On a Tuesday, the weather app promises bright sunshine and 28 degrees for the weekend. So nothing stands in the way of a trip to the lake. But the closer the trip gets, the worse the weather forecast is. And on Friday it finally means: two days of showers and 22 degrees. What is a fictitious scenario here often occurs in reality. Have weather forecasts become less reliable, especially in the apps?

No, say experts. “It is in the nature of things that our weather forecasts cannot be exact – even if our technologies are very modern and are constantly improving,” says meteorologist Peter Knippertz from the Karlsruhe Institute of Technology (KIT).

To understand why this is so, it helps to understand how the predictions are made in the first place. For this, meteorologists need the current state of the atmosphere, as the German Weather Service (DWD) explains. They would need to know the current values ​​for temperature, humidity and air pressure for every point on earth.

These are determined at numerous measuring stations, for example on ships, buoys or aircraft, and supplemented with satellite and radar data, as DWD meteorologist Tobias Reinartz explains. “But that’s not nearly enough for a really complete knowledge of the current state of the atmosphere.” In addition, the atmosphere is a so-called chaotic system, Knippertz points out, “which sometimes makes it difficult for us to make any useful predictions at all.”

Highly complex physical equations

Meteorologists need the state of the atmosphere for their weather model. According to the DWD expert, this consists of highly complex physical equations that ultimately cannot be solved but have to be simplified. “So we’re feeding an inaccurate weather model with an inaccurate initial state of the atmosphere and now we’re demanding that something exact come out of it? Of course that doesn’t fit,” says Reinartz.

So is the increasingly loud claim that forecasts are becoming less and less reliable after all? No, says Knippertz unequivocally, the opposite is the case. Despite some restrictions, the weather can be predicted much better today than it was about ten years ago. Because the technologies and models have become much better and more accurate in comparison.

Reinartz also confirms this. With a view to the development of the air pressure forecast, he states: Today’s forecast for the next seven days is on average more precise than a 24-hour forecast in 1970.

The prospects are quite reliable, especially for the first three days, says the DWD expert. However, the reliability always depends somewhat on the weather conditions. According to Reinartz, it is particularly difficult to accurately predict a thunderstorm. Because these are very small-scale phenomena, especially in their formation, which can only be partially resolved by the weather models.

How are the percentages created?

According to Reinartz, you can use a model to forecast the weather hours in advance days in advance. However, as time goes on, the uncertainty increases – and therefore makes no sense. “To put it bluntly, you can confidently roll the dice for the amount of precipitation in seven days between 2 p.m. and 3 p.m..” Looking at the weather apps, he adds: “App forecasts often pretend to be accurate that doesn’t exist.”

The DWD expert explains that apps try to present the complex weather conditions as simply as possible so that many people can relate to them. “As a user, however, it is best to always get some context information, for example read a prediction text for better classification.”

KIT meteorologist Knippertz can also imagine that using many different weather apps gives the impression that the forecasts have become less precise or contradict each other.

Apps usually display the forecasts with symbols and percentages. Is it clear to everyone what the “50 percent” note on the rain symbol means? The information relates to a statistical calculation: if (within the framework of the existing uncertainties) 100 realistic predictions were made and 50 of these predicted rain for the relevant place and time, then the 50 percent would result, as Knippertz explains.

It is therefore important that people correctly interpret predictions. Because: “The better the weather forecast, the better we can deal with renewable energies, the better we can plan our agriculture and transport,” says Knippertz. It also saves lives in extreme weather situations.