Near Surface Soil Moisture Estimation through Fusion of UAV-Enabled Thermal, Optical, and Multispectral Hyperspatial Imagery at the Oak Ridge Earthflow

High-resolution mapping of near surface soil moisture is important for the characterization of landslides. This study focuses on estimating the near surface soil moisture through UAV-enabled high-resolution topographic, visible, multispectral, and thermal remote sensing data collected over four acquisitions in a six-month period. The spatial distribution of surface soil moisture is estimated via remotely sensed parameters on an active, slowly moving landslide in the San Francisco Bay Area. Regression analysis shows statistically significant correlation between the diurnal temperature difference and measured in situ volumetric water content. Additional predictive features are also engineered from collected spectral and topographic data and used in multivariate regressions. The findings of this study show promise in the prediction of near surface moisture via remote sensing on a variety of surface types and diurnal heating conditions.