Sensitivity to Horizontal Resolution of Regional Climate Model in Simulated Precipitation over Kyushu in Baiu Season

This study statistically investigated sensitivities of simulated precipitation to horizontal resolution of a regional climate model, instead of focusing on particular cases. We performed long-term integrations of models with horizontal grid spacings of 20, 5, 2, 1, and 0.5 km over Kyushu in the Baiu seasons of 2009−2020. The 2-km grid model improves simulated precipitation to the 5-km grid models with and without cumulus parameterization. Further decreasing the gird spacing from 2 km to 1 and 0.5 km reduces the dependency of the frequency biases on intensities of hourly precipitation and mitigates the excessive concentration of heavy precipitation in small scale, approaching that of the radar/raingauge analysis. The features of individual deep moist convections, specifically the horizontal scale of updrafts, the representation of downdrafts, and the number of convections, start to converge when the grid spacing is reduced from 1 to 0.5 km. The results suggest models with grid spacing of 1 km or less are needed to resolve deep moist convections and to represent the resulting precipitations. The 2-km grid models can partly resolve the deep moist convections, but their effective resolution is still insufficient, requiring some parameterizations to simulate convective precipitations appropriately.