The weather app is putting a damper on America's summer plans

Getty Images; Alyssa Powell/BI

Maya Haddad Miller has long been fixated on weather apps — especially how they hit her bottom line. Haddad Miller and her dad own Brooklyn Beach Shop in Coney Island, New York , which sells souvenirs and beach gear. When a little rain icon pops up on the ubiquitous weather apps on any given day, it can cut business in half, she says, whether any rain falls or not. "When you talk about weather forecasts, that's our life — we eat, breathe, everything the weather," she says.

This became such a hobby horse for Haddad Miller that, for a while, she kept a spreadsheet comparing forecasts with actual weather data to track discrepancies. A decade ago, she took the issue up with then-Lieutenant Gov. Kathy Hochul during the politician's visit to the beachside neighborhood. As they stood on the rooftop of a local restaurant, taking in the ocean view, she asked whether anything could be done about the way apps display the weather.

"You have so many businesses statewide that are affected by these false forecasts," she recalls telling Hochul. "A lot of times, it's beautiful out here, but it's very slow. We're selling less. We're contributing less taxes to the city, the state."

Hochul didn't have much to say back then, but Haddad Miller's crusade continues — and while her complaint may sound niche, she's onto a surprisingly significant predicament.

Every day, millions of people plan their lives around the default weather apps on their smartphones . The thing is, their super-simplified designs flatten what are, in reality, complex, highly localized, probabilistic forecasts. Distilling an entire forecast into basic high-to-low temperature ranges and simple sun or rain icons is a fraught task. And when things go wrong, there are tangible consequences: Empty beaches , canceled tee times, and ruined weekends. The little cloud icon of doom leads to real economic damage for businesses , and a nation of would-be customers sitting on the couch on a perfectly nice Saturday, realizing too late that they could have taken the family to the water park after all.

A sunny day in Coney Island apparently not thwarted by the weather app. Nicolò Campo/LightRocket via Getty Images

The problem with your run-of-the-mill weather apps isn't the meteorology, it's the translation.

Ironically, the science behind forecasting has never been better, says Daniel Swain, a climate scientist at the Institute of the Environment and Sustainability at UCLA. Behind every tiny icon sits an enormous amount of data: physics-based computer models, satellite observations, radar, weather stations, and, increasingly, AI systems that process it all. Forecasters often compare and blend multiple models to give the most accurate reads.

That complexity largely disappears once it reaches your phone. A single image represents the dominant conditions for an entire day, even if those conditions are expected to vary widely. And apps often don't do the type of complex analysis that human meteorologists do.

"Rather than leveraging the sum and totality of this incredible wealth of sophisticated meteorological information that's out there, they're just kind of saying, 'Eh, screw it, let's just pick one of these data sources,'" he says. "When you do that, you're kind of wedded to that one data source, which may or may not be the single best option in a given context."

Basic weather apps are OK for getting a "first glance" at what the weather might be, says Jason Furtado, an associate professor of meteorology at the University of Oklahoma, but they fall short on the details. Weather can be hyperlocal — a storm hits one part of town and not another — and a general sketch of the day often brushes over the small events that matter. Many generic weather apps also aren't good at tracking rapidly evolving weather situations. People don't need to know that it might snow at some point during the day in a 100-mile radius — they need to know what time and how severe.

"These apps, if you actually look at it, let's say hour to hour, it'll change," Furtado says. "That's not really helpful, especially if you're trying to make a decision."

Developers can't be blamed for everything — the apps have a user error issue, too. People have a hard time interpreting weather-related probabilities, whether they come from their phone screen or the local forecaster. When a source says there's a 20% chance of rain, it generally means there is a 20% probability that at least 0.01 inch of rain will fall somewhere in a certain coverage area. Some people will interpret 20% as 0% and get mad when they're caught in the rain. Others will decide it's too risky and stay inside, only to get frustrated when nothing happens.

"You already had that foundational misunderstanding of things like percent chance of rain, the difference between watch and warning," says J. Marshall Shepherd, a professor of geography and atmospheric sciences at the University of George and former president of the American Meteorological Society. "Now you throw these uber-convenient apps in the pockets and purses of everybody on the planet."

People regularly ask Shepherd for a weather forecast, and after he responds, counter with, "....but my app says." He says the counter does get to him occasionally: "I wish I had a dollar for every time someone has said that to me, I'd be retired."

The economy is totally dedicated by an icon on the phone. Some forecasts come with a "wet bias" that can overstate the likelihood of precipitation to manage expectations and avoid an outraged public if they get it wrong. Research shows that weather apps often present excessively detailed, unrealistic forecasts far into the future without making their confidence levels (or lack thereof) clear. In other words, that sun symbol on your phone 10 days from now may be pretty meaningless.

"Rarely do they include any estimate of uncertainty," says Susan Joslyn, a professor emeritus of psychology at the Univers...

Indexed and credited by AIPROPX. Originating outlet: Business Insider. Open at source →