Predicted Rainfall Evaluation Method to Prevent Underestimation of Predicted Small River Flooding

Abstract

A train disaster prevention system is currently being studied, with a view to preventing small river flooding disasters. In this system, synthetic precipitation volumes obtained by radar are input as past rainfall, while blended precipitation predictions obtained by combining numerical meteorological simulations and Nowcast, are input as future precipitation. Rainfall used for input was examined, to find a way to prevent underestimation of immersion depths predicted by calculation, due differences in precipitation volumes caused by the falling of raindrops or by the displacement of precipitation areas. As a result, we found out that it is necessary to input the synthetic rainfall obtained two minutes before the measurement time concerned as the past precipitation amount, and the maximum value of predicted precipitation amount of blending in the area within a radius of 1 km - 5 km from the measurement point concerned as the future precipitation amount.