Why do some meteorologists get the forecast right and some are totally off base? How do forecasters predict the weather? Which forecasts can you trust? Everyday Einstein explores the science behind weather forecasting.
Hi I’m Dr. Sabrina Stierwalt, and I’m Everyday Einstein bringing you Quick and Dirty Tips to help you make sense of science.
Last week winter storm Juno was supposed to bury large portions of New England all the way from Maine down to New York City under several feet of snow. In preparation for the storm, the city that never sleeps came to a halt. The subway trains stopped running. All non-emergency vehicles had to stay off the roads. Airlines canceled flights in and out of LaGuardia, JFK, and even Newark airports.
In the end, Juno delivered as promised in many areas: Boston saw 2 feet as predicted. Other places in Massachussettes, as well as in Long Island and Connecticut, got more than 30 inches of snow! The forecast called for 18-24 inches in Portland, Maine which ended up with 23.8 inches, right on the money.
But New York City’s Central Park was left with less than 10 inches - still a significant amount of snow but far less than the expected 2 feet. The borough of Brooklyn saw only around 7 inches.
So what gives? How can some weather predictions be so accurate while others miss the mark, even for the same storm? How well can we actually predict the weather?
See also: Why Is the Weather So Hard to Predict?
Weather Forecasting Is Challenging
A simple model considers an input (or inputs) and predicts an output based on what is known about that input. For example, when food gets dropped on the floor at my house, I predict that my frisky hound will get to it faster than my lazy bulldog. My prediction is based both on my understanding of their relative speeds and on past experience.
In the case of weather forecasting, there are many complicated, inputs like atmospheric pressure patterns, the direction and speed of the wind, the variation of air temperature at different heights in the atmosphere, and the dew point or how much water vapor is in the air for producing rain or snow.
Weather stations all over the world, including some strapped to buoys in the ocean, track these parameters by the minute. Weather balloons monitor the conditions high up in the atmosphere, and satellites provide a view of the Earth’s weather from space. As you can imagine, powerful super computers are necessary to collect and interpret all of this data.
The behaviors of each of these individual inputs, as well as how the inputs interact with each other, are represented or modeled by mathematical equations.
For example, in the middle of the northern hemisphere, around 30-60o latitude, warm air from the tropics below can meet cold air from the polar region above to create what are called low pressure systems, the fodder for future storms. Exactly when, where, and how intense these storms will be depend on all of the initial conditions that set the stage for each storm.
In most places in the world, weather does not proceed in clear, predictable patterns. In other words, weather is chaotic, and those mathematical equations are quite complex. Even small changes in the initial conditions or inputs can lead to big differences in the output predictions.
Due to the complexity of the models used for weather forecasting, each output or prediction is associated with an uncertainty. In the case of Juno, the storm fell about 50 miles east of its predicted path. Thus, most of New England was hit just as hard as predicted, but New York City escaped relatively unscathed.
Fifty miles sounds like a lot when we’re talking the entire population of NYC, but it’s actually only roughly 10% of the storm’s predicted coverage. In the mathematical modeling of something so complicated as weather, that’s pretty good.
Which Weather Model to Use?
Not all the forecasts were so far off with their predictions for New York City. On Sunday The Weather Channel gave similarly devastating predictions as the National Weather Service and others, but by early Monday they lowered their estimates to 8-12 inches of snow, which turned out to be an accurate representation of the snowfall NYC actually saw.
So what did The Weather Channel do differently?