A study of several configurations of numerical forecast models was conducted to evaluate their skill to forecast tropical cyclones track and intensity. For this, four configurations of the horizontal domain were selected with 27-9 and 18-6 km of resolution for HWRF (Hurricane Weather Research and Forecasting Model) and four configurations for the WRF (Weather Research and Forecasting Model), using the dynamic core NMM (Non-hydrostatic Mesoscale Model) with the vortex tracker option. Simulations corresponding to Hurricane Irma were done from 1st to 12th September 2017, initialized with GFS (Global Forecast System) forecast outputs. No notable differences were observed between the eight configurations, although the forecast of the intensity of Hurricane Irma was deficient, with an error in the forecast of the maximum wind speed above 50 km/h. The comparison of the outputs of each configuration with buoys and surface meteorological stations data showed that the behavior of the wind speed and atmospheric pressure variables has a similar tendency to the registered values in the stations, with errors lower than 3.8 m/s for wind speed and 3 hPa for atmospheric pressure. HWRF 18-6-m (referred to the model and the horizontal domain resolution) was the configuration that showed the best ability for tropical cyclones forecast, although it require large time to compute.
Keywords: NMM; HWRF; tropical cyclones; track; intensity; hurricanes forecast