3.3
Predictability of supercell thunderstorms
Rebecca M. Belobraydich, CIMMS/Univ. of Oklahoma, Norman, OK; and D. J. Stensrud
The possibility of providing warnings upon the forecast of a supercell thunderstorm by a numerical weather prediction model, instead of waiting for the storm to first form and be detected prior to issuing a warning, is currently being investigated by the meteorological community. This “Warn-on-Forecast” approach could substantially increase the lead time of warnings for severe thunderstorms and their associated hazards. However, one unknown is the sensitivity of these high-resolution convective-scale forecasts to errors in the initial environment. In this study, the Weather Research and Forecasting (WRF) Model, run as a three-dimensional, storm-scale model, is used to investigate the sensitivity of supercell thunderstorms, assuming convective initiation, to errors in the environment. An idealized, homogeneous environment derived from an initial sounding is used to produce a control run. Typical operational forecast errors are found for 1-, 2-, and 3-hour forecasts from the 13-km Rapid Update Cycle (RUC) Model, from storm days from January through June 2010, using the RUC 0000 UTC analysis as truth. These errors in temperature, relative humidity, and wind speed are used to perturb the initial environmental condition to produce an ensemble of forecast runs for each forecast lead time. The results of the runs are then analyzed to investigate the effects of the uncertainty in the initial environmental conditions on the evolution of the simulations, and to thus find an upper limit to the current practical predictability of supercell thunderstorms and the level of confidence that can be placed in the prediction of the event by the numerical model.
Session 3, Mesoscale predictability and data assimilation I
Monday, 1 August 2011, 4:00 PM-6:00 PM, Marquis Salon 456
Previous paper Next paper