Radiative Transfer Parameter Estimation for SMOS Soil Moisture Retrieval

The SMOS Level-2 Processor is an operational routine to calculate the SMOS Level-2 product soil moisture from the radiometer brightness temperature (Tb). But, the radiative transfer from measured Tb into soil moisture is influenced by several conditions such as soil surface roughness and vegetation opacity, which are parameterized in a general way only. Surface soil roughness and vegetation opacity cannot be easily measured at the scale of SMOS observation of 30 - 50 km. The absolute values of these parameters for different land surfaces are uncertain, and the degree of this uncertainty is unknown as well. In addition, recent studies found that SMOS overestimates the Tb. In this paper, we present a method to enhance the accuracy of the SMOS soil moisture product by parameter estimation using a data assimilation technique (Sampling Importance Resampling Particle Filter - SIR-PF) with in situ soil moisture observations. Therefore, we developed an approach to analyze the ability of the system to track the temporal evolution of parameters such as vegetation opacity and soil surface roughness. Based on observed soil moisture and soil temperature, the L-MEB forward model was run and perturbed according to the measurement accuracy. L-MEB was integrated into a data assimilation framework using the SIR-PF, which is able to concurrently update L-MEB states and parameters. In addition, we investigate the ability of the proposed approach to account for the SMOS observation bias by introducing a bias factor in L-MEB. The overall advantage of the proposed sequential approach is its ability to be integrated into the operational near real time processing of the Level-2 product. The objectives of this study are: (i) to retrieve radiative transfer parameters and their temporal changes and (ii) to account for a bias and uncertainty in SMOS measurements.