A data-driven approach for bare surface soil moisture estimation using Sentinel-1 SAR data and ground observations

Abstract

Soil moisture is an important variable in the agriculture system. Likewise, accurate information on soil moisture is needed for the effective modeling of many hydrological and climatological processes. Synthetic Aperture Radar (SAR) operates with the competence to acquire data in any weather condition, has been proved to be sensitive to surface soil moisture. This study has attempted to establish simple experimental relationships to estimate volumetric bare surface soil moisture (smv ) using the SAR satellite-based radar backscatter values (σ°). In this study, the in-situ smv measurements of two study sites in the United States were obtained from the SoilSCAPE project whereas σ° data for the same study sites were obtained from the C-band Sentinel-1 satellite. Initially, four experiments were designed based on various radar configurations i.e., combination of polarization and incidence angle(s) at an individual or combined node(s) of each study site. Following this, the statistical analysis in each experiment was carried out using the high volume data i.e., the long-term time-series σ° and in-situ smv that were clustered in these radar configurations. Subsequently, the relationships were established on the basis of outcome of each experiment. Based on the detailed analysis, it was found that out of four experiments, only one experiment outcome in terms of correlations statistics, for a particular radar configuration and study site, was found to be significant and accepted for model development. The derived model was applied and validated over the demo farm located in Pune, India. The comparison between the estimated and in-situ smv measurements shows good agreement, with a mapping accuracy of about 8% observed with the radar configuration- vertical-vertical (VV) polarization with a 43° incidence angle.