Groundwater-sustained baseflow is a vital source of river flow, especially during dry seasons. The proportion of river flow sustained by baseflow—the baseflow index—is essential for assessing fluvial nutrient cycling and contaminant transport. However, the global baseflow index remains highly uncertain, with current Earth system model simulations ranging from 12% to 94%. Here we estimate the global baseflow index to be 59% ± 7% based on an emergent constraint approach, which integrates 50 Earth system models with baseflow indices derived from streamflow observations in 15,567 basins. Our observational constraint indicates that at least 21% ± 3% of precipitation recharges groundwater, which is approximately double the figure reported in the Sixth Assessment Report of the United Nations Intergovernmental Panel on Climate Change. Thus, our research suggests a more active role of groundwater in the global water cycle than most Earth system models currently simulate. We present evidence that the considerable disagreement in simulated baseflow stems from unrealistic and varied model representations of infiltration, aquifer structure and groundwater dynamics. These processes should be prioritized so that models can capture active groundwater–river connections.

地下水维持的基流是河流流量的重要来源，特别是在旱季。由基流维持的河流流量比例（基流指数）对于评估河流养分循环和污染物输送至关重要。然而，全球基流指数仍然高度不确定，当前地球系统模型模拟范围为 12% 至 94%。在这里，我们基于紧急约束方法估计全球基流指数为 59%±7%，该方法将 50 个地球系统模型与来自 15,567 个流域的水流观测得出的基流指数相结合。我们的观测限制表明，至少有21%±3%的降水补给了地下水，这大约是联合国政府间气候变化专门委员会第六次评估报告中报告数字的两倍。因此，我们的研究表明，地下水在全球水循环中的作用比目前大多数地球系统模型模拟的更为积极。我们提供的证据表明，模拟基流中的相当大的分歧源于渗透、含水层结构和地下水动力学的不切实际且不同的模型表示。应优先考虑这些过程，以便模型能够捕获活跃的地下水与河流的连接。