Geomorphology 


The fluvial geomorphology of the Molalla River, which demonstrates features typical of many rivers draining the Western Cascade Range, has unique characteristics and fluvial responses that warrant a focused investigation. The geomorphology of the Molalla River was analyzed using available remotely sensed data sets, including LiDAR and aerial imagery, as well as field-collected data.


Observations from reconnaissance trips to the Molalla River during the early stages of this study led to hypotheses that guided subsequent geomorphic analysis. For example, fresh gravel deposits and abandoned channels upstream from Canby (GR3, fig. 2) suggested increased channel-migration activity might have resulted from excess sediment load arriving from upstream. Elevated gravel and cobble bars upstream of the Highway 211 Bridge suggested the possibility of increased sedimentation in the upstream reaches of the study area. It was hypothesized that the river corridor might be subject to widespread aggradation, which, in turn, was leading to increased flooding and channel migration. Subsequent analysis showed systematic aggradation was not widespread and observed sedimentation was due to localized effects and proximity to channel-constraining bedrock.


Geomorphic Reaches


The geomorphic flood plain of the Molalla River was delineated to provide a static reference frame for analysis of temporal changes in the channel morphology. The geomorphic flood plain was also intended to represent the spatial extent of the lateral movement of the river during the Holocene, typically spanning the valley bottom between bedrock or alluvial bluffs that act to confine channel movement. In analyzing both the geomorphic flood plain and other geomorphic and physical characteristics of the river corridor, six reaches exhibiting unique physical traits were identified. These unique geomorphic reaches were determined using LiDAR-derived hill-shade maps that revealed distinct transitions in channel width, presence of secondary channels, and channel-migration potential. 


For the Molalla River, two separate bare-earth LiDAR data sets were obtained through the Oregon Department of Geology and Mineral Industries. LiDAR data spanning Rkm 0.0–22.36 (northern region) were collected between March 16, 2007, and April 15, 2007; LiDAR data encompassing Rkm 22.52–39.41 (southern region) were collected on October 26, 2008, November 1-4, 2008, and March 18, 2009 (Watershed Sciences, 2009a; 2009b). The LiDAR data sets contained topographic points at 0.91-m grid resolution, with vertical accuracy of 5-8 cm in the northern region and 1-3 cm in the southern region. 


Interfaces between geomorphic reaches were also determined by identifying narrow points in the geomorphic flood plain (fig. 6) that indicated natural transitions. The geomorphic flood plain within an individual reach tended to have a relatively consistent width.


The geomorphic reaches were numbered from one to six from the confluence with the Willamette River to the upstream extent of the study area (table 5; fig. 3). Geomorphic reach 1 (GR1) extended from the Willamette River (FPkm 0.0) to the confluence of the Molalla River with the Pudding River (FPkm 2.0). Geomorphic reach 2 (GR2) extended from the confluence with the Pudding River to Goods Bridge at Highway 170 (FPkm 9.0) just south of Canby and adjacent gaging station, Molalla River near Canby (14200000). Geomorphic reach 3 (GR3) extended from Goods Bridge to a relatively narrow point in the geomorphic flood plain at FPkm 15.4 (fig. 6) located just upstream of where the Milk Creek tributary enters the Molalla River geomorphic flood plain. Geomorphic reach 4 (GR4) extended from FPkm 15.4 upstream to another relatively narrow point in the geomorphic flood plain (FPkm 26.1). Geomorphic reach 5 (GR5) extended from FPkm 26.1 to FPkm 33.0, where the overall geomorphic flood plain narrows markedly (fig. 3 and fig. 6), and geomorphic reach 6 (GR6) encompassed the remainder of the study area upstream to Glen Avon Bridge at FPkm 35.7.


Channel Characterization within Geomorphic Reaches


The upstream-most geomorphic reach (GR6) is a relatively narrow, 2.7-km-long fluvial corridor constrained by sandstone and conglomerate bedrock on both banks. Numerous bedrock exposures in the channel bottom throughout GR6 indicate that alluvial fill is thin and the river has incised into bedrock in recent geologic history or is currently downcutting. As a result of bedrock control, there are no secondary or side channels in GR6; however, pools are frequent in GR6, commonly eroded directly into exposed bedrock just downstream of steep riffles. Some tributaries have deposited large boulder assemblages into the main stem that appear to be eroded debris-flow or landslide deposits. From Glen Avon Bridge to the North Fork of the Molalla River (Rkm 43.8), the river has a plane-bed morphology, that is, a relatively flat cobble-bed channel lacking discrete bars that has a low width to depth ratio and large relative roughness (Montgomery and Buffington, 1997). Downstream of the North Fork Molalla River, large floods with ample stream power have mobilized particles and reworked the coarse-grained sediment into a pool-riffle morphologic structure (Montgomery and Buffington, 1997) with boulder bars establishing and controlling the riffle structure. Boulders are the dominant sediment-size class of the large channel-forming bars. LiDAR data were not available for GR6, which precluded a complete analysis of slope, although a coarse calculation from USGS topographic maps showed reach slope to be about 0.005 m/m. A relatively small number of engineered revetments in GR6 mantle bedrock to protect structures built on top of adjacent strath terraces but do not restrict channel movement any more than the underlying bedrock. The narrow, constrained river corridor in GR6, combined with established and mature riparian vegetation, results in a well-shaded channel with riffles and pools (fig. 7A) that keep the water well oxygenated. Despite the established riparian corridor, large woody debris in GR6 is rare due to limited recruitment from channel migration and efficient transport from the reach during high flows.


Leaving the constrained bedrock corridor of GR6, the river enters the alluvial, 6.9-km-long GR5.The mean river slope is about 0.005 m/m. Compared to GR6, the active channel of GR5 is wider and riparian vegetation is farther from the river centerline, resulting in less overall shade. Compared to downstream reaches, however, GR5 is relatively narrow and well shaded (fig. 7B). The morphologic structure of GR5 is pool-riffle, but sediment particle size in the channel-forming bars declines progressively in the downstream direction: fluvially deposited boulder bars are numerous in the upper extent of GR5 and rare in the lower extent; gravel and cobble deposits are common throughout the reach. Sandstone and conglomerate bedrock crops out along the channel bed throughout GR5, but less abundantly than in GR6. Similarly, laterally constraining bedrock occurs sparsely and only where the main stem approaches the outer extents of the wider geomorphic flood plain on both river-right and left. In contrast to GR6, a high-flow alluvial flood terrace, 1 to 4 m higher than the river, is prevalent on both sides of the river throughout GR5, indicative of a river with some degree of active channel migration over the recent geologic time frame. Revetments are infrequent in GR5, but they do restrict channel migration into the alluvial flood plain in strategic locations. Secondary channels occur in the downstream section of GR5. Pools are numerous in GR5, and the deepest pools are located where the channel approaches lateral bedrock at the periphery of the geomorphic flood plain. There is relatively little large woody debris within GR5.


