Increases in extreme rainfall intensities as a result of climate change pose a great risk due to the possibility of increases in pluvial flooding, particularly in urban and developed areas. But evidence is emerging that the observed increases in extreme rainfall are not resulting in universal increases in flooding. Indeed, on a global scale, studies consistently find more gauge stations with decreasing rather than increasing trends in the annual maxima flood magnitude. Here, we aim to improve our understanding of how rainfall and streamflow extremes are changing and why floods are not always observed to increase despite increases in rainfall extremes. To do so, we examine trends in streamflow events using 2776 stations from the Global Runoff Data Centre, with events chosen to isolate the impact of changes to their respective rainfall and antecedent soil moisture. The analysis is limited to stations with 30 years or more of active record with the majority of stations in North America, Europe, Brazil, Oceania, and southern Africa.

Consistent with physical reasoning, for more frequent events such as the annual maxima, it is found the peak 1-day rainfall event intensity is increasing approximately a rate of 6–7%/°C, with rarer event intensities increasing at a rate exceeding the Clausius-Clapeyron relation. We find that storm volumes are not increasing as greatly as the peak rainfall and storm durations are decreasing, pointing to an intensification of rainfall events and a peakier temporal pattern. While rainfall is intensifying, the magnitude of frequent floods (those expected to occur on average once per year) are in general decreasing, particularly in tropical and arid regions of the world. We find that this is likely due to a dominance of drying antecedent soil moisture conditions. However, the magnitude of rarer floods has, on average been increasing. We suggest this is because, for these rarer events, the increase in rainfall outweighs the decrease in soil moisture. Our results point to a worst of both world’s scenario where small floods, responsible for filling our water supplies, are decreasing, while the large flood events which pose a risk to life and infrastructure, are increasing.

由于洪水泛滥的可能性增加，特别是在城市和发达地区，气候变化导致极端降雨强度的增加构成了巨大的风险。但有证据表明，观察到的极端降雨量增加并未导致洪水普遍增加。事实上，在全球范围内，研究一致发现更多的监测站的年度最大洪水量级呈下降趋势而不是上升趋势。在这里，我们的目标是提高我们对降雨和水流极端情况如何变化的理解，以及为什么尽管极端降雨量增加，但并不总是观察到洪水增加。为此，我们使用来自全球径流数据中心的 2776 个站点来检查水流事件的趋势，选择事件以隔离变化对各自降雨量和前期土壤湿度的影响。分析仅限于具有 30 年或更长时间活跃记录的台站，其中大部分台站位于北美、欧洲、巴西、大洋洲和南部非洲。

与物理推理一致，对于更频繁的事件，例如年度最大值，发现峰值 1 天降雨事件强度以大约 6-7%/°C 的速度增加，稀有事件强度增加的速度超过克劳修斯-克拉佩龙关系。我们发现风暴量的增加并不像峰值降雨量和风暴持续时间的减少那样大，这表明降雨事件的加剧和峰值时间模式。虽然降雨量正在加强，但频繁发生的洪水（预计平均每年发生一次）的程度总体上正在下降，特别是在世界热带和干旱地区。我们发现这可能是由于干燥的先行土壤水分条件占主导地位。然而，较罕见的洪水的规模平均一直在增加。我们认为这是因为，对于这些罕见的事件，降雨量的增加超过了土壤水分的减少。我们的结果指出了世界上最糟糕的情况，负责填补我们供水的小洪水正在减少，而对生命和基础设施构成风险的大洪水事件正在增加。