Tuesday, July 8th
(Keynote speaker) Kazuhito Ichii, Chiba University, Japan
AsiaFlux Community: Current Status and Future Perspectives
…abstract coming…
Sebastien Biraud, Lawrence Berkeley National Laboratory, USA
The AmeriFlux Management Project: a Pillar of the Global Flux Community
AmeriFlux is a network of principal investigator (PI)-managed sites measuring ecosystem carbon dioxide (CO₂), water, and energy fluxes across North, Central, and South America. The network is supported by the AmeriFlux Management Project (AMP). As of April 2025, AmeriFlux has registered its 729th site, contributing to a total of 37,820 site-years of flux and meteorological data from 518 publicly accessible sites.
The AmeriFlux FLUXNET data product, generated using the ONEFlux standard processing pipeline, provides critically needed gap-filled and partitioned net CO₂ fluxes. These fluxes are separated into their primary components: gross primary productivity (GPP), representing carbon uptake (sink), and ecosystem respiration (RECO), representing carbon release (source). The dataset also includes robust uncertainty quantification.
To produce this global-ready data product, AmeriFlux has continued its collaboration with regional flux networks, leveraging collective expertise to uphold consistent data standards, co-develop interoperable products, and facilitate scientific exchange. In particular, the development of the AmeriFlux FLUXNET product has been closely coordinated with regional partners including ICOS and OzFlux/Tern networks, ensuring interoperability and alignment toward a unified global FLUXNET data product.
Here, we present the current status of the AmeriFlux FLUXNET datasets and the ongoing efforts to coordinate with regional networks to produce an updated Global FLUXNET data product publicly available by December 2025.
Amukelani Maluleke, South African Environmental Observation Network (SAEON), South Africa
Regional network update and interdisciplinary
This presentation provides an update on the Expanded Freshwater and Terrestrial Environmental Observation Network (EFTEON), part of the South African Environmental Observation Network (SAEON). It highlights developments in key research themes, including biodiversity, atmospheric chemistry, hydrology, biogeochemistry, social ecology, and data management. Snapshot updates include the establishment of long-term vegetation monitoring plots using the SEOSAW protocol, collaborative resource monitoring with rural smallholder communities, and findings from the EDI-SA project on nitrogen biogeochemical cycling.
Jehn-Yih Juang, National Taiwan University, Taiwan
Leveraging Taiwan’s Diverse and Unique Ecosystems for Global Flux Research and Environmental Adaptation Strategies
The flux research community in Taiwan has established a comprehensive network of long-term observation sites spanning a wide range of ecosystems and landscapes from high mountain area to coastal zone, including montane cloud forests, tropical and subtropical plantation forest, agricultural systems (rice paddy, tea fields), estuarine and coastal marsh wetlands, and built environment (residential area, solar farm). Taiwan’s geographic position, located at the juncture of the Eurasian continent and the Pacific Ocean, and across the tropical–midlatitude climatic boundary, offers an exceptional setting for examining ecosystem-atmosphere exchange processes under sharp climatic and topographic gradients. Leveraging this natural complexity, researchers employ an integrated suite of approaches that combine eddy covariance flux measurements, remote sensing, process-based modeling, ecological monitoring, and social science inquiry. These methodological synergies have produced robust datasets and insights into carbon, water, and energy exchanges, contributing significantly to regional and global understanding of land-atmosphere interactions and biogeochemical cycling.
Current research efforts from the flux studies in Taiwan are increasingly oriented toward applying flux data to address pressing environmental challenges in the Asia-Pacific region. Emphasis is placed on understanding how ecosystem fluxes respond to region-specific meteorological and hydrological phenomena such as typhoons, the Asian winter monsoon, tidal influences in coastal zones, and persistent cloud immersion in high-elevation forests. These studies aim to quantify ecosystem resilience and feedbacks to climate variability, while supporting the development of interdisciplinary frameworks for social-ecological system analysis. In particular, researchers are exploring the utility of flux measurements for assessing ecosystem services, guiding land management decisions, and informing adaptation strategies under scenarios of environmental change. This includes expanding observational capacity to encompass critical zone processes and incorporating stakeholder perspectives to evaluate socio-environmental impacts and trade-offs.
Looking forward, the flux community in Taiwan is committed to deepening collaboration with international flux networks, including FLUXNET and other regional flux communities. These partnerships are instrumental for advancing cross-site synthesis, methodological standardization, and data interoperability. By contributing high-quality observations from ecologically and climatically distinctive sites, Taiwan-based researchers aim to enhance global assessments of terrestrial carbon and water dynamics and strengthen the empirical foundation for Earth system modeling. The integration of local expertise within a global network is essential for addressing emergent questions in ecosystem science and for supporting policy-relevant research on climate mitigation and environmental sustainability.
(Keynote speaker) Dario Papale, University of Tuscia, Italy
What next? Possible FLUXNET developments based on past experiences
Ten years after FLUXNET2015 and almost 30 years since the start of FLUXNET, is our global network still needed and useful? And which are the new challenges we need to face and the developments we need to implement in order to keep FLUXNET alive, visible, and unique as it is? In this presentation the current FLUXNET evolution, the strategy currently implemented to ensure the development of the network toward the current scientific needs, the central role of the Regional networks and the importance of the community will be discussed. Starting from the new activities developed in ICOS, a critical (and personal) view of what the community could focus on in the next years will be discussed, highlighting the critical aspects, the importance of the collaboration among networks and Research Infrastructures but also the need of a proper recognition of the work done by all the station teams to ensure one of the most relevant high quality global databases. I hope that at the end of the presentation you will be convinced that “together is better”.
Gilberto Pastorello, Lawrence Berkeley National Laboratory, USA
Supporting the FLUXNET Shuttle through round-robin evaluations of regional network data products
The centralized model used for creating global FLUXNET datasets in the past, such as the LaThuile or FLUXNET2015 datasets, has always had a sustainability problem, with challenges in funding, logistics, and long-term availability, to name a few. In response to these challenges, a more distributed model was adopted by regional networks, who now create the FLUXNET data product for their member sites, and make this product available on their respective data portals. The standardization of the data products relies on using the common and shared codebase, called ONEFlux, for the main processing and creation of the data products. A shared data policy, based on the CC-BY-4.0 license, simplifies the mechanics for data access and assigning credit to site teams. However, from this new organization, a new challenge to data users arises: to be able to access global data, a user has to visit multiple data portals, with different navigation, access details, and structuring of information. The upcoming FLUXNET Shuttle will allow for the on-demand creation of global FLUXNET datasets, dynamically retrieving the most recent FLUXNET data product generated by regional networks. With this new distributed model in place, an issue that might arise is a divergence in data products generated by different regional networks. To address this challenge, we are proposing a set data comparison and validation protocols to cross-check data processing practices from regional networks. We will rely on a round-robin evaluation of data products, having regional networks regularly put each other’s datasets through their own deployment of ONEFlux, checking the final results for consistency. In this presentation we will show the protocols developed so far for this round-robin evaluation and discuss plans for further enhancing this continuous evaluation and cross-validation of data products.