In GR4, a wide and open active channel within a wider geomorphic flood plain, the Molalla River assumes the characteristics typical throughout most of the lower extent of the study area (fig. 7C). The length of GR4 is 10.7 km and the slope of the river here is about 0.0035 m/m. The same units of sandstone and conglomerate bedrock present in GR5 and GR6 underlie much of the channel throughout this reach. Lateral bedrock constrains the river in GR4 only on the right (northeast) extent of the geomorphic flood plain as the left edge of the geomorphic flood plain in GR4 is exclusively a terrace of Pleistocene alluvium. Revetments to protect property, agriculture, and infrastructure are common throughout GR4; they are prevalent on the left side of the river because confinement on the right side of the river is commonly governed by bedrock. The morphologic structure is pool-riffle throughout GR4, and pools, although present throughout the reach, tend to be deeper near confining bedrock or revetments. Particle size in alluvial channel-forming bars progressively finer downstream, with boulder bars giving way to cobble and gravel bars. Secondary channels are not widespread, but are more common than farther upstream. Shade in GR4 becomes sparse as much of the riparian vegetation along the river channel has been removed, exposing more of the water surface to direct sunlight. Limited assemblages of large wood begin to appear in the lower section of GR4, sourced, presumably, from active channel migration within the reach.


As it enters the wide, alluvial geomorphic flood plain of GR3, the Molalla River is active with historical tendencies toward flooding and rapid channel migration, which presents major challenges for people living and working along its banks. The total length of this reach of the geomorphic flood plain is 6.4 km, and the mean slope along the river is about 0.003 m/m. Whereas bedrock is present in the channel of the upper 1 km of GR3 and at one location along the right bank near FPkm 14.8, bedrock is absent from the lower 5 km of GR3, indicating the river here flows over alluvium of unknown depth. The coarse-grained fan of Lake Missoula Flood deposits (O’Connor and others, 2001) in the lower 1 km of GR3, however, does restrict lateral movement of the river to the north. The morphologic structure of the river in GR3 is pool-riffle with island-braided characteristics (Beechie and others, 2006), and numerous cobble and gravel bars occupy a wide and exposed active channel (fig. 7D). Pools are frequent in GR3. Secondary channels are common, due in part to active channel movement during historical peak flows and a relative paucity of revetments. Few revetments exist in the reach and are predominantly placed on river left to protect property and infrastructure. The largest of these is a 0.6-km-long revetment near FPkm 14 that protects the railroad line on river left. Of all the reaches, riparian vegetation and shade is least common in GR3 (fig. 7D) although recruitment and collection of large woody debris on the channel margin from channel migration appears to be an active process. 


As the Molalla River flows past the city of Canby, it enters the relatively confined 7.0-km-long corridor of GR2 (fig. 7E). The overall active channel width in GR2 is narrow and the mean river slope is about 0.0016 m/m. Much of the eastern boundary of the geomorphic flood plain in the downstream section of GR2 is confined by the coarse-grained fan from Lake Missoula Floods spilling through the Oregon City gap (O’Connor and others, 2001). Away from the Lake Missoula Floods deposits, the river flows between Holocene alluvial terraces. Although geologically speaking the river probably swept across the wider geomorphic flood plain in GR2, strategically placed revetments along the river in this reach have restricted the location of the river during the past four decades. No lateral bedrock exists in GR2, but channel bedrock is common and widespread from FPkm 8 downstream to FPkm 2.6. This channel bedrock consists of weakly cemented, fluvially deposited gravels probably emplaced by a Pleistocene-era Willamette River and is distinctly different from the sandstone and conglomerate units present in GR4 and farther upstream. Although not confirmed, the exposed bedrock in GR2 probably is the Linn Formation (O’Connor and others, 2001). The morphologic structure of GR2 alternates between pool-riffle and a straight channel (Beechie and others, 2006), and gravel bars are present along the channel margins although they are not as large or as common as the gravel bars found in GR3. Pools are frequent in GR2, and the deepest pool recorded in the study (6.6 m) was located near the downstream end of GR2. Riparian vegetation is common, and the narrow channel generates more shade than in GR3. There is almost no large woody debris in GR2, possibly due to the relative lack of channel movement, which in turn reduces local recruitment.


From the confluence of the Pudding River to the Willamette River, the Molalla River is a 2.0-km-long, wide river with an underlying pool-riffle morphology (fig. 7F). Unfettered by lateral bedrock or revetments, the river meanders widely and actively across the geomorphic flood plain with large gravel point and lateral bars throughout GR1. Bedrock is not exposed in GR1. No secondary active channels were apparent in the 2009 river configuration although secondary channels were probably common throughout the Holocene. Because GR1 is relatively wide, there is little shade over the water surface, but recruitment and deposition of ample large woody debris in the channel increase ecologic complexity and provide fish habitat.


Flood-Plain Morphology


Following the methodology of Jones (2006), height above water surface (HAWS) maps were generated where LiDAR data were available to visualize fluvial features along the geomorphic flood plain. Along transects drawn orthogonally to the river centerline, elevation of the flood plain is subtracted from the presumed water-surface elevation—determined from LiDAR where the river centerline intersects the transect of interest—to create an elevation of the flood plain relative to the river’s water-surface elevation. Despite inaccuracies in the underlying LiDAR data and water-surface elevation, the resulting HAWS maps, when viewed as a raster data set, allowed qualitative insight of the position of the river in the recent geologic past. Specifically, the HAWS maps enabled the identification of relict morphologic features in the flood plain created by the river, such as abandoned channels, meander bends, and oxbows.


