Erosion and sedimentation pose ubiquitous problems for land and watershed
managers, requiring delineation of sediment sources and sinks across
landscapes. However, the technical complexity of many spatially explicit
erosion models precludes their use by practitioners. To address this critical
gap, we demonstrate a contemporary use of applied geomorphometry through a
straightforward GIS analysis of sediment sources in the San Francisco Bay
Area in California, USA, designed to support erosion reduction strategies.
Using 2 m lidar digital elevation models, we delineated the entire river network in the Arroyo
Mocho watershed (573 km
Channel incision is a common erosional response to natural or anthropogenic forcing that poses challenges to watershed management across the globe (Schumm et al., 1984; Schumm, 1999, 2007). Incised channels (inner gorges, arroyos, gullies, ravines, etc.) are often created by headward incision of the channel network in response to local or regional base-level lowering (Begin et al., 1981; Schumm, 1993), tectonic uplift (Burnett and Schumm, 1983), or disturbances that change the relative balance between sediment transport and supply (Schumm et al., 1984; Schumm, 1999). For example, incision caused by humans can include urbanization that greatly increases runoff and sediment transport (Booth, 1991), or dams and gravel mining that reduce sediment supply (Kondolf, 1997; Surian and Rinaldi, 2003). Examples of natural causes of incision include change to a wetter climate that increases runoff (Balling and Wells, 1990) and catastrophic events that increase sediment supply from volcanic eruptions (Simon, 1992), extreme precipitation (Miller and Benda, 2000), or following wildfire (Benda et al., 1998). Characteristically, channel incision continues until a new equilibrium grade is achieved, then the channel widens by eroding and failing oversteepened banks, and aggradation begins (Schumm et al., 1984). In some environments, incised channels are part of a natural alternating cycle of aggradation and degradation in response to episodic sediment supply (Bull, 1997). Channel incision creates a variety of problems including destruction of valley bottoms (arable land) and increased sediment yield aggrading downstream reaches (Patton and Schumm, 1975; Poesen et al., 2003).
Delineating the extent of channel incision across a watershed or landscape and quantifying erosion from such sources are necessary to design erosion and channel sedimentation abatement measures (Heede, 1974; Schumm et al., 1984). Mapping gullies using remote sensing began in the 1970s (e.g., Patton and Schumm, 1975) but advancing digital technology in the 21st century now allows for even more detailed mapping using geographic information systems (GIS) and digital elevation models (DEM). Mapping the extent of incised channels can involve visual detection (hand digitizing) in conjunction with delineating a zone around channels based on channel size using GIS (called buffering) (e.g., Perroy et al., 2010) or more automated approaches using digital terrain analysis (Evans and Lindsey, 2010; Castillo et al., 2014) or object-oriented classification of gullies (Shruthi et al., 2011; Johansen et al., 2012). Quantifying erosion rates along incised channels often involves calculating the surface elevation difference between current and pre-gully DEMs (Perroy et al., 2010; Evans and Lindsay, 2010), or using a time series of DEMs (Brasington et al., 2000; Fuller et al., 2003; Martínez-Casasnovas, 2003; Wheaton et al., 2010; Picco et al., 2013). Spatially distributed estimates of sediment yield from incised channels often require highly parameterized and calibrated models (e.g., Van Rompaey et al., 2001; Pelletier, 2012). The technical complexity of such approaches precludes their use by many watershed, land, and resource managers (Guertin and Goodrich, 2011) and consequently such agencies often resort to qualitative evaluations or best professional judgment.
