26th Conference on Hurricanes and Tropical Meteorology


The preliminary analysis of the dropsonde data from DOTSTAR and their impact on the typhoon track forecasts (Formerly Paper Number 6C.2)

Wei-Peng Huang, National Taiwan University, Taipei, Taiwan; and C. C. Wu, P. H. Lin, S. D. Aberson, and K. C. Hsu

Tropical cyclones (TC) generally develop in data-sparse oceanic regions. Few observations are available to depict initial conditions of TC models, and this negatively affects the accuracy of track forecasts. In 1982, the Hurricane Research Division (HRD) began to investigate possible improvements to numerical tropical cyclone track forecasts that could result from the assimilation of dropwindsonde observations taken in the data-sparse TC environment. The observations helped the National Centers for Environmental Prediction (NCEP) and the Geophysical Fluid Dynamics Laboratory (GFDL) hurricane model to reduce track forecast errors significantly. During 2003, twelve of these targeting missions were conducted, mainly in Major Hurricane Isabel. The dropwindsonde data collected during these missions improved the NCEP global model track forecasts by an average of 50%. Considering the potential of dropwindsonde data in improving typhoon forecasts, a research project, Dropsonde Observation for Typhoon Surveillance near the Taiwan Region (DOTSTAR; the overview of this project is also submitted in a companion paper), is supported by National Science Council (NSC) of Taiwan, with strong collaboration with HRD. Typhoon surveillance missions with the Astra aircraft are to be conducted to improve the numerical guidance for typhoons near Taiwan in the typhoon season of 2003 - 2005. GPS (Global Positioning System) dropwindsondes have been released to obtain the wind, temperature, and the humidity in the environment of typhoons. The observations are ingested into the global model of the Central Weather Bureau (CWB), NCEP, and FNMOC in real time. Two missions have been conducted in Typhoons Dujuan and Melor during the 2003 typhoon season. Preliminary results from the comparison of the model runs of Dujuan with and without the dropsonde data show an 18% reduction for 48-h forecast track error of the CWB global model. Global Forecasting System (GFS) forecasts are significantly improved at 6-48 h, and the average reduction is 35%, though only three dropwindsondes out of the 11 released ones were assimilated. Meanwhile, as the dropwindsonde data mostly were removed by the initialization scheme in bogusing the inner part of the typhoon, the GFDL hurricane model forecasts show negligible differences between the runs with and without the dropwindsonde data. Work is still ongoing to evaluate how these dropwindsonde data make such dramatic track improvement in GFS. The initial results of DOTSTAR show a promising opportunity for improving the track prediction of typhoons in the western North-Pacific (near Taiwan). More dropwindsondes will be released into the periphery of typhoons near Taiwan in the typhoon seasons of 2004 and 2005. As the number of observations increases, we will begin to see a more statistically significant achievement of these dropwindsondes in TC track predictions.

extended abstract  Extended Abstract (872K)

Poster Session 1, Posters
Wednesday, 5 May 2004, 1:30 PM-1:30 PM, Richelieu Room

Previous paper  

Browse or search entire meeting

AMS Home Page