Impact of Satellite Data an the CDAS-Reanalysis System

In preparation for the execution of the National Meteorological Center and National Center for Atmospheric Research (NMC/NCAR) Reanalysis Project, which will cover the period 1958–93, the impact of satellite data on both analyses and forecasts has been assessed. This was done by diagnosing two sets of analyses and forecasts made with and without the use of satellite data (SAT and NOSAT) within the data assimilation. The analyses and forecasts were performed using a state-of-the-art global data assimilation system and were evaluated for August 1985. The impact of satellite data is smaller than that obtained in previous impact studies during the First GARP (Global Atmospheric Research Program) Global Experiment (FGGE) that took place in 1979, reflecting the effect of improvements that have been implemented in the global analysis scheme and the model. In the Northern Hemisphere (NH), there are no significant differences between SAT and NOSAT analyses for both primary variables and eddy transports. The satellite impact on the forecasts in the NH is positive but very small, reaching about 1% in the 5-day forecasts, as measured by the average rms errors and anomaly correlations. In the Southern Hemisphere (SH), the difference between the SAT and NOSAT analyses is estimated to be equivalent to the difference between 1.5-day SAT forecasts and the verifying analyses. After 5 days, the SAT forecasts are shown to be superior to the NOSAT forecasts by about 1 day, an advantage apparent whether they are verified against SAT or NOSAT analyses. A comparison of SAT and NOSAT analyses suggests that the NOSAT captures well over 90% of the valiance of monthly mean stationary waves of the SAT analyses in most of the Tropics and Southern Hemisphere from 20° to 60°S. The daily variability is captured at 70%–90% in the Tropics and Southern Hemisphere, except above 200 hPa and south of 60°S. In several earlier satellite data impact studies performed using FGGE (1979) data, it was observed that satellite data, which cannot resolve smaller-scale features, have a damping effect on the apparent atmospheric circulation. With the improvements in data assimilation methods, it is seen that the smoothing effect is much less apparent. A comparison of the SAT and NOSAT monthly tropical precipitation derived from the 0–6-h forecast cycle shows a general agreement with the rain estimates from satellite data. Overall, these results am very encouraging, indicating that a reanalysis spanning the years before and after satellite data was available should be useful. In the NH, the analyses are basically unaffected by the satellite data. Even in the SH a large component of both the monthly and the daily anomalies can be captured in the absence of the satellite data, except in the stratosphere and Antarctic region.