Official projections for global warming may be underestimating reality by a quarter. A new methodological approach developed at ETH Zurich indicates that, for a given emission trajectory, the planet will warm approximately 25 percent more than predicted by currently trusted models. This adjustment is not marginal: it transforms an already critical trajectory into a more severe scenario, with direct implications for climate planning and adaptation policies.
The natural variability bias
The key lies in how climate models are validated. Traditionally, simulations of the recent past are compared with observed data; if a model predicts more warming than actually occurred, it is discarded as unrealistic. However, this methodology ignores the natural variability of Earth's climate, which can temporarily mask or amplify the anthropogenic signal. Gergana Gyuleva, a researcher at ETH Zurich, explains that correcting for this bias reveals a different reality: models previously discarded for their high CO2 sensitivity may, in fact, be the most accurate.
From 2.6 to 3.25 degrees Celsius
The consequences of this re-evaluation are numerical and concrete. The Climate Action Tracker, which monitors the effectiveness of current climate policies, projects a global average temperature increase of 2.6°C by 2100 if countries maintain their existing strategies. Applying the correction proposed by Gyuleva's team, this estimate rises to 3.25°C. For the Intergovernmental Panel on Climate Change (IPCC), the last assessment estimated warming between 1.2 and 2.4°C for a doubling of CO2, with a central value of 1.8°C. The new analysis suggests that the upper end of this range, or even beyond, is more likely.
Implications for climate science
Climate sensitivity—how much the planet warms in response to a given CO2 level—remains one of the largest uncertainties in climate science. Some models predict an increase of about 1.6°C for a doubling of CO2, others up to 3°C. As Gyuleva notes, "the range is wide and it really matters": identifying which models are correct is not an academic exercise but a necessity for planning. The new evaluation method offers a more robust tool to distinguish the climate signal from the noise of natural variability, potentially redefining the basis on which long-term projections are built.
This methodological update arrives in an already tense context, where extreme events like European heatwaves and ocean records are testing the resilience of natural and human systems. If the more pessimistic estimates are confirmed, the window to limit damage to manageable levels narrows further.
No comments yet. Be the first!