Application of Flood Early Warning Using High-Resolution Ensemble Rainfall from Numerical Weather Prediction Model: Case Study of the 2013 Largest Flood Event in Japan

Abstract

On mid-September 2013 heavy rainfalls happened over Japan due to the season's 18th typhoon, 'Man-yi', which caused large flooding and enormous landslide disasters over Japan's Kinki region. The Japan Meteorological Agency (JMA) issued a "special warning" for three western Japan prefectures of Fukui, Kyoto, and Shiga. This paper investigates the applicability of ensemble forecasts of numerical weather prediction (NWP) model for flood forecasting area and reviews the reasons why ensembles of NWP model are so attractive than deterministic model run. In this study, 10 km resolution ensemble rainfalls forecast and their downscaled forecasts of 2km resolution were used in the hydrologic model as input data for flood forecasting and application of flood early warning. Ensemble data consists of 51 members and 48 hr forecast time. Ensemble outputs are verified spatially whether they can produce suitable rainfall predictions or not during the 2013 Typhoon No. 18, 'Man-yi' event. Then flood forecasting driven by ensemble outputs is carried out over the Katusra river basin of the Kinki area, Japan. The results shows flood forecasts driven by ensemble outputs showed that the ensemble flood forecast provides additional information to the deterministic forecast.