Object-oriented assessment of short-term forecast of convective hazardous weather events with the WRF model in Perm region
N.A. Kalinin1, A.V. Bykov1, A.N. Shikhov1,2
1Perm State National Research University, Perm, Russia
2A.M. Obukhov Institute of Atmospheric Physics Russian Academy of Sciences, Moscow, Russia
Keywords: squall, heavy rainfall, short-term forecast, WRF model, scale-dependent convection parameterization, object-oriented assessment, event omission, false alarm
Abstract
This study presents an accuracy assessment of the forecasts of hazardous convective weather events made with the WRF v.4.2 atmospheric model for Perm region and adjacent area for the period from May 4 to August 25, 2021. The WRF model forecasts had a 27-h lead time; the model grid step was 5 km. To reduce the probability of false alarms, scale-dependent parameterization of convection was used. The sample of squalls and large hail events was compiled based on the weather station reports, damage reports, and satellite images of forest damage. It includes 56 events. The sample of heavy rainfall events was compiled from the weather stations data only. For the same period, the WRF model predicted 30 squalls events (³ 25 m/s-1) and 63 heavy rainfall events (³ 30 mm/h). Supercells and tornado events are also considered. For squall events, we performed a cross-validation of the simulated and observed events using the distance (50 km) and time (± 3 h) threshold criteria. The accuracy of heavy precipitation was evaluated using the SCI and EDI indices. Overall, the forecasts accuracy for heavy precipitation was unsatisfactory. For squall events, 36% of them were successfully predicted, including the events that caused most substantial damage. Despite the high proportion of event omission (which prevail over false alarms), the forecasts of squalls with the WRF model in this configuration may be useful in preventing damage.
|
