What's the physics behind "lake effect snow"?
Dear Straight Dope:
What's the physics behind "lake effect snow," which Windy City weather prognosticators are always predicting?
Percy Shelley said, "I love snow, and all the forms / Of the radiant frost." Excuse me while I gag. It's obvious Shelley never lived through a Great Lakes winter. I lived through about 2,100 inches worth of them in my formative shoveling phase (20 years at 105" per), and I can assure you that snow loses its poetic appeal after a while. Don't get me wrong; I don't hate snow, but enough is enough.
Snow is only one of the lake effects--that is, the climatic effects that large lakes have on land areas near them, especially on the leeward side. They're not all as hard to deal with as snow. Temperatures are lower in summer and higher in winter than at the same latitude farther from the lakes. The growing season is longer. Hail is rare, but fog is common. The snowmelt is a dependable source of moisture for plants, making droughts rare. These lake effects make possible valuable crops that would otherwise be difficult to grow in the region: apples near Lake Ontario, grapes near Lake Erie, and peaches, cherries, and strawberries near Lake Michigan.
Back to snow. Most northerly locations get their snow from synoptic (large-scale) weather events, such as the passage of low-pressure systems and frontal boundaries. Near the Great Lakes, the amount of snowfall from synoptic events is often increased by the warmth and moisture of the lakes. This is called "lake-enhanced" snow, as distinct from true lake-effect snow. For example, the passage of cold fronts can cause heavy precipitation anywhere. But in the Great Lakes, they often cause whiteout (zero visibility) conditions because of lake enhancement.
True lake-effect snow occurs under conditions when no snow at all would be expected without the lakes' influence. This sometimes happens when a land breeze develops due to convection because of a temperature difference between the cold land and warm lake. Cold dense dry air from the land flows over the surface of the lake and picks up warmth and moisture from the water. If a synoptic wind is blowing from the opposite direction at the same time, the two moving air masses will meet over the lake and will have nowhere to go but up. (This is the process of convergence, which we will meet with again later.) The moist air cools as it rises, forming clouds. Since the lake breeze blows only near the surface and the synoptic wind usually blows aloft as well as at the surface, the clouds will be pushed by the synoptic winds back over the land where they drop their load of moisture as snow. This sort of lake-effect snow forms what are called "shore-parallel bands," because snow falls in a band that is widest along the shoreline. They can result in moderate to heavy snowfall, but the snow rarely reaches more than a few miles inland. One odd thing about this scenario is the fact that the surface wind is blowing toward the water, even though the precipitation is coming from the water.
A different sort of storm can develop when a cold synoptic wind flows across a lake in the absence of a strong land breeze, forming "wind-parallel bands" of snow. This often happens following the departure of a cold front, when the air is cold and a steady wind can be expected. It can also happen in the presence of the steady winds that flow around a stationary high- or low-pressure system. The air flows across the lake, picking up moisture and warmth as it goes. The warmed and moistened air near the surface becomes less dense and so has a tendency to rise and form clouds. Snow often begins to form over the water but it can greatly intensify when the moist air reaches the shore. The friction of land is much greater than that of water, so the wind has to slow down as it crosses the shoreline. But air is still rushing in from over the lake, and it has to go somewhere. Some of it goes up, which facilitates cooling and cloud and snow formation. This is another example of convergence.
Storms of this sort can take the form of multiple parallel bands of clouds (called "cloud streets"), each a couple of miles wide and spaced several miles apart. Or they can take the form of a single dominant band of snow which may be several miles wide. Wind-parallel snow bands can produce moderate to heavy snow, and can reach inland a considerable distance, very often twenty to fifty miles, and more rarely over a hundred miles. One striking characteristic of wind-parallel snow bands is their localized nature. It often happens that one town may receive a heavy snowfall and the next town along the shore little or none.
