Modeling the initialization and tropical transition Hurricane Alex (2004)

The development and intensification of Hurricane Alex in July 2004 represented a significant forecasting challenge because of the developing storm's hybrid nature during the early portion of its lifecycle. Lower-level easterly wave forcing combined with complex upper-level potential vorticity structures of midlatitude origin to initiate the incipient system's tropical transition (TT) on 30-31 July. The tropical cyclone moved slowly along the Eastern Seaboard and generated hurricane force winds on the Outer Banks of North Carolina before recurving near the northern boundary of the Gulf Stream. Despite 26°C sea surface temperatures, the hurricane continued to intensify and reached major hurricane status (Category 3 on the Saffir-Simpson scale) with winds of 105 kt on 5 August.

Operational numerical forecast models under-predicted the intensity of the hurricane throughout its lifecycle. The initial deepening of the storm following TT was under-forecast by 40 kt in 36 h numerical guidance as the hurricane developed 85 kt winds near Cape Hatteras on 3 August. Despite known issues with deterministic hurricane intensity prediction in numerical models, the environmental forcings that contributed to the system's intensification close to the North American continent should have facilitated numerical prediction in this case. For example, a dramatic reduction in the deep-layer shear resulted from the interaction of two upper-level potential vorticity features. This process should have been well predicted by models in real time.

This study undertakes the re-forecasting of Hurricane Alex during its critical initiation and TT lifecycle stages. A multi-model approach is employed in order to assess the components of the forecast failure modes that are attributable to individual model deficiencies versus those that are fundamental to the case itself. Comparisons between re-forecasts made using different analyzed initial states allows for a similar error decomposition of the analysis fields. Error modes that are of particular diagnostic interest are those that are common to a range of modelling and analysis schemes since they highlight the structures and processes that were poorly represented in numerical forecasting systems during the initiation and TT of Hurricane Alex. It is important to understand and rectify these model and analysis deficiencies that allowed a major hurricane to form close to North America without adequate warning from the numerical guidance.