The impact of assimilating SSM/I precipitable water and rain flag data on humidity analysis and short-term precipitation forecasts

In order to evaluate the impact of assimilating precipitable water content (PWC) and rain flag data derived from the Special Sensor Microwave Imager (SSM/I), an assimilation method is developed which modifies the relative humidity (RH) and divergence of background data provided by the Japan Spectral Model (JSM). The assimilation method consists of the following procedures: (1) JSM is integrated to produce background data for the assimilation of SSM/I data. (2) If the model does not predict precipitation in areas where precipitation is observed by SSM/I, RH is raised to a critical RH so that the model precipitation schemes produces precipitation. (3) In areas where precipitation is not observed by SSM/I, RH increments are calculated on each levels based on difference in PWC between SSM/I and the background data, and they are added to the background RH. In this calculation, the correlation coefficients between JSM forecast error of PWC and that of RH on each level are utilized. (4) After the modification of RH, non-linear normal mode initialization with the precipitation process is performed to adjust divergence of the background data. The assimilation method adopted in the present study reduced the positional error of precipitation forecasts by JSM over 12 hours for the case of 21 UTC 12 July 1988. The results of the forecast experiments for this case suggest that assimilation of PWC is effective in eliminating model precipitation in the observed rain-free areas, and that assimilating rain flag data is effective in generating model precipitation in the observed precipitation areas. Assimilation of PWC also improved RH forecast in the lower-troposphere. This was due to large statistical correlation between the JSM forecast error of PWC and that of RH in this layer.