Multitemporal river flow discharge prediction: A new framework for integrated environmental management and flood...

Area of Science:

• Hydrology and Water Resources Engineering
• Environmental Science
• Data Science and Machine Learning

Background:

• Accurate riverine flow estimation is crucial for water resource management and flood mitigation.
• Traditional machine learning and deep learning models show promise but require further exploration for multitemporal river flow prediction.

Purpose of the Study:

• To introduce and evaluate an advanced universal group method of data handling (AUGMDH) model for river flow discharge estimation across various temporal scales.
• To compare the performance of the AUGMDH model against convolutional neural network (CNN) models.

Main Methods:

• The study employed the advanced universal group method of data handling (AUGMDH) model.
• Performance was evaluated by comparing AUGMDH with convolutional neural network (CNN) models for daily, mean monthly, and maximum monthly flow discharge estimation.
• Key performance metrics included coefficient of determination (R²), Nash-Sutcliffe efficiency (NSE), normalized root mean square error (NRMSE), RMSE-observed standard deviation ratio (RSR), and percent bias (PBIAS).

Main Results:

• The AUGMDH model consistently outperformed CNN models across all performance metrics for daily, mean monthly, and maximum monthly flow estimation.
• AUGMDH achieved higher R² and NSE values (e.g., R²=0.972, NSE=0.972 for daily flow) and lower AIC and uncertainty values compared to CNNs.
• The AUGMDH model demonstrated superior accuracy, reliability, and computational efficiency.

Conclusions:

• The advanced universal group method of data handling (AUGMDH) model offers a more robust and reliable solution for riverine flow estimation.
• AUGMDH surpasses CNN models in predicting river flow discharge across different temporal scales.
• The findings support the adoption of AUGMDH for improved water resource management and flood risk assessment.