Tropical Cyclone Genesis in Global Numerical Models

With the advancement of computer technology, many weather centres are now capable of running computer models for both short-term (0-7 days) and medium-term (up to two weeks) weather prediction, with rather accurate results. In particular, predictions of tropical cyclone (TC) tracks have significantly improved. However, to what extent and under what conditions these models can predict the formation (or genesis) of a TC are still relatively unknown. The predictability of TC genesis is important because some TC formations can occur quite close to shore and a knowledge of whether a TC will form within a relatively short period of time is important for issuing early warnings. In addition, while the atmospheric flow patterns associated with TC formation are generally known, the actual physical processes involved are still not completely understood.The relatively few studies that have previously been carried out to evaluate the performance of these computer models in predicting TC genesis generally focused on the statistics of such performance, and only two studies have used the forecasts to study the possible physical processes involved in the good and failed predictions, while none has tried to examine whether these models have different predictabilities under different atmospheric flow patterns.The objectives of this proposal are therefore as follows:1. To study the ability of a computer model in predicting TC genesis in the western North Pacific at different lead times, and2. To identify the physical processes responsible for TC genesis from the model analyses and predictions.The choice of one computer model and focusing on only one ocean basin is due to the limitation of time and resources. Nevertheless, the results of this study will provide information on (1) the ability of a computer weather prediction model in predicting TC genesis in the western North Pacific, (2) the atmospheric flow patterns under which the model has different rates of success in such predictions, and (3) the physical processes associated with TC genesis. Such results will be very useful and important both in operational forecasting and for an improved understanding of the physical processes responsible for TC genesis, at least from the perspective of model prediction. Such an approach has never been adopted before, and this project therefore represents pioneer work in this area and its success will provide a model for future similar studies of other computer weather prediction models and in other ocean basins where TCs occur.