资源环境科技发展态势分析平台

Remotely sensed soil moisture products are ideal candidates for initializing soil moisture profiles of land surface models via data assimilation. This paper investigates the possibility of using a calibrated Integral Equation Model coupled with a hyperresolution land surface model, called Soil, Vegetation, and Snow (SVS) to simulate backscatter and compares the results with C-band RADARSAT-2 Synthetic Aperture Radar (SAR) backscatter signals in postharvest season when the field is considered bare soil or sparsely vegetated. Modifications to SVS evaporation scheme are shown to improve the comparison against SAR measurements. An improved effective soil roughness calculation scheme was also proposed to focus on the inversion of the root mean square height (H-rms) only. Soil dielectric constant compensation was suggested to reduce the inversion error and expand the dynamic range of Integral Equation Model. The combination of SVS soil moisture and effective roughness is found superior to the absence of either of them. This method is promising considering that it does not require any in situ measurements, and yet it still outperforms the original IEM model, which uses in situ measured soil moisture and soil roughness at point scale.