The HAWS maps show the Molalla River has actively meandered across the width of the geomorphic flood plain in GR1, combining with the Pudding River in the lower 2 km of the river corridor (fig. 8). HAWS maps in GR2 show the distinct morphology of the Molalla River and the Pudding River. The level of sinuosity of the Pudding River, on the western side of GR2, over the late Holocene has been greater than the main-stem Molalla River. Although also meandering, the Molalla River in GR2 has not been as sinuous as its western tributary. Based on the absence of fluvial features in the flood-plain surface between the two rivers toward the southern extent of the geomorphic flood plain, there is no evidence the two rivers have joined upstream of about FPkm 6.1 within the recent geologic past. The HAWS maps of GR3 show the dominant geomorphic influence of the railroad, and its long, fortified structure dissecting the flood plain. The HAWS maps show active movement of the river before construction of the railroad line within the upstream section of GR3. This railroad line was constructed in 1913 (Cole and others, 2004). The maps also show a pre-settlement river that swept across the flood plain with braided or island-braided characteristics. It is apparent from the HAWS maps that the Molalla River in GR3 has been active, with migrating meanders and channel avulsions (that is, the rapid abandonment of a river channel and formation of a new channel), even before European settlement. Similarly, the Molalla River in GR4 has migrated actively across the geomorphic flood plain, but the process of movement has been dominated by avulsions in a pool-riffle or island-braided morphology and less so by lateral meandering of the main channel (fig. 8). The HAWS maps in GR5 indicate that the river in this reach has been a predominantly pool-riffle or island-braided system prior to development, not unlike the river today, although contemporary channel movement in GR5 was likely suppressed by bank stabilization projects. LiDAR data were not available for GR6, precluding the generation of HAWS maps; however, field observation confirmed that the river in GR6 is bedrock controlled and not subject to active channel migration.


Key Longitudinal Trends along the River Corridor


Longitudinal trends along the river corridor were analyzed on the basis of a combination of LiDAR data, aerial imagery, and field data collection. A longitudinal profile of water-surface elevation was generated by projecting the river centerline onto LiDAR data. Three kayak float trips covering the Molalla River from Glen Avon Bridge to the Willamette River were made in July 2010 to gather river-depth data, catalogue the locations of confining bedrock and revetments, and to photograph river characteristics. Measured depths, combined with the LiDAR-derived water-surface elevation profile, were used to generate a bed-elevation profile and characterize pools in the river. Finally, field work in September 2010 provided particle-size data and bathymetric data for multiple locations in the study reach. 


Water-Surface Elevation


A water-surface elevation profile was constructed from LiDAR data from Rkm 0.0 to Rkm 39.41 (fig. 9). An available 10-m digital elevation model (DEM) upstream of Rkm 39.41 was too coarse to accurately determine water-surface elevation. Airborne LiDAR systems work by firing a laser at the ground and measuring the return time of the beam reflected off the target surface (Wehr and Lohr, 1999). Although some systems were designed specifically for the purpose of mapping aqueous and subaqueous surfaces, the LiDAR flown over the Molalla River was tuned to measure terrestrial relief, and LiDAR returns from aqueous surfaces were probably discarded and topography at water features were likely interpolated from the nearest known ground points. Thus, the vertical accuracy of points in the main stem of the river is subject to errors associated with these interpolated surfaces. Differences in river discharge during different flight days over different sections of the river also create uncertainty in the final water surface. To generate a water-surface elevation profile of the river centerline, the river centerline was projected on the LiDAR data and elevation points were sampled at 1-m increments and then plotted in an x-y coordinate system. From the raw centerline elevation data, outlier points, those points that obviously misrepresented the water-surface profile, were removed, and depressions and convexities were smoothed to create a realistic profile. The vertical accuracy of the constructed water-surface profile could not be determined independently, however, the profile at any given point along the river centerline probably is within 1 m of the actual water surface. Moreover, for the analyses completed for this study, the degree of accuracy in vertical elevation is adequate to discern general trends of the water-surface profile with respect to the flood-plain scale in terms of slope, profile convexities, profile concavities, and residual pool depth. 


The longitudinal water-surface elevation profile of the Molalla River (fig. 9) is generally concave, consistent with graded rivers that have adjusted their slope to sediment transport supplied from upstream (Macklin, 1948; Knighton, 1998). If the profiles are examined closely, however, they show that the Molalla River deviates from the concave profile of a graded river near Rkm 18 along the interface of GR3 and GR4 (fig. 9A). This change in curvature is more visible when the longitudinal profile is plotted on a semi-logarithmic scale (fig. 9B). The slope of the average water-surface (fig. 9C), computed as a 5-km moving average, shows the steepest portions of the river in the study area to be in the upstream extent with the gradient generally decreasing in the downstream direction to the Pudding River tributary. The increased slope at the upstream end of GR3 is also apparent in the slope data shown in figure 9C.


Water Depth and Residual Depths


A series of three float trips, using a two-person inflatable kayak, was completed in July 2010 to measure water depth and to map bedrock and revetments on the Molalla River from Glen Avon Bridge to the confluence with the Willamette River. The section of the Molalla River from Highway 211 to Goods Bridge was floated on July 13, 2010 (the mean discharge for the day was 6.60 m³/s, or 233 ft³/s). The section of river from Goods Bridge downstream to its confluence with the Willamette River was floated on July 14, 2010 (the mean daily discharge was 6.06 m³/s, or 214 ft³/s). The section of river from Glen Avon Bridge to the Molalla River at Highway 211 was floated on July 22, 2010 (the mean daily discharge was 4.79 m³/s, or 169 ft³/s).


During each float, water depth was measured using a global-positioning system (GPS) referenced fathometer mounted on the boat, with a specified horizontal accuracy of less than 15 m and an estimated depth accuracy of 0.2 m. The GPS-referenced water depths were recorded every 5 seconds and were limited by the instrument capability and boat dimensions to values greater than about 0.3 m. During each float, a conscious effort was made to pilot the boat as close to the thalweg (channel invert, or minimum elevation in the river bed) as possible.


During the floats, bedrock outcrops and engineered revetments were mapped using a hand-held GPS, with a horizontal accuracy of about 10 m. All observed outcrops of bedrock along the banks of the river and along the bed of the river were identified during the float trip and categorized as being within the channel or on the left or right bank. Also, lengths of observed hardened embankments, revetments, and other man-made structures visible from the river were mapped to quantify the current degree of the anthropogenic influence on the river. It is important to note that revetments away from the river or those obscured by vegetation were not mapped; thus the revetment data presented in this report does not represent a rigorous inventory of all revetments in the river corridor. 