In this case study of a San Francisco Bay tributary, we delineate chronic
annual erosion from the sides of incised channels and estimate sediment
yield at multiple scales using an approach readily understandable and
accessible to planners that improves the spatially explicit representation
of sediment supply through the channel network. We use a topographic index
that combines slope and planform curvature to predict generic erosion
potential (GEP) through the process of shallow failures on planar and
convergent slopes (Miller and Burnett, 2007; Benda et al., 2011). GEP values
are calibrated to sediment yield from gage data and then reported to the
channel network and aggregated downstream. Using a terrain mapping platform
(NetMap) (Benda et al., 2007, 2015a; Barquin et al., 2015), we derived a
digital river network directly from a 2 m light detection and ranging
(lidar) digital elevation model (DEM) and couple it to the terrestrial
landscape via flow direction and accumulation grids. Each
The study objective was to delineate and quantify the spatial distribution of chronic sediment supply from incised channels across the Arroyo Mocho watershed in support of erosion reduction strategies. Ultimately, this study demonstrates a straightforward GIS analysis of important geomorphic processes (erosion and sedimentation) often impacted by land use that can be easily applied by watershed and land managers. Within the context of this special issue, our study provides a contemporary example of applied geomorphomety, one designed to increase communication between science and resource planning.
Arroyo Mocho basin drains 573 km
Arroyo Mocho watershed showing the six major basins, mainstem channels, and flood control channels.
In addition to bank erosion from incised channels, mass wasting processes also occur in the uplands, primarily earthflows (Davenport, 1985; Wentworth et al., 1997; Roberts et al., 1999; Majmundar, 1991, 1996). The southern hills are underlain by the hard meta-sedimentary rocks of the Cretaceous Franciscan Formation with patchy outcrops of the Plio-Pleistocene Livermore Gravels. Here, the steep topography is dominated by deep-seated landslides or earthflows, most of which are old and no longer active. The eastern hills are underlain by the Cretaceous Great Valley Sequence and Miocene sedimentary units that tend to have steep slopes prone to earthflows. The northern hills are comparatively gentler, underlain by the Livermore Gravels and Miocene sedimentary rocks that produce clay rich soils prone to earthflows.
Channels in the Arroyo Mocho watershed are most often characterized by
arroyo or gully forms: an incised topography within a broad valley floor,
with steep and occasionally bare eroding banks. The raw banks appear to
dominate the chronic annual supply of sediment to channels in the study
basin, and thus represent the main source of sediment to the aggrading
channels downstream. In general, the low gradient valley floors above arroyo
banks cannot topographically erode or deliver sediment to the channel. In
the steeper upland channels bordered by colluvial hillslopes (i.e., no valley
floor), most of the sediment production occurs on the channel banks and
hillslope areas adjacent to the channel, including the toes of earthflows
that intersect the channel; hillslopes farther from the channel do not
appear to deliver sediment on an annual basis. Figure 2 shows typical cross
sections of valley floor and hillslope channels. Based on these
observations, we confined the analysis of erosion sources to areas adjacent
to channels (e.g., the arroyo or gully landform). To use a buffer width that
scales regionally with drainage area, we used a GIS buffer width of 6 times
the total bankfull channel width that captures the steep eroding banks of
the incised channel form and areas immediately adjacent to the channel (Fig. 3). To estimate bankfull channel width, we used a San Francisco Bay Area
regional regression relationship based on drainage area (Dunne and Leopold,
1978):
Typical topographic cross sections of a valley floor channel and hillslope channel from the Tassajara Creek tributary.
Example of the buffer used to define the sediment source area from incised channels in the upper Arroyo Seco basin.
We delineated and quantified the spatial distribution of chronic sediment
supply from banks along incised channels that can include earthflow toes
(can include the processes of small shallow failures and raveling) across
the Arroyo Mocho watershed using a terrain mapping platform (NetMap) that
has been applied in similar applications elsewhere (Benda et al., 2007,
2011, 2015a; Bidlack et al., 2014; Barquin et al., 2015; Flitcroft et al., 2016). The primary tasks in the analysis included the following:
Generate a digital and attributed stream layer from the DEM. Estimate erosion potential of incised channel banks using a topographic
index that includes hillslope gradient and planform curvature (topographic
convergence). Modify erosion potential based on vegetation size. Convert erosion potential to sediment yield using river gage data. Aggregate erosion predictions at various scales: buffered channel reaches,
subwatershed, tributary watershed, channel network. Estimate sediment storage potential based on channel constraint and stream
power.
An attributed digital stream network was derived from the DEM using a series
of algorithms described by Miller et al. (2002). The following major steps
were followed in the construction of the DEM and extraction of the channel
network:
More details on methods to derive the channel network from the DEM and algorithms used can be found in Miller et al. (2015), Clarke et al. (2008), and Miller et al. (2002).
