Inverse Dynamic Parameter Identification for Remote Sensing of Soil Moisture From SMAP Satellite Observations

In the soil moisture active passive (SMAP) mission's soil moisture retrieval algorithms, the effects of surface roughness and vegetation scattering on the brightness temperature are conventionally modeled using time-invariant parameters: roughness intensity ( h ) and effective scattering albedo (ω). Such simplification neglects the variability of h and ω over time, potentially compromising the accuracy of soil moisture estimates at the satellite footprint scale. This study aims to derive dynamic, pixel-scale h and ω parameters specifically for the SMAP single-channel algorithm (SCA) and the regularized dual-channel algorithm (RDCA). This is achieved through an iterative inverse procedure that minimizes the differences between the simulated brightness temperatures from spatially representative 9 km soil moisture and SMAP observations across the SMAP core validation sites. The results demonstrated that the incorporation of dynamic h and ω parameters, derived on a daily scale, markedly enhanced the soil moisture retrieval performance with an average unbiased root-mean-square error (ubRMSE) of 0.01 (0.02) m 3 /m 3 and Pearson correlation ( R ) of 0.95 (0.90) for the SCA (RDCA) algorithms, indicating that dynamic parameterization holds significant promise for improving retrieval accuracy. The daily scale h parameters are generally above the static values utilized in the SMAP SCA. Within the SMAP SCA framework, the accuracy of soil moisture estimates employing daily scale h and ω parameters-randomly selected from the SCA range ( h ∊ [0.03, 0.16] and ω∊ [0, 0.08])-demonstrates notable stability and is comparable with the SMAP level 3 product. Furthermore, the daily scale parameters were temporally contracted to generate a monthly climatology for h and ω. While soil moisture values derived from these climatological h and ω parameters exhibit reduced absolute bias, their ubRMSE and R slightly degrade relative to SMAP level 3 product. This degradation likely suggests that the climatological parameters' gradual variations are insufficient to capture the fluctuations of those daily parameters. Moreover, the static h and ω values for the RDCA are systematically higher than those for the SCA. However, there is no consistent trend in the magnitudes of dynamic h and ω between different algorithms. Identifying the most effective dynamic h and ω parameters within the SMAP algorithmic framework necessitates not only selecting an appropriate parameter range but also accuratel