Predicting soil depth to bedrock in an anthropogenic landscape: a case study of Phewa Watershed in Panchase region of Central-Western Hills, Nepal

Abstract

The Soil Depth to Bedrock (SDtB) parameter is highly variable over the landscape and is an important input for soil surface modeling including various types of mass movement that most often ignores the variability of soil depth. In landscapes that have been heavily modified by humans, SDtB parameters do not necessarily depend on natural processes, making it difficult to apply physically-based approaches. This study explored the possibility of using topographic attributes to model the SDtB in the complex topography of Phewa watershed in the Panchase region of Nepal. In June 2017, 865 SDtB points were surveyed along the excavated rural roads (approximately 300 km) within the watershed (111 km2). Topographic attributes such as slope, curvatures, altitude and compound terrain index were derived from a Digital Terrain Model along with Land use land cover attributes derived from recent Landsat remote sensing images. Utilizing these attributes as explanatory variables, we implemented simple kriging (SK) and multiple linear regressions (MLR) to predict SDtB for the watershed. The results showed that the MLR predicted SDtB map was statistically significant (R2 = 68%, EC = 0.93) over the SK (R2 = 12%, EC = -0.46). We concluded that the MLR approach in predicting SDtB for the complex geo-morphologic landscape such as the Phewa watershed in the Panchase region of Nepal can effectively be used, helping to inform the evolution of land-use as the region continues to develop.