A series of data withholding experiments was conducted at Center for Weather Forecasting and Climate Studies of the National Institute for Space Research (CPTEC/INPE), with the Global Physical-space Statistical Analysis System (GPSAS) - a combination of the Spectral Atmospheric Global Circulation Model (CPTEC/COLA) with the Physical-space Statistical Analysis System (PSAS). These experiments indicate how efficiently the observations are used in the particular assimilation system. The observation tested were from conventional data (SYNOP, BUOY, SHIP, radiosonde, aircraft, pilot balloons), and satellite data (ATOVS and AIRS/AMSU retrievals, QuikScat wind, Cloud Track Wind and Total Precipitation Water from SSM-I sensor). All observing systems tested have a notable positive impact on the forecast performance. The conventional data have a large impact on North Hemisphere region and it is associated the high density of these information in this region. On the other hand, the AIRS/AMSU retrievals had the largest impact in the Southern Hemisphere. Over the South America, AIRS/AMSU retrievals contribute with the same order of magnitude as conventional data and have a positive impact on all range forecast (1-5 day). Wind and TPW satellite-derived data had the largest impact in the tropical and South America regions at short (1-3 day) range forecast. However, the use of the major observing system it is crucial to obtain operationally a good initial condition of the atmospheric state for initializing of the numerical weather models at CPTEC/INPE.
observation system; data assimilation; PSAS; weather prediction