To provide a relative estimate of erosion potential within the buffered
incised channel network of the Arroyo Mocho watershed, we used a topographic
index called GEP (Miller and Burnett, 2007; Benda et al., 2011) that combines
slope steepness with slope convergence, recognized topographic indicators of
shallow landsliding, gully erosion, and sheetwash (Dietrich and Dunne, 1978;
Sidle, 1987; Montgomery and Dietrich, 1994; Miller and Burnett, 2007; Parker et al., 2010). Slope steepness is a fundamental control on erosion potential,
while convergence causes surface and subsurface flow to become concentrated
and contribute to erosion potential. GEP (Eq. 2) is calculated from
topographic attributes of slope gradient and topographic convergence
(planform curvature) derived from the DEM:
Specific contributing area is used for erosion potential rather than the entire upslope drainage area because (1) conceptually, pore pressures measured during storms in shallow soils correlate poorly with topography (e.g., Dhakal and Sullivan, 2014), but convergent areas tend to exhibit persistently high water content, deeper soils, and are highly responsive to rainfall events, and (2) empirically, specific contributing area has been shown to better predict shallow failures than total contributing area in several studies (Miller, 2004; Miller and Burnett, 2007). While upslope drainage area is often used as a surrogate for subsurface flow, most locations on the hillslope receive contributions from a small proportion of the upslope contributing area due to the low velocity of subsurface flow (Barling et al., 1994; Beven and Freer, 2001; Borga et al., 2002).
We specifically use GEP as a relative index of erosion potential for shallow failures along steep banks of incised channels in the Arroyo Mocho study area. GEP is not used to estimate erosion potential for downcutting the channel bed or bank erosion of outside river bends. As mentioned earlier, the incised channels in the Arroyo Mocho watershed do not appear to be actively downcutting, rather sediment is now supplied to most channels by shallow slides and failures of oversteepened banks (channel widening), a common stage in the evolution of incised channels following downcutting (Schumm et al., 1984).
Because the analysis was confined to the buffered incised channel network, we assume that all sediment eroded from the inner gorges, arroyos, and gullies is delivered directly to the channel network, as confirmed during 2 days of field observations across the watershed. When practitioners want to include other forms of erosion further from the channel, the proportion of sediment delivered to the channel should be estimated based on different topographic attributes (e.g., Mitasova et al., 1996; Miller and Burnett, 2007; Cavalli et al., 2013), which can be automated within the NetMap terrain mapping platform (Benda et al., 2007) and other similar approaches.
We observed that arroyo and gully bank erosion was often reduced by vegetation in the Arroyo Mocho basin, where larger and denser vegetation created stable channel bottoms and banks. Reaches that had little to no vegetation had more exposed, actively eroding banks compared to reaches that had shrubs or trees established on the banks. Where present, the effect of shrubs and trees in reducing erosion in these arroyo channel systems is particularly pronounced because there is little organic groundcover other than the dominant vegetation of annual grasses. Vegetation reduces erosion by lessening raindrop impact, providing increased soil strength through the root network, and thus reducing surface erosion, rill erosion, and shallow bank slumps and slips (e.g., Thornes, 1985; Thorne, 1990; Prosser and Dietrich, 1995; Simon and Darby, 1999; Abernethy and Rutherford, 2001; Micheli and Kirchner, 2002). In addition, increasing tree age (and thus rooting extent and depth) is related to increasing stability of the soil (Sidle, 1987), and tree height and canopy width are proportionally related to rooting width and depth (e.g., McMinn, 1963; Smith, 1964; Tubbs, 1977; Gilman, 1989). These relationships provide a basis for using tree height as a proxy for root spread and thus soil stability as described below.
