Among its many impacts, climate warming is leading to increasing winter air temperatures, decreasing ice cover extent, and changing winter precipitation patterns over the Laurentian Great Lakes and their watershed. Understanding and predicting the consequences of these changes is impeded by a shortage of winter‐period studies on most aspects of Great Lake limnology. In this review, we summarize what is known about the Great Lakes during their 3–6 months of winter and identify key open questions about the physics, chemistry, and biology of the Laurentian Great Lakes and other large, seasonally frozen lakes. Existing studies show that winter conditions have important effects on physical, biogeochemical, and biological processes, not only during winter but in subsequent seasons as well. Ice cover, the extent of which fluctuates dramatically among years and the five lakes, emerges as a key variable that controls many aspects of the functioning of the Great Lakes ecosystem. Studies on the properties and formation of Great Lakes ice, its effect on vertical and horizontal mixing, light conditions, and biota, along with winter measurements of fundamental state and rate parameters in the lakes and their watersheds are needed to close the winter knowledge gap. Overcoming the formidable logistical challenges of winter research on these large and dynamic ecosystems may require investment in new, specialized research infrastructure. Perhaps more importantly, it will demand broader recognition of the value of such work and collaboration between physicists, geochemists, and biologists working on the world's seasonally freezing lakes and seas.

Ice Cover in the Great Lakes has significant impacts on regional weather, economy, lake ecology, and human safety. However, forecast guidance for the lakes is largely focused on the ice-free season and associated state variables (currents, water temperatures, etc.) A coupled lake-ice model is proposed with potential to provide valuable information to stakeholders and society at large about the current and near-future state of Great Lakes Ice. The model is run for three of the five Great Lakes for prior years and the modeled ice cover is compared to observations via several skill metrics. Model hindcasts of ice conditions reveal reasonable simulation of year-to-year variability of ice extent, ice season duration, and spatial distribution, though some years appear to be prone to higher error. This modeling framework will serve as the basis for NOAA’s next-generation Great Lakes Operational Forecast System (GLOFS); a set of 3-D lake circulation forecast modeling systems which provides forecast guidance out to 120 h.

Interannual variability of ice cover, circulation, and thermal structure in Lake Erie for 2003–2012 was investigated using a three‐dimensional hydrodynamic model coupled with ice processes. The model reproduced minima of ice extent in the winters of 2006 and 2012 (mild winters), as well as maxima in 2009 and 2011 (severe winters) in agreement with the observational analysis. The model reasonably captured ice thicknesses, seasonal variation of the mean surface temperature, and lake circulation. The model results showed early onset of stratification in March after the almost ice‐free winter of 2012. In the mild winters, the coastal current speed was significantly higher than the 9 year mean, since the larger open water region due to less ice cover allowed the more effective wind driven circulation. In the severe winters, the lake circulation was slowed because the packed ice reduced wind stress on the water surface. Seasonal means of coastal current speed ranged from 3.9 cm/s in the severe winter (January to March mean) of 2009 to 7.2 cm/s in the mild winter of 2012. The variation was much larger than in the other seasons (±0.6 cm/s). The results imply that decreasing ice cover could lead to a more energetic coastal circulation in winter, which could influence lake turbidity, material transport, and nearshore waves. Finally, the interannual variation of ice cover is discussed in relation to teleconnection patterns. The ice minimum (maximum) in the winter of 2006 (2009) can be explained by the intermittent positive (negative) North Atlantic Oscillation that occurred in January (December to January).

Precipitation impacts on ice cover and water temperature in the Laurentian Great Lakes were examined using state‐of‐the‐art coupled ice‐hydrodynamic models. Numerical experiments were conducted for the recent anomalously cold (2014–2015) and warm (2015–2016) winters that were accompanied by high and low ice coverage over the lakes, respectively. The results of numerical experiments showed that snow cover on the ice, which is the manifestation of winter precipitation, reduced the total ice volume (or mean ice thickness) in all of the Great Lakes, shortened the ice duration, and allowed earlier warming of water surface. The reduced ice volume was due to the thermal insulation of snow cover. The surface albedo was also increased by snow cover, but its impact on the delay the melting of ice was overcome by the thermal insulation effect. During major snowstorms, snowfall over the open lake caused notable cooling of the water surface due to latent heat absorption. Overall, the sensible heat flux from rain in spring and summer was found to have negligible impacts on the water surface temperature. Although uncertainties remain in overlake precipitation estimates and model's representation of snow on the ice, this study demonstrated that winter precipitation, particularly snowfall on the ice and water surfaces, is an important contributing factor in Great Lakes ice production and thermal conditions from late fall to spring.

Proper modeling of the turbulent heat fluxes over lakes is critical for accurate predictions of lake-effect snowfall (LES). However, model evaluation of such a process has not been possible because of the lack of direct flux measurements over lakes. The authors conducted the first-ever comparison of the turbulent latent and sensible heat fluxes between state-of-the-art numerical models and direct flux measurements over Lake Erie, focusing on a record LES event in southwest New York in November 2014. The model suite consisted of numerical models that were operationally and experimentally used to provide nowcasts and forecasts of weather and lake conditions. The models captured the rise of the observed turbulent heat fluxes, while the peak values varied significantly. This variation resulted in an increased spread of simulated lake temperature and cumulative evaporation as the representation of the model uncertainty. The water budget analysis of the atmospheric model results showed that the majority of the moisture during this event came from lake evaporation rather than a larger synoptic system. The unstructured-grid Finite-Volume Community Ocean Model (FVCOM) simulations, especially those using the Coupled Ocean–Atmosphere Response Experiment (COARE)-Met Flux algorithm, presented better agreement with the observed fluxes likely due to the model’s capability in representing the detailed spatial patterns of the turbulent heat fluxes and the COARE algorithm’s more realistic treatment of the surface boundary layer than those in the other models.

Turbulent fluxes of latent and sensible heat are important physical processes that influence the energy and water budgets of the North American Great Lakes. These fluxes can be measured in situ using eddy covariance techniques and are regularly included as a component of lake–atmosphere models. To help ensure accurate projections of lake temperature, circulation, and regional meteorology, we validated the output of five algorithms used in three popular models to calculate surface heat fluxes: the Finite Volume Community Ocean Model (FVCOM, with three different options for heat flux algorithm), the Weather Research and Forecasting (WRF) model, and the Large Lake Thermodynamic Model. These models are used in research and operational environments and concentrate on different aspects of the Great Lakes' physical system. We isolated only the code for the heat flux algorithms from each model and drove them using meteorological data from four over-lake stations within the Great Lakes Evaporation Network (GLEN), where eddy covariance measurements were also made, enabling co-located comparison. All algorithms reasonably reproduced the seasonal cycle of the turbulent heat fluxes, but all of the algorithms except for the Coupled Ocean–Atmosphere Response Experiment (COARE) algorithm showed notable overestimation of the fluxes in fall and winter. Overall, COARE had the best agreement with eddy covariance measurements. The four algorithms other than COARE were altered by updating the parameterization of roughness length scales for air temperature and humidity to match those used in COARE, yielding improved agreement between modeled and observed sensible and latent heat fluxes.