Estimating the Spatial Distribution of Soil Properties Using Environmental Variables at a Catchment Scale in the Loess Hilly Area, China.

Abstract

A comprehensive understanding of the spatial distribution and dynamic changes of soil properties are the basis for sustainable land management. Topography and land use types are key factors affecting soil property variability. This study analyzed the effects of land use types and landscape locations on soil properties, based on data from 111 points of surface soil (0–20 cm) in the Zhujiagou catchment on the Loess Plateau of Northwest China. Soil properties include clay, silt, bulk density (BD), soil organic matter (SOM), total nitrogen (TN) and total phosphorus (TP). Land use types include slope farmland (SFL), terrace farmland (TFL), check-dam farmland (CDL), woodland (WL), shrub land (SL) and grassland (GL). Landscape locations include crest (CT), upper slope (US), middle slope (MS), lower slope (LS) and flat valley (FV). Topographical attributes were divided into primary and secondary (or compound) attributes. Correlation analyses were carried out between soil properties and terrain attribute, and multiple-linear regression models were established to estimate soil properties using land use types and topographic attributes as independents. Results indicated that significant differences in soil properties existed between six land use types, except for bulk density. Higher values of clay, silt, SOM and TN occurred in soils from check-dam farmland, but lower values in soils from shrub land. Significant differences among landscape positions were observed for clay, BD, SOM and TN. Clay, SOM and TN contents on flat valley (FV) positions were higher than those of other positions. Different correlations were found between soil properties and terrain attributes. The regression models explained 13% to 63% of the variability of the measured soil properties, and the model for Clay had the highest R2 value, followed by TN, silt, BD, SOM and TP. Validation results of the regression models showed that the model was precise for soil bulk density, but the variation was large and a high smoothing effect existed for predicted values of other soil properties. For TP, the predicted result was poor. Further observations suggested that land use was the dominant factor affecting soil chemical properties. But for soil physical properties, especially for BD, topography was the dominant factor.