Menu
Overview
In recent years, climate change has caused rainfall to intensify, frequently resulting in serious damage around the world. The importance of support for accurate flood control in dams and early evacuation of river basin residents is increasing, and accurate forecasting information has become indispensable. Yachiyo Engineering's dam inflow forecasting and river level forecasting services implement a predictive learning system to improve prediction accuracy during operation.
A predictive learning system for dam inflow and river level prediction
A predictive learning system for dam inflow and river level prediction
Conventional prediction models in Japan predict rainfall based on historical rainfall experienced to date. However, with the impact of climate change, rainfall is expected to increase to levels never before experienced, putting limitations on building historical rainfall-based models. Thus, to ensure accuracy in dam inflow and river level predictions, we eliminate data environment disparities during the building and operation of our prediction models and train the model to learn the uncertainties in rainfall forecasts (defined as "prediction learning").
Highly practical dam inflow and river level forecasting from uncertain rainfall prediction data using a predictive learning system based on deep learning
A model can be built immediately from normally measured dam inflow observation data and river level data