While approaches linking vegetation and associated rooting strength to
specific types of erosion have been developed for highly localized scales
(e.g., Roering et al., 2003), such empirically based quantitative approaches
to reduce erosion potential using remotely sensed vegetation attributes have
yet to be developed; consequently, many erosion models simply use broad
categories of land cover. For example, the Universal Soil Loss Equation
(USLE) (Wischmeier and Smith, 1978), the revised USLE (RUSLE) (Renard et al., 1997), and similar approaches (e.g., Booth et al., 2014) simply use a generic
reduction (C-factor) based on classes of vegetation cover (e.g., crop type,
forest, scrub, etc.) to adjust erosion predictions over vast areas
regardless of the individual size of vegetation within the categories. Other
approaches infer similar generalized relationships, for example, Pelletier (2012) assumed a linear relationship between vegetation (leaf cover) and
sediment detachment. We also assume the occurrence of vegetation reduces
erosion potential, but use a relationship that includes both the effects of
vegetation cover and rooting width and depth in each 2 m grid cell, using
tree height as a proxy for root spread and related soil stability. During
1 day of field observations across the watershed, we observed less erosion
on banks with taller trees (and larger root spread) compared to banks with
smaller trees and shrubs. Here, we visually estimated the average riparian
vegetation height on a given bank and the proportion of the bank eroded, the
inverse of the latter gives an estimate of erosion reduction. These
observations were plotted to derive the best fit equation (Eq. 3, Fig. 4)
that governs how erosion potential is reduced by vegetation height:
Relationship between tree height (as a proxy for root spread) and erosion reduction (bank stability) used to reduce GEP estimates. Relationship based on field observations.
To provide a more meaningful view of spatially explicit erosion across the
watershed, we converted the dimensionless GEP index to sediment yield
following approaches used by GMA (2007) and Benda et al. (2011). To limit
assumptions, the independently estimated sediment yield rate is linearly
scaled with GEP values. High values of GEP represent higher erosion rates
and lower values of GEP represent lower erosion rates. GEP is converted to
sediment yield by multiplying each GEP grid cell by the following conversion
factor (Eq. 4):
To evaluate the spatial distribution of chronic sediment sources from incised channels across the study basin, we calculated, aggregated, and mapped GEP and sediment yield at four scales: pixel, buffered reach area, subwatershed, and tributary basin. We also estimated the specific sediment yield for each reach to illustrate how sediment yield varies downstream through the channel network. The total sediment yield value at the watershed outlet equals the basin average sediment yield. To estimate sediment supplied to each reach (specific sediment yield), the buffered channel network was discretized to define the buffered drainage area on each side of the channel reach, or drainage wings. Using algorithms within NetMap, the total GEP and sediment yield within the drainage wings were attributed to each segment (reach) of the channel network. To estimate the specific sediment yield at a given stream segment, GEP and sediment yield was cumulatively added moving downstream and divided by total upstream drainage area of the buffered incised channel network. The delineation of erosion from incised channels was checked with direct field observations and by draping erosion predictions over air photos.
As indicated previously, sediment supply from the incised Arroyo Mocho
channel network is aggrading portions of the engineered flood control
channels on the Livermore Valley floor. Sediment supply and aggradation of
the flood control channels could be reduced by promoting sediment storage at
upstream locations, for example by reconnecting channels to floodplains. To
help land managers identify ideal upstream locations for sediment storage,
we developed a relative sediment storage potential index (Eq. 5) that is
calculated for each stream reach:
Viewing the GEP layer draped over high-resolution aerial imagery along roughly 50 % of the channel network, we consistently observed steep eroding banks (bare of vegetation) in areas with high GEP values throughout the watershed. Similar observations were confirmed during 2 days of fieldwork, where additionally we consistently observed much more stable banks with vegetation in areas with lower GEP values (Table 1, Fig. 5, also see extensive photo documentation in Bigelow et al., 2012a). In the field, higher GEP values generally corresponded to steeper, more convergent terrain, while lower GEP values corresponded to flatter and divergent terrain. These observations qualitatively indicate that GEP provides reasonable estimates of relative erosion within a watershed. To quantitatively access the accuracy of GEP requires multiple long-term sediment gages within a watershed, which is not yet possible in this watershed. However, when comparing erosion predictions to field inventories in the Oregon Coast Range, the index performed better than hillslope gradient alone or other available erosion models (Miller and Burnett, 2007). As mentioned previously, GEP is a relative measure of erosion within a basin, which alone is highly useful to practitioners, and the conversion to sediment yield is intended to give more meaningful values that likely corresponds to the correct order of magnitude within the context of sediment budgeting technology (e.g., Reid and Dunne, 1996).
