H11H-0856
Width-Function based Instantaneous Unit Hydrograph rainfall runoff model: relationship between Width Function, hillslope flow velocities and concentration times
The Width Function (WF), measuring the number of channel links or the portion of contributing area at the same hydrologic distance (i.e. the distance measured along the downhill flow path) to the outlet, is an important river basin hydrologic parameter. The rescaled WF, obtained by weighting the flow distances with the corresponding flow velocity, characterizes the residency time distribution, that represents the WF Instantaneous Unit Hydrograph (IUH) or WFIUH, a useful tool for the prediction of the hydrologic response in ungauged basins. The rescaled WF is usually obtained by hypothesizing a constant flow velocity for the entire basin neglecting the heterogeneity of surface flow patterns throughout the basin. This works aims to verify the eventual improvements in the characterization of the flow hydrograph of the WFIUH considering the spatial distribution of hillslope flow velocities as a function of local terrain properties (e.g. slope, land use etc) or river basin lumped parameters (e.g. concentration time).
H11H-0857
On the Determination of Surface Flow Paths From Gridded Elevation Data
Multiple flow direction algorithms are commonly thought to be a useful means for determining drainage areas from gridded elevation data. It remains however unclear whether these algorithms can be used to describe surface flow paths and gravity-driven processes across a terrain without causing unrealistic artificial dispersion of flow. To explore this issue, a unified algorithm for the determination of flow directions is developed and new methods for the validation of surface flow paths are introduced. The unified algorithm allows, by setting appropriate parameters, to perform local or path-based analyses, and to experiment different combinations of single and multiple flow directions in a morphologically significant manner. The new validation methods use drainage systems delineated from contour elevation data as a reference and consider the overlapping between these drainage systems and the corresponding drainage systems obtained from gridded elevation data. A purely morphologic analysis is carried out. The obtained results suggest that dispersive methods may be preferred over nondispersive methods if the computation of the spatial pattern of drainage area, especially along divergent terrains, is the main focus. On the other hand, the results reveal that path-based nondispersive methods should be preferred over dispersive methods if the delineation of drainage systems and surface flow paths is an important focus. Path-based nondispersive methods are found to be a reliable means for the determination of surface flow paths from gridded elevation data, and to provide therefore a sound basis for the distributed description of gravity-driven processes. Future work is needed to formulate models of physical dispersion for water, sediments and solutes upon this purely morphologic basis.
H11H-0858
Topographic Analysis for the Prairie Pothole Region
The unique topography of the pothole region of the North American prairies creates challenges for properly determining basin contributing area. Retreating Pleistocene continental glaciers deposited a glacial till that produced a hummocky terrain with numerous depressions or potholes within the landscape that impound runoff. These potholes impound a great deal of runoff and display intermittent 'fill and spill' behaviour which can produce large step-functional shifts in the land surface area contributing to stream flow. Both the storage capacity and the topological relationships between potholes determines the nature of the shifts in basin contributing area and runoff. Most currently available automated methods such as TOPAZ or ArcGIS treat depressions as artefacts that are removed by depression-filling and subsequent routing of flow across the resulting flat areas. This effectively defines the threshold storage value, that when satisfied, allows 100% of the basin to contribute runoff to the basin mouth. However, most runoff events in the prairie pothole region are sub-threshold events that allow only a portion of the watershed to contribute surface runoff to the outlet. We propose an automated method for determining contributing area that incorporates the dynamic fill-and- spill behaviour of prairie potholes. We apply the algorithm in prairie pothole basins both to demonstrate their efficacy and to test the potential for using conceptual curves to describe a hypothesized non-linear relationship between decreasing potential storage in the landscape and contributing area. Results indicate that the proposed conceptual curves represent the relationship between potential storage and contributing area in the test basins very well.
