Satellite Remote Sensing Data Assimilation Experiment In Mesoscale Model Study

More and more attention is paid to using satellite data in numerical weather predict model. There are some countries whose operational weather forecast center has implemented direct assimilation of satellite radiance data in the four or three dimentional variational (4D/3D-Var) data assimilation system. We carry out two groups of assimilating/simulating experiments. The data assimilation scheme uses incremental form of 3D-Var method and the MM5 mesoscale model is used as assimilating model as well as the forecasting model. In the first group of simulating experiments, SOUND experiment, which assimilates radiosonde observations only is in comparison with SOUND+AMSUA experiments, which assimilates radiosonde observation and AMSU-A channel bright temperature simultaneously. Two experiments conducted foreward integration with assimilation analysis as model initial condiction. The comparison results shows that direct assimilating AMSU-A channel bright temperatures has a positive impact on MM5 model prediction as a whole. The second group of experiments is conducted to compare direct assimilating TOVS radiances data with assimilating TOVS retrieval temperature and moisture profiles. The comparison results shows that direct assimilating TOVS radiances data will result in smaller forecast root mean square errors than the errors of assimilating TOVS retrievals.