Towards LPRM-based soil moisture retrievals with multi-angular microwave observations from SMOS

Passive microwave remote sensing by the Soil Moisture and Ocean Salinity (SMOS) satellite enables observations of surface soil moisture at global scale. This involves a retrieval process that uses a radiative transfer model to relate satellite-observed radiances to geophysical variables such as soil dielectric constant and soil moisture. The L-MEB (L-band Microwave Emission of the Biosphere) based retrieval products of SMOS have previously been evaluated to show reasonably good agreement with ground measurements. An alternative product is based on the Land Parameter Retrieval Model (LPRM), which has been adapted for a wide range of Xand C-band sensors to retrieve soil moisture from single-angle observations. From the implementation viewpoint, LPRM is attractive in its limited requirements for additional auxiliary parameters and in its simple geophysical parameterization. The research gaps lie in the limited previous study on retrieving SMOS multi-angular observations using LPRM. The development of a LRPM-retrieved SMOS product is also motivated by establishing an algorithmic consistency across multiple C-, Xand L-band sensors, allowing them to be compared more directly for relative performance. This study considers the LPRM-like retrieval algorithm based on generic Radiative Transfer Equation (RTE) and uses a controlled numerical experiment to determine the properties of LPRM when acting on synthetic multi-angular SMOS observations. The influences of observational uncertainties, model parameter uncertainties and multi-angle observations (c.f. single-angle) are evaluated on the retrieval performance. In particular, the Markov Chain Monte Carlo (MCMC) algorithm is used to quantify the uncertainties in, and correlations between, retrieved parameters, under these influences. The main findings of this study are listed as follows, • According to retrieval sensitivity analysis, vegetation optical depth (τ) and surface temperature (Ts) are more critical parameters in the retrieval model compared to surface roughness and scattering albedo, and they may be retrieved simultaneously with soil moisture in a 3-parameter retrieval configuration; • The uncertainty in brightness temperature (TB) has more significant impact on a vegetated-wet scenario than on a bare-dry case: vegetated-wet surface can tolerate 2K brightness temperature uncertainty to achieve the target retrieval accuracy whereas bare-dry surface can tolerate as much as 8K TB uncertainty; • MCMC results demonstrate the advantages of LPRM soil moisture retrieval with multi-angular TB observations over single-angle retrievals in terms of higher robustness and less uncertainty in retrieval results. This work therefore provides guidance to adapting LPRM for SMOS data and soil moisture retrieval at continental scale.