Zhi Chen, Chinese Academy of Sciences, China
ChinaFLUX Development: Key Scientific Advances and Future Directions
ChinaFLUX is a comprehensive network of long-term observation sites measuring ecosystem carbon, nitrogen, water, and energy fluxes across China. As of July 2025, ChinaFLUX has registered 102 sites, spanning a wide range of ecosystems including forests, croplands, grasslands, shrublands, wetlands, lakes, deserts and urban. ChinaFLUX has continued its collaboration with global and regional flux networks. This presentation provides developments in key research themes, including regional carbon budget, climate responses, atmospheric nitrogen deposition, and an update on the current status of the ChinaFLUX FLUXNET datasets 2025. In the future, ChinaFLUX continues to commit to deepening collaboration with FLUXNET and other regional flux networks. Future research efforts in China are exploring the advanced and integrative facilities of flux measurements for assessing ecosystem
processes and functions. This includes expanding observational capacity to encompass trace-gas exchange, near-ground remote sensing, and underground critical processes, to contribute to regional and global understanding of land-atmosphere interactions and biogeochemical cycling.
Charuni Jayasekara, The University of Melbourne, Australia
Mountainous Sphagnum peatlands as strong carbon dioxide sinks: insights from long-term eddy covariance study
Peatlands are a type of wetland ecosystem that stores about one-third of the world’s soil carbon due to anaerobic conditions that limit microbial decomposition. However, increasing global temperatures and changing precipitation patterns due to climate change, the CO2 sequestration capacity of peatlands and current carbon stocks are facing a severe threat. We investigated the variation of Net Ecosystem Exchange (NEE) from an Australian mountain peatland using a six-year-long Eddy Covariance data set. The peatland was a net carbon sink during the growing season, with considerable variation in NEE flux associated with changes in meteorological conditions and vegetation growth stages. During the non-growing season, peatland becomes a net CO2 source due to snow cover inhibiting photosynthetic CO2 fixation, while microbial decomposition continues at above-zero soil temperatures, thus CO2 emissions. The non-growing season NEE was most strongly correlated with soil temperature, whereas the growing season NEE was primarily driven by Photosynthetic Photon Flux Density. The long growing season, typical of the Australian mountain region, might be the reason for peatland being an ongoing annual net CO2 sink. Despite no significant correlation between seasonal water table fluctuations and NEE, extreme droughts with very low water table levels significantly reduced CO2 uptake. Our findings suggest that increased drought frequency under future climate scenarios may weaken the peatland’s sink strength. Conversely, earlier snowmelt extending the growing season could enhance CO₂ sequestration. This study highlights the vulnerability of mountain peatlands to climate change and emphasises the importance of hydrological stability in maintaining their carbon sink function.
Mousong Wu, Nanjing University, China
Modelling methane fluxes along the gradient of Boreal-Arctic peatland ecosystems with a process-based model
Northern peatlands serve as critical carbon reservoirs and natural methane (CH₄) sources. Warming-induced permafrost thaw and associated vegetation-soil changes are altering CH₄ emissions, yet uncertainties persist due to complex water-heat-carbon coupling and freeze-thaw cycles. We developed the CoupModel with respect to more comprehensive representation of gas processes in soil and plants, and used it to simulate O 2 , CO 2 , CH 4 as well as energy and water fluxes at nine northern pristine peatland sites along the Boreal-Arctic gradient. These sites, categorized into seasonal frost, discontinuous permafrost, and continuous permafrost zones, comprised four sites from the FLUXNET-CH4 product, three from the ICOS, and two sites sourced from the principal investigator (PI) and publications. Multi- year (≥3 years) CO₂/CH₄ flux as well as soil water and temperature observations were used to constrain the model. The calibrated model captured hourly CH₄ fluxes (R²: 0.18±0.005 to 0.60±0.02), with good performance on shoulder seasons. We found that CH 4 emissions in the shoulder seasons substantially contribute to the annual budget. Shoulder season (autumn freeze, AF; Spring thaw, ST) emissions contribute up to 24.8% of the total annual emissions, and larger methane emissions exhibit during AF than ST in all sites, up to 15 times higher. Despite variability in AF/ST phase duration along the freeze-thaw gradient, methane emission intensity was consistently higher during AF phases. Our results reveal spatial heterogeneity and seasonal asymmetry in CH₄ fluxes, emphasizing the need to integrate freeze-thaw processes into models for robust peatland-climate feedback predictions. We advocate for enhanced observational focus on shoulder seasons to constrain modeling uncertainties and improve annual CH₄ budget estimates. The integration of CH 4 measurements by FLUXNET-CH4 and other flux communities highlights the pivotal role of collaborative networks in advancing high- latitude methane research.
Nicola Lieff, Airborne Research Australia and University of Adelaide, Adelaide, Australia
Assessing the optimal drivers for flux data gap-filling using random forest networks
The Terrestrial Ecosystem Research Network (TERN) Ecosystem Processes group operates a network of eddy covariance stations that collect long-term atmospheric and soil measurements for monitoring and understanding changes in climate and the environment. Ideally, all data collected would be gap-free, however, all real data has gaps where instruments have not recorded measurements or data has been discarded due to low turbulence. To allow this data to be used as a continuous time-series in further analysis, the missing data is gap-filled using PyFluxPro. The standard community approach uses a predefined set of variables (drivers) for gap-filling, which are the same variables for all stations irrespective of location. However, the stations are located in a large range of climate zones, hence the standard gap-filling drivers might not be ideal for all sites. This is because the drivers were chosen for a small set of initial sites and might not be representative for a heating and drying climate.
To identify which drivers were best suited for each station, we developed a random forest model to objectively assess the relative importance of input variables used to gap-fill ustar, carbon, and energy fluxes. We trained this model on the published data for all available Australian sites using a large range of input variables. This model then determined the relative importance of variables, mean absolute errors, and R2 for the accuracy of the model prediction for a target variable at each site. Next, we grouped the variables into atmospheric, energy, turbulence and soil categories of drivers, which highlighted a distinct variation in the contribution of each category of driver across sites. To assess the ecological significance of these trends, the model importances were sorted by the aridity index and grouped by the Köppen-Geiger classification of each site. There is a notable shift in the importance of energy, turbulence, and soil groups with decreasing aridity, and driver contributions were generally consistent within Köppen-Geiger classifications. Reprocessing the gap-filling of a representative subsample of sites demonstrated a marked improvement in predicting the gap-filled target variables, highlighting that this approach can inform driver selection at new and established sites and will improve our understanding of the ecological significance of different drivers in various climate regions.
Wednesday, July 9th
(Keynote Speaker) Licheng Liu, University of Minnesota, USA
Advancing Ecosystem Carbon Cycle Understanding through Knowledge-Guided Machine Learning and Eddy Covariance Integration
Understanding the terrestrial carbon cycle is essential for managing climate risks and sustaining ecosystem services. Yet, current modeling approaches face critical limitations—process-based models struggle with representation and scalability, while black-box machine learning lacks interpretability and physical consistency. In this keynote, I present recent advances in Knowledge-Guided Machine Learning (KGML), a hybrid AI approach that systematically integrates domain knowledge into data-driven models. Leveraging eddy covariance measurements, remote sensing, and process-based simulations, we demonstrate that KGML enhances both predictive performance and interpretability in estimating ecosystem carbon fluxes. Case studies from agricultural and natural ecosystems reveal how KGML improves spatial generalization, reduces uncertainty in carbon budget estimation, and supports scalable applications such as carbon market verification and methane monitoring. I will also introduce emerging tools like PyKGML, designed to facilitate accessible AI model development for ecosystem science. This talk highlights how KGML, combined with flux networks, offers a path toward more robust, transparent, and actionable ecosystem modeling.
