Currently, numerical simulations of hurricanes normally require bogusing of a vortex into the large-scale analysis field as an initialization procedure. A scheme was developed to produce hurricane vortices using the Advanced Microwave Sounding Unit (AMSU) data (Zhu et al. 2002). Recently, several improvements were made to the scheme. Previously, the temperature retrievals were only made from all AMSU sounding channels under clear and non-precipitating cloud conditions where the scattering of cloud droplets is negligible. Under precipitation conditions, tropospheric temperatures below 500 mb were obtained from those AMSU channels that are not affected by precipitation. In the new algorithm, the AMSU measurements at channels 3 to 6 are first corrected to clear radiances under cloud and precipitation conditions, using the relationships derived from radiative transfer modeling including cloud hydrometers. This AMSU derived temperature fields are then assimilated into the NCEP GFS data assimilation system (GDAS) to produce a best analysis of three-dimension atmospheric temperature field within the storm and its environment.

Assimilation of the AMSU derived temperatures into NWP analysis fields can be generalized by using AMSU temperature anomaly fields, rather than the temperature itself. This will make our scheme applicable for any NWP model outputs. Using the GDAS, we found the present scheme significantly improve the GDAS analysis by adding obvious vortex features. These features are mostly needed when the hurricanes are simulated at higher spatial resolutions. Impact studies are undergoing with a mesosale model such as Weather Research and Forecasting (WRF) model.