Water-depth data measured during the study were related to the river centerline and plotted for the entire study area in figure 10. By joining the measured depth data to the nearest 1-m stationing point along the river centerline and the corresponding longitudinal water-surface elevation, a longitudinal bed-elevation profile was generated. Residual pool depth is defined as the difference between the maximum water depth of a pool and the limiting elevation of the riffle at the downstream extent of the pool (Lisle, 1987). After generating the bed-elevation profile, residual pool depth was computed from Rkm 0 to Rkm 39.41 using an algorithm and software developed by Madej (1999). The Madej algorithm identifies only pools deeper than a user-specified threshold, which was set to 1.0 m for the current study consistent with the threshold used by Madej and Ozaki (2009) for Redwood Creek, California. Redwood Creek, in the California Coastal Range where hydrology is similar to that in northwestern Oregon, drains 720 km², a drainage comparable to the Molalla River. 


Residual pool depth statistics generated by the Madej (1999) software showed the average residual pool depth was consistent throughout the river corridor, and ranged from 1.8 to 2.1 m (table 6). Pool frequency per river kilometer ranged from 1.6 to 3.4, with the largest pool frequencies in GR1 and the smallest pool frequency in GR5. Pool frequency generally increased in the downstream direction. The maximum residual pool depth ranged from 3.6 to 6.3 m, with the maximum residual pool depth in GR2.


River-Profile Convexities


Convexities in long river profiles are sometimes used to identify regions of alluvial accumulation (for example, Hanks and Webb, 2006) or to identify incision knickpoints (for example, Stock and Montgomery, 1999). To gain insight into morphologic evolution of the Molalla River, a second-order polynomial trend line representing a graded profile (Mackin, 1948) was fit to the longitudinal profile of water-surface elevation, and the elevation differences between the long profile and the idealized fit were plotted (fig. 11). The bed elevation profile and observations of channel bedrock are also shown in figure 11. Where observed, the channel bedrock was assumed to have the same elevation as the channel bottom, determined by water-depth measurements. The placement of bedrock points in figure 11 is not intended to represent the precise elevation of individual outcrops of bedrock; rather, the points are intended to show generally where bedrock is exposed along the river corridor. The plot shows a strong convexity centered near Rkm 18 that spans GR3 and GR4. A first interpretation might attribute this convexity to the presence of an alluvial wedge; however, the proximity of channel bedrock to the water surface from Rkm 18 to Rkm 38 shows there is little accumulated sediment upstream of Rkm 18, and that the river from GR6 downstream through GR4 flows over a terrace of bedrock positioned just below the water surface. Farther downstream, the outcrops of channel bedrock in GR2 indicate that from Rkm 4 to Rkm 10, the river also flows over a bedrock terrace with a relatively thin mantle of alluvium. Although not determined, it appears that 2–3 m of sediment mantles bedrock in GR1, between the Pudding River and the Willamette River. Between Rkm 10 and Rkm 18, some unknown thickness of alluvium covers the bedrock, but the differenced elevation profile (fig. 11) shows no convexity between Rkm 10 and Rkm 18 that might indicate accumulating sediment. Instead, the Molalla River in this section is well graded to the long profile between bedrock outcrops, suggesting the river can effectively transport sediment from upstream through this reach. Furthermore, as a transport reach, the section of river between Rkm 10 and Rkm 18 is not accumulating new sediment to significant depths.


Though beyond the scope of this study, a possible explanation for the lack of bedrock exposures in GR3 may be the presence of a previously unmapped thrust fault near Rkm 18, where the eastern side has risen relative to the western side. Prior work by the USGS identified an area of geomagnetic anomaly near Rkm 18, possibly indicating a geologic fault (Richard Blakely, U.S. Geological Survey, written commun., 2010). Another explanation for the lack of bedrock exposures is that unique bedrock units upstream and downstream of GR3 led to differential incision rates east and west of Canby that is manifested in the river profile as a change in slope at Rkm 18. Observations during float trips that the sandstone and conglomerate units upstream of GR3 are distinctly different from the conglomerate lithology of bedrock in GR2 support this second hypothesis. The two hypotheses are not mutually exclusive and both mechanisms could have affected the underlying geologic structure.


Particle Size


Particle-size data were collected from exposed bars in September and October 2010. Using the methodology of Wolman (1954), pebble counts were conducted at nine sample locations in all geomorphic reaches except GR1 (fig. 3; table 2). At each site, prominent, unvegetated, recently deposited gravel or cobble bars near the main-stem river were selected for characterization using an approach similar to the one employed by Wallick and others (2010a). For each site, particle-size data were collected along two parallel transects placed on the sampled bars, and a total of 200 particles were measured. The pebble-count transects were typically 2-4 m from the water’s edge and parallel to the flow direction of the river. The particle data were then bracketed into respective phi classes and analyzed (Garcia, 2008). It is important to note that sampled bars were generally at a lower elevation, thus representing bars deposited or reworked by flows within the past one to two years. As a result, the magnitudes of the flows depositing or reworking the sampled bars are much smaller than the peak flows that control the geomorphology of the river and would deposit the highest elevation bars and coarsest particles. Thus the particle-size data presented herein well represents the clasts typically mobilized in a bankfull event.

Figure 12 shows the median particle-size (D₅₀), the 90th percentile particle-size (D₉₀), and the 10th percentile particle-size (D₁₀) data for the nine sampled bars in the Molalla River (appendix A). GR6 had the coarsest collection of particles, and median size decreased generally in the downstream direction, with GR2 having the finest particles. The farthest upstream location, just upstream of Glen Avon Bridge at Rkm 44.32, had a median particle size of 139 mm and a D₉₀ of 405 mm. The bar just downstream of Glen Avon Bridge, at Rkm 44.12, had a median particle size of 76 mm and a D₉₀ of 227 mm. This section of river, upstream of the North Fork Molalla River, had a plane-bed morphology with few recently deposited cobble bars. The sampled region upstream of Glen Avon Bridge, in particular, was a higher, exposed part of the main channel-forming bed composed of coarse particles and was not an accurate representation of the particles mobilized in a bankfull event. Instead, the sampled bar downstream of Glen Avon Bridge, with a smaller overall particle-size distribution, better represented the material mobilized in a bankfull event. Indeed, from this bar downstream of Glen Avon Bridge through GR3, the median grain size on the bars remained relatively constant, suggesting efficient transport, although there was a general downward trend in D₉₀ (fig. 12). The one sampling location in GR2 showed a median size of 42 mm and a D₉₀ of 77 mm.


