Using sequential Gaussian simulation to assess the uncertainty of the spatial distribution of soil salinity in arid regions of Northwest China

Abstract

ABSTRACTAccurately mapping and monitoring the spatial distribution pattern of soil salinity is essential for sustainable soil management and decision-making. The kriging-based interpolation technique is generally used to map the spatial distribution of soil salinity; however, this technique neglects the variation caused by interpolation for each unsampled location. The sequential gaussian simulation (SGS) is an effective tool to collect mapping uncertainties at several locations simultaneously, which is not possible in the kriging-based technique. Soil electrical conductivity has been widely used as an index for soil salinity. Based on 0–100 cm soil profile from 117 locations in the Manas River basin, Northwest China, the SGS algorithm was used to assess the uncertainty of the spatial distribution of soil electrical conductivity. It was found that the SGS algorithm was reliable in reproducing the spatial distribution of soil electrical conductivity. The SGS algorithm reproduced the sample statistics reasonably well. The standard deviations of the samples generated by the SGS algorithm (0.463–0.508 (dS m−1)) were closer to the actual samples (0.675 (dS m−1)) than those generated by kriging (0.454 (dS m−1)). Most of the study area was lightly affected by salinity. Around 30% of the study area was moderately affected, and the heavily affected areas were sporadically scattered across the study area. The spatial uncertainty at multiple point presented a declining trend as the critical probability at a single point increased. The spatial estimation of the soil electrical conductivity in multiple point was more robust than that in the local location because of the low uncertainty.