Summary of agreement between erosion (GEP) predictions and erosion observed on air photos and in the field.
Incised channel on lower Tassajara Creek basin showing GEP and estimated specific sediment yield at the reach scale (upper) and air photo (lower left) with pixel scale GEP draped over image (lower right).
The spatial distribution of erosion (GEP) and sediment yield reveals strong
patterns at the scales of pixel, reach, subwatershed, and tributary
watershed scales. Starting at the smallest scales, the spatial distribution
of GEP and estimated sediment yield can be evaluated at the level of
individual pixels (4 m
Scaling up to the subwatershed distribution of erosion (mean 2.7 km
GEP and estimated specific sediment yields at the subwatershed scale.
The spatial distribution of GEP and estimated sediment yield at the
tributary basin scale (mean 51 km
GEP and estimated specific sediment yields for the 11 major basins in the Arroyo Mocho watershed. Values in parentheses are the percentage of the total estimated sediment yield for Arroyo Mocho contributed by each basin.
The specific sediment yield for each reach is the buffered upstream sediment supply divided by the upstream buffered drainage area. Estimated specific sediment yields aggregated downstream illustrate how sediment yield varies through the channel network (Fig. 8). Similar to the spatially explicit distribution of GEP and sediment yield across the terrain (Figs. 5–7), this channel segment scale analysis illustrates higher sediment supply from the more dissected steep terrain of Arroyo Mocho canyon. We also aggregated the specific sediment yield downstream by reach and divided it by the total sediment yield at the basin outlet to show the percentage of the total sediment yield incrementally downstream through the channel network (Fig. 9).
Estimated specific sediment yield aggregated down through the
stream network for mainstem streams (draining areas > 2 km
Percentage of total estimated sediment yield aggregated downstream
through the stream network for mainstem streams (draining
areas > 2 km
The spatial distribution of GEP and estimated sediment yield at the various scales across the terrain and through the channel network provides a physical basis for evaluating and prioritizing sediment reduction strategies within a large watershed or region. The spatial distribution of erosion at the subwatershed scale (Fig. 6) is perhaps the most useful for prioritizing potential source control activities, where as the spatial distribution of sediment yield through the channel network (Fig. 7) can focus source control at a finer scale, showing which channels to focus on, rather than entire subwatersheds. After a subwatershed has been prioritized for source control, the reach and pixel scale maps of erosion (Fig. 5) can be used to target specific valley segments, reaches, and banks within the subwatershed. This spatially explicit representation of erosion allows watershed managers to target limited funds to areas where they will achieve the most reduction in sediment supply. It should be noted that source control efforts would help reduce chronic sediment supply, but will not prevent massive aggradation from century-scale extreme events (e.g., Dettinger and Ingram, 2013) that have previously filled valley bottom channels in the Alameda Creek watershed (Bigelow et al., 2008).
