Tropical storm genesis in operational model forecasts: problems and challenges

Since 1995, we have noticed that the operational global forecast model (the Medium- Range Forecast model) will periodically take a weak tropical easterly wave and intensify into a fairly strong tropical storm. While many of the predicted storms actually verified as real tropical storms and Hurricanes, there are also many false alarm storms. Although it is very satisfying to see a coarse-mesh global model (~105 km) simulating storm-genesis, this has also brought many problems to operational forecasters. On many occasions, the model generated storms persist from long-term prediction to short-term prediction and eventually contaminate the first guess to our data assimilation system. Among the problems we are working on are: reducing the number of false alarm storms, improving the initialization technique to re-locate the guess storm (in the short run), and using satellite precipitation observations to correct the guess storm (in the long run). We will present cases of successful storm predictions as well as false-alarm predictions from our operation archive. We will also discuss our efforts to reduce the problems caused by the ability of the models to generate tropical storms. Having the operational forecast model routinely predicting tropical cyclone genesis can also be a good research tool for the understanding of tropical storm genesis. We welcome the research community to work with us in this area.