Soil texture is considered as essential properties that influence infiltration rates and other hydrological properties. The characteristic of the landscape as the main driving forces of soil formation are affected the spatial distribution of soil texture. Therefore, the estimation of the spatial distribution of soil texture can differ on different scales. To improve the accuracy, we conducted a hyper-scale morphometric approach to determine the extent of the scale of the land characteristics to predict the spatial soil texture precisely. We performed hyper-scale digital soil mapping following morphometric analysis using second order polynomial 3x3 plane square-gridded moving window and growing the window to conduct the hyper-scale analysis at Kalikonto area (East Java, Indonesia). Statistical analysis is used to examine the soil texture content over land characteristics at the different level of scale. We proposed stepwise regression to build a model by adding or removing predictor variables based on the test statistics of the estimated coefficients. The results show that the R~2 values were 0.28 for predicting the sand content, 0.18 for the silt and 0.23 for the clay content. The relationship between soil properties and soil formation on each landscape characteristics are determined while the spatial distribution processes were interpreted successfully.
展开▼
