Analysis of a rapid soil erosion assessment tool

Abstract

Soil erosion is a serious problem resulting in degradation of soil systems and nonpoint source (NPS) pollution of water resources. Concentrated overland flow is the primary transport mechanism for many NPS pollutants including soil, and locating areas where sheet flow transitions into concentrated flow is useful for assessing the potential for soil erosion. The ability to predict areas where overland flow transitions to concentrated flow and soil erosion potential is high assists land managers in implementing best management practices (BMPs) to reduce soil erosion and NPS. An erosion model, called the nLS model, was developed to identify transitional overland flow regions. The model is based on the kinematic wave overland flow theory and uses Manning’s n values, flow length, and slope as inputs to determine where overland flow transitions to sheet flow and soil erosion potential increases. Currently, the model has only been tested and validated for watersheds within Kansas. In order to assess model uncertainties and evaluate the model’s applicability to other regions, a sensitivity analysis on key input parameters was conducted. To assess model operations, several sensitivity analyses were performed on model inputs, including digital elevation models (DEMs) and landuse/landcover data (LULC). The impact of slope was assessed using two methods. First, by modifying the DEMs in a stepwise fashion from flatter to steeper terrains, and second, by modifying the elevation of each DEM cell based on the associated elevation error. To assess difficulties that might arise from the parameterization of surface roughness, LULC classes were assigned Manning’s n values within the suggested range using a Monte Carlo simulation. In addition, the critical threshold value used for locating erosion potential sites was modified, and alternative model calculations were used to assess the potential for improving model accuracy. Finally, the model was run using data from multiple sites, including two study areas in Hawaii and two in Kansas. The outputs for each site were analyzed in an attempt to identify any trends caused by site characteristics. Results from this study showed that the nLS model was sensitive to all of the inputs. Modifying the Manning’s roughness coefficient significantly altered the final nLS values and shifted the critical threshold points, especially in areas of the upper watershed. Changes in the slope value modified the nLS model outputs in a predictable manner, but there was some variability, especially in areas with lower slope values. In addition, discrepancies in the DEM, which may be present due to measurement or processing error, were shown to significantly alter the flow paths of a watershed. These findings suggest that accurate roughness coefficients and LULC data are especially important for regions with a steeper topography, and accurate elevation data is important for regions with lower slope values. The results also suggest that the threshold value for the model plays a vital role in locating potential soil erosion sites, and adjustments to this value could possibly be used as a method for calibrating the nLS model. Finally, the alternative model calculations used in this study did not significantly improve the accuracy of the nLS model, so the existing model is sufficient for obtaining accurate nLS estimates. The information gained from this study can improve the assessment of soil erosion processes due to concentrated overland flow. By successfully implementing a land management program that makes use of the nLS models, it should be possible to improve BMP placement and design, helping to improve water and soil quality.