Analysis of Polarimetric Radar Data and Soil Moisture From Aquarius: Towards a Regression-Based Soil Moisture Estimation Algorithm

Abstract

Many soil moisture radar retrieval algorithms depend on substantial amounts of ancillary data, such as land cover type and soil composition. To address this issue, we examine and expand an empirical approach by Kim and van Zyl as an alternative; it describes radar backscatter of a vegetated scene as a linear function of volumetric soil moisture, thus reducing the dependence on ancillary data. We use 2.5 years of L-band Aquarius radar and radiometer derived soil moisture data to determine the two polarization dependent parameters on a global scale and on a weekly basis. We propose a look-up table based soil moisture estimation approach; it is promising due to its simplicity and independence of ancillary data. However, the estimation performance is found to be impacted by the used land cover classification scheme. Our results show that the sensitivity of the radar signal to soil moisture changes seasonally, and that the variation differs depending on vegetation class. While this seasonal variation can be relatively small, it must be properly accounted for as it impacts the soil moisture retrieval accuracy.