Estimated relative sediment storage potential for mainstem
channels (draining areas > 2 km
In combination with the spatial distribution of erosion (Figs. 5–9), watershed managers can also use other parameters extracted from the DEM to prioritize areas for sediment storage downstream of the most erosive areas. In our study basin, the Livermore Valley was historically highly depositional, where tributaries deposited sediment as broad coalescing fans across the valley floor (Williams, 1912). Today, most valley floor channels are physically disconnected from their floodplain by channelization (engineered channels or ditches), channel incision that migrated upstream from that channelization (base level lowering), and other causes. Where there is sufficient space to allow channels to reoccupy floodplains, we estimated ideal locations for promoting sediment storage (e.g., reconnection of channels to floodplains). Using Eq. 5, sediment storage potential is estimated to vary considerably across the stream network, where certain portions of the mainstem streams are estimated to have a higher potential for sediment storage compared to other segments (Fig. 10). These estimations do not capture local forcing of sediment storage from channel constrictions (e.g., bridges) or tributary confluences that may exert primary controls on aggradation. While valley floor channels with high potential for sediment storage are found throughout the watershed (Fig. 10), there are more rural undeveloped valley floors in the eastern and southern portions of the basin, and locations here may be more conducive to restoration projects. As an example application of the sediment storage potential index in this region, Fig. 11. shows an ideal area for sediment storage in lower Cayetano Creek, where the channel could be plugged and diverted onto a new channel created near grade with the former floodplain, or with a series of ponds and plugs, where the channel is widened (creating ponds) and the excavated material from the channel widening is used to plug the channel (e.g., Rosgen, 1997). The sediment storage potential here is moderately high because the valley is wide and the channel gradient is low. Where appropriate, sediment storage from incised streams could also be promoted with check dams (Geyik, 1986) and stone weirs (Shields et al., 1995), or more natural analogs using large wood (Shields et al., 2003), beaver dams (Pollack et al., 2014), and Native American techniques (Norton et al., 2002). In the Arroyo Mocho watershed, initial strategies to reduce erosion from incised channels include construction of a step pool channel (Chin et al., 2009), sediment retention ponds, and earthen check dams that also serve as cattle ponds (Bigelow et al., 2012a). In addition to sediment supply and sediment storage potential, practitioners should also consider the causes, current evolutionary stage, and history of channel incision within the watershed when determining the location of restoration projects to promote sediment storage (Schumm, 1999). The Arroyo Mocho basin has undergone several cycles of incision from tectonic uplift (Sloan, 2006), channelization of valley bottoms (Williams, 1912), changes in vegetation and runoff from land use (Rogers, 1988), and possibly gravel mining. Still, the incision cycles, history, and causes remain poorly understood and such knowledge would better target restoration locations and perhaps prevent or mitigate future cycles of incision from anthropogenic causes.
Example in the Cayetano Creek tributary showing how sediment storage potential can help identify locations to reconnect channels to floodplains.
Like most geospatial tools and models, this approach can be adapted and improved for specific applications and objectives. For instance, erosion rates could be adjusted based on other factors not accounted for in this study such as higher erosion rates for weaker lithologies or precipitation gradients across large basins. As an example, we describe two potential improvements to refine estimates in our study basin to characterize longer term, decadal sediment yield and the bedload component. We also include other considerations for practitioners in their watersheds, including combining analyses of sediment supply with other concerns in a watershed.
The estimated sediment yield in this study represents an average condition using only topographic and vegetation attributes that characterize chronic (annual or persistent) sediment supply from bare steep channel banks. However, sediment supply in many landscapes is highly variable in both space and time resulting from episodic processes driven by interactions among storms, vegetation, and topography (Benda and Dunne, 1997). In the San Francisco Bay region, episodic mass wasting triggered during El Niño storms can dominate decadal sediment supply (Ellen and Wieczorek, 1988; Bigelow et al., 2008). For example, Arroyo Mocho is a tributary to Alameda Creek, where a single flood event comprised 48 % of the total load over a 13-year period (Brown and Jackson, 1973). Episodic mass wasting sources in decadal sediment supply can be estimated through a more detailed sediment budget approach (e.g., Reid and Dunne, 1996), or more simply by increasing the sediment yield in active mass wasting areas based on regional literature values. For example, two similar studies (GMA, 2007; Bigelow et al., 2012b) used digitized maps of active earthflows and local literature values of earthflow rates to appropriately increase the sediment yield estimates at these discrete locations within the watersheds. This approach in part accounts for differences in lithologic erosion rates across the basin, as higher sediment yields from earthflows often occur in specific formations that produce clay-rich soils (e.g., Keefer and Johnson, 1983).
This study approach characterizes total sediment supply to the stream network without regard to the proportion of bedload to the overall yield. Some characterization of the chronic bedload yield variation throughout the stream network would better constrain source control efforts to areas that have a higher bedload yield aggrading valley floor flood control channels. Primary controls on bedload yield are typically drainage area and lithology, where the proportion of bedload generally decreases downstream due to particle attrition (abrasion and breakage) (e.g., Madej, 1995; Benda and Dunne, 1998) and attrition rates will also vary based on lithology, where harder rocks in a catchment produce a larger bedload component (Madej, 1995; Turowski et al., 2010; Mueller and Pitlick, 2013; O'Connor et al., 2014). While approaches characterizing bedload yield variation across a watershed can range in complexity (e.g., Dietrich and Dunne, 1978; Collins and Dunne, 1989; Benda and Dunne, 1997; Madej, 1995; O'Connor et al., 2014), a simplified approach is likely appropriate considering that limitations of sediment budgeting technology typically constrain estimates to the correct order of magnitude (e.g., Reid and Dunne, 1996).
