The Weather Research and Forecast Model (WRF) is being developed as a next generation mesoscale forecast model and data assimilation system with an overall goal of advancing both the understanding and prediction of important mesoscale weather, and promoting closer ties between the research and operational forecasting communities. The specific goal is to improve the forecast accuracy of significant weather features across scales ranging from cloud to synoptic, with priority emphasis on horizontal grids of 1-10 kilometers. Convective weather remains a real challenge for numerical weather prediction systems, and is recognized as a major contributor to poor forecasts of warm season QPF. One of the primary objectives of the WRF developmental effort is to improve our ability to represent and forecast convective systems in the 6-12 hour time frame. The success of such an effort depends on many factors, including using sufficient resolution to represent the convective processes explicitly, accurately representing the mesoscale environment of the convective system, especially the diurnal changes to the boundary layer, and appropriately forecasting the timing and location of significant convective triggering.

In order to test the current capabilities and limitations of the WRF model for such scenarios, 2 km and 4 km resolution simulations have been run for several well documented convective cases observed during the International H2O Project (IHOP) field experiment, which took place over the Southern Great Plains of the United States from 13 May to 30 June 2002. The cases observed during IHOP cover a wide range of convective triggering and organizational scenarios, including dry lines, cold fronts, elevated versus surface-based convection, supercells, squall lines, etc. Simulations have been completed for five of the IHOP cases, and compared to coarser resolution (10 km) simulations with parameterized convection, using various microphysical packages (e.g., LIN, NCEP3, NCEP5), land-surface representations, etc. These comparisons suggest that a significant improvement in the representation of the larger convective systems can be achieved when using even such marginally explicit resolutions, including more detailed and physically accurate convective system structure and propagation, as well as more detailed and accurate QPF. The timing and location of convective triggering was also improved with such finer resolutions, but more spurious isolated convective activity was noted as well. Also, while a convective trigger function is necessary when using convective parameterization schemes at 10 km horizontal resolutions or larger, such a triggering function was not found to be necessary for the 2 km and 4 km resolution simulations. Results from these, as well as finer (1 km) resolution simulations will be presented at the conference.