Why are weather forecasts inaccurate?

Over the years, weather forecasts have become much more accurate, which is helped by the continuous collection of data such as temperature, atmospheric pressure and precipitation. Then there are computers that process this data with the help of physical laws and thus make accurate weather forecasts for a certain period of time.

But our atmosphere is a “chaotic system”. As in the butterfly effect, in which even infinitesimal changes can later cause dramatic changes in a system, “even minute variations in temperature, atmospheric pressure, or wind over relatively distant areas can have a significant effect on weather, even with some time delay,” meteorologist Peter Knipperts from the Karlsruhe Institute of Technology (KIT) tells DW.

So “even with big computers and ever-better satellites or other measurement systems there will always be a degree of uncertainty,” explains Knipperts. Since it is impossible to make absolutely correct predictions, meteorology works with the probabilities of whether there will be any rain or storm.

Are weather forecasting apps reliable?

Weather apps available for smartphones are supposed to be very accurate and “give 10-day weather forecasts based on the zip code that the user enters,” says Knipperts. But for apps there is the same uncertainty in predictions, although this is often not really communicated to users.

Furthermore, there is no global regulatory body for these applications. “It’s an open market, anyone can program their own app, put it on the market and make money, for example from ads. Where exactly the information comes from and how the application processes it are two things that are usually not made public,” explains Knipperts.

Improving predictions through artificial intelligence?

Until now weather forecasts are based on physical models. In contrast, AI-assisted predictions work by collecting data and are more statistical. The artificial intelligence examines patterns and sequences in the already collected weather data and makes weather predictions through an algorithm. Therefore, the application of natural laws is considered rather incidentally.

The developments in AI are also impressive in the field of meteorology, however, given the way it operates and that it is based on past weather patterns, AI reaches its limits in cases of extreme weather phenomena, as Knipertz emphasizes. “That is why in the future we should try harder to form hybrid weather forecasting systems, combining conventional methods with those of artificial intelligence, in order to further reduce the risk of incorrect forecasts.”

As there are areas worldwide where weather data is not continuously collected, many are pinning their hopes on artificial intelligence. But Knipertz is somewhat cautious: “We need to try harder to fill the gaps in data collection in these areas. Because in the face of extreme weather events it is important for global meteorology to improve the observation and recording of weather conditions in Africa, Latin America or Southeast Asia. And I don’t think that artificial intelligence can make it easier for us.”

How does climate change affect predictions?

Climate change does nothing to change what is true about the laws of nature and the fundamental problems in weather forecasting. But climate zones differ, as sometimes happens with extreme weather phenomena. “Tornadoes, heavy rains or even droughts may be even more intense than in the past, with correspondingly greater impacts on people.”

There arise “new challenges for meteorology, both in terms of early warning of the world in the face of some extreme phenomenon, and the readiness of the population to take such warnings seriously and behave accordingly”, points out Knipperts.

Why are extreme weather reports so dramatic?

Sometimes the authorities’ forecasts for a heavy rainfall or storm are much more dramatic than the phenomenon that eventually hits the area, which can lead many people not to take such forecasts so seriously. “When people move away and nothing happens, there’s often a lot of backlash on social media and the internet,” says Knipperts.

As the expert reports, warnings are always a difficult balancing act: “The cost of an evacuation seems to me to be much less than drowning in one’s own basement. But in my opinion, there is very limited awareness and understanding of the population about this.”