Characterization and evaluation of global solar-induced chlorophyll fluorescence products: estimation of gross primary productivity and phenology

As a proxy of vegetation photosynthesis, solar-induced chlorophyll fluorescence (SIF) contains rich photosynthetic information that can reveal the physiological state of vegetation and its response to the environment. Current publicly available SIF products vary in accuracy, spatiotemporal resolution, and coverage due to the different inversion algorithms and sensor characteristics. Although awareness of their performances is essential for researchers to select and use data rationally, no systematic comparative analyses of these products have been conducted. In this paper, 8 sets of widely used SIF products were systematically evaluated in terms of spatiotemporal agreement with gross primary productivity (GPP) against 3 GPP datasets (FLUXNET observations, FLUXCOM–GPP, and random forest–GPP) and the derived phenology metrics against the phenological observation data (Pan European Phenological database). Results showed that the GOSIF (757 nm) and CSIF datasets best encapsulate the spatiotemporal variability of global GPP and characterize the spatial distribution of GPP-derived phenology. The ability of SIF products to explain GPP variation changed according to ecosystem type. The ability was strong for deciduous broadleaf forests, mixed forests, and evergreen needleleaf forests, whereas it was poor for evergreen broadleaf forests. Regarding consistency with phenological observations, SIF products were substantial better at predicting the start of the growing season rather than the end or length of the growing season. The systematic evaluation of the widely used SIF products serves as a reference for subsequent studies and may also provide comprehensive information for further refinements and future development of the new SIF products.