We evaluate the forecast skill of a simple deterministic hurricane intensity prediction model that provided intensity forecasts on an experimental basis to forecasters at the National Hurricane Center during the 2000 and 2001 seasons. The model consists of a simple, balanced axisymmetric atmospheric model coupled to an equally simple ocean model comprised of a set of one-dimensional ocean columns positioned along the forecast track of the storm. These ocean columns are vertically mixed in proportion to the wind stress applied from the atmospheric model. This coupled model is capable of simulating all the physics thought to be important in controlling hurricane intensity, with the major exception of atmospheric environmental interactions such as interactions with a sheared environmental wind flow. Hindcasts with the model show that it is capable of simulating with great accuracy the evolution of major Atlantic storms that were relatively unaffected by atmospheric environmental interactions. Beginning in 2000, a proxy for environmental shear was added to the model physics.

During the 2000 season, the model outperformed other deterministic hurricane models, but the skill at all lead times was below that of the Statistical Hurricane Intensity Prediction System (SHIPS) as well as that of the official forecasts. At small lead times, the model suffered from inadequate initialization, so that the initial intensity often differed substantially from the observed intensity. At larger lead times, the model performed well in storms that were not strongly affected by shear, but less well in highly sheared cases.

For the 2001 season, a new initialization procedure was introduced, resulting in much smaller initial intensity errors, and the model runs were automated at NHC, resulting in a much larger set of model predictions. We will present the 2001 results in detail at the conference.