Soil thickness and affecting factors in forestland in a karst basin in Southwest China

Abstract

Soil erosion is a process of land degradation worldwide, and it is a serious problem in karst mountainous areas. To clarify the leading factors contributing to soil erosion in forestland in a karst mountainous basin in Southwest China, the environmental characteristics of 424 designated sites in the Houzhai River basin were investigated and analysed. Artificial neural networks (ANNs) were used to predict soil erosion based on environmental factors. The results indicated that soils in karst mountainous areas are highly heterogeneous (shallow and discontinuous), and soil thickness ranged from 7 to 100 cm with a mean value of 31.05 cm. Correlation analysis revealed that soil gravel, slope gradient, soil organic matter and vegetation were the basic driving forces for soil erosion in forestland in this karst mountainous area. In addition, we found that ANNs could be employed for predicting soil thickness (the correlation coefficients between observed values (validation data) and predicted values ranged from 0.73 to 0.96, with a mean value of 0.85). The normalized importance analysis indicated that altitude (0.13), soil gravel (0.13), soil organic matter (0.12), rock outcrop (0.12), slope gradient (0.12) and soil bulk density (0.11) should be prior considered for soil erosion prediction. Our study suggested that enhance vegetation coverage could be the effective ways to the aspect of prevention and controlling of soil erosion in Karst areas.