The hurricane WRF-ARW modeling system: Tests results for the nested automatic hurricane tracking implementation

In the Weather Research and Forecast (WRF) model, moving two-way nested grids have been implemented in the recent release of the Advanced Research WRF (ARW) dynamical core. The two-way interactive nested grids are engineered such that they can be efficiently integrated in parallel computing architectures that use distributed memory, shared-memory, and hybrid (distributed/shared) memory confgurations. We have implemented a grid-moving algorithm within the WRF-ARW model that allows the moving nests to automatically track hurricanes during an integration. The automatic tracking algorithm periodically checks on the location of the hurricane and moves the nested grid such that it is centered on the hurricane. There are a number of ways to determine the center of the hurricane, and our algorithm uses the centroid of the 500 HPa height field perturbation in the near-storm region. Also, coarser nested inner grids are moved such that the hurricane stays within the center of the multiply-nested grid system.

Preliminary tests using convection-permitting resolutions (dx <=3 km) on the finest grid show acceptable overhead for automatic hurricane tracking and moving two-way nests on varying numbers of processors. Test forecasts for some 2004 Atlantic hurricanes show that the fine grids produce realistic convective-scale structure and might possibly be capturing important hurricane land-sea interactions that are not well resolved in coarser models. The forecasts also demonstrate that convection-permitting resolutions for real-time forecasting using automatic moving nested grids are feasible on present-day computer architectures given that the finest grid need only cover the hurricane and a small portion of its environment. This capability will also make it possible to explore high-resolution ensemble forecasts of huricanes in real time.