Statistics for Meteorology: What's the weather like in 10 years?News article in Der Tagesspiegel

21 February 2024

Photo: picture alliance / Sipa USA

The Coming Decade project was featured in the German daily newspaper Der Tagesspiegel on Feb 20, 2024 with the title "Statistiken für die Meteorologie : Wie ist das Wetter in 10 Jahren?" by Pepe Egger. Link to original article in german: https://www.tagesspiegel.de/statistiken-fur-die-meteorologie-wie-ist-das-wetter-in-10-jahren-11241967.html

Translation by DeepL.com (free version):

Statistics for meteorology : What will the weather be like in 10 years?

In the research project "Decadal Climate Outlook for Europe", Henning Rust is working on forecasts that are relevant for the economy.

By Pepe Egger

Henning Rust is a physicist and meteorologist, but his forecasts do not relate to the next few days as in the weather report on the Tagesschau. Rust also does not create long-term climate models, for example on how much the Earth's atmosphere will heat up by 2050 or 2100. Instead, he works on predictions of what the climate will look like in ten years' time.

His specialty within meteorology: statistics. "The Coming Decade - Decadal Climate Prospects for Europe" is the name of a joint project in which Henning Rust is involved. It is led by Professor Johanna Baehr from the Institute of Oceanography at the University of Hamburg; funding comes from the Federal Ministry of Education and Research, and the Max Planck Institute for Meteorology Hamburg, the University of Bonn, the Karlsruhe Institute of Technology, the German Climate Computing Center and the German Weather Service are also involved.

But how do you create such a "decadal" forecast? Why do you do it and for whom? To understand how a climate forecast for the next decade works, it helps to understand how a "normal" weather forecast works: Since the beginning of the 20th century, explains Henning Rust, meteorology has been a natural science that attempts to explain what happens in the atmosphere using physical equations. "However, this has only been applied in practice and used regularly since the 1960s, because it was only with the development of computers that numerical weather forecasts became possible on a large scale." Such a forecast is based on the current state, the existing highs and lows and their location, and simulates the processes in the atmosphere for the future on this basis.

However, this method only works really well for the coming days: the further into the future the forecast lies, the less accurate the forecasts become. Because the weather model cannot depict reality in its full complexity, but always simplifies it, it is inevitably uncertain. Meteorology solves this problem by using statistical methods and predicting probabilities - and this is Henning Rust's specialty. This is how, for example, the "35% chance of rain next Friday", as predicted in the weather apps on smartphones, comes about.

However, for a projection of the climate in 50 or 100 years, it is of secondary importance where, for example, low Jitka or high Dario are currently located. Such a forecast is based firstly on knowledge of the state of the ocean and its development and secondly on the composition of the atmosphere - in particular the change in greenhouse gas concentrations and their probable development.

The decadal climate forecasts that Henning Rust is working on are positioned precisely between these two extremes: short-term weather forecasts and long-term climate projections. For forecasts that extend decades into the future, the events in the oceans are the most important factor, says Henning Rust - neither highs and lows nor the exact carbon dioxide concentration, but rather the temperature of the oceans, the transatlantic overturning circulation and the El Niño phenomenon of changing ocean currents. Processes that take place much more slowly than the processes in the atmosphere. This is why decadal predictions can also be described as predictions of the "weather in the ocean", says Rust. The aim is not to predict local weather patterns for every hour or every day. Instead, the forecasts refer to large areas such as the whole of Germany and to annual or even multi-year averages. Similar to the short-term weather forecasts, the decadal forecasts are formulated in probabilities.

Statistics optimize predictions

But how do researchers go about producing a forecast of this kind? Henning Rust says that a "coupled atmosphere-ocean model" serves as the basis, in this case a further development of the global ICON model used by the German Weather Service. First of all, the starting conditions in the ocean and in the atmosphere would have to be defined, i.e. the conditions and thus the basis of the model. They should be as close as possible to the observed conditions. The model is then refined and tested by attempting to make predictions based on past data, i.e. to create so-called hindcasts.

Naturally, there are discrepancies in the hindcasts between the results of the model and the actual measured values: this is what Henning Rust is dealing with in a narrower sense in his sub-project of the research project "The coming decade". Using statistical methods, he can describe systematic deviations of the model from reality and thus correct them, which significantly improves the quality of the predictions. For example, a model is being designed that predicts the climate in ten years' time based on the current conditions in the oceans.

But what about the extreme temperature values that were measured in the North Atlantic in particular last year? The global average surface temperature of the oceans - with the exception of the polar seas - was half a degree Celsius higher in July 2023 than the average over the years 1991 to 2020. Do such drastic deviations from long-term averages possibly indicate that fundamental balances have been disturbed here?

Economy needs forecasts

Henning Rust does not want to speculate on this: On the contrary, his work as a meteorologist actually begins precisely with finding explanations for such developments. In addition to warming due to a higher concentration of greenhouse gases, he mentions several possible factors whose influence should not yet be considered quantified: for example, a returning El Niño phenomenon, i.e. the periodic significant warming of the Pacific Ocean off the west coast of Central America. As a second factor, Rust mentions the eruption of an underwater volcano in the Pacific, which currently has a minor or even cooling effect.

Another possible explanation: the reduction in the amount of aerosols. This is a side effect of new regulations for better air quality and cleaner marine fuels: Because there are fewer aerosols as a result of the regulations, their cooling effect also fails - the oceans could thus warm up more because more solar radiation hits their surface.

But who are the addressees and customers of the decadal climate forecasts? Henning Rust says that they are also important because their supplies and requirements are included in the research project. Thanks to the expertise of the Max Planck Institute for Human Development, prediction products are developed and tested together with the addressees using social science methods as part of a co-design approach.

This includes groups of people and institutions that draw up long-term plans and want to know what climatic conditions they will have to adapt to in the near future: Air conditioning manufacturers, companies from the construction industry, the energy sector, agriculture and public authorities. They need key data on the temperatures and precipitation levels to be expected over the next ten years, or how many frost or heatwave days there are likely to be. They can rely on the research ambition of Henning Rust and the entire "The coming decade" team.