Watershed Scale Soil Texture and Uncertainty Prediction using Geostatistical Approaches Incorporating Geophysical Information

Soil texture is a key control for the partitioning of precipitation at the soil surface between water infiltrating or running off. Knowledge of the spatial distribution of soil textural properties at the watershed scale is important for understanding spatial patterns of water movement, and in determining soil moisture storage and soil hydraulic transport properties. Capturing the heterogeneous nature of the subsurface without exhaustive and costly sampling presents a significant challenge. Geophysical methods, such as electromagnetic induction (EMI), provide the possibility of obtaining high resolution images across a landscape to identify subtle changes in subsurface properties. In this work we advance the geostatistical analysis of EMI data to predict both the clay % and its uncertainty across the landscape, using EMI subsurface images from the -38 ha Reynolds Mountain East Watershed near Boise, Idaho. We use EMI maps as a surrogate to predict clay% at unsampled locations using kriging approaches that integrate different levels of information such as clay percentage, apparent electrical conductivity (ECa), and spatial location. Our results show that the multivariate estimation methods incorporating the information in the better sampled ECa data exhibit lower RMSE of estimation. Leave-one-out cross-validation showed that cokriging and regression kriging, integrating ECa data, were able to improve the RMSE by 7% and 28% respectively, relative to ordinary kriging that used only clay percentage data. Electromagnetic induction measurements provide an important spatial exhaustive layer of information that can improve the quality and resolution of spatial soil property information used in ecohydrological, environmental, and agricultural research