Extreme storms are becoming more intense and frequent under climate change. Although these extreme wet events are smaller in extent and duration than drought events, recent evidence suggests the global impact of both extremes is similar. However, the impact of individual extreme storms on photosynthesis-and therefore on vegetation and the carbon cycle-remains difficult to predict, as photosynthesis may be suppressed via waterlogging or increased by the alleviation of moisture stress. Here, we use random forest models to calculate daily photosynthesis anomalies attributable to extreme soil moisture using data from 54 FLUXNET sites across the globe. We hypothesize that photosynthesis' response to a given extreme event is primarily controlled by storm intensity, and to a lesser degree by site vegetation, climate, soil, and topography. However, we find instead that photosynthesis responses are better explained by site characteristics (soil texture, climate, topography, and vegetation density) than by storm intensity, such that the likelihood of waterlogging from a given storm is heavily site-dependent. Although storms that induce waterlogging are roughly as common as those that induce stress alleviation overall, photosynthesis rarely declines at sites not prone to waterlogging. Instead, photosynthesis anomalies at these sites show a much weaker relationship with storm intensity. Increasingly intense storms are therefore unlikely to impact all locations equally. This highlights the potential to use site characteristics to enhance prediction of storm effects on ecosystems and the land carbon sink.

Keywords: carbon uptake; climate extremes; ecohydrology; eddy covariance; extreme storms; machine learning; photosynthesis; waterlogging.

© 2025 The Author(s). Global Change Biology published by John Wiley & Sons Ltd.

极端气候条件下，风暴变得更为剧烈和频繁。尽管这些极端潮湿事件在范围和持续时间上不如干旱事件大，但最近的证据表明两种极端情况对全球的影响相似。然而，个别极端风暴对光合作用的影响——从而影响植被和碳循环——仍然难以预测，因为光合作用可能因水淹而受到抑制，也可能因水分胁迫缓解而增加。在这里，我们使用随机森林模型计算了54个全球FLUXNET站点每天由于极端土壤湿度变化而导致的光合作异常值。我们假设光合作用对特定极端事件的反应主要受风暴强度控制，并在较小程度上由站点植被、气候、土壤和地形决定。然而，我们的研究发现，光合作用响应更多地取决于站点特征（如土壤质地、气候、地形和植被密度），而不是风暴强度，也就是说，某一地点发生水淹的可能性高度依赖于该地点的特性。尽管引发水淹的风暴与缓解水分胁迫的风暴在总体上差不多常见，但在不常发生水淹的地点，光合作用很少会下降。相反，在这些站点，光合作异常值与风暴强度的关系较弱得多。因此，越来越强烈的风暴不太可能对所有地区产生相同的影响。这突显了利用站点特征来提高对未来风暴影响生态系统和陆地碳汇预测的可能性。

关键词：碳吸收；气候变化极端事件；生态水文学；涡度相关技术；极端风暴；机器学习；光合作用；水淹。

版权所有 © 2025 The Author(s). Global Change Biology由John Wiley & Sons Ltd出版。