Conference Agenda

Estimating gross primary productivity (GPP) from top-of-canopy (TOC) optical measurements remains a major challenge due to the complex interactions among canopy structure, physiological regulation of plant photosynthesis across diurnal and seasonal timescales, and the influence of observation scale. These factors complicate the interpretation of vegetation spectral signals and limit the understanding of remotely sensed reflectance and fluorescence into reliable and direct indicators of photosynthetic function.

This study presents a strategy for transferring spectral modelling approaches previously developed under controlled laboratory conditions to field-scale applications. These methods are designed to capture subtle spectral features associated with non-photochemical quenching (NPQ) processes, which play a central role in the regulation of photosynthesis. The first approach is based on spectral fitting techniques using least-squares regression, combined with targeted refinements of spectral inputs, to quantify the contribution of photoprotective pigments at the leaf level. The second approach employs partial least squares regression (PLSR) applied to hyperspectral TOC laboratory measurements to identify spectral features and wavelength regions correlated with photoprotective mechanisms. To enable application at the ecosystem scale, we investigate short- and long-term spectral variability in relation to GPP and NPQ dynamics, as well as the representativeness of single-point FLOX measurements.

Diurnal variations in canopy reflectance are analysed in relation to concurrent GPP dynamics, with the aim of disentangling short-term physiological responses from longer-term trends. To support this separation, is investigated the influence of canopy structure on the bidirectional reflectance distribution function (BRDF) in TOC FLOX measurements. Spectral analyses employ short temporal-window PLSR models to capture early photosynthetic adjustments associated with subtle pigment dynamics, including rapid physiological responses during recovery phases that temporarily alleviate stress conditions. These short-term approaches are contrasted with seasonally aggregated models that integrate both short- and long-term spectral drivers, resulting in more complex representations of cumulative physiological regulation and structural canopy changes over time. Variable importance in projection (VIP) scores spectra from PLSR models are used to compare dominant spectral contributors across temporal scales and environmental conditions, highlighting the relative importance of specific wavelength regions and sensitivities associated with spectral co-variation with NPQ, and providing insight into how different physiological mechanisms are expressed in the canopy reflectance signal.

Spatial representativeness of point-based FLOX observations is further evaluated by comparing single-tree canopy reflectance dynamics with ecosystem-scale measurements. Further, reflectance relationships are examined between Sentinel-2 reflectance at the pixel scale and ecosystem scale from eddy-covariance tower GPP footprint, as well as between MODIS vegetation products and FLOX-derived vegetation indices, in order to assess cross-scale consistency.

Building on previous findings obtained under laboratory conditions that link changes in photosynthetic efficiency to multi-peak variations in green (500-600 nm) and red-edge (680-750 nm) canopy reflectance, this work further investigates how co-varying spectral behaviour relates to light-use efficiency and NPQ at the canopy level. Although direct temporal correlation between fluorescence signals and traditional vegetation indices is not consistently observed, preliminary modelling results indicate meaningful relationships between spectral dynamics and GPP on the short-term scale. Overall, this study aims to start the development of a GPP modelling framework based on spectral unmixing and the exploitation of co-varying spectral drivers, with potential implications for improving optical monitoring of ecosystem productivity under variable environmental conditions.