Kim Novick, Indiana University – Bloomington, USA
Use tower data to address a potentially widespread underestimation of forest carbon uptake
While forest-based climate solutions (FbCS) are widely viewed as the most promising natural carbon removal pathways, at scales ranging from individual sites to the entire globe, estimates of forest carbon removal vary widely. Globally, this results in petagram-scale uncertainty that stymies efforts to conduct accurate greenhouse gas inventories and robustly evaluate the potential of FbCS.
Much of this uncertainty stems from the measurement methods used to quantify forest carbon removal. Most operational protocols for measuring, monitoring, and predicting forest carbon removals are based on changes in tree biomass estimated from allometric equations and inventory data. This allometric approach has been foundational for understanding spatial and temporal patterns of biomass stock change. However, important carbon pools in soils, deadwood, and the understory are usually neglected or imprecisely estimated. And allometric equations themselves are imprecise – in some places, estimates of biomass can vary by a factor of two depending on which set of allometric equations is used. In contrast, eddy covariance flux towers continuously monitor carbon fluxes at the ecosystem-scale, integrating over all carbon pools and eliminating the need for allometric equations provided lateral losses to harvest or runoff are small or quantified. However, tower data have their own biases, especially in areas of complex terrain where wind regimes prevent a straightforward application of the methodology.
Here, we use co-collected forest inventory data and flux tower data from across the world to show that tower-derived estimates of forest carbon removals are 15-50% higher than those derived from traditional allometric approaches. This is a large bias with major implications for our understanding of the role forests play in governing the pace of climate change, now and in the future. We contend that it reflects a general underestimation of forest carbon removals by allometric approaches, and a likely overestimation by flux towers due to suboptimal wind conditions at night. We end with a discussion of the near-term measurement and monitoring initiatives that could allow us to better integrate both sources of data together to generate more accurate and policy-relevant information on local to global forest carbon removals.
Chandra Shekhar Deshmukh, APRIL, Indonesia
Bridging Research and Policy: The Impact of Tier 3 Emissions Estimates on Tropical Peatland Management and Climate Goals
Nature-based solutions are now key topics at international forums such as the United Nations Conference of the Parties, aiming to limit global warming to below 1.5°C. Tropical peatlands in Southeast Asia are major contributors to regional land-sector greenhouse gas (GHG) emissions and play a critical role in national emissions reduction targets under Nationally Determined Contributions (NDCs). Yet, existing estimates of GHG emissions from tropical peatlands vary significantly and remain debated. Thus, accurate estimates of contemporary GHG balances and their controlling factors are essential for understanding the global climate impact of these peatlands.
Since 2016, we have been measuring GHG exchanges in intact forest, degraded forest, and Acacia crassicarpa fiber wood plantations within the same peat landscape in Sumatra, Indonesia. Using the eddy-covariance technique, we measure net exchanges of CO2 and methane, and employ flux chambers for soil nitrous oxide exchanges. In 2021, we began measuring methane and nitrous oxide emissions from tree stems in plantations and peat swamp forests using flux chambers. Additionally, we have included measurements of lateral carbon export using water sampling. Last year, we added measurements of diffusive GHG emissions using floating flux chambers and ebullitive GHG emissions using inverted funnels from natural rivers and artificial canals. Our study aims to improve knowledge of emission factors to establish high-resolution (Tier 3) estimates and their links with environmental factors.
Surprisingly, we find that net CO2 fluxes at different groundwater levels from eddy-covariance studies are substantially lower than those derived from previous flux chamber and subsidence studies. Contemporary emissions from Acacia plantation on peat are lower than what the IPCC and Indonesian’s FREL report. Acacia plantations have lower GHG emissions than degraded sites with similar average groundwater levels. Thus, establishing plantation forests on degraded sites reduces long-term GHG emissions; further, using tree biomass for bioenergy avoids fossil fuel burning. Finally, our results confirm that the net avoidance of GHG emissions through the conservation of remaining intact peat swamp forests makes a significant contribution to Indonesia’s NDC.
Aaron Wall, University of Waikato, New Zealand
Carbon fluxes in New Zealand agricultural ecosystems: Learnings from 90+ site-years of measurements on Waikato dairy farms
Understanding the carbon cycle in agricultural ecosystems, particularly determining land management impacts that result in gains or losses of soil carbon, has taken on increasing importance and interest in the context of climate resiliency. Over the past ~20 years, >90 site-years of carbon balance measurements have been made over dairy farming systems on mineral and organic soils in the Waikato region of New Zealand. Eddy covariance (EC) measurements of CO2 fluxes were coupled to measurements and estimates of the major non-CO2 flows of carbon into and out of the ecosystem boundary – typically 2-3 paddocks (6-10 ha) in size – to calculate the net ecosystem carbon balance (NECB). The calculated NECB is assumed to represent the change in soil carbon stocks with much greater precision and uncertainty than soil carbon sampling. Key learnings included: (1) EC as an experimental tool allows for quantification of management effects on ecosystem carbon (e.g. by using treatment-control experiments); (2) while sites were often a CO2 sink, this did not always indicate an ecosystem carbon sink; (3) under standard continuous rotational grazing, sites on mineral soils had nearly steady state NECB, but sites on drained organic soils were a large source of carbon (300-1100 g C m−2 y−1); (4) management matters: pasture renewal (renovation/restoration) and periodic cropping of permanent pastures can result in substantial carbon loss (up to 800 g C m−2 for a cropping event), while the traditional ryegrass/clover sward was better than alternative swards and compositions for ecosystem carbon stocks. For the Waikato dairy farming systems that operate on lands generally characterised by high soil carbon stocks, our EC/NECB measurements have demonstrated that this carbon can be easily and quickly lost through management interventions but is difficult and slow to be (re)gained.
Fan Liu; Yucui Zhang, Chinese Acadamy of Sciences, China
Achieving grain security and carbon neutrality: Challenges from carbon allocation
Climate change and management practices influence crop allocation of carbon (C), and consequently can alter grain yield and the magnitude of C sequestration (or release) from agroecosystems. However, few in situ longitudinal studies are available to quantify these changes. Here, we combined the results from 13 years (from October 2007 to September 2020) of eddy covariance data and detailed crop production measurements to
investigate changing climate and C allocation in a typical wheat (Triticum aestivum L.) and maize (Zea mays L.) double cropping agroecosystem in the North China Plain. We found that the agroecosystem on average acted as a
slight C sink, i.e., net ecosystem carbon balance (NECB) is 36 g C m 2 yr 1) across the study period. Increased CO2 led to a rising trend of gross primary production (GPP, 72 g C m-2 yr-2), ~35% of which led to increased NECB (the slope is 25 g C m-2 yr-2). However, concomitant increases in temperature and decreases in surface soil moisture caused higher partitioning of GPP to autotrophic respiration, leading to lower increases in net primary production and grain yield. Summer maize experienced a greater risk of C source increase, as well as greater grain yield reduction than winter wheat, most likely due to higher temperatures and drought in summer. Overall, our observational evidence suggests that current management and ongoing climate change increase the
ability of the agroecosystem to increase NECB, but does not enhance crop production in this intensively managed high yield agroecosystems. However, C allocation strategies are unlikely to maintain constant in the future as multiple climate change factors act on the agroecosystem.