Sand-size particles (diameter less than 2 mm) were noted in the pebble counts, although not counted in the computations of particle-size statistics. Sand counts in the eight upstream-most pebble-count locations were less than 2 percent. The farthest downstream sample site in GR2 had a sand count of 5.5 percent. Overall, the Molalla River is a unimodal, gravel-bedded system with little sand. 


Cross-Section Data


During September 2010, five bathymetric cross sections were surveyed near bridges throughout the study area to characterize channel geometry of the Molalla River. The cross section near Highway 213 was surveyed on September 1, 2010 (the mean daily discharge was 2.49 m³/s, or 87.9 ft³/s). The other cross sections (at the Glen Avon Bridge, Feyrer Park Road, Highway 211 Bridge, and Goods Bridge) were surveyed on September 2, 2010, when the mean daily discharge was 3.40 m³/s, or 120 ft³/s. All cross sections were surveyed using a total-station survey instrument and reflective prism on a stadia rod. The relative accuracy between surveyed points was about 0.03 m. The five cross sections were surveyed relative to the centerline of the bridges and referenced to benchmarks near the bridges. The cross sections were aligned and positioned using aerial imagery and, therefore, were not accurately geo-referenced to a global horizontal or vertical datum. The cross section at the Glen Avon Bridge was surveyed to the top of the banks. For the other cross sections, only the unvegetated channel was surveyed, and the topographic data for the remainder of the cross sections in the wider flood plain were obtained from the LiDAR data.


The five surveyed cross sections (fig. 3; fig. 13) show a general trend of increasing channel width downstream. At the Glen Avon Bridge, the river is relatively confined within the canyon. At the other cross sections, the river is less confined, with gravel and cobble bars (near the right bank at Feyrer Park, Highway 211, and Highway 213) and low terraces or benches (near the right bank at Highway 211 and Highway 213, and near the left bank at Goods Bridge) within the river banks.


Analysis of Stage-Discharge Relation at
Gaging Station


Following the approach of Klingeman (1973) and Smelser and Schmidt (1998), long-term aggradation trends in the Molalla River were analyzed using stage-discharge relations at the Molalla River near Canby gaging station (14200000). By analyzing the reported stage-discharge relation for the period of record that extends discontinuously back to 1928, the stage for a given discharge was determined. Stage at the gaging station is dependent on downstream hydraulic control and can be used to infer trends in aggradation or incision in the section of river downstream of the gaging station. 


For the analysis, four target discharge values were used: 570 m^3/s (20,100 ft³/s), which represented the approximate discharge of a 5-year recurrence-interval peak event in the lower Molalla River; 110 m³/s (3,800 ft³/s), the value of discharge exceeded 5 percent of the time; 19 m³/s (670 ft³/s), the value of discharge exceeded 50 percent of the time (median discharge); and 1.7 m³/s (60 ft³/s), the value of discharge exceeded 95 percent of the time. Reported changes in the gaging-site location or the reference datum were noted and used to correct the raw stage data to a common datum.

Figure 14 shows the overall trends in stage in the Molalla River near Canby (14200000) from 1928 to 2010. Each data point represents the reported stage for a given discharge value on that date. Two gaps in the data show when the gaging station was not in operation: November 1958 to October 1963 and November 1977 to October 2000. All data were corrected to accommodate station moves or shifts in the datum such that all the points shown in figure 14 are displayed relative to a common datum. As a result, all stage values for a given discharge are directly comparable, even across data gaps when the gaging station was not operating. From 1928 to about 1948, stage at all analyzed discharge values was relatively constant, shifting by no more than a few centimeters. From 1948 to 1953, stage decreased about 0.25 m for all discharge values, indicating a general trend of incision. This decrease may have been associated with gravel mining commonly practiced along the Molalla River in the middle 20th century (Cole and others, 2004). Between cessation of gaging station operation in 1958 and reactivation of the gaging station in 1963, low-flow stage increased about 0.25 m and high-flow stage decreased a similar amount. From 1958 to 1977, stage decreased at a steady rate. Between 1977 and 2000, the second period when the gaging station was not in operation, stage decreased by another 0.25 m. From 2000 to 2010, stage has fluctuated slightly, but there has been no consistent upward or downward trend. Taken as a whole, the gaging record at Canby shows a general trend of incision for the latter half of the 20th century. More importantly, there was no appreciable aggradation downstream of the gaging station in the past 10 years. Notably, the river downstream of the gaging station did not change appreciably after the December 1964 peak of record. 


Spatial Analysis of Flood Plain


Morphologic trends along the Molalla River were determined from historical aerial imagery (table 7). Orthorectified aerial imagery collected in 1994, 2000, 2005, and 2009 was used to digitize channel features and quantitatively analyze changes in the active channel during the last two decades. Note that 1994 imagery for the Molalla River was available as part of the 1995 USGS DOQ imagery set (table 7). Older aerial imagery, from 1936 to 1988, that were not orthorectified allowed for the qualitative analysis of river response over a longer period. Knowledge of the Molalla River flood hydrology between 1994 and 2009 allowed the quantitative analysis of river response to unique peak-flow conditions. For example, the 1994–2000 imagery bounds the large February 2, 1996, high flow that has an estimated discharge of about 900 m³/s (32,000 ft³/s). Similarly, the 2005–2009 imagery bounds a smaller, yet still significant, January 2, 2009, peak event of 691 m³/s (24,400 ft³/s). In contrast, the 2000–2005 imagery bounds a series of modest peak flows, the largest of which was the February 1, 2003, event of 467 m³/s (16,500 ft³/s). 


Methodology of Digitization


Areas of channel migration and flood-plain morphology were digitized using methodology previously developed by the USGS for analysis of the Chetco and Umpqua Rivers. Mapping guidelines and definitions are briefly summarized below; however, a more detailed description of the standard mapping methodology can be found in Wallick and others (2010a; 2010b). 


Digitization of channel morphology was completed at a scale of 1:3,000 and only channel features larger than 300 m² were digitized. Features smaller than 300 m² were integrated into larger, adjacent features. Mapping of morphologic features was primarily confined to the regions adjacent to the active channel, defined as the area typically inundated during annual high flows (Church, 1988). Each digitized data set was reviewed by another team member for quality, accuracy, and consistency. 


Base layers used to map geomorphic features within the active channel of the Molalla River include LiDAR-derived imagery (Oregon Department of Geology and Mineral Industries, 2010), digital orthophoto quads from 1994 (1995 DOQ Molalla River imagery) and 2000 (Oregon Geospatial Enterprise Office, 2010); and orthoimagery from 2005 and 2009 (U.S. Department of Agriculture, 2010). For each of the four imagery sets (1994, 2000, 2005, 2009), regions along the river corridor were mapped into one of five general morphologic features: (1) the wetted channel, (2) gravel bars, (3) secondary water features, (4) bedrock outcrops, and (5) flood plain. 


