Remote-Sensing-Based Streamflow Forecasting Using Artificial Neural Network and Support Vector Machine Models

In this study, we aimed to investigate the hydrological performance of three gridded precipitation products—CHIRPS, RFE, and TRMM3B42V7—in monthly streamflow forecasting. After statistical evaluation, two monthly streamflow forecasting models—support vector machine (SVM) and artificial neural network (ANN)—were developed using the monthly temporal resolution data derived from these products. The hydrological performance of the developed forecasting models was then evaluated using several statistical indices, including NSE, MAE, RMSE, and R2. The performance measures confirmed that the CHIRPS product has superior performance compared to RFE 2.0 and TRMM data, and it could provide reliable rainfall estimates for use as input in forecasting models. Likewise, the results of the forecasting models confirmed that the ANN and SVM both achieved acceptable levels of accuracy for forecasting streamflow; however, the ANN model was superior (R2 = 0.898–0.735) to the SVM (R2 = 0.742–0.635) in both the training and testing periods.

Monthly Streamflow Forecasting Using Convolutional Neural Network

Monthly streamflow forecasting is vital for the management of water resources. Recently, numerous studies have explored and evidenced the potential of artificial intelligence (AI) models in hydrological forecasting. In the current study, the feasibility of a relatively new AI model, namely the convolutional neural network (CNN), is explored for forecasting monthly streamflow. The CNN is a method of deep learning, the unique convolution-pooling mechanism in which creates its superior attribute of automatically extracting critical features from input layers. Hydrological and large-scale atmospheric circulation variables including rainfall, streamflow, and atmospheric circulation factors (ACFs) are used to establish models and forecast streamflow for Huanren Reservoir and Xiangjiaba Hydropower Station, China. The ANN and ELM with inputs identified based on cross-correlation analysis (CC) and mutual information analysis (MI) are established for comparative analysis. The performances of these models are assessed with several statistical metrics and graphical evaluation methods. The results show that CNN performs better than ANN and ELM across all the statistical measures. Moreover, CNN shows better stability in forecasting accuracy.