Lianhong Gu, Oak Ridge National Laboratory, USA
A physical theory of eddy covariance for measuring Earth-atmosphere mass and energy exchanges
Energy processes on the Earth’s surface critically affect life in our biosphere. Yet even eddy covariance (EC) – the state-of-the-art measurement technique – cannot achieve the Earth’s surface energy balance. This problem has puzzled Earth system scientists for decades, casting doubt on the reliability of data used to validate Earth system models, and questioning whether our understanding of energy processes in our living environments is complete. Here we show that the conventional theory guiding EC measurements contains two flaws. First, it neglects the close coupling between mass and total (internal, kinetic, and potential) energy transfers in turbulent flows on the Earth’s surface, an open thermodynamic system. Second, it inadequately constrains offset errors in vertical wind velocity measurements. These two flaws lead to underestimation of the magnitude of diurnal Earth-atmosphere exchanges of sensible heat while the second flaw also leads to biases in the measured exchanges of gases (e.g., CO2). We form a physical theory of EC from the fundamental equations of coupled mass and energy transfer derived from the first principles of physical fluid mechanics and thermodynamics. Contrary to the conventional theory, the physical theory simultaneously conserves mass and total energy. New approaches to constraining wind velocity offset errors are also proposed. We demonstrate the improvements brought about by the physical theory at contrasting EC sites. EC measurements around the world should be conducted according to the new theory while past measurements should be corrected. Our development of the physical EC theory removes a major uncertainty in Earth system research.
(Keynote Speaker) Caitlin Moore, University of Western Australia, Australia
Lessons from long-term monitoring of carbon gains and losses in cropping systems
*Caitlin E Moore, Bethany Blakely, Taylor L Pederson , Nuria Gomez-Casanovas, Christy D. Gibson,
Anya M. Knecht, Guler Aslan-Sunger (Rojda), Evan H. DeLucia, Emily A. Heaton, Andy VanLoocke,
Tilden Meyers, Carl J Bernacchi
Contact:[email protected]
Long-term monitoring of carbon, water and energy fluxes enables detailed understanding of how climate
and land management influence ecosystem productivity and greenhouse gas emissions, including in
agricultural systems. It is commonly assumed that despite heavy carbon losses during land conversion,
agricultural soils reach a carbon steady-state under standard practices and even gain soil carbon under
improved management practices or through optimal species selection. The U.S. Midwest agricultural
region epitomizes these assumptions, whereby native prairie grasslands were extensively cleared by the
early 1900s for annually tilled maize-soybean production. Today conservation tillage practices and
incorporation of alternative crop varieties are being explored as ways to reduce carbon losses from these
systems. However, testing these assumptions with real-world observations remains a challenge due to a
lack of reliable data at scale. Fortunately, several long-term (7-17 years) eddy covariance stations have
been operating in different cropping systems in the U.S. Midwest, facilitating assessment of whether these
assumptions hold for those systems. The flux towers show that after more than 100 years, conventionally
tilled maize-soybean systems still lose substantial carbon and have not reached steady-state. Conservation
tillage practices reduce carbon losses compared to ‘business as usual’ tilled systems but show minimal
evidence for long-term ecosystem carbon storage. Where substantial carbon gains can be made is through
incorporation of perennial crop varieties, with Miscanthus, Switchgrass, and restored tallgrass prairie
systems all increasing ecosystem carbon uptake within their first year of establishment. These findings,
although focused on one region, suggest that the assumptions of steady-state soil carbon and increased
storage from conservation practices do not universally apply and that significant changes to
agroecosystems are required to increase their carbon storage. What remains to be fully understood is how
a changing and more variable climate might alter the current patterns of carbon sequestration in these
systems, necessitating the continued monitoring of agroecosystems into the future.
Yoshiaki Hata, The University of Tokyo, Japan
Conversion from Bornean tropical rainforest to oil palm had contrasting effects on the drivers of carbon and water fluxes
Bornean tropical rainforests have suffered severe deforestation over the past 50 years, mainly due to the rapid expansion of oil palm plantations. Because the rainforests play a key role in regional and global climate by acting as significant carbon sinks and stimulating the hydrological cycle, it is essential to determine the impact of this land-use change on the controlling factors of carbon and water fluxes. However, few comparative studies based on field measurements have been conducted in this region. In this study, we performed paired tower flux measurements over three years in a Bornean tropical rainforest and a mature oil palm plantation on mineral soils in northeast Borneo. Then, we constructed gradient boosting machines for reproducing net ecosystem carbon dioxide exchange (NEE) and evapotranspiration (ET) in both sites and interpreted the model with explainable artificial intelligence. Our results showed that the land-use conversion altered the primary driver of NEE from photosynthetic capacity to solar radiation, indicating that the carbon flux has become more controlled by abiotic factors. On the other hand, the drivers of ET were unchanged and mainly controlled by solar radiation, canopy conductance, and vapor pressure deficit. These results implied that the impact of converting tropical rainforests into oil palm plantations in Borneo on the local material cycle differed between carbon and water, which could inform discussions about future forest conservation in the region.
Luri Nurlaila Syahid, National University of Singapore, Singapore
A cost-efficient and robust approach to monitor ecosystem photosynthesis using near-infrared cameras
Photosynthesis is a critical ecosystem function that determines the land carbon cycle, supports agricultural production, and sustains biodiversity. The tropics contribute the majority of global photosynthesis, however, this region also faces the greatest challenges in monitoring photosynthesis using either ground-based method (e.g., logistics in deploying eddy covariance flux towers) or remote sensing (e.g., cloud contamination). In this study, we propose to use low-budget near-infrared cameras to monitor ecosystem photosynthesis (i.e., gross primary productivity; GPP) across various ecosystem types, including a tropical one. Across 24 sites with collocated near-infrared cameras and eddy covariance towers, we found that the camera-based near-infrared reflectance and photosynthetic active radiation (NIRvP) is strongly correlated with GPP, with the coefficient of determination (R²) ranging from 0.33 to 0.89. One unit of camera-based NIRvP corresponds to 50–90 mol CO2/m2/s in GPP, largely depending on the radiation intensity and plant functional types (PFT) of the site. Our study demonstrates the potential of using near-infrared cameras as a cost-effective and reliable method for estimating GPP, offering a valuable alternative for monitoring ecosystem productivity, particularly in tropical regions with sparse eddy covariance observations.
David Rowlings, Queensland University of Technology, Australia
Using the Australian Long-Term Agroecosystems Research network (ALTAR) to build industry tools for natural capital markets and guide the Australian carbon credit market.
Australia is unique within OECD countries in both its large areas of tropical climates and the paucity of data available for understanding agricultural management impacts on carbon and natural capital. As a result, there are limited known economically viable management strategies available to farmers to increase natural capital, poor understanding of current baselines, and no accurate measurement and verification tools that are economically viable at scale. Australian agricultural supply chains therefore face severe barriers to entry into global natural capital markets.
ALTAR was established to leverage TERN ecosystem processes capability to provide Australian agricultural industries a technological foundation to overcome these barriers. Here we present a framework to test the readiness level of this technology across northern Australian agricultural landscapes. Partnering with industry, we have deployed 18 eddy covariance towers in a paired agricultural management approach across Queensland and NSW. Towers are combined with phenocams, acoustic sensors, GHG chambers, drones, and remote sensing, and at some sites automated livestock weighing, GPS livestock tags and in-field enteric methane sensors. Data is not only testing specific management, but will form the basis of a regionally specific reference network to help set the reasonable bounds for low-carbon Australian agriculture.