For mapped wetted channels, two discrete units were classified. First, the “primary channel” of the Molalla River was mapped on the basis of the visible extent of the wetted perimeter. Then for large tributaries, such as the Pudding River and Milk Creek, the wetted perimeter of the “tributary channel” was mapped roughly 500 m upstream of its confluence with the Molalla River. Next, gravel bars, defined as features greater than 300 m² containing exposed bed-material sediment, were delineated and classified as either “flood-plain bar” (adjacent to flood plain) or “island bar” (surrounded by wetted channel). In addition, the amount of vegetation was estimated for all flood-plain-bar and island-bar features as containing bare (less than 10 percent), moderate (10–50 percent), or dense (greater than 50 percent) vegetation cover. Secondary water features, including side channels, backwater sloughs, and disconnected water bodies, were also mapped. Bedrock outcrops, as visible in the aerial imagery, were mapped along the river’s active channel. However, only a few outcrops were identified and mapped, which demonstrates the limitation of using aerial imagery to identify bedrock along a vegetated river corridor where subsequent field observations identified extensive bedrock outcrops along most of the corridor. The remaining area that might typically be inundated during annual high flows, based on vegetation distribution and development, was classified as “flood plain,” allowing for baseline comparison of morphologic features between different time periods. In addition, the approximate channel centerline of the main-stem Molalla River was digitized for each of the four sets of imagery.


Mapping of channel features was affected by the quality and resolution of available imagery. Some areas of imagery had varying degrees of glare, shadows, or local obstructions of channel features. Errors were introduced by imprecise line placement and orthorectification inaccuracies. To minimize errors and increase the overall precision of the interpretive mapping, all delineated features were regularly checked and quality controlled by other members of the project team. 


An important consideration when viewing results from the digitization is that biases in the size of respective morphologic features may exist due to differences in discharge. Although all imagery was acquired during periods of relatively low flow, even a small change in discharge can affect the width of the wetted channel and the area of exposed bars. Discharge data for the Molalla River were not available for all four imagery time periods, so the digitization was not corrected to accommodate for differences in discharge. As a result, no comparative analysis was completed using only the area of the wetted channel.


Channel sinuosity was determined for 1994, 2000, 2005, and 2009 along the main channel of the Molalla River with respect to the geomorphic flood plain. Sinuosity values were determined for each of the geomorphic flood-plain reaches by dividing the reach-segregated channel centerline length by the corresponding geomorphic flood-plain centerline length for that reach. 


The active channel, that section of the river corridor relatively free of vegetation that conveys the majority of the water and sediment during high flow, was defined to include the wetted channel, flood-plain and island bars (less than 50 percent vegetation), secondary water features, and bedrock outcrops forming low benches within the channel. Morphologic features that did not contribute to the active channel included flood-plain or island bars with dense (greater than 50 percent) vegetation, flood plain, bedrock forming steep banks, and tributary features. Also excluded from the category of active channel were any secondary water features, island bars, or flood-plain bars that did not form a collective, continuous arc with an upstream and downstream connection to the main active channel. Active-channel width was then determined by dividing the total active-channel area in a given 1-FPkm subreach of the geomorphic flood plain by the length of that subreach. The average active-channel width was determined for each 1-FPkm subreach segment for each of the four years.


Channel centerlines from 1994, 2000, 2005, and 2009 were used to compute average annual migration rates for each of the three time periods between aerial imagery with channel migration averaged over each 1-FPkm subreach segment. The sum migration distance between two sets of aerial imagery was divided by time elapsed between images to determine an average annual migration rate. 


The percentage of bank confinement due to bedrock outcrops and revetments, as digitized using observations from the float trips, was quantified for each of the six geomorphic flood-plain reaches. For each geomorphic flood-plain reach, the total length of hardened banks was divided by twice the length of the 2009 channel centerline (to account for the left and right banks separately) in order to determine the percentage of the 2009 channel margin that was confined by bedrock, revetments, or coarse-grained Missoula Flood deposits. 


Specific bar area for a given 1-FPkm segment was defined as the total bar area divided by the length along the geomorphic flood-plain centerline. Specific bar area was determined for both the island bars and flood-plain bars, and the three classifications of bar vegetation density (bare, moderate, and dense) were reported along 1-FPkm subreach segments.


Quantitative Trends


Channel sinuosity was highest in GR1, GR3, and GR4, where the channel meandered significantly during the study period (fig, 15). Sinuosity was lowest in GR2, where the channel was relatively straight and ran along the right extent of a wide geomorphic flood plain.


Average active-channel widths were largest and most variable from 1994 to 2009 in GR3 (fig. 16). The smallest and least variable active-channel widths were in GR2 and GR6. The relatively narrow active-channel width in GR2 was due to the numerous confinements, and the narrow active-channel width in GR6 was due to bedrock confining both edges of the channel. GR1 was characterized by relatively large active-channel widths that remained relatively constant through time. In general, active-channel widths in GR3, GR4, and GR5 were variable over time. Indeed, the continuous river corridor from GR5 downstream to GR3 exhibited a large variability in active-channel width from 1994 to 2009 consistent with the pool-riffle morphology, active meandering, and relatively few revetments or confining bedrock walls. Although observed changes in active-channel width of GR1, GR2, and GR6 were largely constant for the four sets of imagery, active-channel width in GR4 and GR5 was largest in the imagery acquired in 2000, after the large February 1996 high flow. By 2005, the active-channel width in GR4 and GR5 had decreased to a value similar to that in 1994. Although the active-channel width in 2009 in GR4 and GR5 did not increase greatly in response to the January 2009 high flow, the active-channel width in GR3 increased significantly.


Consistent with the observed trends in active-channel width, calculated channel-migration rates were largest from GR5 downstream to GR3, and overall channel-migration rates were largest in GR3 (fig. 17), the reach of river prone to extensive management issues for residents living along the banks. In general, channel-migration rates were smallest from 2000 to 2005 throughout the entire study reach, which is indicative of a hydrologic period with smaller peak flows (fig. 5). Channel migration was largest within GR3 for all three periods, with the largest channel-migration rates occurring between 2005 and 2009. From 1994 to 2000, the period bracketing the large February 1996 high flow, channel-migration rates in GR4 and GR5 were generally larger than during the other two periods (fig. 17).