Most solar-induced fluorescence models treat Photosystem II in isolation, yet Photosystem I contributes substantially to both fluorescence emission and photosynthetic regulation. This gap limits our mechanistic understanding of the fluorescence-photosynthesis relationship, particularly at ecosystem scales where observational constraints on Photosystem I dynamics remain absent. The Johnson-Berry model addresses this gap by resolving excitation balance, cyclic electron flow, Photosystem I fluorescence, and dynamic absorption cross-section; however, it lacks ecosystem-scale validation with in-situ measurements. Here, we replaced the photosynthesis module in the Breathing Earth System Simulator with the Johnson-Berry model and incorporated a fluorescence radiative transfer module, creating a PSI-explicit framework that simulates both gross primary production and solar-induced fluorescence at the canopy scale. We evaluated this framework at a rice paddy site over 3 months. Our validation targeted both the core Photosystem I-related mechanisms and the accuracy of simulated gross primary production and solar-induced fluorescence. The framework successfully simulated both carbon and fluorescence dynamics, explaining 94% of daily gross primary production variability and 74% of daily solar-induced fluorescence variability, with Photosystem I contributing nearly half of the total fluorescence emission at 760 nm. The simulated quantum yield partitioning among photochemistry, fluorescence, and heat dissipation agreed with field measurements from pulse-amplitude modulated fluorometry. Field observations validated two PSI-explicit mechanisms in the framework: the fractional changes in P680 and P700 states maintained proportional balance, with observed maximum cytochrome b6f turnover rate reaching 210 mol e⁻ mol⁻¹ s⁻¹, supporting the excitation balance assumption; and electron transport measurements revealed cyclic electron flow activation beyond approximately 100 μmol m⁻² s⁻¹. We found that the dynamic absorption cross-section mechanism substantially influenced both non-photochemical quenching and fluorescence simulations: enabling this mechanism produced near-zero Photosystem II non-photochemical quenching, while disabling it led to fluorescence underestimation. Additionally, we found the coupling between Photosystem II reaction center openness and stomatal conductance held only under low vapor pressure deficit. These findings validate key mechanisms in the PSI-explicit framework under field conditions and demonstrate that the framework accurately simulates ecosystem-scale carbon and fluorescence dynamics, providing a mechanistic basis for understanding the relationship between solar-induced fluorescence and gross primary production.

Sun-induced chlorophyll fluorescence (SIF) has been widely used as a proxy for gross primary productivity (GPP) based on their strong linear relationship at large spatiotemporal scales. However, whether this correlation reflects direct physiological coupling or shared dependence on light absorption remains unclear, limiting confidence in SIF-based GPP estimation under environmental stress. Here, we examine the origin of the SIF-GPP correlation by separating contributions of absorbed photosynthetically active radiation by chlorophyll (APAR) as a shared driver, quantum yields of fluorescence (ΦF) and photochemistry (ΦP) regulated through nonphotochemical quenching, and structural versus biochemical components represented by escape fraction (fesc) and electron use efficiency (EUE). For this, we used multi-year, half-hourly observations of canopy-scale SIF and eddy covariance fluxes from five different ecosystem types in Korea including deciduous broadleaf forest, evergreen needleleaf forest, cropland, wetland, and mixed forest. We found that the positive correlation between SIF and GPP became negative after controlling for APAR, and this pattern was consistent across all five ecosystem types. The initially negative relationship exhibited distinct patterns that varied across ecosystem types under different environmental conditions, reflecting trade-offs in energy partitioning between fluorescence emission and carbon assimilation. We also found that the relationship between ΦF and ΦP with canopy conductance differed across ecosystem types. In forest ecosystems, ΦP decreased with declining canopy conductance while ΦF showed little response or even increased, suggesting that stomatal closure primarily limits photosynthetic efficiency rather than fluorescence emission. In contrast, cropland and wetland showed similar responses in both ΦF and ΦP to conductance changes. Our findings indicate that the apparent SIF-GPP relationship arises predominantly from shared APAR absorption rather than from physiological coupling. While this pattern holds consistently across contrasting ecosystem types, the site-specific variation in conductance effects suggests that universal SIF-GPP relationships may not adequately capture carbon uptake across diverse ecosystems, particularly when stomatal regulation limits photosynthesis.

