Estimation of global transpiration from remotely sensed solar-induced chlorophyll fluorescence

Accurate estimation of ecosystem transpiration (T) is critical to understanding of global land-atmosphere water, energy and carbon fluxes. However, the complexity of processes governing canopy conductance to transpired water vapor (Gc) poses a substantial challenge for modelling T. Canopy conductance governs the uptake of CO2 for photosynthesis and the release of water vapor through transpiration, and so satellite observations of solar-induced chlorophyll fluorescence (SIF), a new proxy for plant photosynthesis, may create a good opportunity to estimate T at large scales. In this study, a new SIF-based model is proposed to estimate T at the canopy scale: photosynthesis rate, a key input for the coupled photosynthesis–stomatal conductance framework, is mechanistically approximated from top-of-canopy (TOC) SIF and readily-available meteorological data. One major improvement makes our work conceptually novel, compared to previous T modelling efforts: SIF is mechanistically translated into photosynthetic CO2 assimilation such that the regulation of Gc and eventually T can be realized in a more physiologically realistic way. We evaluate the modeled T forced by satellite SIF observations and globally gridded meteorological information at 31 eddy covariance flux sites covering ten vegetation types in three continents. The model performs well in various vegetation types, particularly in ecosystems with dense canopies, explaining nearly 76% of the variability in their daily flux-derived T. We apply the model to obtain the spatial and temporal distributions of global T at a daily time step for the period 2019–2020, and diagnose the response of T to vapor pressure deficit (VPD) and soil water content (SWC). At the global scale, increasing VPD exhibits a negative correlation with Gc but a positive correlation with T, and their correlations decrease with decreasing SWC. In contrast, a decrease in SWC is not associated with a clear reduction in Gc and T as long as the SWC is not low. Our proposed model better unleashes the potential of SIF in modelling T, and thus opens a new pathway to better understand the mechanisms impacting the coupling of carbon and water cycles.