Global climate models (GCMs) are run on a large temporal scale, requiring relatively coarse resolution, and the use of convective parameterization (CP). Even below horizontal grid spacing on the order of 10 km, the sensitivity to spatial resolution is significant, with the changes in simulated TC minimum central pressure as large as 20 hPa (Gentry 2007). High-resolution simulations are necessary in order to produce a simulation with adequate representation of physical processes important to TC intensity. For the purpose of obtaining a realistic seasonal representation of TC intensity, high-resolution simulations of the Atlantic main development region (MDR) are performed using the Weather, Research and Forecasting (WRF) model. Then, these high-resolution WRF simulations can be repeated with environments based on future climate scenarios as obtained from GCM results in a downscaling manner.
As a first step in this effort, the ability of WRF to reproduce past seasonal TC activity must be established. A proof-of-concept simulation is performed where WRF is run for the month of September, 2005. Horizontal grid spacing of 6-km is used to simulate the Atlantic MDR. The Global Forecast Systems (GFS) model is used for initialization and daily boundary conditions, with a fixed sea-surface temperature (SST) field. Initial model results compare favorably with verification. During September of 2005, three TCs of category 1 strength on the Saffir-Simpson scale occurred along with two major hurricanes, Maria and Rita. Low-level model-simulated winds indicate that, in the WRF simulation, two major TCs appear and take similar tracks to Maria and Rita. However, the dates during which the model-simulated storms appear do not correlate well with the time periods during which Maria and Rita occurred. Also, three disturbances of tropical storm or depression strength appear in the model simulation.
Although the climatology was adequately reproduced, there is much room for improvement. A time-varying SST field will be used in future simulations. The possibility of running the simulation over multiple months of the season will also be explored. Also, the option of using moving, vortex-tracking nests will be further examined, as a means of increasing the horizontal resolution even further. Nudging towards a reanalysis field is also an option. After the ability of WRF to simulate both active and inactive seasons has been adequately assessed, WRF will be used to simulate an Atlantic hurricane season with an environment modified according to future climate change scenarios.