Sun-Induced chlorophyll Fluorescence (SIF) is the most promising remote-sensing proxy of Gross Primary Production (GPP) in terrestrial ecosystems. However, the estimation of GPP using SIF is challenging when plants experience stress, particularly during extreme climatic events whose frequency is projected to increase in the future. Recently, the feasibility of canopy-level active chlorophyll fluorescence measurements (LED-Induced chlorophyll Fluorescence (LIF)), which directly measure the apparent fluorescence yield (F_yieldLIF), has provided new perspectives on detecting the responses of plants to abiotic stress. This study was conducted during the summer 2022 European heat waves in a mixed temperate deciduous broadleaf forest, located in the French Fontainebleau-Barbeau station. Continuous measurements of carbon dioxide (CO₂) and energy exchanges, SIF, F_yieldLIF, and ancillary environmental variables were acquired. We investigated how high atmospheric dryness induced by heat-waves, measured as Vapor Pressure Deficit (VPD), affected canopy chlorophyll fluorescence (both SIF and F_yieldLIF) and GPP, as well as their relationships. At the half-hourly scale, our results revealed a decrease of the correlation between SIF and GPP (R² decreased from 0.49 to 0.17) at high atmospheric dryness. In contrast, the correlation between F_yieldLIF and GPP increased significantly under high atmospheric dryness (R² increased from 0.07 to 0.43). However, at the daily scale, the correlations between SIF and GPP and between F_yieldLIF and GPP showed an overall increase compared to the half-hourly scale, suggesting a time-scale-dependent response of these relationships to atmospheric dryness. Our tiered analysis further demonstrated that F_yieldLIF provides a significantly more robust proxy for the maximum photosynthetic rate (Amax) than SIF normalized by Photosynthetically Active Radiation (PAR), SIFy, under atmospheric stress. Specifically, under clear sky and high VPD conditions, the R² between Amax and F_yieldLIF reached 0.85, a substantial improvement compared to the R² of 0.56 observed for Amax and SIFy. This enhanced relationship is attributed to the advantage of F_yieldLIF, which, due to its stable excitation source and fixed geometry, is directly proportional to the true chlorophyll fluorescence quantum yield (), thereby directly capturing physiological regulation such as Non-Photochemical Quenching (NPQ). In contrast, SIFy remains confounded by structural and radiative transfer components, including the fraction of emitted SIF that escapes from the canopy (f_esc) and fraction of absorbed PAR by chlorophyll (f_APARchl), whose variability weakens its correlation with Amax. This study highlighted F_yieldLIF's advantage in detecting plant responses to high atmospheric dryness. We concluded that integrating canopy-level active fluorescence (like F_yieldLIF) with passive SIF measurements is essential for the accurate mechanistic interpretation and physiological validation of SIF signals, especially under future, more frequent extreme climate events.

Recent advances in passive remote sensing of solar-induced chlorophyll fluorescence (SIF) have spurred the development of SIF modules within terrestrial biosphere models (TBMs), creating a new generation of TBMs that explicitly simulate fluorescence emissions. The integration is motivated by the mechanistic link between fluorescence and photosynthesis, enabling SIF observations from tower, airborne, and satellite platforms to directly constrain photosynthetic carbon assimilation—quantified as gross primary productivity (GPP) at the ecosystem scale—and related ecosystem processes. However, substantial discrepancies in SIF simulations remain across models. The SIF Model Intercomparison Project (SIF-MIP) was designed to systematically quantify these differences by comparing tower-based SIF observations with simulations from six TBMs (BEPS, CLIMA, CLM, JULES, ORCHIDEE, and TECs) across diurnal, seasonal, and interannual timescales. Large inter-model variability was identified, with coefficients of variation (CVs) of 0.360–0.684 for absorbed photosynthetically active radiation (APAR), 0.559–1.099 for GPP, and 0.603–0.891 for SIF. Significant inconsistencies also emerged in key ecological relationships, including the GPP–SIF relationship, the coupling between GPP yield (GPP/APAR) and SIF yield (SIF/APAR), and the link between fluorescence efficiency (ϕF) and photochemical efficiency (ϕP). Assimilating satellite-based SIF into TBMs substantially improved model performance, reducing root mean square errors (RMSEs) for GPP by 30–50% and for SIF by more than 50%. These findings highlight the urgent need to improve TBM-SIF parameterizations from the leaf to canopy scale, and to better exploit satellite SIF as a model constraint—both of which are critical for improving projections of terrestrial carbon dynamics under ongoing climate change.

