Original title: Sensitivity of the ECMWF seasonal forecast model to CO2 and anthropogenic aerosol forcings: Experimental design and impact on climate trends
Michael Mayer, Daniel J. Befort, Jacob Maddison, Antje Weisheimer · August 2026
Standard climate models struggle to match real decadal climate trends. To address this problem, scientists turned to a weather forecasting model from the ECMWF. These forecasting models are regularly reset with real-world data, which gives them lower biases and higher prediction skill. The researchers ran tests using tweaked historical conditions. They altered atmospheric gases like carbon dioxide and adjusted ocean starting temperatures by adding or removing human-caused warming. They also tested the effect of aerosols. They found that boosting human-caused warming strengthened several observed trends that standard setups underestimate, such as tropical Pacific winds and energy fluxes at the top of the atmosphere. However, the model still failed to maintain the ocean temperature gradient across the tropical Pacific through natural air-sea feedbacks. Aerosols showed little impact, likely because indirect aerosol effects were left out. Overall, this method creates a useful way to study climate drivers and model flaws.
Imagine baking a cake every day using a recipe that always produces a slightly flat result. You suspect the baking powder is the issue, but you cannot be sure because you also alternate between old flour and fresh flour. One day you bake two identical cakes: one with double the baking powder, one with none at all. The cake with extra baking powder rises better, but still not as much as you expected—telling you something else about the oven or mixing is limiting the rise. In this paper the researchers baked multiple histories of the climate, doubling CO₂ or removing it, to see which trends changed and which did not.
Climate scientists have struggled to match observed temperature trends with their models, partly because they could not test alternate histories with the most accurate tools. This paper shows how a seasonal forecast system—the same kind that predicts next winter's weather—can rerun decades as if CO₂ and aerosols never changed. That lets researchers isolate which trends humans actually caused and which the model gets wrong, pointing to where the model itself needs fixing.
The abstract states that the coupled model's Bjerknes feedback fails to sustain a strengthened tropical Pacific temperature gradient, suggesting a fundamental model limitation. Additionally, aerosol forcing has little impact likely because indirect aerosol effects are omitted. The study does not validate the counterfactual hindcasts against observed trends not mentioned in the abstract, nor does it assess whether the approach works for variables beyond temperature, radiative fluxes, and tropical Pacific circulation.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Sensitivity of the ECMWF seasonal forecast model to CO2 and anthropogenic aerosol forcings: Experimental design and impact on climate trends — Michael Mayer, Daniel J. Befort, Jacob Maddison et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.