H11H-0859
Hydrological landscape analysis based on digital elevation data
Topography is a major factor controlling both hydrological and soil processes at the landscape scale. While this is well-accepted qualitatively, quantifying relationships between topography and spatial variations of hydrologically relevant variables at the landscape scale still remains a challenging research topic. In this presentation, we describe hydrological landscape analysis HLA) as a way to derive relevant topographic indicies to describe the spatial variations of hydrological variables at the landscape scale. We demonstrate our HLA approach with four high-resolution digital elevation models (DEMs) from Sweden, Switzerland and Montana (USA). To investigate scale effects HLA metrics, we compared DEMs of different resolutions. These LiDAR-derived DEMs of 3m, 10m, and 30m, resolution represent catchments of ~ 5 km2 ranging from low to high relief. A central feature of HLA is the flowpath-based analysis of topography and the separation of hillslopes, riparian areas, and the stream network. We included the following metrics: riparian area delineation, riparian buffer potential, separation of stream inflows into right and left bank components, travel time proxies based on flowpath distances and gradients to the channel, and as a hydrologic similarity to the hypsometric curve we suggest the distribution of elevations above the stream network (computed based on the location where a certain flow pathway enters the stream). Several of these indices depended clearly on DEM resolution, whereas this effect was minor for others. While the hypsometric curves all were S-shaped the 'hillslope-hypsometric curves' had the shape of a power function with exponents less than 1. In a similar way we separated flow pathway lengths and gradients between hillslopes and streams and compared a topographic travel time proxy, which was based on the integration of gradients along the flow pathways. Besides the comparison of HLA-metrics for different catchments and DEM resolutions we present examples from experimental catchments to illustrate how these metrics can be used to describe catchment scale hydrological processes and provide context for plot scale observations.
H11H-0860
Assessment of PEM4PIT parameters by analyzing catchment form and processes
A physically based approach (PEM4PIT, Physical Erosion Model for PIT removal) was recently introduced to correct hydrologic spurious depressions (pits) and flat areas in Digital Elevation Models. Despite PEM4PIT proved to be more suitable than commonly used geometric methods to reconstruct hydrologically connected topography and reliable stream network metrics, fundamental in rainfall/runoff modeling, the best choice of the three model parameters (slope-area exponent, (theta) erodibility (beta), and diffusivity (D)) remained an open issue to be investigated. In this work a methodology to select the optimal set of parameters is described. In particular two methods are illustrated for theta estimation, the former using Horton ratios (Flint, 1974) and the latter analyzing the "approximate characteristic form" of slope profile (Kirkby, 1971). Regarding beta and D, they are estimated applying separately a simplified topographic equilibrium equation for the basin domains interested by the fluvial erosion and diffusion processes, respectively. ASTER DEMs of several case studies watersheds are used as input dataset; finally PEM4PIT results are compared, in terms of extracted network, with the results of standard approaches and with the digitized bluelines.
H11H-0861
Deriving Global Drainage Networks from SRTM Elevation Data
SRTM elevation data are widely used to produce hydrographic derivatives such as river networks and watershed delineations. Its seamless, near-global coverage makes the data particularly interesting for large scale applications, but standard GIS software is typically ill prepared for dealing with the large data quantities when processing SRTM data at a global scale. After developing various customized GIS tools, a new hydrographic database has been created by World Wildlife Fund, termed HydroSHEDS, which is derived from SRTM elevation data at 3 arc-second (90 m) resolution at global extent. The original SRTM data have been hydrologically conditioned using a sequence of automated procedures and supported by various ancillary data sources. Several of the processing steps used in the development of HydroSHEDS represent standard GIS techniques for hydrographic applications, such as sink detection and filling, stream burning, and deriving flow directions. However, because of the unique characteristics of SRTM data and the large-scale aspect of the applied procedures, some new techniques and algorithms were developed, including customized methods of data filtering, valley carving, and iterative processing. Extensive manual corrections were made where necessary. Finally, an algorithm was developed to upscale HydroSHEDS data to coarser resolutions which are better suited for global applications. Quality assessments indicate that the accuracy of HydroSHEDS significantly exceeds that of existing global watershed and river maps. Nevertheless, various issues remain, including the filling of large data gaps, and the filtering and removal of vegetation cover which poses a particular source of error in large, low-relief wetland and coastal areas. HydroSHEDS data is offered by USGS free of charge for non-commercial applications at http://hydrosheds.cr.usgs.gov/
H11H-0862
Effect of DEM resolution and threshold area on the hydrologic response at catchment scale
