Drivers of vegetation activity during European summers: A causal inference approach applied to solar-induced fluorescence observations

Abstract

<p> Summer weather in Europe has become more extreme in recent years. Several studies have focused on unraveling the influence that this extreme weather may have on ecosystem dynamics. However, traditional optical indices characterise the state of vegetation in terms of greenness or structure, but fail to capture short term impacts on vegetation activity caused by water or heat stress. Being a byproduct of photosynthesis, solar induced fluorescence (SIF) represents an exception, since its dynamics may reflect an integral of the environmental stressors that have immediate influence on ecosystem water, energy and carbon exchanges during droughts or heatwaves. Spaceborne datasets of SIF have not only been used to monitor crop photosynthetic activity and GPP at global scales, but also as a proxy of transpiration dynamics, or even biogenic volatile organic compound emissions. Additionally, numerous case studies have indicated the potential of using SIF for early drought detection and monitoring of ecosystem impacts. </p><p>However, as with most earth science applications, the majority of previous studies rely on correlations or linear regressions to establish these cause–effect relationships, which implies that the actual drivers of drought and periods of vegetation stress remain largely unresolved.</p><p>Here we examine the underlying causality and interactions between vegetation activity (represented by changes in SIF) and potential environmental drivers of vegetation stress over Europe during the summer months. Using satellite observations of  photosynthetically active radiation (PAR) and the fraction of absorbed PAR (fPAR), the SIF signal is decomposed into the component that relates to fPAR and the component that relates to the fluorescence yield, which represent different physical and biochemical responses to vegetation stress. Using recently developed methods for causal inference applications in Earth science (https://causeme.uv.es/), the dynamics of SIF and its deconstructed components are evaluated against satellite observations of soil moisture, vapor pressure deficit and temperatures. Common causal relationships and dynamics are observed when grouping regions by aridity index and fractions of vegetation cover. Results help establish direct and indirect links of potential drivers of vegetation activity during periods of heat and water stress.</p>