Hydroclimate modeling using AI/ML

This project applies advanced machine learning techniques to reconstruct and forecast hydroclimatic variables—specifically wave height and ice cover—in the Great Lakes. Due to the limited availability of long-term observational data, particularly before the satellite era, understanding the historical impact of climate variability on lake dynamics has been challenging. To overcome this, a suite of deep learning models—including Multi-Layer Perceptrons (MLP), Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN)—was developed using historical meteorological inputs such as wind speed and air temperature (Abdelhady & Troy, 2024; Abdelhady & Troy, 2025). These models demonstrated performance on par with, and in some cases superior to, physics-based models. The resulting framework enabled the reconstruction of wave height and ice cover time series back to 1950, providing a critical foundation for assessing past climate impacts and informing future projections.