Zaher Al Salmani, Sultan Qaboos University, Oman
Enhancing Water Management in Oman’s Date Palm Cultivation with Eddy Covariance Systems
In arid regions like Oman, where annual rainfall is less than 100 mm, efficient water management is critical for sustainable agriculture, particularly in date palm cultivation, which represents 80% of the country’s fruit production. This study evaluates the integration of advanced technologies for real-time estimation of crop water requirements, focusing on the performance of the Eddy Covariance (EC) system compared to remote sensing and weather station-based methods. The research was conducted at Samail Farm, where actual evapotranspiration (ET) was measured using an EC system, remote sensing data (SEBAL model), and the FAO56 Penman-Monteith equation.
Results indicate significant discrepancies among the methods, with EC measurements ranging from 0.8 to 1.3 mm/day, remote sensing estimates between 0.3 and 1.5 mm/day, and FAO56 estimates substantially overestimating water needs at 2.6 to 3.7 mm/day. The EC system demonstrated the highest accuracy, offering real-time, direct measurements of water loss, making it a valuable tool for optimizing irrigation scheduling. Findings highlight the potential of EC-based monitoring in mitigating drought stress, improving water use efficiency, and supporting sustainable date palm production. This research underscores the need for integrating modern measurement technologies into Oman’s agricultural practices to enhance water resource management and ensure long-term sustainability.
Ossénatou Mamadou, Institut de Mathématiques et de Sciences Physiques, Bénin
The West African Flux Network (WAF-Net): A new regional community fostering opportunities and collaboration in West Africa
The West African region has contributed little to global FLUXNET data over the past 20 years. Whilst many micrometeorological tower sites monitor CO2, water vapor and energy fluxes between ecosystems and the atmosphere via eddy covariance, the absence of a regional network has limited the scope for coordinated actions to enhance the region’s visibility. In 2024, the West African Flux Network (WAF-Net) was established to coordinate research teams involved in flux measurements in this region. West Africa, known to be highly vulnerable to climate change, is facing numerous challenges including land use-cover change, aerosol-induced health issues, and weather extremes (drought, floods, heatwaves, etc.); all of which have an impact on population well-being and biodiversity. Moreover, the increasing frequency and intensity of extreme climatic events may heighten the vulnerability of certain ecosystems, while others may respond differently depending on limitations imposed by water availability and exposure to high temperatures. Understanding these differential responses is key to assessing ecosystem resilience under multiple stressors. The region’s expected temperature increase and probable impact on rainfall will also influence its ability to mitigate climate change through terrestrial carbon sequestration and prevention of emissions through e.g., reduced deforestation. The WAF-Net encompasses a large range of climates (from the arid Sahelian to the humid tropical zones) and ecosystems ranging from sahelian lands to closed forests and natural vegetation to cultivation and plantation lands, with a large range of crop diversity and specific agro-forestry systems. Currently, WAF-Net has at least sixteen (16) operational sites spread across five countries (Benin, Burkina-Faso, Ghana, Niger and Senegal). It offers, therefore, a great opportunity to study these specific, understudied ecosystems to fill knowledge gaps and to contribute valuable data to the global FLUXNET network. In this talk, we will present the sites comprised in the WAF-Net and discuss results from select research teams. We will also highlight activities and the strategic workflow planned to strengthen West Africa’s contribution to resolving global challenges. WAF-Net is also seeking partnerships to enhance research capacity, to address the regional shortage of in-situ observations, to co-produce novel and transformative knowledge and to secure funding for a long-term program.
Thursday, July 10th
(Keynote Speaker) Belinda Medlyn, Western Sydney University, Australia
How is Australian vegetation changing, can we predict how it will change in future, and how can flux data help?
…abstract coming soon…
(Keynote Speaker) Masahito Ueyama, Osaka Metropolitan University, Japan
Bridging Asia and FLUXNET with the JapanFlux2024 Dataset: A Foundational Step Forward
Long term eddy covariance observations have been conducted across Asia for many years, yet data sharing has remained relatively limited in the region. In Japan, the JapanFlux regional network was launched in 2006 to integrate flux observations conducted by Japanese research institutions. As a result, JapanFlux has developed the first publicly available regional flux dataset in Asia — JapanFlux2024. This dataset contains observations collected between 1990 and 2023 across Japan and surrounding areas, comprising 683 site years of data from 83 sites. Its format is inspired by FLUXNET data products, but has been streamlined and adapted to a dedicated processing pipeline tailored for the Asian context. JapanFlux2024 is now participating in the ongoing next generation FLUXNET initiative, making a significant contribution to the global understanding of energy, water, and carbon cycles. We hope that JapanFlux2024 will serve as a model for AsiaFlux, fostering closer collaboration and a more open data-sharing culture across the Asian flux community.
Jacob Nelson, Max Planck Institute for Biogeochemistry, Germany
X-BASE: terrestrial carbon and water flux products from FLUXCOM-X
Mapping in-situ eddy covariance measurements (EC) of terrestrial carbon and water fluxes to the globe is a key method for diagnosing terrestrial fluxes from a data-driven perspective. We describe the first global products (called X-BASE) from a newly implemented up-scaling framework, FLUXCOM-X. The X-BASE products cover the globe at 0.05° spatial resolution for every hour and include estimates of CO2 net ecosystem exchange (NEE) and gross primary productivity (GPP).
Compared to previous FLUXCOM products, the new X-BASE NEE better reconciles the bottom-up EC-based NEE and estimates from top-down atmospheric inversions (global X-BASE NEE is -5.75±0.33 PgC yr-1). The improvement of global NEE was likely only possible thanks to the international effort to improve the precision and consistency of eddy covariance collection and processing pipelines, as well as to the extension of the measurements to more site-years resulting in a wider coverage of bio-climatic conditions. However, X-BASE NEE shows low inter-annual variability, which is common to state-of-the-art data-driven flux products and remains a scientific challenge. With 124.7±2.1 PgC yr-1, X-BASE GPP is slightly higher than previous FLUXCOM estimates, mostly in temperate and boreal areas, and temporal patterns agree well with TROPOMI-based SIF.
Moving forward from X-BASE, we outline how the underlying site selection impacts global fluxes with regards to geographical and bio-climactic representation. Initial test show that site selection can have a significant impact on some ecosystems, such as arctic/boreal or agricultural areas. Furthermore, we explore the limits of advanced machine learning methods in accounting for limitations in bioclimactic representativity. As an outlook, we will outline the potential for new remote sensing products to give more information site characteristics and thus improve our ability to scale from eddy covariance to global and regional fluxes.
