Real-Time High-Resolution MM5 and WRF Forecasts during RAINEX

Hurricane intensity forecasts continue to be a challenging problem. Recent advancements in high-resolution hurricane prediction models, in which the inner core and rainbands convections can be explicitly resolved at 1-2 km grid spacing, and computational capabilities have greatly improved both storm intensity and track forecasting. There remains to be a significant need for high resolution models to help predict and understand the hurricane structures and intensity change. During the Hurricane Rainband and Intensity Change Experiment (RAINEX), an experimental real-time high-resolution “mini-ensemble” was conducted using the 5th generation Pennsylvania State University-NCAR non-hydrostatic Mesoscale Model (MM5) and the Weather Research and Forecasting model (WRF). Both models include the vortex-following nested grids developed at the University of Miami. The 3-5 day forecasts were conducted on a daily basis during the months of August and September in 2005 using four different initial and lateral boundary conditions from the large-scale and global models including GFS, GFDL, NOGAPS, and CMC. The high-resolution model forecasts were extremely valuable in the RAINEX mission planning in real-time. This study focus on the model forecasts of the track, intensity, structure, and rainfall for Hurricanes Katrina, Ophelia and Rita (2005). The model forecasts were able to capture the rapid intensification of Hurricane Katrina and the eyewall replacement cycle in Hurricane Rita. The model forecasts will be evaluated and validated using the RAINEX observations of storm structure in addition to the conventional best track data. There are a number of limiting factors in the model forecasts, especially the lack of accurate model initial conditions. The high-resolution model forecasts are also depended on the lateral boundary conditions from the global models. One of the challenges in RAINEX and the 2005 hurricane season in general is that most storm develop in the western Atlantic, the Caribbean Sea and the Gulf of Mexico region. Forecasting tropical cyclone (TC) genesis becomes more important for the RAINEX mission planning. The RAINEX model ensemble has shown a fair amount of skill in predicting TC genesis. Preliminary results show that it appears to have an advantage over the global models. This study also provides a statistical approach to validate the models performance on TC genesis and an attempt to diagnose the model biases.