Deep learning model on rates of change for multi-step ahead streamflow forecasting

Abstract Water security and urban flooding have become major sustainability issues. This paper presents a novel method to introduce rates of change as the state-of-the-art approach in artificial intelligence model development for agenda. Multi-layer perceptron (MLP) deep learning long short-term memory (LSTM) models were considered flood forecasting. Historical rainfall data from 2008 2021 at 11 telemetry stations obtained predict flow confluence between Klang River Ampang River. The initial results MLP yielded poor performance beneath normal expectations, which was R = 0.4465, MAE 3.7135, NSE 0.1994 RMSE 8.8556. Meanwhile, LSTM generated 45% improvement its R-value up 0.9055. Detailed investigations found that redundancy input multiple target values had distorted performance. Qt introduced into parameters solve this issue, while Qt+0.5 value. A significant detected with 0.9359, 0.7722, 0.8756 3.4911. When employed, an impressive seen plot actual vs. forecasted flow. Findings showed could reduce forecast errors helpful additional layer early detection.