The Making of a Weather Forecast Headline

These days a lot of thought goes into social media messaging and news headlines, but maybe not enough. While the ultimate goal is often to generate clicks, at what point are we doing our audience a complete disservice by drawing them to our site with a wildly exaggerated message?

Yesterday, The

UPDATE: Up to 15 inches of snow possible in #Denver by Friday night: http://t.co/taMI3Fp6Qd #COWX

— The Denver Post (@denverpost) January 30, 2014

The operative wording here is "up to." But to the average consumer it sure looks like a forecast for 15 inches of snow in Denver. While 15 inches may not have been completely out of the question, the probability we would be anywhere near that number in the city itself by Friday night was completely misleading.

Denver Post Article

To make matters worse, the article itself had the correct forecast from the National Weather Service which called for 5 - 10 inches in Denver, with "locally up to a foot of snow possible." It was in the eighth paragraph in the piece. So where did 15 inches come from?

This is less a critique on the Denver Post, and more of a comment on how we, those creating forecasts and communicating the weather, present our users with the information they need to prepare for a storm. We should always present the most probabilistic forecast, whether it's a hit-generator or not.

The problem here is that the public assumes you're presenting them with a deterministic forecast. Furthermore, they are more likely to latch onto the greater extreme, rather than the lesser. If you forecast 4 to 8 inches of snow, most people will expect 8 inches, and often if there's only four, there might as well have been zero. The same can be said if you forecast "up to" 8 inches of snow, meaning 1 inch is within your forecast range, why would any consumer assume you don't mean 8 inches is a high probability, or at the very least at equally as likely as 1 inch.

Social media users often share the most exaggerated headlines, and those media outlets receive criticism after the weather has happened and their forecasts turn out to be wrong. Readers then tend to lump all meteorologists into one group -- as if Weather5280, Denver Post, etc are all in on it together. For the outlets that stuck to more realistic forecasts rather than sensational attention-grabbers, there's often a sense of being cheated.

Whether this was the case with the Denver Post headline or not (it's unclear where they got that number), misleading the public for the sake of traffic can lead to a general feeling of mistrust toward meteorologists -- as if the weather itself doesn't already do a good enough job at that. Exaggerated headlines in the end could cause more damage than good when a big storm comes around.

In the end, Denver ended up with ~3 inches of snow overnight, with another 1 - 3 inches possible today. At Weather5280 we had a forecast of 4 - 8 inches for most of Denver (5 - 10 on the west side), with a "Bust Index" of 5 going into Thursday night (Bust Index means potential totals will be lower than forecast on a scale of 1 - 10).

Brendan Heberton

Brendan is founder of Weather5280. He is co-founder of FreshyMap, and develops software for geospatial data analysis and visualization.

Denver, Colorado
You can unsubscribe at anytime.