Parameterization of vegetation backscatter in radar-based, soil moisture estimation Rajat Bindlish a , Ana P. Barros b, * a SSAI, USDA/ARS Hydrology Laboratory, Beltsville, MD, USA b Division of Engineering and Applied Sciences, Harvard University, 118 Pierce Hall, 29 Oxford Street, Cambridge, MA 02138, USA Received 22 May 2000; accepted 28 October 2000 Abstract The Integral Equation Model (IEM) was previously used in conjunction with an inversion model to retrieve soil moisture using multifrequency and multipolarization data from Spaceborne Imaging Radar C-band (SIR-C) and X-band Synthetic Aperture Radar (X-SAR). Convergence rates well above 90%, and small RMS errors were attained, for both vegetated and bare soil areas, using radar data collected during Washita 1994. However, the IEM was originally developed to describe the scattering from bare soil surfaces only, and, therefore, vegetation backscatter effects are not explicitly incorporated in the model. In this study, the problem is addressed by introducing a simple, semiempirical, vegetation scattering parameterization to the multifrequency, soil moisture inversion algorithm. The parameterization was formulated in the framework of the water ± cloud model and relies on the concept of a land-cover (land-use)-based dimensionless vegetation correlation length to represent the spatial variability of vegetation across the landscape and radar-shadow effects (vegetation layovers). An application of the modified inversion model to the Washita 1994 data lead to a decrease of 32% in the RMSE, while the correlation coefficient between ground-based and SAR-derived soil moisture estimates improved from 0.84 to 0.95. D 2001 Elsevier Science Inc. All rights reserved. Keywords: Vegetation; Backscatter; Soil moisture; Radar; Inverse methods; Retrieval 1. Introduction Previously, Bindlish and Barros (2000) used the Integral Equation Model (IEM) developed by Fung, Li, and Chen (1992) in conjunction with an inversion algorithm to retrieve soil moisture using multifrequency and multipolar- ization data from Spaceborne Imaging Radar C-band (SIR- C) and X-band Synthetic Aperture Radar (X-SAR). The RMS error in the estimated soil moisture was of the order of 0.05 cm 3 /cm 3 for data collected during the Washita 1994 experiment (Starks & Humes, 1996), which is comparable to the effect of noise in the SAR data. The original IEM model was, however, developed for bare soil conditions only, and although the retrieval algo- rithm performed well even for vegetated areas (convergence rates were well above 90%), a valid concern is whether it would be equally successful when dense vegetation is present. Vegetation canopies complicate the retrieval of moisture in the underlying soil, because the canopies con- tain moisture of their own. Thus, the retrieved surface water content corresponds to the combined signatures of vegeta- tion and soil water. Due to multiple scattering effects, the interaction between the two contributions and the observed backscatter is highly nonlinear. The key research question is, therefore, how to separate the contribution of vegetation backscatter and absorption from that of soil moisture. Use of remote sensing techniques in the optical domain (visible and shortwave infrared) to monitor vegetation canopies over space and time has been well-documented in the literature (Asrar, Kanemasu, Jackson, & Pinter, 1985; Sellers, 1985; Tucker, Vanpraet, Sharman, & Van Ittersum, 1985). Cloud cover, however, strongly limits the number of available optical images. In addition, these techniques are limited by the observed saturation of the observed signal with increasing biomass. Radar provides a useful tool for assessing biomass, since it is unaffected by cloud cover or low solar zenith angles * Corresponding author. Tel.: +1-617-495-2858; fax: +1-617-496- 1457. E-mail address: barros@deas.harvard.edu (A.P. Barros). www.elsevier.com/locate/rse Remote Sensing of Environment 76 (2001) 130 ± 137 0034-4257/00/$ ± see front matter D 2001 Elsevier Science Inc. All rights reserved. PII:S0034-4257(00)00200-5