Uncertainty assessment of soil erodibility by direct sequential Gaussian simulation (DSIM) in semiarid land uses

Abstract

Erodibility strongly affects both detachment and transport processes of soil erosion, and probability functions are commonly used to characterize empirically estimated point data of soil erodibility with a spatial distribution on a map. The aim of this study is to assess spatial uncertainty of soil erodibility factor resulting from differences of soil properties influenced by land uses, using kriging and Direct Sequential Simulation (DSIM). The study was implemented in Asartepe Dam Basin located in İlhan Çayi sub-catchment of the Sakarya Basin, Ankara, Turkey. Soil surface samples were collected using grid sampling method from five different lands uses, fallow-crop, oak forest, grassland, alluvial cropland, and colluvial cropland. The soil erodibility values (K factor) were calculated for soils of the sites by the Nomograph equation of Revised Universal Soil Loss Equation (RUSLE). The findings demonstrated that kriging provided the best K estimates at unknown locations while DSIM successfully allowed K simulation without any transformation unlike other simulation methods. Kriged estimates falling inside a 95 % probability interval of the simulated values were studied to assess the uncertainty of the soil erodibility factor. The simulation results showed that 99.4 % of the estimated values fell in the probability interval, implying a very low spatial uncertainty level.