Author(s): Naoki Koyama; Tadashi Yamada
Linked Author(s):
Keywords: Flood forecasting; Time series analysis; Multivariate auto regressive model; Water level; Rainfall
Abstract: The purpose of this study is to verify the accuracy of water level prediction in flood disasters using a multivariate autoregressive model, which is one type of time series analysis. In recent years, serious flood disasters have frequently occurred in all over Asia almost every year. the forecasting information on the water level is the most important information for residents to evacuate. The major problem is to decide how long the lead time (Time needed to evacuate before a disaster occurs) should be. Therefore, flood prediction was performed using a multivariate autoregressive model with water level and rainfall data during heavy rains. In the present study, we studied two cases, in the first case, when calculating using only the water level, it was possible to predict the time of concentration at observation station of used observed water level data and the reference station. The second case, when calculating using the water level and rainfall, it can predict with high accuracy several hours ahead compared to using only water level.
Year: 2020