The bias correction schemes introduced by Dee and da Silva (1998) and Dee and Todling (2000) have been incorporated in a land data assimilation system. In addition to providing model state estimations, this system also allows the model bias estimated and corrected efficiently as a parallel computation through data assimilation processes. Assimilation experiments are performed by assimilating real satellite soil moisture data into a land surface model over the North American domain. Under perfect observation assumption, the improvement to the soil moisture estimates shows significant reductions of RMS errors resulted from bias correction experiments with respect to the one without bias correction. In general due to the bias correction methods, the RMS errors are largely reduced in the western part of the US, and some regions to the South East. The rest areas show marginal improvements. Our results demonstrate the feasibility of the bias correction schemes for land surface modeling, and the great potential in compensating model deficiencies arising from model physical parameterizations, initial conditions, as well as atmospheric forcing inputs. Thus the bias correction method is proved to be a potential in improving the land data assimilations.