Sensitivity of localized heavy rainfall in Northern Japan to WRF physics parameterization schemes

This study examined the sensitivity of cloud microphysics and planetary boundary layer (PBL) parameterization in the simulation of localized heavy rainfall in northern Japan such as the August 2022 event. Accurate simulation/forecast of such quasi-stationary line-shaped mesoscale convective systems is a pressing issue for the region. The sensitivities of 17 different cloud microphysics parameterization schemes, including single and double moments, and eight different PBL schemes, including nonlocal and local closure schemes, were investigated. Overall, the simulated heavy rainfall in August 2022 was notably sensitive to the microphysics and PBL schemes. Among the microphysical parameterization comparisons, the Goddard and Eta (Ferrier) schemes exhibited superior performance in simulating the magnitude and spatial distribution of heavy rainfall events, evaluated using various precipitation statistics and object-based measure (Structure-Amplitude-Location (SAL) method). Considering the sensitivity of the PBL schemes, the combination of the Goddard microphysics scheme and the Grenier-Bretherton-McCaa PBL scheme performed the best in the 48-h basin-average precipitation depth and spatial distribution RMSE. Ensemble correlation analysis revealed that a better representation of the strong water vapor flux convergence was the key to simulating localized heavy rainfall. The optimal combination of physical schemes reasonably simulated the July 2020 and July 2023 heavy rainfall events in the Tohoku region, suggesting their potential for future RCM studies in the region.