Some of these numbers are the worst. (iStock)

Bryan Norcross is a hurricane specialist at WPLG-TV, Miami, and the Weather Channel.


Imagine making a forecast for the first really warm day in spring. Let’s say the expected high temperature is 86. But would it be easier for the viewers to imagine what it’s going to feel like if the forecast said “90 degrees?” Suppose the humidity is going to be high, as well. Does that make a better case for forecasting 90?

Should weather forecasters fudge the forecast if they think it’s going to better convey the upcoming weather?

Or consider this question: Which is a better forecast, 82 or 83? A meteorologist’s first response would be, “What do the models say?” But if you consider the psychological “odd effect” — people take longer to react and judge odd digits — the clear answer is 82. Even numbers and numbers ending in 5 feel better than odd numbers. Research has shown that people process odd numbers more slowly, so why make forecasts harder to digest when a fix is so easy?

Efforts to quantify the even/odd feel-good ratio put it around 60/40. Reaction times are slower and less favorable when a person encounters a discordant number ending in 1, 3, 7 or 9.

Back in the day, I had a rule that forecast temperatures should end only in even numbers or 5, unless the point was to communicate a trend over at least three days. In that case 82, 83, 84 would be okay. The sequence of 81, 82, 83 would not, however. That would be too much negativity for one forecast.

Additionally, meteorologists know that measuring the high temperature in a city on any given day is a dicey proposition. But the Fahrenheit temperature you see on your phone is already approximate — converted from Celsius, which the models use, and rounded up or down. And there is systematic uncertainty in the measuring system, which easily amounts to half a degree. If you add these “errors” up, odd-numbered forecasts are not worth it. Pinpointing a high of 83 degrees is no more exact than changing it to 82.

Of course, the computer models that produce most of the forecasts today don’t give a whit how we humans feel. They crank out 79s and 83s all day with no attention to artful communications.

The fix is easy. In 10 minutes, a programmer could tune up a computer so that it quits forecasting unfriendly numbers. But it’s the bigger question that is more important. If users gain confidence in the forecaster, either because they favorably remember the heat alert while sweating through the day for which 90 degrees was forecast, or because the 82 simply sits better in their mind, should we fight against that reality?

This question is much bigger than the numbers used in temperature forecasts. The word salad we use to communicate tornado risk requires people to know if enhanced is worse than moderate, for example. (It’s not, by the way.) During winter, which is worse: an advisory or a watch? The answer is neither, and it would take too long to explain why.

Meteorologists should ask their neighbors, “What does today feel like to you?” The next time the same weather pattern is developing, the answer will make a good forecast. And if an even-number temperature soothes the soul, so be it.

That the best modern science should underpin every forecast goes without saying. But, the path to good communication has to include artists who don’t need psychological research to know that enhanced, moderate, advisory and watch are intrinsically confusing words, and an 83-degree forecast feels off.

Read more from Bryan Norcross:

Facebook and Google have a responsibility during disasters

Hurricane forecasts are great — as long as the storms behave

It’s time to rethink our unintelligible weather terminology