Hurricane Vortex Initialization with AMSU and AMSR-E Measurements

While the skill of hurricane track prediction has been significantly improved during the past decades, the forecast of hurricane track and intensity is still a challenging task. An accurate initial vortex is one of essential factors for a successful prediction. Because of the lack of observation data over ocean, a hurricane initial vortex given by a global model is often too weak and sometimes misplaced. In this study, we developed a new scheme to construct initial vortex for weak tropical cyclones (maximum wind speed around 30 knots) based on the AMSU retrieved temperature profiles and AMSR-E retrieved sea surface wind (SSW).

In the vortex initialization scheme, the sea surface wind (SSW) speed under hurricane atmospheric conditions is derived with a new physical based algorithm using AMSR-E measurements (Yan and Weng, 2005). The wind direction is assigned with the NCEP Global Data Assimilation System (GDAS) 10-m wind direction. The atmospheric temperature profiles are retrieved from AMSU measurements (Zhu and Weng, 2004), and then are used to enhance the GDAS temperature field through an objective analysis process. The 3-D wind field is derived from sea surface wind field and atmospheric temperature profile based on thermal wind relation.

Several hurricane cases in 2004 were selected to assess the new vortex initialization scheme when a storm is at formation or dissipation stage. It is found that the temperature structure under hurricane system is improved after incorporating the AMSU retrieved temperature. The AMSR-E measurements can provide more accurate SSW as verified against the National Hurricane Center (NHC) Best Analysis data. The derived hurricane initial wind field shows reasonable vortex structure at all levels.

As a natural extension of above method, we then use 4DVAR technique to assimilate AMSR-E SSW and AMSU temperature data into hurricane initial vortex. Compared with the above static vortex initialization scheme, the 4DVAR method can generate an initial vortex with balanced fields among all model variables and give a smooth start for the model integration. The impact study of the AMSU and AMSR-E data is undergoing for 2005 hurricane season by using the 4DVAR method. The detailed description of the scheme and more cases studies will be given in the presentation.