Improving soil moisture retrieval from GNSS-interferometric reflectometry: parameters optimization and data fusion via neural network

Abstract

ABSTRACT Soil moisture is a vital surface physical quantity in studying the earth’s ecology. It plays a crucial role in the hydrological cycle, crop yield estimation, and ecological monitoring. Global Navigation Satellite System Interferometric Reflectometry (GNSS-IR) technology inversion to obtain high accuracy soil moisture is a hot topic of current research. However, due to the limited available sites, it’s difficult to obtain an extensive and continuous range of soil moisture based on this technique. It is necessary to build algorithms for encryption based on known sites’ data, combined with the corresponding geographic environmental elements. This paper extracted the surface environmental factors affecting soil moisture using high-precision optical remote sensing images. The contribution of each surface environmental element to the soil moisture inversion was analysed using back propagation (BP) neural network optimized by the genetic algorithm (GA). Based on this, ten surface environmental elements (latitude and longitude information, precipitation, temperature, land cover type, normalized difference vegetation index (NDVI), elevation, slope, slope direction, and shading) were identified as critical factors, and a multi-data fusion soil moisture inversion model was constructed. The results showed that the constructed model could better describe the relationship between soil moisture and these elements, and the Pearson correlation coefficient R reached 0.8724, and the RMSE was 0.0346 cm3 cm−3. GNSS-IR technology provides an effective technical means for inversing soil moisture over a large area with high spatial and temporal resolution.