Using SAR-Derived Vegetation Descriptors in a Water Cloud Model to Improve Soil Moisture Retrieval

The water cloud model (WCM) is a widely used radar backscatter model applied to SAR images to retrieve soil moisture over vegetated areas. The WCM needs vegetation descriptors to account for the impact of vegetation on SAR backscatter. The commonly used vegetation descriptors in WCM, such as Leaf Area Index (LAI) and Normalized Difference Vegetation Index (NDVI), are sometimes difficult to obtain due to the constraints in data availability in in-situ measurements or weather dependency in optical remote sensing. To improve soil moisture retrieval, this study investigates the feasibility of using all-weather SAR derived vegetation descriptors in WCM. The in-situ data observed at an agricultural crop region south of Winnipeg in Canada, RapidEye optical images and dual-polarized Radarsat-2 SAR images acquired in growing season were used for WCM model calibration and test. Vegetation descriptors studied include HV polarization backscattering coefficient (sigma(degrees)(HV)) and Radar Vegetation Index (RVI) derived from SAR imagery, and NDVI derived from optical imagery. The results show that sigma(degrees )(HV)achieved similar results as NDVI but slightly better than RVI, with a root mean square error of 0.069 m(3)/m(3) and a correlation coefficient of 0.59 between the retrieved and observed soil moisture. The use of sigma(HV)degrees can overcome the constraints of the commonly used vegetation descriptors and reduce additional data requirements (e.g., NDVI from optical sensors) in WCM, thus improving soil moisture retrieval and making WCM feasible for operational use.