Evaluation Study for Predictability of a Heavy Localized Rainfall Using WRF and a Climate Change Database

A heavy localized rainfall, whose maximum daily precipitation of 545.5mm was historically recorded on 5^th of July 2017, hit northern Kyushu, Japan. It triggered landslide disasters that swept massive volumes of driftwood and soil into communities, leaving 36 human fatalities and damaging more than 600 residential buildings. It strongly required better understanding of the meteorological processes as well as the predictability of the rainfall event.

In this study, sensitivity of the precipitation over this area is evaluated to investigate the better combination of physical schemes in the meteorological numerical model with different downscaled grid resolutions in the nesting method. The Weather Research and Forecasting (WRF) Model with NCEP-FNL datasets as an initial data were used for the sensitivity analysis.

The simulated meteorological parameters, such as relative humidity, sea level pressure, temperature, wind velocity time series are agreed reasonably well with the observation results among the physical schemes. The Thompson microphysical and YSU or MYNN Level 2.5 PBL scheme (PBL: Planetary Boundary Layer) with the downscale grid resolution of 2km or less showed quite good agreement with the radar-AMeDAS observed results, for the hourly and 24-hours accumulated precipitations.

It is also found that the warm moist air from the southwest and the cold air from the northwest flowed into the location of localized rainfall. In here, the cumulonimbus clouds are repeatedly generated and developing furiously from the west to east, in a phenomenon known as back-building storm.

Climate Change database“Database for Policy Decision-Making for Future Climate Change” (d4PDF) was also used to investigate the changing of precipitation in the area under global warming effects. The frequency distribution of the daily precipitation is quite agreed with the past recorded results. Although the future daily precipitation of the area is found to be increased under global warming, but its value is smaller than the precipitation which was recorded from this heave localized rainfall event. It could be resulted from the insufficient resolution of grid spacing of 20km which is used in the d4PDF.