Gross primary productivity (GPP), the carbon fixed by plants through photosynthesis, is fundamental to ecosystem function and the global carbon cycle. Satellite-observed sun-induced chlorophyll fluorescence (SIF) is used as a proxy for GPP because fluorescence is linked to photosynthetic energy allocation. However, the SIF-GPP relationship varies among vegetation types and environmental conditions, while satellites detect only fluorescence escaping the canopy.
Leaf arrangement, canopy density, viewing geometry, and radiative transfer can obscure physiological information. Forest height influences light interception, yet its role in SIF interpretation remains insufficiently quantified. Based on these challenges, in-depth research is needed into how forest height regulates the SIF-GPP relationship.
Researchers from Nanjing University, Fujian Normal University, the University of Toronto, and China University of Geosciences published (DOI: 10.34133/remotesensing.1061) the study on August 19, 2026, in Journal of Remote Sensing. The work addresses a persistent satellite remote-sensing challenge: converting SIF observations into reliable estimates of ecosystem photosynthesis despite major differences in canopy architecture. Combining global forest-height observations with SIF angular normalization and a two-leaf model, the team tested whether canopy height explains systematic variation between fluorescence signals and carbon uptake across forests.
The study identifies two mechanisms by which forest height reshapes the SIF-GPP relationship. Taller canopies reduce the probability that fluorescence photons escape to space because longer pathways increase scattering and absorption. Across the analyzed forest types, escape probability consistently declined as height increased. Taller forests also contained proportionally fewer sunlit leaves and more shaded leaves, with both trends significant at P < 0.001. Because sunlit leaves dominate SIF emission while shaded leaves contribute strongly to GPP, this structural shift raises the GPP-to-SIF ratio, providing a clear physical explanation for height-dependent biases in satellite SIF observations at the global scale overall.
The GPP-to-observed-SIF ratio, K, increased consistently with forest height across different forest types. After correcting SIF for escape probability, the height dependence of Kt = GPP/SIFtotal became weaker, showing that photon escape is a major structural driver. The two-leaf analysis revealed another asymmetry: sunlit leaves produced stronger fluorescence, whereas shaded leaves generated higher average GPP and usually larger GPP-to-SIF ratios. Sunlit leaf area was generally constrained to values of 2 or less, while shaded leaf area continued increasing as total leaf area rose in taller, denser forests. The correction also changed regional interpretation. In the Amazon, raw satellite SIF was markedly lower than in Southeast Asia or eastern North America, but corrected total canopy SIF became comparable with Southeast Asia and closer to eastern North America. This indicates that structural correction can recover physiologically meaningful information otherwise suppressed by complex canopies across highly diverse canopy structures at the global scale.
Highlighting the practical implication, the authors write that "future remote sensing correction frameworks for SIF should incorporate forest height as a structural factor." They argue that canopy-height information can refine SIF-GPP models and improve satellite-based estimates of global photosynthetic productivity. Future studies, they add, should test the framework at finer spatial scales with higher-accuracy observations.
The researchers combined a global 30-m forest-height product derived from GEDI observations with TROPOMI near-infrared SIF data for 2020, aggregated to 0.2° resolution. They applied a geometric-optics-based angular-normalization model that separates sunlit and shaded leaf fluorescence and estimates SIF escape probability. A two-leaf light-use-efficiency GPP product supplied corresponding productivity components. MODIS leaf area index and clumping index data, plus ESA-CCI land-cover information, supported canopy characterization. SIF observations with cloud cover of 0.2 or greater were excluded to reduce clear-sky bias.
Adding forest height to satellite photosynthesis models could improve global GPP estimates, particularly in tall, dense forests where canopy structure strongly suppresses observed fluorescence. Better GPP monitoring would strengthen assessments of ecosystem carbon uptake, forest productivity, and climate-carbon feedbacks. Future work can combine finer canopy-height maps with next-generation SIF observations and higher-resolution ecological datasets. Such integration could improve regional and global carbon accounting while helping researchers distinguish structural changes in vegetation from physiological responses to climate change, disturbance, or forest management.