It is widely recognized that catchment geomorphology relationships can be used as predictors of catchment flood properties. Starting from this consideration, this paper want to analyze the effect of DEM resolution and threshold area on the hydrologic response at different basin scale, by using the approach of the geomorphologic instantaneous unit hydrograph (IUH) in two different formulations. The analysis is carried out on seven different catchments located in Sicily (Italy) with area ranging from 30 to 1800 square km, using different seven threshold area values and four different grid size of DEM, obtained from the original 20 m DEM after a resample process. The assessment of hydrological response needs the calculation of some geomorphic indexes based on two different stream ordering (Horton–Strahler and Shreve): the Horton's ratios, the length of the outlet stream, the magnitude, topological diameter, the mean length of the internal links. The geomorphologic model (GIUH), based on Horton – Strahler stream ordering, and the topological model (TIUH), based on Shreve stream ordering, are used to simulate the hydrological response of the basins, starting from the impulsive response (IUH), forced with a synthetic rainfall. The results show that the influence of the grid DEM size and the threshold area depends on the model used, since the sensitivity of the two model to the geomorphic parameters is different. The TIUH model provides flood discharges more or less invariant with the resolution of DEM and the threshold area. This is due to the invariance of the two parameters of the model, the magnitude and the mean length of the internal links, with the resolution of DEM while the same parameters are variable, as a power function, with the threshold area, providing peak discharges characterized by low variance. On the contrary, the parameters of TIUH model (the length of the outlet stream and the Horton's ratios) are more sensitive to the two scale factors here analyzed, providing, in this way, highly variable peak discharges. In conclusion, while evaluation of flood discharge with GIUH could be affected of remarkable errors, especially in the case of large grid size and high threshold area, the TIUH model provides hydrological response more stable and more or less invariant with the two above mentioned scale factors.
H11H-0863
Impact of Digital Elevation Model (DEM) Aggregation Methods on Watershed Properties for a Flat Regional Basin
Digital Elevation Models (DEMs) are widely used in hydrological modelling to describe the geomorphological variability of watersheds. However, due to limited computational and memory resources, the resolution of the most precise available DEMs is often too fine to run models over regional scales. DEMs therefore need to be aggregated to coarser resolutions, affecting the representation of the land surface. In the flat Lake Chad basin (2.5 M km2), six algorithms are assessed to aggregate the SRTM (Shuttle Radar Topography Mission) DEM from 3 sec (90 m), the resolution of the released product, to 5 min (10 km), a resolution directly usable by a regional hydrological model. The results show that the suitability of aggregation methods strongly depends on the hydrological entity under consideration. First, concerning Lake Chad bathymetry, the level-area and level-volume relationships are preserved with all the methods assessed except the minimum and maximum which distort the lake bottom topography. Second, regarding the drainage network, point oriented methods such nearest neighbour, minimum and maximum induce local channel capture effect. In contrast, methods which smooth the topography such mean and median result in a consistent river network representation. Third, in order to preserve the small depressions governing the floodplain dynamic, only the nearest neighbour and the maximum are relevant. Indeed, with these methods the simulated evaporated volume over the surface waters of the Lake Chad basin is close to the estimations, whereas with other methods it is importantly underestimated. So as to optimize the preservation of these three hydrological entities during the aggregation procedure, a double-step method is therefore proposed. The first step consists of a preliminary median filtering of the 3 sec DEM in order to smooth the very local variations in elevation including SRTM speckle artefacts. The filter window size (39 sec) corresponds to the average range of variograms computed over flat areas of the basin. In the second step, the filtered DEM is aggregated to 5 min via the nearest neighbour method. With the resulting DEM, the Lake Chad bathymetric curves are close to those extracted from the 3 sec DEM (RMS deviation < 3.5%) and there is no anomaly on the extracted drainage network. Moreover the simulated floodplain hydrology is consistent with estimations (3% underestimation for simulated evaporation volumes).
H11H-0864
Generalized Methods for Terrain-based Flow Analysis of Digital Elevation Models
Topography is an important land surface attribute for hydrology that, in the form of Digital Elevation Models
(DEMs), is widely used to derive information for the modeling of hydrologic processes. Much hydrologic
terrain analysis is conditioned upon an information model for the topographic representation of downslope
flow derived from a DEM, which enriches the information content of digital elevation data. This information
model involves procedures for removing spurious sinks, deriving a structured flow field, and calculating
derivative surfaces. We present a general method for recursive flow analysis that exploits this information
model for calculation of a rich set of flow-based derivative surfaces beyond current weighted flow
accumulation approaches commonly available in Geographic Information Systems, through the integration of
multiple inputs and a broad class of algebraic rules into the calculation of flow related quantities. This flow
algebra encompasses single and multi-directional flow fields, various topographic representations, weighted
accumulation algorithms, and enables untapped potential for a host of application-specific functions. We
illustrate the potential of flow algebra by presenting examples of new functions enabled by this perspective
that are useful for hydrologic and environmental modeling. Future opportunities for advancing flow algebra
functionality could include the development of a formulaic language that provides efficient implementation
and greater access to these methods. There are also opportunities to take advantage of parallel computing
for the solution of problems across very large input datasets.
