Estimating Surface Soil Moisture Using Radarsat-2

Extremes in soil moisture, either too much or too little, present a significant risk to agricultural productivity. Managing and mitigating risk requires information and knowledge to assess risk potential, implement risk reduction strategies and deliver responses to this risk. Synthetic aperture radars (SARs) are sensitive to the dielectric properties of soils and are thus well suited to provide quantitative soil moisture estimates to support effective risk assessment and mitigation. With the launch of RADARSAT-2 in 2007, Agriculture and Agri-Food Canada (AAFC) began testing the accuracy of fully polarimetric SAR data to estimate surface soil moisture. One model under investigation is the Integral Equation Model (IEM), including the calibrated IEM. The calibrated IEM introduces an optimum correlation length ( opt2) to improve the performance of the IEM. To test the ability of the IEM and the calibrated IEM to accurately estimate surface soil moisture, AAFC collected three quad-polarization RADARSAT-2 images in 2008 over their western Canadian test site. Coincident with each SAR acquisition, AAFC collected approximately 2000 in situ soil moisture measurements using hand-held soil moisture probes. Surface roughness was measured using a 1-metre needle profiler. Overall, a better agreement was found between the calibrated IEM results and SAR-based backscatter coefficients compared to the original IEM results. The calibrated IEM also reduced the impact of variation in incidence angle on both the HH and VV backscatter coefficients. Inversion of the calibrated IEM was implemented using a look up table (LUT) approach. The LUTs were generated by simulating HH and VV backscatter coefficients using the opt2 formulation. When sample sites were averaged by soil texture, the calibrated IEM was able to estimate volumetric soil moisture with an RMSE of 5.37%.