This study develops a surrogate-based method to assess the uncertainty within a convective permitting integrated modeling system of the Great Lakes region, arising from interacting physics parameterizations across the lake, atmosphere, and land surface. Perturbed physics ensembles of the model during the 2018 summer are used to train a neural network surrogate model to predict lake surface temperature (LST) and near-surface air temperature (T2m). Average physics uncertainties are determined to be 1.5°C for LST and T2m over land, and 1.9°C for T2m over lake, but these have significant spatiotemporal variations. We find that atmospheric physics parameterizations are the dominant sources of uncertainty for both LST and T2m, and there is a substantial atmosphere-lake physics interaction component. LST and T2m over the lake are more uncertain in the deeper northern lakes, particularly during the rapid warming phase that occurs in late spring/early summer. The LST uncertainty increases with sensitivity to the lake model’s surface wind stress scheme. T2m over land is more uncertain over forested areas in the north, where it is most sensitive to the land surface model, than the more agricultural land in the south, where it is most sensitive to the atmospheric planetary boundary and surface layer scheme. Uncertainty also increases in the southwest during multiday temperature declines with higher sensitivity to the land surface model. Last, we show that the deduced physics uncertainty of T2m is statistically smaller than a regional warming perturbation exceeding 0.5°C.

There are large uncertainties in our future projections of climate change at the regional scale, with spatial variabilities not resolved adequately by coarse-grained Earth System Models (ESMs). In this study, we use pseudo global warming simulations driven by end of the century upper end RCP (Representative Concentration Pathway) 8.5 projections from 11 state-of-the-art ESMs to examine changes in summer heat stress extremes using physiologically relevant heat stress metrics (heat index and wet bulb globe temperature) over the Great Lakes Region (GLR). These simulations, generated from a cloud-resolving model, are at a fine spatiotemporal resolution to detect heterogeneities relevant for human heat exposure. These downscaled climate projections are combined with gridded future population estimates to isolate population versus warming contributions to population-adjusted heat stress in this region. Our results show that a significant portion of summer will be dominated by critical outdoor heat stress levels within GLR for this scenario. Additionally, regions with higher heat stress generally have disproportionately higher population densities. Humidity change generates positive feedback on future heat stress, generally amplifying heat stress (by 24.2% to 79.5%) compared to changing air temperature alone, with the degree of control of humidity depending on the heat stress metric used. The uncertainty of the results for future heat stress are quantified based on multiple ESMs and heat stress metrics used in this study. Overall, our study shows the importance of dynamically resolving heat stress at population-relevant scales to get more accurate estimates of future heat risk in the region.

The Great Lakes of North America form one of the largest freshwater systems on Earth, exhibiting seasonal water level fluctuations that exceed 0.5 meters. These fluctuations pose substantial challenges for coastal resilience, flood risk management, and navigation planning. Accurate seasonal forecasting of lake levels using traditional mechanistic models is challenging due to the complex physical mechanisms and coupled hydroclimatic processes involved. Recently, deep learning has gained prominence in geoscience applications for its ability to recognize intricate patterns within multiphysical datasets. Here, we introduce a novel Dual-Transformer deep learning framework, tested on Lake Superior—the largest of the five Great Lakes. This architecture integrates two modified Transformer models: the Prophet, which predicts underlying trends, and the Critic, which refines the Prophet’s predictions. The final lake level prediction is derived by weighting the outputs of both models through a Multi-Layer Perceptron (MLP), jointly trained with the Prophet and Critic to enhance overall accuracy. Our results demonstrate that the Dual-Transformer model, which uses seven atmospheric and lake features, achieves unprecedented accuracy in seasonal forecasting in the testing dataset, attaining a correlation coefficient of 0.97 and a root mean square error (RMSE) of 4 cm for forecasts up to six months ahead. Additionally, the Dual-Transformer model runs six orders of magnitude faster than conventional mechanistic models, producing results in less than one second on a typical personal computer. These findings suggest our deep learning framework provides an efficient and reliable tool for real-time lake level forecasting, with significant implications for water management and disaster mitigation.

To understand the future summer precipitation changes over the Great Lakes Region (GLR), we perform an ensemble of regional climate simulations through the Pseudo-Global Warming (PGW) approach. We found that different types of convective precipitation respond to the PGW signal differently. Isolated deep convection (IDC), which is usually concentrated in the southern domain, shows an increase in precipitation to the north of the GLR. Mesoscale convective systems (MCSs), which are usually concentrated upstream of the GLR, shows a shift to the downstream region with increased precipitation. Thermodynamic variables such as convective available potential energy (CAPE) and convective inhibition energy (CIN) are found to be increased in almost the entire studied domain, providing a potential environment more (less) favorable for stronger (weaker) convection systems. Meanwhile, changes in lifting condensation level (LCL) and level of free convection (LFC) show a strong correlation with variations in convective precipitation, underscoring the significance of these thermodynamic factors in controlling precipitation over the domain. Results show that decreased LCL and LCF over places where convective precipitation is increased, is mainly contributed by the atmospheric moisture increase. In response to the prescribed warming perturbation, MCSs show more frequent occurrences downstream, while localized IDCs show more intense rain rate, longer duration, and larger rainfall area.