The most impressive lake-effect storms occur when the synoptic wind blows along the length of the lake, rather than across it or at an oblique angle. A single "mid-lake band" can form if a surface-level land breeze develops at the same time, leading (once again) to convergence and lifting. Where this band intersects the land at the leeward shore, a huge amount of snow can be dropped. Snow continues to fall as long as sufficiently cold air flows in the right direction. It was just such a storm that dumped almost seven feet of snow in eight days at Buffalo in late 2001. The same wind system dumped more than ten feet on Montague, New York (east of Lake Ontario). See a detailed description of that event here.
The amount of snow dumped by lake-effect storms depends on several factors, including the temperature difference between the air and water. A difference of at least 20 F to 25 F is usually needed, but an extreme difference (say, 50 F) may not mean heavier snow, since very cold winds are less conducive to evaporating large amounts of water from the lake surface. If all the ingredients for a lake-effect storm are in place, but the temperature is too warm for snow, lake-effect rain can occur. This happens most often in September and October and can cause flooding.
Wind speed is another important factor. If the wind is too slow (less than about 10 m.p.h.), most of the snow may fall over the water and not the land. If the wind speed is too high (over about 40 m.p.h.), the air may not pick up enough moisture on its quick trip over the lake to produce a lot of snow.
Another important factor is the fetch over water. No, it's not a command you give your Labrador retriever. Fetch is the distance the wind travels over open water before it hits land again. The longer the fetch, the more moisture the air can receive and more snow it can form. One reason that mid-lake bands are so impressive is that the fetch is maximized. Major storms rarely develop unless the fetch is at least 50 miles.
Fetch across a frozen stretch of lake doesn't contribute since evaporation is much reduced. Lake Erie is the most likely of the Great Lakes to freeze because it is by far the shallowest. Erie will be 90% ice-covered by February in a normal year and it frequently freezes over completely. This explains why lake-effect snow is primarily an early-winter phenomenon off Lake Erie. The three upper lakes are deeper (though farther north) and will be half to three-quarters ice-covered in a normal year, and they occasionally freeze over completely. Lake Ontario is the least likely of the Great Lakes to freeze, because it is relatively far south and relatively deep, and because winter temperatures are moderated by the presence of the upper lakes. In a normal year, Lake Ontario will be only a quarter ice-covered, and it has completely frozen over only twice since records have been kept. The last time was in 1934.
This brings us to the question of why some cities on the lakes (like Buffalo) get much more lake-effect snow than others (like Chicago). The heaviest lake-effect storms occur when a cold synoptic wind flows over a long stretch of warm open water before reaching land. In the Great Lakes region, the wind is usually from the west. With winds from that quadrant, the fetch off the nearest lake is zero at many of the larger cities (population over 200,000) in the region (within 40 miles of the lakes): Chicago (37" of snow a year), Toledo (also 37"), Detroit (43"), Milwaukee (51"), Toronto (53"), and Hamilton, Ontario (60"). These cities receive significant lake-effect storms only on those rare occasions when a cold wind is blowing from a quadrant other than the western. Toronto and Hamilton are close enough to Georgian Bay (an arm of Lake Huron) and Lake Erie that they receive some lake-effect snow from those lakes, in addition to some from Lake Ontario.
A study of the ten biggest snowstorms in Chicago and southeastern Wisconsin from 1970 to 1994 found that not one of them was a true lake-effect storm, and only three were probably or possibly lake-enhanced events. Variations over relatively short distances are telling. South Bend is only 75 miles from Chicago, but gets twice as much snow, because it has a significant fetch over water when winds are from the northwest.
