Prediction of Lightning Activity over Bangladesh Using Diagnostic and Explicit Lightning Parameterizations of WRF Model

Abstract

Lightning discharge from thunderstorms is a major weather hazard and the loss of lives and properties caused by lightning is high in Bangladesh due to frequent lightning activity in the pre-monsoon season (March-May). In this study, numerical simulations in predicting the lightning flashes using diagnostic and explicit lightning parameterization options in WRF (Weather Research and Forecasting) model are performed over Bangladesh for three selected pre-monsoon lightning events (01 April 2019, 04 April 2019 and 20 May 2021). WRF model sensitivity for five microphysics and three planetary boundary layer schemes are also investigated in this work. The combination of Morrison and YSU (Yonsei University scheme)is found to be the best configuration by comparing the RMSE (Root Mean Square Error) of hourly area averaged rainfall. The lightning flash counts are estimated by using four diagnostic methods: (1) PR92 (Price and Rind 1992) (based on wmax), (2) PR92 (based on 20 dBZ top), (3) PR92 (based on level of neutral buoyancy), (4) LPI (Lightning Potential Index) based on cloud hydrometeors and updraft and (5) an explicit: physics-based method from cloud electrification referred to as WRF-Elec. The WWLLN (World Wide Lightning Location Network) and NASA LIS (Lightning Imaging Sensor) observations are used to compare the simulated lightning flashes for the three events. The 24 hr (hour) accumulated rainfall are also analyzed and the results show good consistency with the observations from NASA GPM datasets. Evaluations based on FSS (Fraction Skill Score) and performance diagrams are carried out for a better understanding of rainfall of how model simulations perform in forecasting rainfall. In qualitative assessment framework, the spatial patterns of WRF-Elec based simulations of lightning flashes to predict the primary regions of lightning occurrence exhibit good agreement with observations in most of the studied cases.