Youngryel Ryu, Seoul National University, South Korea
GPP-net: a robust high-resolution GPP estimation network for Sentinel-2 using only surface reflectance and photosynthetically active radiation
High-resolution gross primary productivity (GPP) estimation is crucial for ecological and agricultural applications that require fine spatial details to capture GPP heterogeneity. Satellite-based GPP estimation usually relies on land cover and meteorological data. For the high-resolution GPP estimation, the misclassification of land cover data and coarse resolution of meteorological data greatly increase the uncertainty. Here, we propose a robust high-resolution GPP estimation deep learning (DL) network, named GPP-net, using only satellite surface reflectance (SR) in Sentinel 2 and photosynthetically active radiation (PAR). Specifically, GPP-net is based on a fully 1-D convolutional encoder-decoder network combined with a spectral band importance estimation module. To enhance the generalization of GPP-net, we ran the soil-canopy energy balance radiative transfer (SCOPE) model using the FLUXNET2015 dataset to simulate surface reflectance. We then combined these simulated reflectance data with GPP and PAR data extracted from FLUXNET2015 to pre-train GPP-net. Compared to three benchmark models including near-infrared reflectance of vegetation multiplied by incoming sunlight (NIRvP), partial least squares (PLS) and random forest (RF), GPP-net improved half-hourly GPP retrieval across seven plant functional types (PFTs) including four forest types, cropland, grassland and wetland, with RRMSE values reduced by 21.6%–26.5%, 9.2%–17.0% and 4.3%–7.9%, respectively. Owing to its robust nonlinear feature learning capabilities, GPP-net facilitates robust GPP estimation across both C3 and C4 vegetation. We found that GPP-net could reliably estimate GPP under drought and heatwave conditions, with minimal improvement from including VPD as a predictor. Furthermore, compared to the benchmark models, GPP-net showed better agreement with time-series in-situ half-hourly GPP, demonstrated greater robustness to soil in GPP mapping, and had strong ability in capturing inter-annual variability of GPP. Our model dispenses with land cover data and minimizes the requirements of coarse-resolution meteorological data for high-resolution GPP estimation, and could serve as an effective model for global high-resolution GPP estimation.
Jacqueline Reu, University of California, Berkeley, USA
A Review: Long-term trends and emergent dynamics in the eddy covariance record
Long-term eddy covariance records provide critical insights into ecosystem-climate interactions by capturing seasonal, yearly, and decadal variability in carbon, water, and energy fluxes. Here, we synthesize findings from eddy covariance studies with time series ≥7 years, examining drivers of both interannual variability (IAV) and long-term trends in ecosystem fluxes. We first provide a comprehensive overview of the environmental drivers of gross primary production (GPP), ecosystem respiration (RECO), and evapotranspiration (ET). From there, we discuss the causes of emergent patterns in ecosystem water and light use efficiencies respectively (WUE, LUE). We then assess how changes in these ecosystem fluxes and resource use efficiencies in turn drive the IAV and long-term trends in net ecosystem exchange (NEE), as changes in NEE often reflect de-couplings between its component fluxes (GPP and RECO) rather than direct responses to environmental drivers. Phenological changes and extreme events (e.g., droughts, fires, logging) are discussed in their own sections due to their complex impacts on ecosystem fluxes. We identified several key knowledge gaps and discuss the relevance of long-term data records for resolving them. Our results highlight the critical need to extend and expand our long-term flux records, as eddy covariance provides unparallelled continuous measurements of natural processes we must understand in the face of climate change.
Yi Yu, The University of Sydney, Australia
Assessing multi-scale agricultural water stress dynamics through the integration of remote sensing and OzFlux eddy covariance data
The integration of eddy covariance measurements with remote sensing data provides critical insights into agricultural water stress dynamics across spatial and temporal scales. By leveraging surface-atmosphere exchange processes quantified at OzFlux sites, this study establishes mechanistic linkages between field-scale water flux measurements and satellite-derived drought indices. Here, we present a multi-scale drought assessment framework that employs OzFlux measurements as validation benchmarks for a remotely sensed drought index, namely the Soil Moisture Agricultural Drought Index (SMADI). We integrated multi-sensor information of soil moisture, land surface temperature and vegetation indices to generate downscaled SMADI (dSMADI) estimates across Australia’s agricultural regions. At the continental scale, SMADI demonstrated moderate correlations with monthly Standardised Precipitation Evapotranspiration Index across multiple timescales (median absolute R = 0.57 at 12-month scale for agricultural regions). Spatiotemporal fusion techniques preserved fine-resolution spatial patterns while maintaining temporal consistency, with validation against OzFlux tower measurements revealing robust agreement for both thermal (R > 0.91) and vegetation (R > 0.93) components. The dSMADI effectively captured field-scale heterogeneity during the 2019 “Tinderbox” drought that was obscured in coarser resolution products. Validation against semi-independent flux tower SMADI derived from OzFlux measurements enabled rigorous assessment of component contributions across spatial domains. Scale-dependent analysis revealed differential contributions: thermal stress and lagged vegetation response co-dominated drought variations at continental scales, while moisture dynamics exhibited mediating influence at field scales and OzFlux tower footprints. This framework established quantitative linkages between satellite-derived indices and surface-atmosphere exchange processes, providing insights for precision agriculture and targeted drought management in dryland ecosystems.
Xiangzhong (Remi) Luo, National University of Singapore, Singapore
The spatiotemporal variations in ecosystem photosynthetic quantum yield and their drivers
The quantum yield (α) of photosynthesis represents the maximum light use efficiency (LUE) as indicated by the initial slope of photosynthetic light response curves. α is an important variable in LUE-based models which are widely used to simulate gross primary productivity (GPP) from regional to global scales. However, the spatiotemporal variations in α at the ecosystem scale remain elusive despite its importance. Here, we leveraged long-term eddy-covariance observations from 90 sites globally and examined the spatiotemporal variations in α and their drivers, using statistical and machine learning approaches. We found significant spatial variability in α across and within biomes, primarily driven by atmospheric vapor pressure deficit (VPD) and soil moisture variations. Meanwhile, the temporal changes in α are primarily driven by the negative effect of VPD, which weakens the positive effects of elevated CO2 and leaf area index (LAI). Our results highlight the dominant role of VPD in controlling the spatiotemporal variations of α and the unneglectable impacts of soil moisture, CO2, and LAI on α. These new results provide insights for improving the representation of α in LUE-based models for GPP simulations.
_____________________
Poster abstracts
Gab Abramowitz, UNSW Sydney, Australia
Flux tower data and land model evaluation
This presentation will discuss the application of flux tower data in land model evaluation, briefly illustrate its importance and recent results in this area, before detailing an attempt to connect the OneFlux pipeline to flux tower data used in the modelevaluation.org platform used by the land modelling community. This connection aims to avoid data duplication, address issues of version control and provenance, as well as work towards consistency of tower measurements, processing and ancillary information between the two communities.
Manuel Acosta, Global Change Research Institute, Czech Academy of Sciences, Czech Republic
soil CO2 emissions from a rotational cropping soil – sugar beet (Beta vulgaris) and maize (Zea mays L.) plantation.
In agricultural cropping systems, the larger part of the carbon is stored in the soil. Improving agricultural practices has great potential to increase the amount of carbon sequestered in cropland soil. Promising approaches in recent years are changes in management practices and biochar application. We conducted soil CO2 emissions (SCO2) measurements on four identical plots with different management practices (tillage, no-tillage, biochar, no-biochar) during two growing seasons (2021 and 2022) on a rotational cropping soil, the first crop was sugar beet and the second crop was maize. In all the investigated management during both growing seasons, SCO2 showed a positive correlation with soil temperature but not with soil water content. The highest and the lowest SCO2 were measured at the conventional tillage management with biochar application (10.7 molCO2m-2s-1) and no-tillage management without biochar application (0.4 molCO2m-2s-1 ), respectively, both during the maize crop cover. The SCO2 coefficient of variation (CV) ranged between 0.4 – 0.6 for both growing seasons. The Kruskal-Wallis test did not reveal a statistically significant difference between treatments. However, the Dunn test showed a relevant difference in the R10 value and the Q10 coefficient between management in both investigated years. Nevertheless, in our study, the highest calculated SCO2 (up to 26 tha-1) was obtained at the maize plantation in the conventional tillage without biochar application management, indicating that this kind of soil agricultural management is not appropriate when SCO2 is taken into account.