The increased channel-migration rates observed from GR5 downstream to GR3 were due, at least partially, to the presence of fewer channel confinements (fig. 18). The percentage of channel margins confined by bedrock outcrops, revetments, or coarse-grained Missoula Flood deposits varied from 0 percent in GR1, where there were no significant revetments or bedrock outcrops on the banks, to nearly 45 percent in GR6 (fig. 18). In GR2, there were numerous revetments on both the left and right banks. Additionally, the right-bank channel abuts Missoula Flood deposits throughout much of GR2. Although there are revetments in GR3, most notably along the railroad bridge, along Missoula Flood deposits, and at strategic locations along the left bank, much of the GR3 is free of confinement, which allowed a more active channel between 1994 and 2009. The channel in GR4 is moderately confined by numerous bedrock outcrops along the right bank and by numerous revetments, primarily along the left bank. However, strategic placement of revetments, acting in concert with lateral bedrock outcrops on river right, has helped to constrain channel movement in GR4. There is a general tendency for the river to remain positioned along prominent lateral bedrock outcrops or along strongly armored revetments with relatively little hydraulic roughness. Qualitative observations of pool depths suggest a spatial correlation between deeper pools and lateral bedrock or well-engineered revetments, which in turn indicates that these erosion-resistant features create hydraulic conditions during large flows that promote effective sediment transport along the bedrock or revetment face, thereby holding the position of the river against the lateral feature over decades. GR5 is confined by a mix of bedrock and revetments on both the left and right banks distributed throughout the reach. GR6 has the highest percentage of confined banks due to bedrock on both the left and right banks, occurring concomitantly on both banks in several sections. Because the revetment data used to generate figure 18 were based on field observations from 2010, the data best describe plan-form conditions of the rivers as determined from the 2009 imagery. Although not evaluated in this study, it is possible that many left-bank revetments, plentiful in GR5 and GR4, were constructed or enhanced following the 1996 flood, which in turn suppressed channel movement in GR5 and GR4 during the 2009 flood.


Specific vegetated bar area (the total height of the columns on the graph) was greatest in GR3, indicative of active channel migration and a large riparian corridor (fig. 19). In contrast, specific bar area was smallest from restricted channel migration in GR2 and GR6 for all four time periods. Specific bar area is variable temporally in portions of GR3, GR4, and GR5. In 1994, most of the flood-plain and island bars were densely vegetated. From 1994 to 2000 there was a shift toward a larger proportion of bare flood-plain bars throughout most of the study reach as well as a greater area of island bars in GR3, a river response to the February 1996 high flow. Progressing to 2005, after a period of smaller high flows, the flood-plain bars shifted toward a greater proportion with moderate or dense vegetation coverage as vegetation grew and reclaimed some of the active channel. By 2009, following the January 2009 high flow, some channel-margin vegetation was removed and fresh gravel-bar deposits resulted in a larger proportion of exposed morphologic features.


Qualitative Trends after the 1930s


Geomorphic changes between 1936 and 1988 were qualitatively assessed using historical aerial imagery (table 7). These older images were not rectified and digitized, but general trends of channel width, bar growth, vegetation, and channel migration were analyzed. This qualitative assessment of historical imagery, along with the quantitative analysis from 1994 to 2009, demonstrates the qualitative trends of geomorphic change from 1936 to 2009.


Because problems caused by flooding and channel migration have been most acute in GR3 in recent years, the qualitative analysis was focused here. Figure 20 shows paneled mosaics of aerial imagery for all the available aerial images. In 1936, the Molalla River in GR3 was meandering with established riparian vegetation, extensive gravel point bars, and a wide active channel that appeared to be in a state of continual migration into the adjacent flood plain consistent with the fluvial mechanics of other alluvial rivers in the Willamette River Valley and the greater Pacific Northwest. Qualitatively, the Molalla River of 1936 in GR3 had a geomorphic plan-form appearance similar to the contemporary river (in 2009). The plan-form characteristics of the river in GR3, including sinuosity and active-channel width, changed little from 1936 to 1948, although large areas of exposed gravel bars appeared in the upper extent of GR3 just upstream of the railroad bridge (fig. 20B). These new gravel bars were probably deposited by the relatively large 711 m³/s (25,100 ft³/s, with a recurrence interval of about 10 years; table 4) flow that occurred in January 1948, six months prior to acquisition of the 1948 imagery (table 7). 


By autumn 1964 (fig. 20C), the river in GR3 had straightened noticeably as a number of meanders were cut off; however, active-channel width and the extent of vegetation was not greatly different between 1948 and 1964. The imagery collected in 1964 predates the large December 1964 peak of record of 1,240 m³/s (43,600 ft³/s), and channel straightening probably occurred in response to the high flow of November 1960 (fig. 5), estimated to be 980 m³/s (34,500 ft³/s) on the basis of a 7.62 m high-water mark (relative to the gage datum) recorded by the Canby Fire Department (Andy Bryant, National Weather Service, written commun., 2010). The discharge based on this high-water mark was determined using the current (2011) stage-discharge relation at the Molalla River near Canby gaging station (14200000), but because the trends in stage elevation have been decreasing since 1960 (fig. 14), this discharge estimate is likely too large. The geomorphic effects of the December 1964 peak of record are apparent in figure 20D. Imagery taken in 1970 shows a wide active channel with large, extensive gravel bars free of vegetation in a relatively straight channel of low sinuosity. 


By 1980, the active channel was starting to become constricted by encroaching riparian vegetation and meanders were again developing. From 1988 to 1994, a hydrologically quiet period of time with relatively small peak flows (fig. 5), vegetation growth was extensive and the river increased its sinuosity and channel migration. Images acquired in 2000 (fig. 20H), show that the February 1996 high flow again widened the active channel but did not straighten the channel like the 1960 high flow, as evidenced by the 1964 imagery. As discussed earlier, vegetation encroached upon the active channel through 2005 and was subsequently removed by the relatively large January 2009 flood (fig. 5). The active-channel width in the 2009 imagery in GR3 increased, and at least two prominent meanders were cut off (fig.20J). These meander cut offs appeared to supply a large volume of gravel that was deposited locally within 1 km of the cut off, as evidenced by exposed gravel bars in GR3. This locally derived gravel also appears to have caused the marked increase in active-channel width in GR3 near FPkm 12.5 (fig. 16). 


