Microwave remote sensing of surface soil moisture and its application to hydrologic modeling

This thesis addresses the problems associated with the retrieval of surface soil moisture distributions from microwave remote sensing measurements, and the application of this information to hydrologic simulations. A qualitative analysis of aircraft radar and radiometer data collected from two field campaigns is conducted to examine the sensors' behavior under various land surface conditions. For the Slapton Wood catchment in Devon, England, the analysis results indicate that existing soil moisture retrieval algorithms and theoretical scattering models often produce biased estimates; and there is a need to develop new algorithms for the NASA Jet Propulsion Laboratory airborne Synthetic Aperture Radar (AIRSAR)--the sensor used in the experiments. A signal simulation procedure based on a calibrated coupled vegetation-surface scattering model, is developed to enhance the limited experimental data set. Two different techniques (stepwise regressions and artificial neural networks) are employed to devise semi-empirical retrieval algorithms for grass-covered areas. Results from a verification study based on 250 hypothetical conditions indicate that the average root mean square error of volumetric soil moisture estimates is approximately 3 {dollar}sim{dollar} 7% when assuming no a priori information concerning the illuminated areas.; For the Mahantango catchment in central Pennsylvania, linear regression models are developed to relate AIRSAR backscatters with soil moisture. The microwave derived soil moistures are compared with ground measurements and predictions from hydrologic models. The results suggest that both passive and active sensors correctly reflect the temporal variations of soil moisture. The model predictions based on the standard streamflow-derived initial condition significantly overestimate the surface soil moisture content. A two-layer process-based hydrologic model is developed to improve the simulations. This model uses remotely sensed soil moistures as a feedback to adjust the catchment average water table depth and obtains satisfactory results in good agreement with field measurements. The simulation results point out that for small areas such as the studied catchment, the advantage of finer spatial resolution soil moisture information upon hydrologic simulations is not decisive. Finally, a systematic framework is constructed to fully incorporate multi-temporal soil moisture data from future satellite sensors into the developed hydrologic model.