The University of British Columbia Department of Statistics

We present a flexible, integrated statistical-based modeling approach to improve the robustness of soil moisture data predictions in a pilot study in Canada with a small data set with 44 locations and 3 time points. We apply this approach in exploring the consequence of choice of leading predictors and model structures. Competing predictors (covariates) and spatial covariance are often ignored in empirical analyses, validation and modeling studies. But an optimal choice of covariates and consideration of spatial variance may provide a more consistent and reliable explanation of the high environmental variability and stochasticity of soil moisture observational data. Our modeling integrates active polarimetric satellite remote-sensing data (RADARSAT-2, C-band) with ground-based in-situ data for an agricultural monitoring site in Canada, alongside semi-empirical relationships between the dielectric constant and remote sensing data which are commonly utilized to predict near-surface soil moisture. We apply a grouped step-wise algorithm to iteratively select best-performing predictors of soil moisture. In this way, such models may better account for observed uncertainty and be tuned to different applications that vary in scale and scope, while also providing greater insights into spatial scaling (upscaling and downscaling) of soil moisture variability from the field to regional scale. We show that the statistical approach can substantially reduce the uncertainty in field-scale soil moisture for more reliable multi-scale prediction. We find that the use of ground-based data,