High-resolution WRF simulations of a monsoon event (2019) over the Badulu Oya Catchment, Sri Lanka: Role of cumulus parameterization condition and microphysics schemes

Numerical weather modelling has piqued the attention of the hydrological community because precise predictions from the models might lessen the extreme hydrological repercussions. Despite the paucity of existing studies, significant tropical storms frequently affect the Asian island of Sri Lanka. This research investigates the Weather Research Forecast (WRF-ARW) model's cumulus parameterization condition and physical parameterization schemes for a 2019 northeast monsoon event over the Badulu Oya Basin, Sri Lanka. Three cumulus schemes (Kain–Fritsch (KF), Betts–Miller–Janjic (BMJ) and Multi-scale Kain–Fritsch (MKF)) and four microphysics schemes (WRF single-moment 5-class (WSM5), WRF single-moment 6-class (WSM6), Kessler (KSL) and WRF double moment 6-class (WDM6)) were evaluated for their impact on modelled rainfall. The model performances were assessed using 24-hr accumulated model rainfall and observed rainfall with various model configurations at a horizontal grid resolution of 3 km using categorical and two quantitative comparison techniques. The study concluded that the activated KF scheme with a finer domain resolution (3 km) would be preferred for cumulus parameterization in the study region. The KF-WSM5 combination was the best since it produced the highest statistics: ETS is 0.38, B is 0.95, r is 0.76, NSD is 1.06, NRMSE is 0.72, and CCPA is 75%. Research highlights We simulated an extreme northeast monsoon precipitation event over the Badulu Oya catchment, Sri Lanka. Simulations were performed using the Weather Research and Forecasting model (WRF-ARW). Sensitivity studied to cumulus parameterization condition and microphysics schemes (CPSs- KF, BMJ, MKF and MPSs-WSM5, WSM6, KSL and WDM6). The activated Kain–Fritsch cumulus scheme at 3 km resolution was found to be the most accurate. Combination of KF-WSM5 with activated cumulus scheme in the finer domain could be preferable option for heavy rainfall simulations over the study area.