Investigations into the role of solar-induced fluorescence (SIF) as an indicator of photosynthesis and vegetation stress across multiple spatiotemporal scales have gained momentum in recent years, particularly in preparation for the upcoming FLEX mission. In this context, the exploitation of complementary information from the Sentinel-3 (S3) mission—flying in tandem with FLEX—is crucial for the correct interpretation of SIF dynamics. Here, we investigate the synergy between S3-derived vegetation products (S3VPs) and SIF derived from TROPOMI (TROPOSIF) as a baseline framework for the future exploitation of FLEX observations. Our objective is to assess the global capacity of SIF, in combination with S3VPs, to predict photosynthetic activity as quantified by gross primary productivity (GPP), and to explore the extent to which environmentally driven GPP dynamics are embedded within SIF and S3VPs.

We first analyze the dominant environmental drivers of GPP dynamics across major global biomes. Second, we quantify Pearson correlations (i) between an eddy-covariance-based GPP product and both TROPOSIF and S3VPs, to assess their predictive power for GPP estimation, and (ii) between TROPOSIF/S3VPs and key meteorological variables, to evaluate their role as proxies for environmental drivers in GPP prediction. We then intercompare two global GPP products derived using the same machine-learning framework (Gaussian Process Regression): one driven by satellite observations (TROPOSIF and S3VPs), hereafter referred to as the hybrid SCOPE-GPR-GPP model (1Reyes-Muñoz et al., 2024), and another driven by meteorological inputs, referred to as the data-driven EC-GPR-GPP model (2Reyes-Muñoz et al., 2025). Finally, we assess the consistency and robustness of GPP estimates derived from TROPOSIF and S3VPs at a spatial resolution of up to 300 m by analyzing predictive uncertainties across multiple variable configurations.

Results indicate that the dominant drivers of GPP in the EC-GPR-GPP model are: leaf area index (LAI), latent heat flux, incoming shortwave radiation, and soil and air temperature. In contrast, LAI, fraction of absorbed photosynthetically active radiation (FAPAR), and SIF emerge as the primary predictors in the SCOPE-GPR-GPP model. The two GPP products show strong global agreement, suggesting that SIF and FAPAR can effectively proxy a comprehensive set of meteorological drivers for GPP prediction. Furthermore, global correlation maps reveal predominantly positive Pearson correlations when each variable (SIF, LAI, and FAPAR) is individually correlated with GPP, yielding modal correlation coefficients of approximately 0.85 for SIF–GPP, 0.90 for LAI–GPP, and 0.87 for FAPAR–GPP. Regions exhibiting weak or negative correlations are primarily associated with arid ecosystems (e.g., central Australia) or complex tropical environments (e.g., the Amazon basin). These patterns suggest that: (i) limiting conditions such as drought stress or light saturation may decouple SIF emissions from carbon assimilation, highlighting SIF’s role as an energy-regulation mechanism, and (ii) tropical ecosystems exhibit subtle and rapid dynamics that may fall within the uncertainty bounds of current models, leading to weak correlations.

We conclude that canopy-leaving SIF is a robust global predictor of GPP dynamics and that TROPOSIF, in combination with S3-derived vegetation products, can effectively serve as proxies for a full suite of environmental drivers. Future work will benefit from improved spatiotemporal resolution and is expected to further advance once the FLEX–Sentinel-3 tandem mission becomes operational.

1Reyes-Muñoz, Pablo, et al. "Inferring global terrestrial carbon fluxes from the synergy of Sentinel 3 & 5P with Gaussian process hybrid models." Remote Sensing of Environment 305 (2024): 114072.

2Reyes-Muñoz, Pablo, Dávid D.Kovács, and Jochem Verrelst. "Tower-to-global upscaling of terrestrial carbon fluxes driven by MODIS-LAI, Sentinel-3-LAI and ERA5-Land data." Ecological Indicators 177 (2025): 113597.

