A comparison of cumulus parameterization schemes in the WRF model
Erin K. Gilliland, Univ. of Nebraska, Lincoln, NE; and C. M. Rowe
Cumulus parameterization schemes (CPSs) in atmospheric numerical models represent thermodynamic and dynamic processes of moist convection occurring at sub-grid scales. No universal framework exists for CPSs, which led to the development of numerous different schemes. The Weather Research and Forecasting (WRF) model, which was designed to run both idealized and case study simulations at mesoscale grid spacings, currently supports three CPSs: the Kain-Fritsch and Betts-Miller-Janjic schemes and the Grell-Devenyi ensemble scheme.
Past studies show that CPSs struggle with predicting precipitation for warm-season events with weak large-scale dynamic forcings. The WRF model was used to investigate the different CPSs and how they handle warm-season convection. An idealized supercell case, examined at grid spacings of 2 and 4 km with all three schemes, showed that all the schemes, except the Grell-Devenyi ensemble, simulated classic supercellular features with varying degrees of success. At 4 km, the no-CPS simulation represented convection explicitly demonstrating that a CPS might not be needed at 4 km grid spacings.
The 28-29 July 2005 summertime convective system with weak synoptic-scale forcings that occurred in South Dakota and Nebraska was simulated using a 12 km spacing to represent current operational model grid spacings and a 4 km spacing to determine if the WRF model could resolve the convection explicitly. The WRF model was unable represent convection explicitly at 4 or 12 km and needed a CPS to represent sub-grid scale precipitation. The only CPS unable to simulate precipitation was the Betts-Miller-Janjic scheme due to the dryness of the environment. The Kain-Fritsch scheme represented the isolated nature of the system at 4 km, while Grell-Devenyi could only simulate the general area. All three CPSs performed consistently at both grid spacings. This study demonstrates that the use of a CPS in a numerical model is not clearly defined and caution needs to be employed when choosing whether or not to use a CPS and which one.
Extended Abstract (1.7M)
Supplementary URL: http://weather.unl.edu/~crowe/AMS2007
Poster Session 2, Land-Atmosphere Interactions Posters
Wednesday, 17 January 2007, 2:30 PM-4:00 PM, Exhibit Hall C
Browse or search entire meeting
AMS Home Page