A new article by Rahim Barzegar, Ehsan Raei, and Jan Franklin Adamowski has been published in The Conversation, exploring how explainable artificial intelligence (AI) can improve both the prediction and understanding of water-level variability in the Great Lakes.
The work shows that AI can go beyond accurate forecasting by helping identify why lake levels change. By combining machine-learning models with explainable-AI techniques, the study examines the influence of key hydroclimatic factors and reveals how their contributions vary among the Great Lakes and over time.
The findings highlight the important roles of precipitation, evaporation, temperature, and other climatic variables in controlling water-level fluctuations. Rather than treating the prediction model as a “black box,” the explainable-AI framework helps determine which climatic factors are responsible for individual predictions and provides insight into the processes driving periods of high and low water levels.
These results demonstrate the potential of combining hydrology, artificial intelligence, and explainable machine learning to provide more transparent and scientifically interpretable forecasts. Such information can support water-resource planning and management as climate variability and climate change continue to affect the Great Lakes system.