Abstract:
Satellite remote sensing observations of land surface temperature (LST) and surface soil moisture are available from a variety of sources. Assimilating such retrievals into a land surface model (that is either driven by observed meteorological forcing data or coupled to an atmospheric model) should improve estimates of land surface conditions. Such estimates may subsequently be used in the initialization of weather or seasonal climate forecasts.

Our land surface assimilation system is based on the Ensemble Kalman filter (EnKF). While the satellite and model data may contain consistent and useful information in their anomaly signals that can be merged and maximized in a data assimilation system, it is well known that the climatologies of state estimates from retrievals and from land surface models typically differ. The assimilation of satellite retrievals into a land surface model may thus benefit from a scaling approach whereby the retrievals from each sensor are scaled to the model's climatology before they are assimilated into the land model. After assimilation, the merged land surface product may be scaled back into the climatology of the satellite retrievals if the application calls for it. A simple and effective scaling method is to utilize a mapping between the cumulative distribution functions of the satellite and model data.

In this paper, we demonstrate the feasibility of this approach by assimilating LST from the International Satellite Cloud Climatology Project (ISCCP) along with surface soil moisture retrievals from the Advanced Scanning Radiometer for EOS (AMSR-E) into the NASA Catchment land surface model. The estimates from the assimilation are evaluated against available in situ measurements, and the impact of the satellite retrievals on the assimilation estimates is documented.