Research on Watershed Runoff Forecast Based on Deep Reinforcement Learning

DOI

10.2991/978-94-6463-336-8_82How to use a DOI?

Keywords

Watershed runoff forecast; Deep reinforcement learning; Deep Q-Network (DQN); Long Short-Term Memory (LSTM)

Abstract

With the drastic changes in global climate, the frequent occurrence of extreme weather events has led to an increasing number of flood events, highlighting the crucial importance of accurate watershed runoff forecasting for disaster prevention, water resource management, and environmental protection. However, conventional machine learning methods have often struggled to establish precise models when facing complex hydrological environments, resulting in significant forecast deviations. In contrast, deep reinforcement learning methods have shown remarkable performance in handling complex problems and achieved substantial success in various domains. In this context, our study takes the pioneering step of applying deep reinforcement learning methods to watershed runoff forecasting, aiming to enhance forecast accuracy and reliability to address the challenges posed by global climate change and frequent floods. We conducted comparative experiments on commonly used machine learning methods, namely Long Short-Term Memory (LSTM) and Deep Q-Network (DQN), for watershed runoff forecasting. The experimental results demonstrate that DQN, with a forecast accuracy of approximately 91.3%, outperforms LSTM significantly in terms of forecast accuracy and reliability. Even when DQN encounters forecast errors, the deviation does not exceed one runoff level.

Copyright

© 2023 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Xuhong Fang
AU  - Jiaye Li
AU  - Qunfeng Liu
AU  - Hongguan Chen
AU  - Zihao Luo
PY  - 2023
DA  - 2023/12/30
TI  - Research on Watershed Runoff Forecast Based on Deep Reinforcement Learning
BT  - Proceedings of the 2023 9th International Conference on Architectural, Civil and Hydraulic Engineering (ICACHE 2023)
PB  - Atlantis Press
SP  - 715
EP  - 724
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6463-336-8_82
DO  - 10.2991/978-94-6463-336-8_82
ID  - Fang2023
ER  -