Extended Abstract (156K)9.2
Testing and Verification of the Weather Research and Forecast (WRF) Model
Stanley B. Trier, NCAR, Boulder, CO; and C. A. Davis and W. Wang
We summarize the results from three areas of testing and evaluation of the WRF model; idealized simulations, case studies, and real-time forecasts. We evaluate two WRF dynamical cores, the Eulerian height and mass coordinates. In general, both formulations exhibit nearly identical performance when integrated with identical physics and initial conditions. We therefore focus on the importance of variations in grid spacing and physics options over a wide variety of flows within a single core.
Idealized two-dimensional and quasi three-dimensional (e.g., periodic) WRF simulations at high horizontal resolution (1-6 km) are being performed to test sensitivity to microphysical and planetary boundary layer schemes for stratified flow over simplified topography and deep convection over a range of vertical wind shear and thermodynamic stability than spans the observed midlatitude-tropical parameter space. These simulations will be validated using analytic solutions, converged fully three-dimensional simulations, and observations, as appropriate.
Test simulations of observed cases have been performed using WRF for a wide range of disparate meteorological regimes. These include an intense synoptic cyclone, a developing tropical cyclone, a shallow arctic cold frontal passage leeward of steep terrain, and a long-lived midlatitude mesoscale convective system. We shall present an example from WRF simulations of tropical storm Barry (4-6 August 2001). A 48-h WRF simulation with 10-km horizontal grid spacing accurately captured the intensification of Barry, and closely approximated its observed northward track toward the southern Gulf coast. The simulated precipitation structure and cyclone intensity were strongly sensitive to both the cumulus parameterization and strength of horizontal diffusion used.
Real-time forecasts were performed with WRF include twice daily, 48-h integrations on a 22-km grid covering the continental U.S., and a 36-h forecast initialized from 0000 UTC over the Central U.S. Statistics of precipitation forecasts will be presented for both the 2001-2002 cool season and the 2001 and 2002 warm seasons, with focus on June 2002, the International H2O (IHOP) experiment period.
Session 9, WRF Model Development
Wednesday, 14 August 2002, 3:30 PM-6:00 PM
Previous paper Next paper