Radiative forcing is a fundamental quantity for understanding anthropogenic and natural drivers of past and future climate change¹, yet significant uncertainty remains in our quantification of radiative forcing and its model representation^2,3,4. Here we use instrumental measurements of historical global mean surface temperature change and Earth’s total heat uptake, alongside estimates of the Earth’s radiative response, to provide a top-down energy budget constraint on historical (1861–1880 to near-present) effective radiative forcing of 2.3 W m⁻² (1.7–3.0W m⁻²; 5–95% confidence interval). This represents a near 40% reduction in the 5–95% uncertainty range assessed by the IPCC Fifth Assessment Report². Although precise estimates of effective radiative forcing in models do not widely exist, our results suggest that the effective radiative forcing may be too small in as many as one-third of climate models in the fifth phase of the Coupled Model Intercomparison Project. Improving model representation of radiative forcing should be a priority for modelling centres. This will reduce uncertainties in climate projections that have persisted for decades^4,5.

辐射强迫是了解过去和未来气候变化^{1 的}人为和自然驱动因素的基本量，但我们对辐射强迫的量化及其模型表示^2,3,4仍然存在很大的不确定性。在这里，我们使用对历史全球平均地表温度变化和地球总热量吸收的仪器测量，以及对地球辐射响应的估计，为历史（1861-1880 年至今）有效辐射强迫提供自上而下的能量预算约束。 2.3 W m ^-2（1.7–3.0W m ^-2；5–95% 置信区间）。这意味着 IPCC 第五次评估报告²评估的 5-95% 不确定性范围减少了近 40%. 尽管模型中有效辐射强迫的精确估计并不广泛存在，但我们的结果表明，在耦合模型比对项目第五阶段的多达三分之一的气候模型中，有效辐射强迫可能太小了。改进辐射强迫的模型表示应该是模型中心的优先事项。这将减少持续数十年的气候预测的不确定性^4,5。