http://www.engineering.usu.edu/dtarb/taudem
H11H-0865
Evaluating topographic and hydrologic attribute sensitivity to upscaled resolution DEMs from LIDAR data
Raster-based Digital Terrain Models (DTMs) have been extensively used for determining topographic attributes used for hydrologic modelling topographically based. Several studies have been reported that the hillslope hydrology response is strongly affected by the local topography. Despite the increasing availability of fine resolution topographic data captured by Light Detection And Ranging (LIDAR) technique, some drawbacks arise, both from the computational point of view, and also because the higher detail does not match with the other spatial attributes (e.g. land use, vegetation cover, climate etc.). A compromise is then needed to satisfy the computational effort, and at the same time make the spatial input homogeneous, by either downscaling the coarsest ones or by upscaling the finest ones. Usually, during resampling of original DTM, topographic details could be lost because of smoothing effects. For this reason it is necessary to investigate whether and how a coarser resolution DTM can preserve hydrologic information, crucial for modeling performances and reliability, as uncertainties in the inputs will be propagated into the output prediction, producing biases. In this work two case studies are presented using 1 m LIDAR DTMs. A series of DTMs having 5, 10 up to 20 m grid size are derived from the finest DTM of 1m, this applying standard resampling methods. Several topographic and hydrologic characteristics are tested at different grid cell sizes e.g. the wetness index, and the flow path length of the main channel, in order to test the changes in the lag time between precipitation and flow peak discharge, resulting in different hydrographs. Even if some of these attributes prove to have few differences in basin averages by changing DTM resolution, it is here shown that, unlike for the lumped models, where the heterogeneities are ignored, for the semi-distributed and distributed models the input spatial variability can affect significantly the results.
H11H-0866
The Impact of the Dachaoshan Dam on Seasonal Hydrological Dynamics in the Main Stream of the Mekong River
In the Mekong River watershed, traditional social and industrial systems have long existed in harmony with water and biological resources. Since the 1950s, many dam-construction projects have been started to develop power and water resources to meet increasing demand for energy and food production. Since the 1970s, there have been temporary interruptions to these projects because of civil war or regional volatility of international relations. Many of these projects have been restarted in the last 15 years. This raises international interest, as there are transboundary issues cross-border issues related to both development assistance and environmental conservation. By 2008, two Chinese dams had already been completed (the Manwan dam in 1996 and the Dachaoshan dam in 2003) on the Mekong River in Yunnan province. Dam construction has some positive impacts, such as electricity production, management of water resources, and flood control. However, upstream control of water discharge can have negative impacts on traditional agricultural systems and fisheries downstream from the dams, such as drastic changes in flow volume and sediment load. We used hydrological simulation of the watershed to quantify the impact of the construction of the Dachaoshan dam by comparing annual water discharge and sediment transport before and after the dam was completed. Our main objectives were to use watershed hydrologic modeling to simulate changes to annual hydrological parameters and sediment transport, and to map spatio-temporal changes of these data before and after dam construction. Our study area covered the part of the Mekong River main channel that extends about 100 km downstream from the junction of the borders of Myanmar, Thailand, and the Lao People's Democratic Republic. We used five data validation points at 25-km intervals along this section of the river and calculated model parameters every 1 km. The years we modeled were 1990 (began dam construction) and 2006 (after dam completed). We used the MIKE-SHE and MIKE11-Enterprise (developed by DHI) to calculate seasonal changes of water level, water velocity, and sediment transport. These models provided both water discharge and sediment transport dynamics at each modeled point along the river. The sediment budget was calculated as the difference of sediment load by volume between adjacent modeled points. All parameters used in the model were calibrated with field survey data; the river structure and water flows were measured in November 2007. To validate our simulated results we used historical water-level records from the towns of Chensean and Chencone. To determine the relationship between water discharge and sediment load, we analyzed the turbidity of monthly river water samples collected in the study region between November 2007 and November 2008. Our watershed runoff models simulated water discharge and sediment load at 1-km intervals and 1-h time steps for 1990 and 2006. The model results were compiled in GIS format and maps were produced to provide simple spatial displays of modeled parameters. Our simulations show that after construction of the dam, there was a moderate decrease in peak discharge volume and water velocity during the rainy season from August to September.
