Impack of TOMS ozone observations on hurricane track prediction

Impact of TOMS Ozone Observations on Hurricane Track Prediction

Yonghui Wu and X. Zou Department of Meteorology Florida State University

Level 2 along track non-gridded TOMS ozone data in hurricane environments is assimilated into the MM5 in an attempt to improve track prediction. Impacts of TOMS ozone on hurricane track prediction are evaluated through six case studies. For each case, two numerical experiments are carried out with a nested domain: One is initialized by assimilating a bogus sea level pressure in the nested inner domain while the NCEP reanalyses is used as the environmental fields in the outer domain (hereafter BDA, Zou and Xiao, 2000); the other is initialized by assimilating the same bogus sea level pressure in the nested inner domain and the TOMS ozone in the outer domain (hereafter BDAO3). Two regime-regression models approach developed in Zou and Wu (2005) are utilized for the assimilation of total ozone through its link to the vertically integrated potential vorticity. Among the six cases, significant improvements in hurricane track forecast are obtained in three cases due to the use of TOMS ozone data, while the remaining 3 shows no improvements and no degradation. After 12 hours the relative track errors of the improved cases is reduced by half, and the 3-day track forecast error is less than 100 km.

The improvements in the model track forecast after assimilating TOMS ozone data into MM5 model initial conditions are related to the improvement of the environment flows of hurricane. In the BDA experiments, the simulated hurricane tracks often have a slight southward shift and lagged behind the observed tracks. Compared to that of BDA experiments, the meridional steering flow of BDAO3 experiments is about 0.5 to 1.2 m/s greater than that of the BDA cases, which corresponds to about 10 to 20 km northward movement in a six hour time period. Moreover, the model forecast of temperatures from the lower troposphere up to 200 hPa of the BDAO3 experiments are about 0.5 to 2 degrees higher than those of the corresponding BDA experiments to the north and northwest of the simulated hurricanes. Compared to TOMS ozone observations, it is found that both the mean and the root mean square errors of the 24-h and 48-h forecasts of the BDAO3 experiments are smaller than those of the BDA experiments in the improved cases, while it is vice versa in the no improvement cases. Further diagnosis of model forecasts are being conducted and findings and results will be presented at the conference.