Prediction of the Variability of Soil Depth Using Qualitative and Quantitative Geomorhological Information: Sierra Nevada, CA, USA

Abstract

The variability of soil depth at the catchment scale is an essential input for environmental modeling. If the model requires high resolution inputs, the inherent variability of soil depth in granitic and forested regions poses a challenge to capture representative values. To help to address this problem, I propose that, in soil studies, qualitative and quantitative geomorphological information at the catchment scale should be used as inputs of soil-landscape models. Qualitative information is presented as a geomorphological map, and quantitative data are generated by algorithms applied to the high resolution digital elevation model.

In this dissertation, the term soil depth is used to mean the weathered subsurface layers composed of mobile regolith and saprolite. Measurements of soil depths obtained from 204 hand drillings until refusal, and 6,645 estimations of depths collected using ground penetration radar (GPR) were compared, Alone, the hand drilling data showed lack of autocorrelation, however, a significant correlation was found between GPR and auguring measurements (r=0.9, p<0.001), validating this method to capture the spatial variability in a continuum. Therefore, GPR resulted to be a valid, inexpensive and quick tool to survey soil depth in this complex environment.

Contrasting several geomorphological variables and measurements of soil depths, significant correlation with lineaments, slope and wetness index suggest that geomorphology is a significant factor in the distribution of soil depths at the catchment scale in this environment. This relationship was then assumed for modeling the soil depth in a catchment scale. Because the linear regression model and geostatistical methods are not valid approaches for the available sampling design, the use of a simple soil-landscape model, based in the work of Dahlke et al. (2009), is proposed to map the measurements of soil depths to a catchment scale. Indeed, the main strength of this model lies in its independence of sampling design. The resultant prediction maps at different resolutions showed the importance of selecting an appropriate scale of work and adjust the density of GPR survey in the performance of predictions.