Satellite-observed solar-induced chlorophyll fluorescence (SIF) provides a direct radiative link between solar radiation, photosystem de-excitation and vegetation photosynthetic activity. As multiple satellite missions now deliver global SIF products, machine learning (ML) has become a key tool for: (i) flexible nonlinear SIF retrieval, (ii) spatial reconstruction and downscaling of SIF fields, (iii) full-spectrum SIF reconstruction beyond narrow absorption windows, and (iv) data-driven analysis of the SIF–gross primary production (GPP) relationship. In addition, ML methods are increasingly used for: (v) uncertainty quantification (UQ) along the SIF information chain, and (vi) emulation (i.e., surrogate modelling) of radiative transfer models (RTMs) to accelerate computationally demanding SIF workflows. This review provides a conceptual and methodological survey of recent ML applications across the satellite SIF processing chain, summarises emerging products and methods, and highlights open challenges in uncertainty treatment, spectral reconstruction, and hybrid RTM–ML approaches. Particular emphasis is placed on the upcoming ESA FLEX mission, planned for launch in 2026, which will deliver multi-band SIF observations optimised for photosynthesis monitoring. While FLEX Level-2 (L2) operational processing will be based on physically grounded retrieval algorithms developed within ESA projects, ML is expected to play an important role in scientific exploitation and in the development of higher-level products (L3/L4), supporting high-resolution, uncertainty-aware SIF and GPP products and helping to bridge scales from leaf to ecosystem.

Sun-induced chlorophyll fluorescence (SIF) is a critical indicator of photosynthetic activity. Yet, existing satellite SIF products typically suffer from coarse spatial resolutions, generally coarser than 500 m, which limits their utility for fine-scale ecosystem studies. Here, we present a cloud-computing framework designed for the generation of a downscaled SIF product (S3-SIF₇₄₃) derived from Sentinel-3 (S3) Ocean and Land Colour Instrument (OLCI), with a spatiotemporal resolution of 300 m and 4 days. Our approach uses the Google Earth Engine (GEE) cloud-computing platform to integrate SIF produced from TROPOspheric Monitoring Instrument measurements within the 743–758 nm retrieval (TROPOSIF743), S3 radiances, S3-based vegetation traits, latitude and longitude within a Random Forest (RF) regression framework. Model training over Europe achieved robust performance against TROPOSIF743 reference data (R2 = 0.767, RMSE = 0.137 mW m⁻² sr−1nm−1), and the approach was subsequently extended globally. Validation against ground-based tower observations confirmed that S3-SIF₇₄₃ effectively reproduces seasonal dynamics across diverse ecosystems. Comparisons against TROPOSIF743 demonstrated strong spatial consistency in temperate agricultural regions, with the highest R2 values in croplands and lower agreement in sparsely vegetated or persistently cloudy regions. Global mapping revealed coherent patterns of photosynthetic activity, with peak values in tropical rainforests and major agricultural zones. Importantly, S3-SIF₇₄₃ reduces retrieval noise relative to TROPOSIF743 and provides unprecedented insights into sub-kilometer spatial heterogeneity. By combining S3’s rich spectral capabilities with GEE’s scalable computing environment, our approach bridges the gap between current coarse-resolution SIF products and ESA’s upcoming FLEX mission, offering a flexible and operational pathway for high-resolution monitoring of terrestrial photosynthesis.

Solar-induced chlorophyll fluorescence (SIF) provides a direct optical proxy of photosynthetic activity and has shown strong empirical relationships with gross primary productivity (GPP). However, the spatiotemporal extent to which satellite-derived SIF and vegetation traits encode meteorological constraints for prediction of GPP remains insufficiently understood. In this study, we analyze the relationships among SIF, vegetation structural variables, meteorological drivers, and GPP using global satellite and reanalysis datasets. Two global Gaussian process regression (GPR)-based GPP products were intercompared: (i) a hybrid model driven by TROPOMI-derived SIF (TROPOSIF), Sentinel-3-derived vegetation traits, and three ERA5-Land key meteorological variables; and (ii) a tower-driven empirical model based on additional meteorological predictors and MODIS-derived leaf area index (LAI). The results showed strong agreement between the two products (median $R=0.76$), suggesting that satellite-derived SIF and vegetation traits capture variability associated with meteorological forcing that is relevant for GPP prediction. Next, Pearson correlation worldwide and multivariate Granger causality over Europe were analysed in 2019 to investigate dependencies among meteorological variables, vegetation traits, and TROPOSIF. The results indicate that incoming shortwave radiation, temperature, soil moisture, latent heat flux, and leaf area index, in combination, significantly influence SIF dynamics over large regions. Together, the findings highlight the complementary role of satellite SIF and vegetation traits as proxy predictors of ecosystem productivity monitoring, and provide insights relevant for the recently launched FLEX mission.