In recent years high-impact weather events have grown more “visible,” as the market for 24-hour coverage (through both traditional and social media channels) expands. The question, Was this storm due to climate change?, often arises during such coverage—especially after storms like Hurricane Sandy or the lake-effect snowfall event that dumped more than five feet of snow on Buffalo, New York, in November 2014.
In considering the question, we must first distinguish between climate and weather. Weather means the conditions and variability of the atmosphere at a specific location or region over a short time scale, such as minutes or days. Climate, on the other hand, means the aggregate of weather over a long time scale, such as years or decades. Weather is therefore aggregated in the climate of a given region. However, that view is reversed when people in the media ask if a given weather event was caused by changes in the climate.
Lake-effect snow is evident in this image, notably on the eastern shores of Lake Michigan. The image was taken in January 2013 by NASA's Terra space craft. CREDIT: NASA
Of course, what the reporters really want to know is whether physical connections exist between what happens on a daily basis and what is changing on timescales of many years. It may be then, that the original question is poorly phrased. As Kevin Trenberth of the National Center for Atmospheric Research stated in 2012, “All weather events are affected by climate change because the environment in which they occur is warmer and moister than it used to be.”
The key part of the quote is Trenberth's use of the term “affected” instead of “caused.” So the question pertaining to high-impact weather events should instead be phrased, How was this weather system affected by climate change? It is hard to say whether a current event happened due to changes in the climate, but it might be practical to consider how the intensity and dynamics of the system were influenced by changes to the ingredients that make up a storm.
To address that question, scientists conduct sensitivity studies of a given weather event by isolating different components of varying values. In that way, we can begin to understand how changes in the frequency or nature of the components change the strength, morphology, and lifetime of a system.
Although some of the consequences of climate change could be assessed by running climate models, such simulations have limitations. First and foremost, the current spatial resolution of climate simulations is on the order of tens of kilometers. Small-scale processes influencing the flow of the atmosphere cannot therefore be explicitly resolved; they need to be parameterized. The parameterizations work to create a more realistic large-scale flow, but they also tend to simplify the physics of these processes.
Parameterization can cause problems because processes happening on the sub-grid scale, such as the creation of cumulus clouds or the evolution of the planetary boundary layer, are the features typically associated with weather. Therefore, running simulations of weather—as opposed to climate—at a fine grid resolution allows for more explicit representation of processes like the creation of cumulus clouds.
A second limitation of climate models lies in the nature of weather. Because the weather is variable on short times cales, climate simulations model all variables changing simultaneously. Consequently, the models could fail to sample all the combinations of variables possible in a future climate. For example, the Buffalo snowfall event occurred because the waters of Lake Erie were warm compared to unseasonably cold air that remained over the lake for a long period of time. In climate simulations, that combination of factors may not be generated, leaving the dynamics of certain scenarios unrealized.
Cold air flowing over a warm lake body can create lake-effect snowfall in the winter. CREDIT: COMET, the Cooperative Program for Operational Meteorology, Education, and Training
The Great Lakes region of North America provides a rare opportunity to investigate the connections between climate and weather. The region’s lakes feature importantly in the weather patterns and climate of the region, mainly through the relatively high heat capacity and low diurnal temperature variability of the lake water.
During the late spring and early summer, temperature contrasts between the colder lakes and the warmer air and land surfaces and can be as high as 15 kelvin. In the winter, that contrast is reversed, with warmer lake temperatures compared to the overlying air and surrounding land. If the air is much colder than the surface water, the difference provides a strong forcing that enhances precipitation by increasing local instability and convection. Conversely, if the air is much warmer than the water, the difference provides a strong forcing that inhibits precipitation by increasing local stability.
One of the most common and dramatic types of lake-induced precipitation is lake-effect snow. The phenomenon occurs early in the winter season when cold air moves over the warm, open surface waters of the lakes. The air near the lake surface destabilizes through a combination of both sensible and latent heat fluxes from the lake surface. That condition, in combination with convergence of wind over either the center of the lake or downwind of the lake, results in enough vertical motion and mixing to create snowfall.
Lake-effect precipitation is crucial to the region’s moisture budget, as it provides up to 50% of the local annual wintertime precipitation downwind of the lakes. Later in the winter, ice forms over the lakes and the sensible and latent heat fluxes are reduced, along with the chance for snowfall. In recent years, the lake ice coverage has varied drastically from record low ice coverage in 2011–12 to near record high ice coverage during the 2013–14 winter. The variability in ice cover has led to the question of how changes to surface water temperature and ice coverage can change the overall dynamics and intensity of a single snowfall event.
To answer that question, my colleagues and I performed sensitivity studies using varying lake surface properties from a January 2009 lake-effect snowfall event in which all five of the Great Lakes—Superior, Michigan, Huron, Erie, and Ontario—produced snowfall. The lakes were initialized in two sensitivity simulations to investigate the change in lake surface characteristics in a potential warmer climate. In one simulation, all the ice was removed from the lakes. In the other, lake temperatures were uniformly raised by 3 kelvin. Simulations have shown that the number of cold air outbreaks in the future will likely decrease, but will not disappear completely. That finding implies that conditions of concurrently warmer lakes and colder air could still be possible.
When the lake ice was removed in our simulations, the amount of snowfall expanded along the lakeshore, increasing the area receiving snowfall during the event. When the lake temperatures were increased, the intensity and downwind propagation of snowfall increased, while the area along the shoreline that received precipitation did not change from the previous simulation. Those results showed that the ice plays a key role in determining the overall placement along the lakeshore of the snowfall, while the lake temperatures dictate the intensity and downwind extent of the snowfall.
One exception to those general conclusions was snowfall downwind of Lake Ontario, which had forcing to generate snowfall not only from warm lake water and the convergence of air over the lake, but also from orographic lifting caused by the Tug Hill Plateau. During the snowfall event of 2009, the changes in lake ice cover and lake temperature resulted in changes to the midlake convergence area, shifting the band of snow to the north of the plateau, resulting in less snowfall due to the reduction in orographic lift.
Because this research included only a single simulation, deriving claims about seasonal snowfall would be overstating the results. Still, the research does illuminate some of the nonlinear processes operating in lake-effect snow, and the processes’ sensitivity to variable climate conditions.
Those complex, fine-scale interactions make it key to study how different storm components react to changes in intensity. Our sensitivity studies have shown how important the lakes are to determining the downwind weather, and what a weather event in a changing climate may entail in terms of snowfall intensity and placement. By finding out how these different ingredients influence the weather, we can begin not only to answer the question, Was this storm affected by climate change?, but also to help prepare society for climate change and its potential dangers.
David Wright is a PhD candidate at the University of Michigan, where he studies the sensitivity of precipitation processes to lake surface characteristics and the connections between climate change and daily weather systems.
