The Laurentian Great Lakes, one of the world’s largest surface freshwater systems, pose a modeling challenge in seasonal forecast and climate projection. While physics-based hydrodynamic modeling is a fundamental approach, improving the forecast accuracy remains critical. In recent years, machine learning (ML) has quickly emerged in geoscience applications, but its application to the Great Lakes hydrodynamic prediction is still in its early stages. This work is the first one to explore a deep learning approach to predicting spatiotemporal distributions of the lake surface temperature (LST) in the Great Lakes. Our study shows that the Long Short-Term Memory (LSTM) neural network, trained with the limited data from hypothetical monitoring networks, can provide consistent and robust performance. The LSTM prediction captured the LST spatiotemporal variabilities across the five Great Lakes well, suggesting an effective and efficient way for monitoring network design in assisting the ML-based forecast. Furthermore, we employed an explainable artificial intelligence (XAI) technique named SHapley Additive exPlanations (SHAP) to uncover how the features impact the LSTM prediction. Our XAI analysis shows air temperature is the most influential feature for predicting LST in the trained LSTM. The relatively large bias in the LSTM prediction during the spring and fall was associated with substantial heterogeneity of air temperature during the two seasons. In contrast, the physics-based hydrodynamic model performed better in spring and fall yet exhibited relatively large biases during the summer stratification period. Finally, we developed a statistical integration of the hydrodynamic modeling and deep learning results based on the Best Linear Unbiased Estimator (BLUE). The integration further enhanced prediction accuracy, suggesting its potential for next-generation Great Lakes forecast systems.

This paper presents a vegetated flow and sediment transport model for investigating flow–vegetation–sediment interactions. Vegetation-induced turbulence is simulated by a $k-\epsilon$ turbulence closure, in which shear and stem-scale wake turbulent kinetic energies are solved separately. The model simulates both unidirectional and oscillatory flows as well as turbulence structures reasonably well with the model skill greater than 0.8 and the coefficient of determination greater than 0.95, and better agreements between simulations and measurements are observed in cases with large stem Reynolds number. To account for sediment suspension induced by wake turbulence within the canopy, a modified Shields parameter based on turbulence intensity is introduced. Compared to the traditional Shields number model in which sediment suspension is related to mean flow, the proposed model can more reasonably simulate sediment suspension in a dense vegetation canopy with the model skill greater than 0.8 and the coefficient of determination greater than 0.75, where mean flow is weak and stem-scale wake turbulence is dominant. The model uncertainties are evaluated by conducting sensitivity tests on the new coefficients introduced in the turbulence and sediment models. The influence of vegetation density on sediment suspension is also investigated. It is found that suspended sediment concentration is generally higher with higher vegetation density within the canopy, and lower with increasing vegetation density above the canopy.