Assimilation of SSM/I radiances in the NCEP global data assimilation system

A technique for the assimilation of Special Sensor Microwave Imager (SSM/I) data in the National Centers for Environmental Prediction’s (NCEP’s) global data assimilation and forecast system is described. Because the radiative transfer (RT) model used here does not yet allow for cloud/rain effects, it is crucial to properly identify and exclude (or correct) cloud/rain-contaminated radiances using quality control (QC) and bias correction procedures. Both procedures require an accurate estimate of the total column cloud liquid water (CLW). The CLW is used in a frequency dependent cloud detection component of the QC and as a predictor in the bias correction algorithm. A two-month data assimilation experiment from July to August 2004 using the Gridpoint Statistical Interpolation (GSI) analysis system and the NCEP operational forecast model reveals that SSM/I radiance assimilation has significant positive impact on the analyses and forecasts. It adds moisture in the NH and tropics, and slightly reduces the moisture in the midto high-latitudes of the SH. Although the moisture added appears to be slightly excessive in the tropics, it reduces model spin-up of precipitation and substantially improves the dynamic fields. In addition, it improves forecast skill globally, especially in measures of the 200hPa vector wind error in the tropics. In terms of hurricane track, the SSM/I radiance assimilation produces more cases with smaller errors and reduces the average error. No disruption of the Hadley circulation is found by introducing the SSM/I radiance data.