Simulation Experiments for Testing the Assimilation of Geostationary Satellite Temperature Retrievals into a Numerical Prediction Model

Abstract

Recent diagnostic studies using retrievals from the Visible Infrared Spin-Scan Radiometer (VISSR) Atmospheric Sounder (VAS) indicate that there are limitations of gestationary satellite sounding data (poor vertical resolution of temperature and moisture profiles) as well as advantages (fine horizontal resolution and time resolution better than the conventional network and polar-orbiting satellites). A simulation experiment is presented which tests a procedure that assimilates temperature soundings from geostationary satellites using a method developed by Gal-Chen. The method takes advantage of the strength of this observing system (Increased temporal and horizontal resolution) while avoiding its major weakness (poor vertical resolution). The assimilation technique is based on multiple insertions of satellite data over an analysis cycle. In the experiment, the insertion rate is varied from every half hour as an upper limit to every six hours as a lower limit. In the updating procedure, we ensure that the fine vertical structure that might be generated by the model is not destroyed by the coarse vertical resolution of the inserted satellite data. At the same time, the model is constrained to reproduce thickness compatible with those that would be inferred from the satellite sounding (assuming the radiance measurements are exact). The results of these simulation experiments indicate that, for the case of a baroclinically unstable wave, considerable improvements in the short-range forecast (?48 h) may be realized if geostationary satellite data is inserted with a frequency near 1 h during a 6-h analysis cycle compared with a single insertion. On the assumption that cloud cover would limit the ability to obtain satellite soundings, an experiment was conducted showing that, while poor vertical resolution apparently can be tolerated, a nearly complete horizontal coverage is necessary to maintain the positive impact on the numerical simulations. This implies that in order to make geostationary satellite data useful for initializing numerical models, microwave sounding channels should be considered for geostationary orbit so that frequent observations of the surface-500 mb thickness can be estimated for cloud-covered (but nonprecipitating) areas as well.