What is a climate model?

Using a climate model, researchers try to simulate the natural processes in the Earth's atmosphere and the ocean as realistically as possible over longer periods of time. Sometimes they look several thousand years into the past or the future.

Ultimately, a climate model is a gigantic computer program, which has to be fed with basic meteorological and physical knowledge. These are, in particular, physical quantities such as the temperature or the radiation power of the sun, as well as formulas which, for example, describe wind which is the flow of air associated with high and low pressure areas. Often even processes like the exchange of greenhouse gases between the atmosphere, the land surface and the ocean are simulated. This also includes chemical and biological processes such as the absorption of carbon dioxide by plants.

In order to obtain an accurate picture of these natural processes, the model divides the atmosphere and the oceans into a network of cubes, so-called grid boxes. Depending on the application, the spatial resolution of this network is finer or coarser. In global climatic models, a grid box usually has an edge length of approximately 100 kilometres. For each grid box, the model calculates how the physical values change from one point in time to the next. If, for example, high air pressure prevails in a grid box, the air mass is pushed to the adjacent grid cells where lower pressure prevails, which in turn leads to pressure changes.

Step by step, the model calculates the state evolution into the future. A climate model can, for example, jump in 15-minute steps, whereby the calculation of such a time step requires only fractions of a second depending on the computer's performance. Since many physical formulas and values have to be processed for each grid box at each time step, the data volume is immense. Therefore, high-performance computers are required for climate modelling.The AWI experts use supercomputers such as the one at the German Climate Computing Centre (DKRZ) in Hamburg for their climate simulations. They can access it via an Internet connection.

Climate models can illuminate various questions, for example how the climate changed during the Ice Age or how it could change in the future due to the man-made greenhouse effect. For this purpose, climate simulations are started at a certain time in the past, such as a date in the pre-industrial era. Day by day, month by month the model calculates the change in the physical variables in the grid boxes. If a simulation is to provide information not only about the long-term development of the climate, but also about short-term changes in weather, the model can be fed with actual meteorological and oceanographic observations, as in a classical meteorological forecast. Experts call this start of a climate simulation initialization.