Numerical weather prediction and tropical cyclone track forecasting in the Caribbean using MM5 and ARPS. Case studies of hurricane Dean (2007), Omar (2008) and Paloma (2008)

The results of the application of the mesoscale models MM5 and ARPS to the local conditions of the island of Guadeloupe (FWI) and its neighboring region and the simulation of the case of hurricane Dean (2007) are reported, including an evaluation of the performance of both models. MM5 was run in a set of three nested domains with horizontal resolutions of 60, 20 and 6.6 km, while ARPS was run in a 20 km reslution domain. The results of both models were satisfactory for very short terms, but after 24 hours, the MM5 track forecasts were better than those obtained with ARPS with the applied set of domains and choices of parameterizations of physical processes. With the purpose of evaluating the possibility of using operationally an unique set of domains to simulate the behavior of the meteorological fields in the region, including all the Caribbean Sea and adjacent coasts, a two domain set was defined, including both the Lesser and Major Antilles and simulations were made for hurricanes Omar and Paloma (2008). The dependence of the type of interaction between the nested domains on the error of estimation of the track was investigated. It was concluded that MM5 produced a good track forecast up to 36-48 hours and simulated well the change of intensity, even if the central pressure was smoothed and over-predicted. Further, the forecast tracks produced by MM5 were more precise than those produced by ARPS, possibly because of the more effective nesting procedure, allowing the MM5 output to be relatively independent of the external GFS data. Neither one-way or two-way nesting proved to be clearly superior in performance, even if the results may be slightly different. It can be concluded that two domains are enough to obtain good track and intensity change predictions. However, the use of a third domain with a space step less than 10km is important to obtain a detailed description of the predicted wind and precipitation distributions, that are very important parameters, specially for small islands.