Soil erosion topographic factor (LS): Accuracy calculated from different data sources

Abstract

The availability and precision of topographic data determines the reliability of the calculated slope length and slope gradient (LS) factor, which limits the use of the universal soil loss equation (USLE) and its adapted versions in region-scale soil erosion risk assessment. One of the common complications is the effectiveness of topographic data with different resolutions or the compatibility of topographic data originated from different sources. In this study, the topographic data from five common sources, 5-m digital elevation model (DEM) based on 1:10,000 topographic maps (5-m topo DEM of 1:10,000), 25-m DEM derived from 1:50,000 topographic maps (25-m topo DEM of 1:50,000), 30-m ASTER GDEM, 30-m and 90-m SRTM DEMs, were individually applied to calculate the LS factors of five catchments with distinct terrain characteristics. Our results show that the computation accuracy of the LS factors in the five study catchments decreased with greater grid sizes derived from the five topographic data sources. Compared to the most precision 5-m topo DEM of 1:10,000, the relative computation error of the mean LS factors was less than 10% when calculated from the 25-m topo DEM of 1:50,000, and that was less than 25% from the 30-m ASTER and 30-m SRTM DEMs. For scenarios in gently rolling areas, the 90-m SRTM DEM with the relative computation error less than 15% could be recommended when there are noises on the open sources DEMs surface. Therefore, depending on the requirements of data accuracy, different data sources can be applied individually or in combined to obtain the optimal predictions of the LS factors. However, such recommendations on data sources proposed in this study appeared to be more applicable for regions with complex terrains. Further studies over a range of terrain features and spatial scales are required to validate the effectiveness of different topographic data sources in calculating the LS factor.