Sungsik Cho, National Center for Agro Meteorology, South Korea
Biometric and eddy covariance-based estimates of annual carbon storage to assess the carbon balance of forests in South Korea
Sungsik Cho 1,2, Minseok Kang 3*, Seungwon Sohn 1, Juhan Park 1, Sung-Won Choi 1, Seung-Hoon Lee 1, Su-Jin Kim3
1 National Center for Agro Meteorology, Seoul, South Korea
2 Interdisciplinary Program in Agricultural and Forest Meteorology, Seoul National University, Seoul, South Korea
3 Department of Atmospheric and Environmental Sciences, Gangneung-Wonju National University, Gangneung, South Korea
3 Division of Forest Ecology, National Institute of Forest Sciences, Seoul, South Korea
In the era of the climate crisis, accurately measuring and verifying carbon (C) exchange is essential, as forests act as major C sinks. Traditionally, field-based sampling and biomass estimation have been widely used to assess forest C storage. Additionally, the eddy covariance (EC) technique enables high-frequency (~10–20 Hz) monitoring of CO2 fluxes between forest ecosystems and the atmosphere over large area (~1 km2), making it suitable for comparison with plot-level inventory methods. However, while biomass-based estimates provide net primary production (NPP), EC measures net ecosystem production (NEP). To compare these two methods, NEP must be converted to NPP by adding soil heterotrophic respiration (Rh). In this study, we compared NPP from EC (NPPEC) with biometric NPP (NPPBM) to identify differences and estimate the annual C storage of forests. The NPPEC was calculated by adding NEP, measured at six flux observation sites of the National Institute of Forest Sciences, Korea (Gwangneung, Anmyeondo, Wando, Jeju, Pyeongchang, and Hongcheon), to Rh values derived from a global soil respiration database (SRDB-v5.0) at a 1-km resolution using a random forest algorithm. The mean annual NPPBM from 2016 to 2021 was 660 ± 130 g C m⁻² yr⁻¹ (range: 427 to 944 g C m⁻² yr⁻¹), while the NPPEC was 904 ± 246 g C m⁻² yr⁻¹ (range: 522 to 1283 g C m⁻² yr⁻¹). The multi-year annual mean NPPBM and NPPEC values were strongly correlated (r2 = 0.87, p < 0.001), though the mean NPPEC was 27% greater than that of NPPBM (p < 0.001). Comparison with MODIS-derived NPP (NPPMOD) for the same period revealed that NPPBM was approximately 35% lower, while NPPEC was about 3% higher. This discrepancy likely arises because NPPBM does not represent the total stand productivity, as biomass estimates only include trees with a measurable diameter at breast height (≥ 6 cm), excluding shrubs, herbaceous plants, and smaller trees. To address this limitation, we corrected the values using the non-forest cover probability from MODIS Vegetation Continuous Fields (VCF, MOD44B-v6.1), which improved agreement with NPPEC (r² = 0.75, slope of 0.95). This study highlights methodological differences in forest C estimation, identifies sources of discrepancies, and proposes adjustments to improve accuracy. Integrating biometric, EC, and satellite-based approaches can enhance C stock assessments and contribute to more reliable greenhouse gas inventories.
Acknowledgement: This work was supported by the “Maintenance and Data Quality Control of the Flux Tower Observation Network for Major Forest Ecosystems in South Korea project (2025)” from the National Institute of Forest Science, Republic of Korea.
Yuling Fu, East China Normal University, China
Fungi and bacteria trade-off mediates drought-induced reduction in
wood decomposition
The frequency and intensity of drought events have significantly increased in recent decades, raising concerns about their impacts on wood decomposition, a critical process in forest carbon cycling. While previous studies have focused on microclimate and wood traits, the role of microbial strategies in mediating drought-induced changes in wood decomposition remains unclear. In this study, we conducted a field drought experiment with three rainfall reduction treatments (control, -35%(D1), and -70%(D2) rainfall reduction) in a subtropical forest to examine the relative contributions of wood traits, microclimate, and microbial biomass on wood CO₂ efflux. Our results showed that wood CO₂ efflux decreased by 28.27% and 47.49% under the D1 and D2 treatments, respectively. Wood N density was positively correlated with deadwood CO2 efflux, while stable C/labile C exhibited negative correlations under the control, -35% and -70% rainfall treatments. Drought-induced reductions in wood CO₂ efflux were primarily driven by shifts in microbial community composition, especially the transition from bacterial-dominated to fungal-dominated microbial communities as drought intensity increased. Structural equation modeling highlighted the importance of microbial biomass, particularly fungi, in regulating wood CO2 efflux. We also observed significant correlations between fungal oligotroph/copiotroph (O/C) ratios and wood CO₂ efflux under both drought conditions, with tree species showing faster decomposition when Basidiomycota were abundant. However, there were no changes were observed in the studied bacterial variables. Our study underscores the necessity of incorporating microbial functional traits (particularly fungal niche differentiation) into predictive frameworks of forest carbon cycling under climate change.
Liam Grace, Queensland University of Technology, Australia
Improving the capture of management events for net ecosystem carbon balance studies in agricultural production systems.
Ecosystem carbon and water mass balance approaches centred on the eddy-covariance method will likely see increased utilisation in agricultural systems given new availability of instruments that are lower-cost and easier to maintain. In cropping and grazing lands, management events, for example – grazing, tillage, mechanical harvest and burning often represent (1) a considerable lateral transport of carbon and water from the control volume and/or (2) a period when critical assumptions of the method may not be met. Deployments that do not accurately capture this information may not only provide misleading results, but also neglect a valuable opportunity to improve understanding on management effects. In practice, management event records are often inconsistent between producers and contain sensitive information – requiring a consistent approach to translation before use or sharing publicly. For integration into simulation models, historical management event insight may be crucial, but sources may be highly uncertain. This presentation examines ideal, and realistic methods for capturing, interpreting and sharing management events focusing on an Australian grazing rangeland context. Progress towards the implementation of this approach will be demonstrated in the context of the establishment of the Australian Long-Term Agroecosystem Research network (ALTAR).
Nina Hinko-Najera, The University of Melbourne, Australia
Soil respiration and its contribution to the forest carbon balance in a temperate eucalypt forest
Forest ecosystems are one of the most important terrestrial carbon sinks but future trends and its strength are uncertain. Limited data are available on respiration processes for temperate Australian forest ecosystems. This study investigated the temporal dynamics of the net ecosystem carbon balance (NEE), soil respiration (RS) and its component fluxes in a dry temperate eucalypt forest in south-eastern Australia. The integration of results from EC-flux measurements, automated RS chamber measurements and manual measurements of soil heterotrophic respiration (RH) via root exclusion allowed the partitioning of ecosystem respiration (ER) into its component fluxes, such that the contribution of autotrophic respiration (RA), RH and RS to ER and the overall NEE at the Wombat Forest TERN-OzFlux tower site (AU-Wom) was determined.
