論文ID: 2022-003
We propose a new application method in which radar/raingauge-analyzed precipitation amounts (RAP) produced by the Japan Meteorological Agency are spatially converted into 5km-resolution data, in addition to a three-hourly accumulation procedure, in order to statistically analyze localized heavy rainfall areas (HRAs) for a long period. A long-term trend and homogeneity in the appearance frequency of RAP with 5km-resolution converted by several methods, including the conventional method, are statistically evaluated in comparison with rain-gauge observations. The results indicate that the following application method is the most suitable to represent long-term variations in the appearance frequency of HRAs; (1) the converted value of RAP from 1 km to 5 km resolution is set to the 90th percentile value in 30 segments with 1km-resolution included in a grid with 5km-resolution, and (2) this spatial conversion is conducted after accumulating original RAP with 1km-resolution for three hours. Statistical analyses were performed for the appearance frequency of HRAs extracted from 5km-resolution RAP that were produced by the new application method, which indicates that the number of HRAs of the linear-stationary type could be increased without compromising the characteristics of HRAs extracted by the conventional method.