The latest generation of climate models now matches observed reality so closely that the models are no longer the main limitation in climate calculations — the quality of observational data is. That is the conclusion of a study led by Dr Lukas Brunner of the CLICCS cluster of excellence at the University of Hamburg, published in Communications Earth & Environment.
Brunner and colleagues examined 176 climate models spanning the past 30 years and compared their temperature maps against ten observational datasets rather than a single reference. On average, the models have become significantly more accurate, though the researchers note the progress is uneven: “If you take a really good model from the 1990s, its performance is actually similar to that of a mediocre model today.”
The study highlights the newest kilometre-scale models, which simulate the climate system on grids of five to ten kilometres rather than the older standard of around 100 kilometres. That resolution allows processes such as the formation of thunderstorm cells to be simulated directly instead of approximated. The work also dispels a common assumption: simply doubling resolution does not improve results unless model physics and tuning are adapted too.
For the Baltic, where wind, precipitation and storm patterns shape both energy yield and coastal risk, the reliability of regional climate projections is more than academic. Better models feed the IPCC reports and the regional adaptation planning that energy and infrastructure decisions increasingly rest on.
