Abstract
Monthly streamflow forecasting is vital for managing water resources. Recently, numerous studies have explored and evidenced the potential of artificial intelligence (AI) models in hydrological forecasting. In this study, the feasibility of the convolutional neural network (CNN), a deep learning method, is explored for monthly streamflow forecasting. CNN can automatically extract critical features from numerous inputs with its convolution–pooling mechanism, which is a distinct advantage compared with other AI models. Hydrological and large-scale atmospheric circulation variables, including rainfall, streamflow, and atmospheric circulation factors are used to establish models and forecast streamflow for Huanren Reservoir and Xiangjiaba Hydropower Station, China. The artificial neural network (ANN) and extreme learning machine (ELM) with inputs identified based on cross-correlation and mutual information analyses are established for comparative analyses. The performances of these models are assessed with several statistical metrics and graphical evaluation methods. The results show that CNN outperforms ANN and ELM in all statistical measures. Moreover, CNN shows better stability in forecasting accuracy.
Highlights
Long-term hydrological streamflow forecasting has been considered an important basis for the design, construction, and management of water conservancy and hydropower projects
It can be noted that the forecasting for Xiangjiaba is more accurate than that for Huanren with Nash-Sutcliffe efficiency coefficient (NSE) varies in the ranges of 0.13-0.84 and -0.55-0.37, respectively
The results demonstrate that the convolutional neural network (CNN) can be successfully applied to forecast the monthly streamflow
Summary
Long-term hydrological streamflow forecasting has been considered an important basis for the design, construction, and management of water conservancy and hydropower projects. Data-driven models find relationships between system state variables without explicit knowledge of the physical behavior using statistic or machine learning algorithms (Ghorbani et al 2016) Due to their advantages including simplicity in design and implementation, minimum information requirements, and relatively high accuracy, the data-driven models are becoming increasingly popular in hydrological forecasting (Yaseen et al 2016; Adamowski and Sun 2010). The SVM proposed by (Vapnik 1995) is a more advanced AI model, which is based on the principle of structural risk minimization, theoretically minimizes the expected error of a learning machine and reduces the phenomenon of overfitting (Yu et al 2017) This AI model has gained the attention of many researchers and been applied to streamflow forecasting with promising results (LIN et al 2006; Samsudin et al 2011; Shabri and Suhartono 2012).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
