Ghosts from summers past: quantifying the role of vegetation legacy to climatic extremes

在过去的二十年里，欧洲各地在夏季经历了创纪录的干旱和高温，对生物多样性、公共卫生和土地吸收的碳量产生了广泛的影响。预计气候变化将增加未来干旱和极端高温的频率、规模和持续时间。然而，受影响的生态系统从极端事件中恢复所需的相关时间尺度在很大程度上是未知的，在评估未来气候变化影响时很少考虑。如果没有这种关键的恢复时间尺度的知识，我们可能会低估生态系统和相关服务对复合事件（热浪与干旱同时发生）和/或反复出现的极端气候的敏感性。为了有力地预测未来的影响，我们需要新的理论来代表我们对极端气象的生态遗产持续存在的时间尺度（月到年）的理解。该项目将通过将最先进的机器学习技术应用于操纵实验（降雨/变暖）、现场数据和覆盖最近破纪录的欧洲夏季极端天气的卫星数据，来确定多个空间尺度的遗留时间尺度。这些见解的规模（时间和空间范围）的遗产持久性将被用来测试四个假设，管理植被的干旱和极端高温的反应。这种统计机器学习与假设驱动的模型开发的结合，将揭示过去在决定植被对未来极端环境的反应、降低相关风险和促进减灾规划方面的重要作用。我们的最终影响将是修改UKESM的陆地表面部分-英国新的地球系统模型-促进我们评估欧洲和全球极端气候未来影响的能力的变革性提高，这对预测未来陆地碳汇至关重要。

The last two decades have witnessed record-breaking drought and heat extremes during summer months across Europe, with wide-ranging consequences for biodiversity, public health, and the amount of carbon taken up by the land. Climate change is expected to increase the frequency, magnitude, and duration of future droughts and heat extremes. Yet, the associated timescale needed for affected ecosystems to recover from extreme events is largely unknown and rarely considered in evaluations of future climate change impacts. Without this critical knowledge of recovery timescales, we are likely to underestimate ecosystem and associated services sensitivity to compound events (heatwaves coincident with drought), and/or repeated climate extremes. To robustly predict future impacts, we need new theories to represent our understanding of the timescales (months-to-years) over which the ecological legacy to meteorological extremes persists. This project will determine legacy timescales at multiple spatial scales by applying state-of-the-art machine learning techniques to both manipulation experiments (rainfall/warming), field data, and satellite data covering recent, record-breaking European summer extremes. These insights into the scale (both time and spatial extent) of legacy persistence will be used to test four hypotheses that govern the vegetation's response to droughts and heat extremes. This combination of statistical machine learning, alongside hypothesis-driven model development, will unlock critical new insights into the role of the past in dictating vegetation responses to future environmental extremes, reducing associated risk, and facilitating mitigation planning. Our ultimate impact will be the modification of the land surface component of UKESM - the UK's new Earth System Model - facilitating a transformative improvement in our capacity to assess the future impact of climate extremes both across Europe and globally, critical to forecasting the future terrestrial carbon sink.