Anthropogenic effects on GR3 can also be deduced from the aerial imagery. Although agriculture and timber harvest from the riparian corridor is evident in 1936, neither activity is widespread nor well established, and buildings and other structures are relatively rare. By 1948, agricultural fields started to increase in number and extent on the southern edge of the river, reaching a peak of activity around 1970. From 1970 to 1994, imagery appears to show that agricultural activity waned slightly, but a number of buildings and houses were constructed on the southern edge of the river. It is important to note that the increases in development along the river corridor in GR3 coincided with the relative stasis in hydrologic peak-flow activity from the middle 1970s to 1996. As a result of fewer and less severe floods, the river channel in GR3 between 1975 and 1996 was relatively inactive. It is also worth noting that the railroad bridge at FPkm 14 and Goods Bridge at FPkm 9 have been stable structures since at least 1936, controlling channel movement near each end of GR3. As a result, most channel-migration activity in GR3 has occurred in the reach between the railroad bridge and Goods Bridge. 


Downstream in GR2, overall channel-migration activity and geomorphic changes have been relatively modest (compared to that in upstream study reaches) between 1936 and 2009 (fig. 21). These historical trends are consistent with the quantitative trends analyzed for GR2 and can be attributed to significant channel confinement (fig. 18) and a relative lack of gravel accumulation, an observation confirmed with available historical aerial imagery.


Within GR1, the active channel in 2009 was slightly wider than the active channel in 1936, and agriculture has encroached slightly on the outer edges of the wider flood plain (fig. 22). However, the overall geomorphic plan form of the primary river corridor is similar between 1936 and 2009. Riparian vegetation is extensive and well established, and overall sinuosity is similar. One significant change between the dates is that although GR1 in 2009 included the total surface-water flow of the combined Pudding and Molalla Rivers, in 1936 the two rivers were almost separate down to the confluence with the Willamette River (fig. 22A). Although the rivers shared a channel in 1936 for a short distance in the middle of GR1, each river flowed mostly in its own channel and the imagery suggests mixing was negligible. 


The 1936 imagery extended no farther upstream than the middle of GR4, but active-channel width, sinuosity, and extent of riparian vegetation in GR4 in 1936 was similar to the geomorphic state of the river in 2009. By 1948, however, GR4 had noticeably changed relative to 1936. The active-channel width was markedly larger, sinuosity had decreased, and a number of avulsions had occurred (fig. 23). Sections of the river corridor in GR4 near FPkm 20 contained extensive bare gravel bars as wide as 200 m. These gravel bars and the active channel in 1948 were significantly larger than exposed bars observed in any section of the contemporary river corridor. Farther upstream in GR5, active-channel width also appears to have increased by 1948, although the amount of widening was not as severe. Furthermore, the 1948 imagery of GR6, still farther upstream, showed neither a widening of the active channel nor extensive gravel deposits. The 1948 imagery, however, does show aggressive logging and dendritic road networks that were presumably used to access forested lands in the catchment away from the river. Although a large high flow could, in theory, cause the geomorphic response observed in the 1948 imagery of GR4, neither the gaging station record (fig. 5) nor the 1948 imagery from GR1, GR2, GR3 (fig. 20), or GR6 support such a hypothesis. Taken in context, it appears that active logging in the upper Molalla River catchment mobilized substantial amounts of sediment that exceeded the transport capacity of the channel in GR4, resulting in a significantly wider active channel and extensive gravel-bar deposits in 1948. Subsequent imagery in 1964 suggests this pulse of sediment both moved downstream and was extracted by gravel-mining operations that were apparent in the imagery.


In GR5, the 1964 imagery shows that the high flow of November 1960 widened the active channel but did not lead to major avulsions. The channel response in GR6 that is visible in the 1964 imagery was not significant. By 1970, however, imagery shows that there was widespread geomorphic change from GR6 downstream through GR4. Vegetation was removed throughout the upper river corridor, and new and reworked gravel bars appeared in GR5 and GR4. The 1970 imagery of GR6, which showed a removal of vegetation, did not show widespread deposits of new sediment. By 1980, riparian vegetation had begun to constrict the active channel throughout GR4, GR5, and GR6, and this process of revegetation and suppressed channel migration continued through the 1994 imagery set, resulting from the relatively quiet hydrologic period from 1975 to 1996.


Channel Response to Flooding


Inferences about the response of the Molalla River to the magnitude of high flows recorded in the gaging station record were made by comparing aerial imagery acquired from 1936 to 2009. As with other lowland rivers in western Oregon and Washington (for example, O’Connor and others, 2003; Beechie and others, 2006), the Molalla River is subject to a flashy peak-flow regime that acts to promote a relatively wide active channel, large channel migration rates, and frequent avulsions. In turn, vegetation growth works to reduce the active-channel width. In the Molalla River, geomorphic response is dependent on the size of the high flows as well as on anthropogenic constraints, revetments, and the underlying controls of the specific geomorphic reach. Pulses of sediment, whether generated from the catchment and transported through the river corridor or derived locally from channel migration and avulsions, can also drive geomorphic change in some sections of the river. The most upstream reach, GR6, appears to be relatively insensitive to the high flows or pulses of sediment. Although the large recorded high flow of December 1964 widened the active channel slightly within GR6, changes in plan-form position were modest. GR6 is also largely a transport reach as little sediment seems to gather. In contrast, the relatively alluvial reaches of GR5 down through GR3 are sensitive to both high flows and pulses of sediment. Wide gravel bars, active channel migration, and occasional avulsions have been common in GR5, GR4, and GR3 since 1936. In the absence of large flows, vegetation encroachment acts to reduce active-channel width and increase sinuosity. Revetments, however, more effectively stabilize the channel plan form. Although anecdotal accounts of active channel movement in GR3 have suggested increased channel mobility in the past 10–20 years, historical aerial imagery shows that channel mobility of the Molalla River in GR3 is likely no greater in the 21st century than it has been historically back to 1936. Relative to GR5 and GR4, the larger quantitative rates of channel migration and active-channel width in GR3 is a function of the few revetments and sparse bedrock control. Farther downstream, GR2 has been restricted by revetments and channel bedrock that have limited channel mobility relative to that in other reaches. It also appears that GR2 has not been subject to widespread sedimentation. GR1 is a reach subject to gradual channel migration across the geomorphic flood plain that reflects the fluvial processes of the pre-development river.


First posted February 29, 2012