial land classication, 1853 to present, the most salient point
being the transformation of old historical oodplains to new terraces as the result of channel erosion.

permit natural steam migration, bank erosion and downstream Desiree Tullos (Oregon State), Gert Verstraeten (Leuven), and Des
sediment transfer, the Coon Creek basin has lost much of its suitability Walling (Exeter) for the critical reviews of the paper. Ron Shreve
as a natural laboratory of uvial processes. advised on the stream power evolution (Fig. 4A) and David Rigby
advised on the statistical tests. Chase Langford produced the
Acknowledgements excellent gures. I am grateful to Edmund Brick, Jennie Trimble,
Alicia Trimble Gordon, Jon Carson, and Paul Price for the eld
I thank John Boardman (Oxford), Tom Dunne (CaliforniaSanta assistance. The National Geographic Society, Wisconsin Department
Barbara), Robert Meade (USGS), Olav Slaymaker (British Columbia), of Natural Resources, the USDA Natural Resources Conservation
S.W. Trimble / Geomorphology 108 (2009) 823 23

Fig. 19. Proles of Range CV 28 (0.5 km downstream from Range CV 27) showing the burial of a high terrace.

Service, and UCLA Faculty Senate provided the funding and/or other Trimble, S.W., 1974. Man-Induced Soil Erosion on the Southern Piedmont, 17001970.
logistical services. Soil and Water Conservation Society of America, Ankeny, Iowa.
Trimble, S.W., 1975a. Response of Coon Creek, Wisconsin to soil conservation measures.
In: Zakrzewska-Borowiecki, B. (Ed.), Landscapes of Wisconsin: A Field Guide.
References Association of American Geographers, Washington, DC, pp. 2429.
Trimble, S.W., 1975b. Coon Creek watershed eld trip notes. Association of American
Fraczek, W., 1987. Assessment of the Effects of Changes in Agricultural Practices on the Geographers, April 1975 (mimeographed, 19 pp.).
Magnitude of Floods in Coon Creek Watershed Using Hydrograph Analysis and Air Trimble, S.W., 1975c. Denudation studies: can we assume stream steady state. Science
Photo Interpretation, Unpublished MS Thesis, University of Wisconsin, Madison, 188, 12071208.
WI. Trimble, S.W., 1975d. A volumetric estimate of man-induced soil erosion on the
Gebert, W.A., Krug, W.R., 1996. Streamow trends in Wisconsin's Driftless Area. Journal Southern Piedmont. Present and Prospective Technology for Predicting Sediment
of the American Water Resources Association 32, 733744. Yields and Sources, pp. 142154. USDA-ARS Pub. S-40.
Gilbert, G.K., 1917. Hydraulic Mining Debris in the Sierra Nevada. U.S. Geological Survey Trimble, S.W., 1976a. Sedimentation in Coon Creek Valley, Wisconsin. Proceedings of the
Professional Paper, No. 105. Third Federal Interagency Sedimentation Conference, Washington, D.C. pp. 100112.
Glanz, J., 1999. Sharp drop seen in soil erosion rates. Science 285, 11871189. Water Resources Council, sec. 5.
Happ, S.C., 1944. Effects of sedimentation on oods in the Kickapoo valley, Wisconsin. Trimble, S.W., 1976b. Modern stream and valley sedimentation in the Driftless
Journal of Geology 52, 5368. Area, Wisconsin, USA. In: Gerasimov, I.P. (Ed.), Geomorphology and Paleogeo-
Happ, S.C., Rittenhouse, G., Dobson, G.C., 1940. Some principles of accelerated stream graphy. Proceedings of the 23rd International Geographical Congress, Moscow,
and valley sedimentation. U.S. Department of Agriculture Technical Bulletin 695. pp. 228231.
U.S. Government Printing Ofce, Washington, D.C. Trimble, S.W., 1976c. Unsteady state denudation. Science 191, 871.
Happ, S.C., Rittenhouse, G., Dobson, G.C., 1975. Valley sedimentation as a factor in Trimble, S.W., 1977. The fallacy of stream equilibrium in contemporary denudation
sediment-yield determinations. Present and Prospective Technology for Predicting studies. American Journal of Science 277, 876887.
Sediments Yields and Sources, pp. 5760. USDA, ARS Pub. 540. Trimble, S.W., 1981. Changes in sediment storage in the Coon Creek basin, Driftless Area,
Haverland, F.A., 1944. Physical land conditions in Vernon County, Wisconsin: Viroqua, Wisconsin, 1853 to 1975. Science 214, 181183.
Wisconsin. USDA, Soil Conserv. Service, mimeographed, 34 p. Trimble, S.W., 1983. A sediment budget for Coon Creek basin in the Driftless Area,
Kay, P.A., 1973. Hydrological and Morphological Changes During the Historic Period in Wisconsin, 18531977. American Journal of Science 283, 454474.
Ellis Branch, Southwest Wisconsin. Master's Thesis, Department of Geography, Trimble, S.W., 1993. The distributed sediment budget model and watershed manage-
