Global assessment of partitioning transpiration from evapotranspiration based on satellite solar-induced chlorophyll fluorescence data

Abstract

• A model is developed for terrestrial transpiration estimation from spaceborne SIF. • Global mean terrestrial transpiration to evapotranspiration ratio from SIF is 0.57. • Global terrestrial transpiration in 2018 is estimated using our model. An accurate assessment of terrestrial ecosystem transpiration (T) is important to understand the vegetation-atmosphere feedbacks under climate change. Solar-induced chlorophyll fluorescence (SIF) shows great potential to estimate T because of its mechanical linkage with photosynthesis and stomatal conductance. However, a global and spatially estimation of terrestrial T based on remotely sensed SIF remains unresolved and novel strategies are challenged to entail a precise partition of T from evapotranspiration (ET) across various biomes. Here, with far-red SIF from Sentinel-5 Precursor satellite and ground observations for a total of 30 sites encompassing ten primary plant functional types (PFTs), we extend a SIF-driven semi-mechanism canopy conductance (g c ) model for different plant functional types (PFTs), and use the optimized Penman-Monteith model (PM opt ) to calculate T and T/ET. We reveal that the relationship between SIF and the product of g c and 0.5 power of vapor pressure deficit (g c × VPD 0.5 ) is tighter than the relationship between SIF and ecosystem productivity. The SIF-g c × VPD 0.5 linear regressions show improved R 2 and increased magnitude in slopes across PFTs when aggregating daily to 16-day. Our T/ET results show high correlations with the results of the Ball-Berry-Leuning model combined with PM opt at the site scale (R 2 = 0.69), and with the results calculated by leaf area index in a previous study at the PFT scale (0.70). We further determine the global mean T/ET (0.57 ± 0.14), close to the ensemble mean of global averaged T/ET (0.55), using 36 different methods. The global T estimated using the SIF-based approach is compared with two other remote sensing products. Our method provides a valuable tool for T and ET estimation using remote sensing data and is critical to understanding ecohydrological processes under climate change.