Direct Insertion of SMOS Soil Moisture Products into a Numerical Weather Forecast Model

We examine the usefulness of satellite derived soil moisture products from the Soil Moisture Ocean Salinity (SMOS) satellite in short-range convective precipitation forecasts over the United States Great Plains. Typical data assimilation methods require large computational demands and can be time consuming. To increase efficiency, we substituted existing model soil moisture with SMOS soil moisture values, which allowed the satellite observations to have maximum weight and to be taken as solid truth. SMOS attempts to retrieve soil moisture in the top few cm of the soil. The deeper soil moisture levels were adjusted accordingly to preserve gradients between the deeper soil layers. Biases were removed from the SMOS data prior to substitution. Model simulations with the Weather Research and Forecasting Model were initialized with soil moisture values from the North American Regional Reanalysis dataset (the control), with SMOS adjusted values, with artificially high soil moisture values, and with artificially low soil moisture values. Specific cases were chosen with the criteria of having minimal synoptic-scale forcing (so land surface interactions would be important) and having a SMOS pass covering much of the Great Plains region. Results showed no significant changes when SMOS soil moisture values were used compared to control runs, despite larger changes that did occur when extreme soil moisture values were used. This highlights the idea that common data assimilation techniques that would place less weight on SMOS observations may not be useful over this region.