Understanding how drought impacts forest photosynthesis is critical for predicting carbon cycle responses to climate change. Solar-induced fluorescence (SIF) has emerged as a promising remote sensing tool for monitoring photosynthetic activity at ecosystem scales, yet the mechanistic relationship between SIF signals and photosynthetic processes under drought stress remains poorly understood, particularly for coniferous forests. In this study we aimed to investigate how tower-based continuous measurements of canopy SIF relate to photosynthetic performance in drought-stressed and watered Scots pine (Pinus sylvestris) canopies over an 18-month period. For this purpose we measured canopy SIF using two JB Hyperspectral FloXbox systems at the Pfynwald long-term drought manipulation experiment in Switzerland, with one system monitoring an irrigated plot and another monitoring a non-irrigated control plot experiencing natural drought conditions. From July 2024 to November 2025, we continuously measured canopy-level SIF at both the red (SIF A, 685 nm) and far-red (SIF B, 740 nm) wavelengths, alongside leaf spectral reflectance measurements. The tower-based observations were complemented by monthly ground measurements including photosynthetic gas exchange (assimilation rates and stomatal conductance), pulse-amplitude modulated (PAM) chlorophyll fluorescence to characterize photosynthetic energy partitioning, and biochemical analyses of needle chlorophyll, carotenoid, and xanthophyll cycle pigment content and composition. Preliminary findings reveal that both SIF A and SIF B successfully captured the seasonal dynamics of canopy photosynthesis, showing characteristic peaks during summer months and pronounced downregulation during winter. Vegetation indices derived from spectral reflectance data, including the near-infrared reflectance of vegetation (NIRv), photochemical reflectance index (PRI), and chlorophyll carotenoid index (CCI), exhibited consistent trends with SIF, supporting their utility as complementary proxies for gross primary productivity and photosynthetic phenology. Our analysis revealed two key findings. First, we detected a consistent increase in SIF A in the irrigated plot compared to the non-irrigated control during the dry summer months, providing clear evidence that SIF A is sensitive to drought-induced reductions in photosynthetic activity. This drought signal was particularly pronounced during periods of high vapor pressure deficit when stomatal limitations on photosynthesis are expected to be most severe. In contrast, SIF B showed limited sensitivity to drought stress, with detectable differences between treatments restricted to a brief period during summer 2025 when temperatures peaked and vapor pressure deficit reached maximum values during the observation period. Secondly, we identified a strong correspondence between temporal variation in SIF A and NIRv throughout the measurement campaign. This relationship suggests that structural changes in canopy greenness and photosynthetic capacity, as captured by NIRv, are reflected in the SIF A signal, supporting the mechanistic link between remotely sensed fluorescence and actual photosynthetic function. Next steps in our analyses will integrate leaf-level physiological and biochemical measurements to validate and mechanistically explain the observed drought responses in canopy SIF in drought stressed Scots pine.

Our current best estimate of carbon sinks under changing climate forcing comes from Earth System Model (ESM) projections, which are highly uncertain. The carbon science community has developed collaborative responses to reducing uncertainty such as the International Land Model Benchmarking (ILAMB) project, which has developed a large body of data to benchmark model projections and parameterizations. While informative, existing ILAMB datasets generally do not provide a strong constraint and leave interpretive ambiguity, do not adequately sample the heterogeneity of processes, or are limited by the proxy nature of the observations. Only Earth Observation data from remote sensing and ground-based networks combined with models can bridge the scale gap between the heterogeneity of the biosphere and global greenhouse gases. Innovative data assimilation and data science techniques can help assess consistency among multiscale observations and models, to reduce uncertainty in predictions. The CARDAMOM model-data fusion (MDF) system for the terrestrial carbon cycle combines vegetation, carbon, and water remote sensing observations with coupled carbon-water cycle processes at multiple scales needed to understand and constrain the heterogeneity of the carbon cycle and carbon-climate feedbacks with sufficient accuracy to confidently reduce uncertainty in ESM predictions. This work presents CARDAMOM as a first-of-its-kind multi-decadal terrestrial reanalysis of carbon, water, and energy fluxes, and their uncertainty, constrained by space-based observing assets, offering a self-consistent benchmarking framework for current and next generation ESMs.