Newly explored machine learning model for river flow time series forecasting at Mary River, Australia

Area of Science:

• Hydrology and Water Resources Engineering
• Artificial Intelligence in Environmental Science
• Data-driven Predictive Modeling

Background:

• Hourly river flow monitoring is crucial for sustainable river engineering, water management, and flood risk reduction.
• Accurate river flow forecasting is essential for informed decision-making by stakeholders.
• Existing predictive models require continuous improvement for enhanced accuracy and reliability.

Purpose of the Study:

• To introduce and evaluate a novel data-intelligent model, the emotional neural network (ENN), for hourly river flow prediction.
• To assess the performance of the ENN model against established methods using historical data from the Mary River, Australia.
• To determine the efficacy of ENN as a simulation tool for near real-time river flow forecasting.

Main Methods:

• Utilized a historical hourly dataset spanning four years (2011-2014) for the Mary River.
• Developed and implemented an emotional neural network (ENN) model for predictive analysis.
• Validated the ENN model's performance against Minimax Probability Machine Regression (MPMR), Relevance Vector Machine (RVM), and Multivariate Adaptive Regression Splines (MARS) models.

Main Results:

• The emotional neural network (ENN) model demonstrated outstanding performance in hourly river flow prediction.
• Performance ranking of the models was determined as: ENN > MARS > RVM > MPMR.
• Numerical and graphical evaluations confirmed the superior predictability of the ENN model compared to benchmark approaches.

Conclusions:

• The proposed ENN model represents a promising strategy for hourly river flow simulation and prediction.
• The ENN model offers significant potential for advancing the state-of-the-art in river engineering and water resource monitoring.
• Further exploration of the ENN model is recommended for near real-time forecasting applications.