The Forecast Applications Branch (FAB) under the Global Systems Division at the Earth System Research Laboratory (ESRL) has been working on a THORPEX project to better assimilate satellite data for targeted data assimilation. Targeted data assimilation (TDA) identifies a region that is important to a future forecast that is local to a specific area upstream of the forecast area in space-time. That area is then provided with higher resolution data for a subsequent model run. Ideally, better data coverage in this critical region will generate a higher quality forecast in the desired area at the desired future time.

The work at ESRL/GSD/FAB is concerned about two aspects of TDA. One is using gridded statistical interpolation (GSI) to provide high resolution satellite data in the region that is of interest upstream in space-time, while maintaining nominal satellite data coverage in all other areas. The second goal of this TDA work is to put the satellite data in to a wavelet framework that is both easily stored and transmitted (compressed) that intrinsically contains high density data in the region of interest and low density information in all other locations. A GSI interface has been created to assimilate the wavelet data.

Wavelet schemes also provide for another benefit. By their nature they can filter data and in our application remove high-frequencies in the regions of low density use, and transition in a smooth manner to high frequency in the region of high data density interest. The transform's ability to make a smooth transition between regions adds to its attractiveness in that numerical noise at the resolution boundaries is reduced.

Current progress on this 3-year project is presented.