線状降水帯豪雨予測に向けた水蒸気のアンサンブル予測情報の更新履歴解析

Abstract

For disaster prevention, it is important to predict the duration and the amount of rainfall brought by line-shaped rainbands. We examined a prediction method of ensemble forecast under a hypothesis that ensemble forecasts for hard-to-predict events, such as line shaped rainband heavy rainfall, show two patterns of characteristic changes when the forecasts are updated. The first pattern is that the ensemble mean does not approach reality even if the forecast is updated. The second pattern is that the dispersion of ensemble does not decrease. We conducted an ensemble forecast taking multiple initial time for the northern Kyushu heavy rainfall occurred in 2017, and examined the prediction information of water vapor, which is the physical quantity preceding precipitation. Pattern 1 was strongly observed around Asakura City. Pattern 2 was observed not only around Asakura City but also in areas upstream of water vapor inflow. These characteristics correspond to the time duration of heavy rainfall to some extent. From these analysis, we showed the possibility of utilization of two patterns for the heavy rainfall exceeding the scenario of ensemble forecast.