Does Great Lakes ice cover predict West Michigan weather?

Posted at 10:32 AM, Feb 26, 2015

GRAND RAPIDS, Mich. — A great deal of media attention has been given to the amount of ice cover on the Great Lakes again this season.  As of February 25, the Great Lakes Environmental Research Laboratory listed the ice cover at 85.5%.  As the image above shows, the greatest ice coverage is on Lakes Superior, Huron, and Erie.

Now, as winter (in theory, at least) starts to wind down, there is speculation about how the cold lakes will influence weather around the region as we transition into spring and summer.  The headlines of several articles, like this one, suggest that lake ice may doom us to several more months of cooler-than-average weather.

So, I decided to see if the data backs up that idea for us here in West Michigan. Now this is, admittedly, a very limited look at the statistics.  A better analysis would use temperatures for a wider area and attempt to correlate exact ice percentages and average temperatures.

Good ice cover data from GLERL goes back to 1973.  I separated each year into categories of above- or below-average peak ice cover, and then calculated average temperatures for Grand Rapids in spring (March-April-May) and summer (June-July-August), comparing them to the 1981-2010 averages.

If ice cover has a significant effect on temperatures in the spring and summer, one would expect above-average ice to lead to below-average temperatures, and vice versa.

So, what do the numbers show?

Great Lakes peak ice cover vs. Grand Rapids spring average temperatures

Avg. Temperature
Below Avg. Above Avg.
Ice Cover Above Avg. 16 years 6 years
Ice Cover Below Avg. 8 years 10 years

Great Lakes peak ice cover vs. Grand Rapids summer average temperatures

Avg. Temperature
Below Avg. Above Avg.
Ice Cover Above Avg. 19 years 4 years
Ice Cover Below Avg. 7 years 10 years

Note: In both spring and summer, there are two seasons with exactly average temperatures.  I treat those years as half a “win” and half a “loss” below.

So, in the spring, ice cover correctly predicts temperatures 27 times, and fails 15 times — a 64% success rate.  In the summer, it is correct 30 times compared to 12 — 71% success.

That sounds pretty solid on the surface, but correlation does not equal causation.  The persistence of a warmer or colder weather pattern due to jet stream positioning, etc. plays a role in both ice formation and average temperatures.

As with all things weather and climate, there is likely some randomness to these results as well.  One might expect that if the ice-temperature connection is a strong one, it would show better results in the spring than summer while there is still ice present on the Great Lakes; however, the correlation is actually higher in the summer here.

In the Chicago Tribune article I linked to earlier, you have to read to the end to get the full scoop.  Words like “usually” and “probably” start to slip into the statements from the experts.  As a climatologist from the Midwestern Regional Climate Center is quoted, “The Great Lakes will help to keep areas cooler, but it’s not the foremost thing that keeps things cool.”  However, uncertainty doesn’t make for good headlines, and so this sentence gets buried as the last words of the article.

When it comes to weather (and science in general), it’s our natural tendency to want to simplify things.  But in long-range forecasting, there is rarely a connection so strong that we can say “X causes Y.”  At best, we can try to say, “X favors Y,” or “X often results in Y.”  It’s a game of playing probabilities, but people don’t process that information very well.  And as any gambler knows, we’re bound to be disappointed if we start counting on a specific outcome.

Long story short — you should be skeptical about anyone (including me) claiming to know what the weather will be like weeks or months from now.  If you asked me to bet on the spring and summer weather, I would put my money on being cooler than average, but because of factors other than ice.