Factors controlling the spatial variability of surface soil particles using GLMs and GAMs

According to the soil particles data, DEM elevation, slope, distance to river, distance along river, land use, and soil type for 95 typical sampling profiles along the Tarim river mainstream, the generalized linear models (GLMs) and generalized additive models (GAMs) are used to identify the spatial variability pattern of soil particles on the watershed scale and associated driving forces. The results show that the drainage basin sand, silt and clay content along the Tarim river mainstream all present moderate variability. The spatial pattern of soil particles is mainly subjected to the influence of gravity, hydrodynamic force, soil parent material and human activities. Modeling results show that the GAMs performs better than GLMs in regard to the explanation of total deviation. The contribution degree of GAMs for total deviation is 44.19 % (sand), 38.68 % (silt) and 37.46 % (clay). This major influence factors of soil particles distribution are land use, hydrodynamic force, soil type and terrain. The study has provided a way to explore aspects of spatial heterogeneity and to discover how spatial pattern controls the distribution of the soil particles.