H11H-0867
Effects of Bathymetric Lidar Errors on Hydraulic Models
Advances in the technology of airborne bathymetric lidars have recently made it possible to map the three- dimensional topography of channels with accuracy and precision similar to that of terrestrial lidar maps of subaerial landscapes. Bathymetric lidar-generated DTMs can potentially define channel boundary conditions necessary to support computational fluid dynamics models (CFDs). However, little is known about the effects of lidar measurement errors on the accuracy of CFD model predictions. We investigated the effects of point errors in bathymetric lidar data on predictions of the USGS two-dimensional CFD model MD-SWMS. Random elevation errors were introduced to each measurement point in bathymetry surveyed with the Experimental Advanced Airborne Research Lidar (EAARL). The errors were produced by sampling from a normal distribution with a mean vertical error of zero and one standard deviation of 10cm. We then compared on a relative basis MD-SWMS predictions of water elevation, stream bottom shear stress and vertical flow velocity as a function of discharge in the original bathymetry and bathymetry with random point errors. We also compared predictions of the same variables in MD-SWMS using lidar and ground surveyed bathymetry. Results show only small differences on the order of a few centimeters in predicted water elevations. Median velocity errors were less than 2% and median shear stress errors less than 1.5%. Bathymetric errors had greater impact at low flow conditions when the flow field tends to more closely follow the bed topography. Shear stress and velocity errors were also greater close to the banks where small elevation errors can cause a few model cells to change from wet to dry or the reverse. Our results suggest this bathymetric lidar will map channel boundary conditions with an accuracy that will support 1D and 2D CFD modeling for most purposes.
H11H-0868
Performances of terrain partitioning methods in shallow landslide modelling
The realistic representation of flow through complex terrain is a major issue to address in developing spatially distributed hydro-geomorphologic modelling. An attempt to evaluate the role of terrain partitioning method in shallow landslide triggering models is here investigated. Shallow landslide susceptibility is determined by using the same spatially distributed model (Rosso et al., 2006) starting from both contour-based and grid-based topographic surface representation. Hillslope hydrology is modelled by coupling the conservation of mass of soil water with the Darcy's law used to describe seepage flow. Soil mechanics is investigated by infinite slope stability analysis. This yields a simple analytical model capable of accounting for the transient effects of precipitation in triggering shallow landslides. Model performances in detecting instability are evaluated by using the Rosso-Rulli-Vannucchi indexes (Rosso et al., 2006) in comparing observed landslide vs. simulated scenarios. Hereafter model based on contour will be addressed as RRV-Contour and the grid based model as RRV-Grid. Models application by considering different durations of a precipitation episode are shown for the Mettman Ridge study area in Oregon, USA. For the study area, the available data set includes high resolution digital elevation data acquired through Airoborne Laser Scanner (ALS), field tests for hydrologic and geomechanic parameterization of soil, observations of hydrologic variables and highly reliable field surveys for shallow landslide mapping. The results show that the two types of topographic discretization have different performances, with hillslope elements based on the flow tube approach better describing the shallow landslides initiation.
H11H-0869
Assessment of SRTM Precision for River Slope and Cross Section by Comparison with Satellite Altimetry
Slope of the river is a widely used parameter for discharge estimation. In poorly monitored basins, SRTM have been used to determine river slope (Le Favour et Alsdorf, 2005). Also, SRTM is expected to constrain long wavelength slope in future altimetry mission, such as SWOT. It is then important to assess the quality of SRTM data over river surface, floodplains and wetlands, in particular in case of dense vegetated cover of the river banks, in order to evaluate if such data can reach modeling requirements. We present two types of analysis : river longitudinal profiles and river cross sections extracted from SRTM compared with altitudes computed from altimetry data (ENVISAT, T/P, ICESAT, GPS surveys).
