Recently, the river flood damage in small-and-medium-sized rivers happens frequently by regional localized torrential rain. In this study, we constructed the water level prediction model of the tidal river, Motohiranumabashi gauging station in the Katabira river in Yokohama City, using the genetic programming that was a kind of the artificial intelligent technique, and evaluated accuracy using the real-time data. Studying the water level, the rainfall of radar, the astronomical tide level, the wind direction and the wind speed data for the past two years, the water level prediction was done by using the observed and forecasted weather data. It was confirmed to be able to predict the water level including both the tide and the rainfall-runoff, because of the error margin of 10cm or less.
View full abstract