Why Are Weather Forecasts Often Wrong?

Scientists use mathematical models for everything from weather forecasts to predicting who will win the next election. But how do these models work and why do they sometimes get it wrong? Everyday Einstein looks at the science behind weather prediction.

Lee Falin, PhD,
August 3, 2012
Episode #015

Page 2 of 2

All Models Are Wrong

There is a popular saying in scientific modeling circles, "All models are wrong...to a certain extent." This means that no matter how much work you spend in trying to refine the equations of your model to make them perfect, there is always a small degree of uncertainty.

If you've ever heard a newscaster report on an election poll, you might have heard one say something like "38% of voters prefer candidate X, plus or minus 5 percent." That "plus or minus 5 percent" is the uncertainty in the model, sometimes called the "margin of error." Taking the uncertainty into account, a more accurate statement would be "somewhere between 32% and 43% of voters prefer candidate X" but that just doesn’t have the same confident ring to it.

Let's Talk About the Weather

So let's go back to the weather. Weather models are a lot more complicated than predicting childhood juice consumption. The equations for weather forecast models incorporate things like wind speed, temperature, geographic features, and many other factors. The equations become so complicated that supercomputers are required to solve them.

Uncertainty in the results comes from the fact that neither the equations themselves nor the input data is perfect. For instance, each instrument that measures temperature has a small margin of error itself. All of these little errors add up until you end up riding home in a 3:00 thunderstorm that was supposed to be a 7:00 light shower.

That isn't to say that we aren't getting better. Every day scientists gather more data, look at the errors of past models, refine their equations, and build more accurate instruments in order to try and reduce the uncertainty.


So now you know how scientists use models to predict the weather, and why they’re so often wrong. This same process applies to each and every prediction that scientists make. Here are the steps they take:

  1. Gather data.

  2. Come up with a set of equations to describe the data.

  3. Use the equations to predict the outcome you can expect to happen with a given input.

  4. Don't be surprised when your models aren't exactly right.

If you liked today’s episode, you can become a fan of Everyday Einstein on Facebook or follow me on Twitter. If you have a question that you’d like to see on a future episode, send me an email at everydayeinstein@quickanddirtytips.com

Weather Forecast image from Shutterstock


You May Also Like...