Development of a Topographic Filter to Identify Dominant Sediment Source Areas in a Watershed

Many US waters are identified as impaired with respect to fine sediment. Efforts to locate and address sediment sources are widespread, expensive, and uncertain. We propose a simple topographic model that may be useful for characterizing sediment source areas. The model takes advantage of the extraordinary effort by the USDA to estimate and map sediment production rates in the US. When combined with high-resolution topography and a measure of total sediment load from a watershed, a simple topographically driven sediment delivery model can be used to identify those parts of the watershed from which most of the sediment yield is derived. For each raster cell in the watershed with its respective sediment production rate, distance, slope, and curvature to the nearest waterway are calculated, along with distance and slope down the channel network to the watershed outlet. Because the model has only five parameters to represent sediment delivery from many millions of raster cells in the watershed, there are many combinations that provide nearly equivalent fit between predicted and observed sediment yield. We do not attribute particular significance to any particular solution, but instead develop a stochastic simulation approach in which sediment yield distribution is determined for any plausible combination of parameters. Taken together, the results provide a simple, but robust estimate of sediment source location. Beginning with sediment source cell at the mouth of the watershed, we use a search routine to map the upstream extent of 90% of the sediment yield in the watershed. The stochastic approach provides a measure of uncertainty in this boundary. With the increasing availability of high-resolution topography, this simple tool provides a useful filter for estimating the dominant sediment source areas in a watershed. We demonstrate the tool using soil loss and TSS gaging results in subwatersheds of the Minnesota River Basin. Extensive agricultural development and changes in hydrologic and soil systems over the past 150 years have increased sediment yields by an order of magnitude. Multiple turbidity TMDLs have been defined for the watershed and the expense for addressing sediment sources will be enormous. We show that 90% of the sediment yield in a subwatershed occurs in the areas near the basin outlet—about lower one quarter of the drainage area. The inputs of the model are easily obtainable without the requirement of extensive parameter inputs, the process makes intuitive sense in connecting topographic influences to soil loss and sediment delivery, and outputs are easily communicable to a wide range of stakeholders. The identified high-yield areas are presented in a GIS-based map format, and can be used to prioritize management actions to improve environmental management decisions.