Beyond the multimodel ensemble mean (MME): assessing plausible climate risk at a given warming level

Climate risks at a given warming level are still commonly interpreted through the multimodel ensemble mean (MME), yet the MME does not fully represent the range of plausible future outcomes that matter for risk assessment. Recent work shows that even in a 2 °C warmer world, some plausible climate outcomes can produce sectoral hazards—including drought, extreme precipitation, and fire-weather or wildfire-related hazards—that are more severe than the MME response at 3 °C or even 4 °C. This finding challenges the widespread assumption that moderate warming necessarily implies moderate risk. Here we argue that climate-risk assessment should move beyond reliance on the MME alone and pay greater attention to physically plausible high-impact outcomes within the ensemble spread. We emphasize that both model performance and internal climate variability are important for understanding such risks, although their relative roles may differ across sectors and timescales. Future assessments should therefore combine average forced responses with process-based evaluation of extreme outcomes to better inform adaptation planning and climate-risk management.