Global vegetation responses to wet and dry soil moisture extremes

Hydrological extremes are continuing to intensify under climate change. However, the responses of vegetation to dry and wet soil moisture extremes, and the dominant drivers of these responses, have not yet been analyzed consistently. In this study, we utilize long-term observations of Normalized Difference Vegetation Index (NDVI) as a proxy of vegetation responses to soil moisture extremes. We then analyze related drivers with a machine-learning attribution approach to assess the role of pre-extreme vegetation conditions, characteristics of extremes, and of the environmental background. Vegetation generally loses greenness during dry extremes, indicated by widespread and consistent negative NDVI anomalies. This is mainly modulated by pre-extreme vegetation conditions and the characteristics of the extreme (especially seasonal timing) which reflect varying vegetation vulnerability. In contrast, wet extremes lead to more heterogeneous responses, including both positive and negative NDVI anomalies. This is modulated by multiple aspects including pre-extreme vegetation conditions, the characteristics of the extreme (especially seasonal timing) as well as environmental background variables such as climate (e.g., long-term mean air temperature, aridity) and topography (topographic variability). This illustrates that vegetation response to wet extremes is complex and potentially influenced by different processes. Further, regions with negative NDVI anomalies during extremes that are strongly modulated by environmental background indicate localized vulnerability arising from adverse climatic, soil or topographic conditions, such that vegetation stress can occur even under extremes with less severity. These results highlight the roles of seasonal timing and of environmental background conditions for impacts of soil moisture extremes on vegetation. This clarifies the predictability of ecosystem responses to hydrological extremes, and serves as a basis for related management planning.