Algorithm for the retrieval of soil moisture from the radar backscattering coefficient

Abstract

An algorithm based on a fit to the small perturbation method (SPM) was developed so that soil moisture can be derived directly from radar backscattering coefficient data. Using the genetic algorithm with a simulated data set generated from the original SPM model, this algorithm is developed to derive the dielectric constant and then the soil moisture of bare soil surfaces. The fitting algorithm is tested against the original SPM model for incidence angles between 10° and 60°, soil dielectric constants between 3 and 41, and the surface root mean square height between 1 and 20 mm. The fitting algorithm has the same frequency range as the original SPM model. The fitting algorithm computes the backscattering coefficients with an average error of 0.05 dB for horizontal horizontal (HH)-polarisation and 0.15 dB for vertical vertical (VV)-polarisation, where the backscattering observations are taken from the literature. Comparison of the soil moisture derived from the radar backscattering coefficient using the inversion algorithm with the simultaneous measurement shows that the soil moisture retrieved from the inversion algorithm agrees very well for VV-copolarisation (R=0.89, in contrast with R=1 for perfect agreement) and agreement between the calculation and measurement is significant only at the 90% significance level for HH-copolarisation.