The larger cities in the region with a significant fetch when the wind is from the west are Akron (47"), Cleveland (61"), Grand Rapids (71"), London, Ontario (83"), Buffalo (96"), and Rochester, New York (100"). Of these, Akron and Cleveland have a relatively short fetch when the wind is from the western quadrant. More importantly, the fetch is often across the very shallow western end of Lake Erie, which usually freezes early. At Buffalo, fetch is more likely to be across the deeper eastern half of the lake, which freezes later. The unfrozen state of Lake Ontario explains why (averaged over 30 years) Rochester gets more snow than Buffalo, even though it doesn't make the news as often. Buffalo gets much of its snow in huge mid-lake storms early in the season (before Lake Erie freezes). Such huge storms may happen only a few times a decade, but they get lots of attention when they do. Rochester gets smaller lake-effect storms, but it continues to get them throughout the winter. Slow and steady may win the race, but it doesn't get you on the national news. Buffalo did beat Rochester in the cold and snowy 1970s (111" a year vs. 107"). Those were in the days when we were worried about the coming ice age, not global warming. Actually Rochester would lose the race to a number of smaller cities such as Watertown (111"), Syracuse (120"), and Oswego (153") near Lake Ontario in New York; and Sault Ste. Marie (131"), Marquette (180"), and the Hancock-Houghton airport (227") near Lake Superior in Michigan.
Towns at a high elevation above the lakes are pretty well set for snow, too. In addition to convergence, there is another mechanism at work here, called "orographic lifting." As the moist air off the lakes encounters steep terrain, it is forced to rise, which causes it to cool, form clouds and drop precipitation. This is the same mechanism that causes the very high snowfall totals in mountain ranges east of the Pacific Ocean, such as the Cascade Range of Washington state and the Sierra Nevada in California. Between Erie, Pennsylvania, and Buffalo the edge of the Appalachian Plateau approaches within a few miles of Lake Erie. A number of small towns in this region get well over 100". For example, Colden, N.Y. (158"), is twelve miles from Lake Erie and 450 feet above above it, while Corry, Pennsylvania (122") is 25 miles from the lake and 860 feet above it.
At the eastern end of Lake Ontario, the Tug Hill Plateau and the Adirondack Mountains are directly in the path of any mid-lake band that develops, making this one of the snowiest regions in the U.S. east of the Rocky Mountains. For example, Old Forge, New York (227") is sixty miles east of Lake Ontario and 1490 feet above it. This part of the state is sparsely populated for some strange reason, but places like Boonville (219"), Barnes Corners, and Montague occasionally make the news. Snow is an important part of the economy here, attracting downhill and cross-country skiers and snowmobilers. A dot on the map called Searchmont (167") in the highlands east of Lake Superior is said to be the snowiest spot in the province of Ontario, but data for that station is sketchy. The snowiest place in Ontario for which good records are available is Wiarton (156"), at the base of Bruce Peninsula. The peninsula separates Georgian Bay from the rest of Lake Huron, so Wiarton gets lake-effect snow from both directions.
The Great Lakes are the most famous lakes for producing this type of snow, but they're not the only ones. The Caspian Sea, the Aral Sea, and Lake Baikal are all said to produce lake-effect snow. In Vermont, Lake Champlain contributes to the 82" that Burlington receives. Champlain freezes over most years, which limits the amount of snow and must incidentally make life difficult for any lake monsters that might live there. In Utah, the Great Salt Lake contributes to Salt Lake City's 63". That lake's salinity reduces evaporation somewhat, but that is more than made up for by the fact that it prevents freezing (it never freezes completely over).
Speaking of salt water, a phenomenon similar to lake-effect snow, called ocean-effect snow, can occur when cold winds flow from a continent across an arm of the ocean to an offshore island or peninsula. Western Korea receives ocean-effect snow from the Yellow Sea, Hokkaido from the Sea of Japan, Newfoundland from the Gulf of St. Lawrence, and Cape Cod from Cape Cod Bay.
In researching this topic, I learned that a Cleveland brewery makes a Lake Effect Ale. I haven't tried it, but I'm willing to bet it's easier to take than the other sort of lake effect. I never had the pleasure of visiting Montague, New York, which in 1997 got 77 inches of snow in just 24 hours, a new national record. I'd rather have 77 inches of beer any day, and I suspect the people of Montague would too.
Handbook of Snow: Principles, Processes, Management & Use, edited by D. M. Gray and D. H. Male
U.S. snowfall averages are from Weather America (Grey House Publishing). Similar but older data is available from The Midwestern Regional Climate Center.
Ontario snowfall averages are from Environment Canada.