Estimating Soil Moisture Profiles with P-band Low-frequency Radar Measurements Using Regularization

Synthetic aperture radar (SAR) measurements at P-band frequencies (here, ~400 Mhz) allow the sensing of root-zone soil moisture. The NASA Airborne Microwave Observatory of Subcanopy and Soil Moisture (AirMOSS) mission is the first largescale campaign to measure root-zone soil moisture using P-band SAR observations. Radar backscattering coefficients are influenced by vegetation cover and soil roughness as well as soil moisture profile shape. The problem of retrieving the soil moisture profile from single-frequency polarimetric SAR data can therefore be underdetermined. In this presentation we propose to constrain the measurement inversion using Tikhonov regularization. The regularization adds a term to the inversion of the observed backscattering coefficients; this term penalizes solutions that fit the observations but deviate strongly from the long-term mean and covariance of the soil moisture profile. These profile statistics can be determined from hydrologic modelling. The resulting process avoids retrieval of unrealistic soil moisture profiles and, more importantly, makes the solution less sensitive to measurement errors and errors in radiative transfer modelling. The performance of this approach is first investigated using a synthetic case study with one year of simulated soil moisture data representing a grassland in Ione, CA. Additionally, the approach is tested using AirMOSS observations at several locations near the Harvard Forest in Petersham, MA. The soil moisture profile retrievals are insensitive to the accuracy of the long-term profile statistics used a priori in the regularization term, as long as the covariance matrix has a realistic structure.