The exploration of a Temporal Convolutional Network combined with Encoder-Decoder framework for runoff forecasting

• Published in: Hydrology Research Vol. 51; no. 5; pp. 1136 - 1149
• Main Authors: Lin, Kangling, Sheng, Sheng, Zhou, Yanlai, Liu, Feng, Li, Zhiyu, Chen, Hua, Xu, Chong-Yu, Chen, Jie, Guo, Shenglian
• Format: Journal Article
• Language: English
• Published: London IWA Publishing 01.10.2020
• Subjects: artificial neural network; Coders; Concentration time; deep learning; encoder-decoder architecture; Encoders-Decoders; Forecasting; High flow; Horizon; Hydrology; Low flow; Mathematical models; Neural networks; Precipitation; Rain; Rainfall; Runoff; Runoff forecasting; Stability; Stream flow; Support vector machines; temporal convolutional network
• ISSN: 0029-1277; 1998-9563; 2224-7955; 2224-7955
• DOI: 10.2166/nh.2020.100

The Temporal Convolutional Network (TCN) and TCN combined with the Encoder-Decoder architecture (TCN-ED) are proposed to forecast runoff in this study. Both models are trained and tested using the hourly data in the Jianxi basin, China. The results indicate that the forecast horizon has a great impact on the forecast ability, and the concentration time of the basin is a critical threshold to the effective forecast horizon for both models. Both models perform poorly in the low flow and well in the medium and high flow at most forecast horizons, while it is subject to the forecast horizon in forecasting peak flow. TCN-ED has better performance than TCN in runoff forecasting, with higher accuracy, better stability, and insensitivity to fluctuations in the rainfall process. Therefore, TCN-ED is an effective deep learning solution in runoff forecasting within an appropriate forecast horizon.