A Coupling Model for Soil Moisture Retrieval in Sparse Vegetation Covered Areas Based on Microwave and Optical Remote Sensing Data

Abstract

Soil moisture is an important component of natural water cycles and plays a crucial role for the healthy development of ecological environment in arid and semiarid areas. Microwave remote sensing techniques are promising to rapidly monitor regional soil moisture. However, major difficulties associated with retrieving soil moisture via microwave remote sensing are attributed to effects of surface roughness and vegetation cover. The objective of this paper is to investigate the potentials of combined roughness parameters and develop a model that mostly relies on the satellite data and requires minimum a priori information for soil moisture inversion over sparse vegetation covered areas through combing Radarsat-2 synthetic aperture radar (SAR) and GF-1 data. For the purpose, the impacts of vegetation on the radar backscattering coefficient were removed by the water-cloud model (WCM) and a new roughness parameter $Zs$ ( $Zs = S^{3}/L$ is defined as a combined roughness parameter) was proposed by simulating relationships between surface roughness parameters, soil moisture, and backscattering coefficients using advanced integral equation model (AIEM). On the basis, a coupling model of soil moisture inversion for blown-wind areas of the Uxin Banner in Inner Mongolia, China, was in turn developed with Radarsat-2 HH and VV polarization data, and the soil moisture content (SMC) values for sparse vegetation covered surfaces were retrieved with limited use of in situ roughness parameters. The in situ measurements and satellite data set were used to validate the reliability of the developed model. The results showed that the retrieved soil moisture levels fell below in situ soil moisture levels before the vegetation effect was removed, and the precision of retrieved soil moisture was effectively improved when the vegetation effect was corrected by the WCM, with the root-mean-square error and mean absolute error decreasing from 7.45% and 6.24% to 5.12% and 3.44%, respectively. According to the map of retrieved soil moisture levels generated for the study area, most of SMCs were below 35% in accordance with field observations. These study results can serve as a foundation for monitoring soil moisture and water environments in arid and semiarid regions.