Spatial modeling of soil depth and landscape variability in a small, forested catchment.

摘要

The linkage between soil, topography, and landscape processes remains poorly understood. The objective of this thesis was to examine the relationship between soil depth and topography. Depth-to-bedrock was measured at 190 locations using a dynamic cone penetrometer and ground-penetrating radar (GPR) in a 10 ha catchment in the Georgia Piedmont. Depth-to-bedrock was correlated to the slope length index (p = 0.77) and the 30 m profile curvature (p = 0.66). This suggests that the distribution of soils is controlled by mass wasting processes. Analysis of variance indicated the confounding effect of landscape position. Variogram analysis of depth-to-bedrock indicated a high microscale variability. Kriging with topographic variables as an external drift did not significantly improve interpolations compared to ordinary kriging using GPR depths. The hillslope scale corresponds to the broad definition of soil catena but contains a nested level of microscale "random" variability. The correlation with topography remains unresolved at scales less than hillslope units.