University of Wisconsin, Madison, WI. ment in the Paleozoic Plateau of the upper Midwestern United States. Physical
Knox, J.C., 1972. Valley alluviation in southwestern Wisconsin. Annals of the Association Geography 14, 285303.
of American Geographers 62, 401410. Trimble, S.W., 1994. Erosional effects of cattle on streambanks in Tennessee, U.S.A. Earth
Knox, J.C., 1977. Human impacts on Wisconsin stream channels. Annals of the Surface Processes and Landforms 19, 451464.
Association of American Geographers 67, 323342. Trimble, S.W., 1997a. Stream channel erosion and change resulting from riparian forests.
Knox, J.C., 1987. Historical valley oor sedimentation in the upper Mississippi Valley. Geology 25, 467469.
Annals of the Association of American Geographers 77, 224244. Trimble, S.W., 1997b. Streambank sh-shelter structures help stabilize tributary streams
Knox, J.C., 2001. Agricultural inuence on landscape sensitivity in the Upper Mississippi in Wisconsin. Environmental Geology 32, 230234.
River Valley. Catena 42, 193224. Trimble, S.W., 1999. Decreased rates of alluvial sediment storage in the Coon Creek
Knox, J.C., Barlein, P., Hirschboek, K., Muckenhirn, D., 1975. The response of oods and basin, Wisconsin, 197593. Science 285, 12441246.
sediment yields to climate variation and land use in the upper Mississippi Valley. Trimble, S.W., 2004. Effects of riparian vegetation on stream channel stability and sediment
Institute for Environmental Studies Report 52, Madison, WI. budgets. In: Bennett, S.J., Simon, A. (Eds.), Riparian Vegetation and Fluvial
Krug, W.R., 1996. Simulation of temporal changes in rainfall-runoff characteristics, Coon Geomorphology, pp. 153169. AGU Water Science and Application 8, Washington, D.C.
Creek Basin, Wisconsin. Journal American Water Resources Association 32, 745752. Trimble, S.W., Crosson, P., 2000. U.S. soil erosion ratesmyth and reality. Science 289,
Lyons, J.S., Trimble, S.W., Paine, L.K., 2000. Grass versus trees: managing riparian areas 248250.
to benet streams of central North America. Journal of the American Water Trimble, S.W., Lund, S.W. 1982. Soil Conservation and the Reduction of Erosion and
Resources Association 36, 919930. Sedimentation in the Coon Creek Basin, Wisconsin. U.S. Geological Survey
McKelvey, V.E., 1939. Stream and Valley Sedimentation in the Coon Creek Drainage Basin. Professional Paper 1234, Washington.
Master's Thesis, Department of Geology, University of Wisconsin, Madison, WI. Trimble, S.W., Mendel, A.C., 1995. The cow as geomorphic agenta critical review.
Melton, F.A., 1936. An empirical classication of ood-plain streams. The Geographical Geomorphology 13, 233253.
Review 26, 593609. U.S. Department of Agriculture, Soil Conservation Service, ms, 1942, Hydrologic Studies
Potter, K.W., 1991. Hydrological impacts of changing land management practices in a in the Coon Creek Watershed, Wisconsin: Washington, D.C., Tech. Pub. 46, 220 p.
moderate-sized agricultural catchment. Water Resources Research 27, 845855. USDA-SCS, 1985. Engineering Field Manual for Conservation Practices. Reprinted by
Reid, L., Dunne, T., 2003. Sediment budgets as an organizing framework in uvial Peace Corps, Information Collection and Exchange, Department of Agriculture, Soil
geomorphology. In: Kondolf, G., Piegay, H. (Eds.), Tools in Fluvial Geomorphology. Conservation Service, Washington, DC.
Wiley, Chichester. Ward, A.D., Trimble, S.W., 2004. Environmental Hydrology. Boca Raton, Fla.
Rommens, T., Verstraeten, G., Bogman, P., Peeters, I., Poesen, J., Govers, G., Van Rompaey, Woltemade, C.J., 1994. Form and process: uvial geomorphology and ood-ow
A., Lang, A., 2006. Holocene alluvial sediment storage in a small river catchment in interaction, Grant River, Wisconsin. Annals Association of American Geographers
the loess area of central Belgium. Geomorphlogy 77, 187201. 84, 462479.
Schumm, S.A., 1973. Geomorphic thresholds and complex response of drainage systems. Woltemade, C.J., Potter, K.W., 1994. A watershed modeling analysis of uvial
In: Morisawa, M. (Ed.), Fluvial Geomorphology. New York State University Pubs., geomorphologic inuences on ood peak attenuation. Water Resources Research
Binghamton, NY, pp. 299310. 30, 19331942.
Schumm, S.A., 1977. The Fluvial System. Wiley, New York. 333 pp.
Trimble, S.W., 1970. Culturally Accelerated Sedimentation on the Middle Georgia
Piedmont. Soil Conservation Service, Fort Worth.