Leveraging microwave polarization information for the calibration of a land data assimilation system

This letter contributes a new approach to calibrating a tau-omega radiative transfer model coupled to land surface model output with low-frequency (<10 GHz) microwave brightness temperature (TB) observations. The problem of calibrating this system is generally poorly posed because various parameter combinations may yield indistinguishable (least squares error) results. This is theoretically important for a land data assimilation system since alternative parameter combinations have different impacts on the sensitivity of TB to soil moisture and misattribution of systematic error may therefore disrupt data assimilation system performance. Via synthetic experiments we demonstrate that using TB polarization difference to parameterize vegetation opacity can improve the stability of calibrated soil moisture/TB sensitivities relative to the more typical approach of utilizing ancillary information to estimate vegetation opacity. The proposed approach fully follows from the radiative transfer model, implemented according to commonly adopted assumptions, and reduces by one the number of calibration parameters. Key Points Assimilating microwave observations in land surface model requires calibrationCalibrating with noisy data can bias sensitivity of a radiative transfer modelMicrowave polarization is shown to add beneficial constraints on calibration