Bistatic Scatterometer Measurements for Soil Moisture Estimation Using Grid Partition–Based Neuro-Fuzzy Inference System at L-Band

Abstract

The potential of co-polarization ratio P HH of radar system was investigated for the estimation of soil moisture (SM) along specular direction in the present study. The radar data were collected by indigenously designed ground-based scatterometer system for 20°–60° incidence angles at steps of 10° in the specular direction for HH- and VV-polarizations at L-band. The correlation analysis was done between the angle of incidence and P to select the optimum parameters of bistatic scatterometer system. In this study, hybrid machine learning algorithm combined with fuzzy inference system and artificial neural network called neuro-fuzzy inference system was evaluated for the estimation of SM using the L-band bistatic scatterometer measurements. Grid partition–based neuro-fuzzy inference system called G-ANFIS was used for estimation of SM content. The Gaussian membership function (MF) for partitioning of the data sets into grids was investigated to estimate the SM content. The optimum number of MF was chosen by training the algorithm using different number of MF from 2 at the steps of 1 by trial-error method and calculating the Root Mean Squared Error (RMSE) values between observed and estimated values at different number of MF. The performance index RMSE was used to evaluate the estimation efficiency of the G-ANFIS algorithm. © 2021 Scrivener Publishing LLC.