Dual state-parameter estimation of land surface model through assimilating microwave brightness temperature

Besides uncertainties introduced by atmospheric forcing and initial states, land surface simulation results are mainly determined by model structure and related model parameters. Traditional data assimilation approaches, as they only focus on mathematically updating the simulated states when observations become available, have little intrinsic improvement in the model performance. Model parameter optimization will lead to reduced biases in simulation results and then a better forecasting skill can be expected. Therefore, calibrating model parameters and updating states simultaneously in the framework of sequential model-data fusion would be valuable for uncertainty quantification. A dual state-parameter estimation land data assimilation system is implemented in this paper by coupling the Variable Infiltration Capacity(VIC) land surface model, the Tau-Omega Radiative Transfer Model(RTM) and Sampling Importance Resampling Particle Filter(SIR-PF) algorithm. Passive microwave brightness temperature observations from Passive/Active L and S band (PALS) sensor in SMEX02 are assimilated and the results demonstrate that both soil moisture states and model lumped parameters can be estimated simultaneously.