This case study focuses on sediment supply primarily from incised channels, however, the approach can also be applied to most other forms of erosion including landslides, debris flows, and surface erosion across a watershed and including an estimate of sediment delivery (e.g., Miller and Burnett, 2007; Benda et al., 2011). While we have focused on characterizing sediment supply and floodplains to identify potential restoration locations, many land managers have additional considerations when prioritizing stream restoration projects, including aquatic habitat, riparian habitat, and stream temperature. Such considerations can be incorporated using the same type of terrain analysis of additional topographic and vegetation attributes and other appropriate models and data. For example, intrinsic fish habitat potential can be estimated using channel slope, valley constraint (valley width/channel width), and mean annual flow (e.g., Burnett et al., 2007); large wood supply to streams can be estimated using a tree height layer and tree mortality rates (e.g., Benda et al., 2015b; Flitcroft et al., 2016); and thermal loading to streams and effects of riparian shade can be estimated from a vegetation basal area and height layer and bare earth radiation calculations (e.g., Groom et al., 2011). Ultimately the various characterizations of intrinsic fish habitat, sediment supply, large wood supply, stream temperature, and other parameters of interest can be overlain and combined into a robust analysis to delineate and prioritize the most beneficial restoration locations across watersheds and landscapes (e.g., Benda et al., 2007).
This case study shows how practitioners can delineate sediment sources from
incised channels and prioritize potential sediment source control locations
using the following primary steps:
Define the incised stream network and derive an attributed stream network
from the DEM. Characterize erosion potential using a topographic index of slope and
convergence as well as the presence and size of vegetation. Convert erosion potential to sediment yield based on available gage data. Scale up erosion potential and sediment yield from the pixel scale to reach,
subtributary, and tributary, and major basin scales to observe variation in
erosion potential and sediment supply at various scales across the
watershed. Estimate the specific sediment yield for each reach to illustrate how
sediment yield varies downstream through the channel network. Characterize sediment storage potential using a valley width index and
stream power to identify potential areas to reconnect channels to
floodplains.
The spatially explicit erosion modeling approach used here demonstrates a straightforward GIS analysis incorporating contemporary geomorphometric techniques that only requires a DEM and vegetation height layer and characterizes erosion from incised stream banks and the toes of earthflows. The erosion model can also be used to characterize most other forms of erosion and sources further from the channel that require an estimate of sediment delivery (e.g., Miller and Burnett, 2007; Benda et al., 2011). Accordingly, this approach should appeal to those seeking a simpler easily applied erosion model with wider applications compared to more complex models that are highly parameterized or limited to specific erosion processes. The approach can be adjusted for other factors influencing sediment supply (e.g., precipitation gradients, variation in lithology, episodic supply) and combined with analyses of other topographic attributes and models to address additional concerns in a watershed such as fish habitat, riparian conditions, and stream temperature. More broadly, this study illustrates a contemporary analysis of geomorphic processes that can be used by practitioners to solve common problems in watersheds.
The Zone 7 Water Agency funded an initial study that provided the basis for this paper. Dan Miller wrote and compiled the code used for DEM analysis within NetMap. Kevin Andras developed the various NetMap GIS layers. Jamie Kass and Julie Beagle helped obtain and process the lidar data and develop the vegetation height grid. Lester McKee provided a review of the initial Zone 7 study report. Lorenzo Marchi, two anonymous referees, and the associate editor Giulia Sofia provided thoughtful and constructive comments that improved the manuscript. We thank all of these people for their support of this project. We also express our sincere gratitude to David Bowie, for his beautiful and inspiring life and music that make “the stars look very different today”. Edited by: G. Sofia