H11H-0870
Estimating flooding potential associated with dam breach using GIS-based modeling
One of the damaging events attributable to engineering structures is inundation flooding due to dam breach. A variety of programs exist for determining the extent and magnitude of flooding downstream, which are in widespread use to determine the lives and property at risk to such a failure. The U.S. Army Corps of Engineers' Hydraulic Engineering Center's River Analysis System (HEC-RAS) is an example of such a dam break/flood routing model used to breach a dam and route the water downstream. Producing a stable dam break model and inundation map from HEC-RAS is a process requiring days to weeks of the modelers time. Such an effort is reasonable and appropriate for a detailed examination of the risks associated with such a failure, but the demands of such an approach preclude this tool to be used for a large inventory of dams that might fall under the regulatory responsibilities of a government agency. In an effort to minimize the time required for generating dam breach inundation maps, a method was developed to delineate dam inundation using a simplified procedure that is 1) consistent with conventional practice of estimating inundation resulting from a dam breach, 2) capable of being applied broadly to data layers that are available on the scale of entire state, and 3) computationally robust to enable it to be applied across a broad set of topographic and geographic conditions. VADUS Vulnerability Assessment of Dams using Simplifications automates generation of inundation maps by extracting data from a standard set of data Geographic Information Systems (GIS) layers. The VADUS methodology is based on a combination of procedures from 1) United States Bureau of Reclamation (USBR), 2) Washington State's Department of Ecology (WSDE), and 3) United States Geological Survey (USGS). USBR's method for determining flood peak discharges downstream was used to determine the magnitude of dam break flood waves, WSDE's method of estimating flood extent due to roughness and cross-section area was incorporated to determine cross-sectional area through which peak discharge passes, and a USGS methodology for determining the areal extent of inundation was incorporated to find the extent of the flooded area. The modeling tools were applied to over 3000 dams in Mississippi across a range of sizes and geographic conditions.
H11H-0871
Accuracy of Forest Road and Stream Channel Characteristics Derived from LiDAR Terrain Data in Forested Mountain Conditions
Airborne LiDAR data is now commonly used to create digital elevation models with a vertical accuracy and grid resolution that surpasses traditional sources of topographic data. The improved topographic detail of LiDAR-derived DEMs allows for identification and measurement of important topographic features including forest roads, historic landslides and potentially unstable areas. When used conjunction with GIS-based flow accumulation algorithms, high-resolution DEMs can also provide improvements to the positional accuracy, longitudinal profile, stream lengths of topographically-derived stream channels. LiDAR data collected over forested and steep terrain however, requires careful evaluation. The reduced density and variable spacing of LiDAR ground returns, which is often encountered in these settings, reduces the effective resolution of LiDAR DEMs and may limit fine-scale topographic mapping. This study tested the completeness and accuracy forest road and stream channel features mapped using a 1.5 m LiDAR-derived DEM for a 526-hectare watershed in the Santa Cruz mountains in California. Positional accuracy, longitudinal slope and length of LiDAR-derived features were compared to field-survey measurements obtained using total station. LiDAR-derived measurements were also compared to measurements from traditional data sources such as a USGS 10 m DEM and a 1 m orthophoto. The LiDAR- derived road was manually digitized from the LiDAR hillshade and slope layers, with a positional accuracy of 2.2 m normal to the field-surveyed centerline. LiDAR-derived road slope was accurate to within 0.59% of field-surveyed slope at a 95% confidence interval. The entire 4 km road length could be mapped using the LiDAR data. Due to the dense overstory canopy only 18% of that road length could be identified from the orthophoto. Stream channel features derived using Arc Hydro were compared to field-surveyed data at six 100 m study reaches. The average positional accuracy of the derived channel locations was 2.7 m normal to the field- surveyed thalweg. Longitudinal channel slope was measured to within 0.49% compared to the field- surveyed slope while channel length was underestimated by an average of 6 percent. These measurements provided an improvement over the features derived from the USGS 10 m DEM, where channel features had a positional accuracy of 8.7 m, channel length was underestimated by 11% and had a longitudinal slope error was one order of magnitude greater than the LiDAR-derived measurement. These findings indicate that even in densely forested areas, LiDAR has the potential to support accurate, watershed-wide inventories of forest road and stream channel features, which can contribute to our understanding road erosion, sediment transport, and their effects on the stream channel network.