🌡️ Where do these systematic temperature errors in climate models come from?

Despite improvements in model resolution, systematic errors - known as biases - persist in the representation of average oceanic and atmospheric temperature in certain coastal regions. A study published in Geophysical Research Letters has identified the origin of this bias. A promising discovery that will allow scientists to develop even more effective models.

Global climate models are of crucial importance for predicting the impacts of global warming. Their major improvement over recent decades has been the subject of numerous studies, and their degree of realism has often been correlated with the increase in model resolution, i.e., the fineness of the grid point they can represent.


Recently, the unprecedented increase in available computing power has enabled the development of kilometer-scale climate models, which not only allow for a better representation of the climate on a global scale, but also provide projections of changes at the local scale, where adaptation policies and strategies are implemented.

However, a recent study highlighted that, despite the improvement in model resolution, systematic errors - called biases - persist in the representation of average oceanic and atmospheric temperature in certain coastal regions. These biases can be explained by the absence of tide consideration in this type of model.

The tide, although a very short periodic phenomenon compared to the timescales of climate evolution, actually has a non-zero average effect on our climate system via the turbulent mixing it induces. The absence of this phenomenon in models can lead to temperature errors on the order of 5.4°F (3°C) in the ocean and 2.7°F (1.5°C) in the atmosphere, particularly along the coasts of the English Channel, the Irish Sea, and the North Sea.

This study is based on the evaluation of 8 climate models from different research centers around the world (Europe, United States, China, etc.) as well as on satellite observations. It was carried out by scientists from the CNRS (see box), Ifremer, the Met Office (United Kingdom) and the Max Planck Institute for Meteorology (Germany).

Coastal regions being particularly vulnerable to the effects of climate warming, the prediction of temperature in present and future climates by models is of high stakes. This discovery will allow scientists to develop strategies to better include the effects of tides in the next generation of climate simulations that will inform future IPCC reports.

To learn more

Delpech, A., Tréguier, A. M., Marié, L., Ghosh, R., & Roberts, M. J. (2025). Persistent Coastal Temperature Biases in km‐Scale Climate Models Due To Unresolved Oceanic Tidal Mixing. Geophysical Research Letters, 52(19), e2025GL118014.