Enhancing a Multi-Step Discharge Prediction with Deep Learning and a Response Time Parameter

Flood forecasting is among the most important precaution measures to prevent devastating disasters affecting human life, properties, and overall environment. It closely involved with precipitation streamflow data tasks. In this work, we introduced a multi-step discharge prediction framework based on deep learning models. A simple feature representation technique using correlation of backward lags was enhanced time concentration (TC) concept. Recurrent neural networks their variants, coupled TC-related features, provided superior performance over 0.9 Nash–Sutcliffe model efficiency coefficient substantially high values for multiple forecasted points. These results were consistent both Upper Nan Loei river basins in Thailand, which used as case studies work.