When Apple’s weather app crashed twice in a week in early April, causing users to stare at a blank screen instead of weather data for hours, the resulting “noise” is indicative of modern dependence on weather applications.

Many international media rushed to cover the issue. But on Twitter, too, thousands of users were willing to comment: “Apple Weather has been down all morning and I had no idea how many disruptions it would cause,” one user characteristically wrote, according to CNN, while another joked referring to the “Top Pistol”: “The biggest storm of the season is about to hit Fargo, and Apple’s weather app isn’t working. I’m flying blind, Goose.”

Today, less and less “hanging from the lips” meteorologist just before the end of the news to see if it will rain the next day. Now, with the default mobile app, you can see if they need to bring an umbrella or wear short sleeves. But how much can we trust the weather apps we use on our devices?

The role of numerical models

“These applications do not predict the weather. In fact, they visualize and convey to us in a simple form the results of a numerical model that “launched” sometime earlier, at best six hours ago. This means that their reliability depends on the reliability of the numerical model from which they derive their results.” meteorologist Panagiotis Skrimizeas of EMY’s Office of Forecasting Methods, Research and Meteorological Models explains “K”.

“If the numerical models they can’t give us the weather with absolute certainty what will prevail in the future, at any particular moment – and I mean not only the type and intensity of phenomena, but also temperature, humidity, wind, sunlight, percentage of cloud cover, etc. – with what certainty they I can document it chance about what the weather will be like, but also range of possible deviations“, he adds.

During the first 24-48 hours, the probability of success of numerical models can exceed 90%.

During the first 24 to 48 hours of forecasting, the scientists argue, numerical models, if they “work” within scientifically acceptable limits, can be very reliable, with a success rate of up to 90% or higher. “With new technologies, the weather industry has great potential to improve forecasting models and draw more accurate conclusions about weather evolution. It means that we also expect the apps to get even better in the future,” said National Meteorological Service Director Thodoris Kolidas “K”.which encourages users to use weather apps on their mobile phones, emphasizing that the weather services themselves should give preference to some trusted apps.

20% that “burn” users

About 300 days a year, or 80% of all days, the weather app will work correctly, he recently told an American magazine. Atlantic meteorologist Matt Lanza from Texas, who manages Houston’s space weather. “It is the remaining 20% ​​that burn users and create problems,” he says.

“The problem starts when you take model data and load it directly into applications. without interfering with the interpretation of the meteorologist himselfbecause you won’t get any differentiation from those apps. “It will probably rain all day” is different from “it will rain all day,” he says in an interview.

But how can we use weather apps effectively so that we don’t get frustrated every time they “fall out”? According to meteorologist Panagiotis Skrimizeas, we must use the information provided by each application on our mobile phone wisely.

“It is important to understand the information only as an indication of the weather trend. This means that I will take an umbrella with me if it looks like it will rain, but I will not plan an event or any other major event based on this forecast, especially if it is still 24 hours away,” she says.

“You can’t replace the weather forecaster”

The meteorological service forecaster, using his knowledge and experience, is able to evaluate the numerical model and confirm whether it can be trusted. “Numerical prediction models need to incorporate this knowledge and experience of the meteorologist. In this direction, the scientific community is already investing in the application of machine learning (ML) and artificial intelligence (AI) systems, says Mr. Skrimizeas, emphasizing, however, that such systems can by no means replace a scientist. . “A business meteorologist will never cease to be needed. It will evolve to assimilate and use the opportunities that modern technology gives it, as it has been doing for all these years.”