2020 Volume 98 Issue 6 Pages 1163-1182
Ultra-high-resolution numerical weather prediction (NWP) experiments over a large domain have been conducted to investigate the impacts of different factors of an NWP model in simulating the Hiroshima heavy rain event in August 2014. This is a continuation of the study in Part 1 in which similar experiments were carried out for the Izu Oshima heavy rain event in October 2013. We have demonstrated the benefit of using a high-resolution model (500-m grid spacing or less) with a large domain in simulating torrential rain events.
The simulated location and intensity of the rain band in the Hiroshima case has been shown to be sensitive to the model resolution. The simulation at the 2-km grid spacing could reproduce the rain band, but with a shift to the northeast. This displacement error was reduced when the grid spacing was further reduced to 500 m and 250 m. The best simulation both in the location and intensity was obtained at the 250-m grid spacing. The planetary boundary layer schemes had a smaller impact in this case, which is different from the Izu Oshima case.
This study also investigated the dependency of simulated convective cores (CCs) on model resolutions. The local rate of change of the number of CCs with respect to the model resolution was found to start decreasing at very high resolutions that are around the 500-m grid spacing. This implies the number of CCs tends to converge when the resolution goes beyond 500 m.