The forest experienced very high rainfall during the first two years of the study period and was a large and constant carbon sink and all fluxes had a pronounced seasonality. Both GPP and ER peaked during summer whereas NEE peaked in early spring and again in summer. RS rates were lowest in winter and highest in summer but had distinct inter-annual patterns. Soil temperature was the main factor controlling seasonal variation of RS when soil moisture content was not limiting, i.e. above a threshold of 19 vol%, primarily during summer months. Inter-annual variation of RS was considerable and reflected inter-annual variability in rainfall distribution and subsequent changes in soil moisture patterns, that was greatest during summer months. Both, an empirical soil temperature and soil moisture model and the mechanistic Dual Arrhenius Michaelis-Menten (DAMM) model performed equally well and explained 72% of temporal variation in RS. The mean relative contribution of RH to RS was 47% but strongly varied with season and was highest during summer. RS was the main contributor to ER with an overall average RS/ER ratio of 0.65 that did not vary from year to year but varied greatly from season to season, being highest in winter and autumn and lowest in spring and summer. The lower RS/ER ratio in spring and summer indicated a greater contribution of aboveground autotrophic respiration (RAab) to ER. This corresponds to higher GPP and net carbon uptake, i.e. continued growth in spring and summer and associated construction costs. Overall, the dry temperate eucalypt forest was a strong net carbon sink and while RS was the main carbon efflux from the forest, overall ER was dominated by autotrophic respiration processes (70%, RA and RAab) during high rainfall years. The study clearly demonstrates the dynamic nature of the relative contribution of various components of ecosystem respiration. Consequently, the use of a fixed value ratios of RS/ER and RH/RS in many process-based models will misrepresent these processes and highlights the need for empirical data to reduce uncertainties in ecosystem feedback predictions to climate change.
Jason Hupp, LI-COR, USA
Systems as sensors: Hyper-streamlined instruments for eddy covariance measurements
There is increasing recognition of the value of, and need for, high-quality measurements of CO2 and water vapor exchange between the earth’s surface and the atmosphere, for a range of applications from fundamental academic research to carbon MRV frameworks. Eddy covariance is perhaps the best suited approach for direct flux measurements in many applications. However, complexity of traditional eddy covariance systems and subsequent data processing has required specialized expertise to implement, maintain, and make use of the method. This has consequently limited the use of eddy covariance flux measurements outside of academic research.
A new generation of eddy covariance instruments is starting to emerge that greatly reduces system and data processing complexity. These next generation instruments leverage technological and conceptual advancements to re-package the eddy covariance “system” as a single “sensor”, similar in size, cost, and complexity to many widely used meteorological sensors. They pair edge processing at the sensor with connectivity and cloud-based applications to provide an IoE (Internet of the Environment) enabled measurement and node infrastructure, that provides fully processed and quality-controlled flux measurements.
Here we present technical features and field performance of the latest introduction to this new generation of eddy covariance instruments, the LI-720. We discuss benefits, trade-offs and limitations to this sensor, and show performance of the LI-720 in comparison to traditional eddy covariance systems.
Linjie Jiao, Xishuangbanna Tropical Garden, China Academy of Science, China
Evapotranspiration in a seasonal tropical rainforest at the northern edge of Tropical Asia
Xishuangbanna Tropical rainforest, located in the northern edge of Tropical Asia, is a representative seasonal rainforest influenced by Monsoon climate. We observed the evapotranspiration (ET) in this forest through eddy covariance latent heat flux since 2003. This study analyzed the inter-annual and seasonal pattern of ET and their correlation to the changing meteorological factors, elevated CO2 concentration as well as plant growth to clarify the biotic and abiotic process ET. Canopy conductance and water use efficiency are further estimated to evaluate the effect of ET trend on forest water-carbon coupling. The results in this protected rainforest primary convey new insights to forest responding to global change. The long-term continuous eddy covariance observation provides important value for us to understand the water cycle in this special area.
Minseok Kang, Gangneung-Wonju National University, South Korea
Leveraging on-site flux calculations as a surrogate for missing raw high-frequency time-series data
Long-term monitoring of water, carbon, and energy fluxes using eddy covariance towers presents technical challenges, particularly in terms of data storage and processing. High-frequency (10–20 Hz) time-series data required for turbulent flux estimation demand substantial storage capacity and computational resources. While modern hardware enables efficient handling of such data, technical limitations in earlier years often led to the loss of raw data due to storage or backup failures. As a result, many sites relied on on-site flux calculations using data loggers, with pre-processed fluxes often preserved even when the raw data were not. These legacy data typically lack standardized post-processing, resulting in discrepancies compared to fully processed fluxes. This study presents a framework for reprocessing historical on-site covariance data using the spectral correction, sensor separation compensation, and stationarity test. We evaluate how well this approach aligns on-site fluxes with KoFlux-standard results, demonstrating that legacy data can serve as a practical alternative to gap-filling methods when raw high-frequency data are unavailable.
Acknowledgments This study was supported by “Cooperative Research Program for Agriculture Science and Technology Development (Project No. RS-2024-00401659)” of the Rural Development Administration, Republic of Korea and Technology Development Project for Creation and Management of Ecosystem based Carbon Sinks (202300218237) through KEITI, Ministry of Environment, Republic of Korea.
Kuno Kasak, University of California, Berkeley, USA
Methane Flux Dynamics in a California Oak Savanna
Methane (CH4) is a significant greenhouse gas, yet the role of trees in the global CH4 budget remains uncertain. Some studies suggest that wetlands and certain upland trees emit soil-derived or in-tree-produced CH4 at the stem base, while others indicate that upland trees may act as net sinks for atmospheric CH4. This study investigates CH4 exchange in oak savanna in California (AmeriFlux: US-Ton). We measured stem CH4, CO2, and episodically N2O fluxes from six mature oak trees at three heights (0.4 m, 1.3 m, and 2.6 m) along with soil fluxes near each tree and in open areas (without canopy shading) using LI-COR 7810 and 7820 analyzers. We also used the eddy covariance (EC) method to capture ecosystem-level CO2 and CH4 fluxes with open-path LI-COR 7500 and 7700 analyzers, respectively. To explore sub-canopy variability, we installed a second set of EC instruments below the canopy. Three trees were equipped with thermocouples (on both the north and south sides) at all heights to continuously monitor stem temperatures. Preliminary results indicate that tree stems act as small sources of CH4, while the soil mostly functions as a CH4 sink, except in some spots where the ground was waterlogged after the rain. We noticed that both trees and soil showed very limited N2O uptake. The vertical flux profile of stem flux measurements did not suggest a direct link to soil CH4 fluxes, as no statistical differences were seen between different heights. However, we noticed that when air temperatures exceeded 30°C, all stem sampling heights shifted to become CH4 sinks. However, at lower temperatures, most sampling points from the lower to upper heights exhibited clear CH4 emissions. Ecosystem-level EC measurements revealed no consistent CH4 emission patterns. However, since the instruments were installed in mid-autumn, we are still missing the spring and summer periods. Monthly EC flux data showed that both tower and sub-canopy EC are in a similar range, indicating that the ecosystem is a CH4 sink. However, tree stem + soil flux did not equal the tower fluxes, and the sum of these was almost two times lower. On the other hand, both tree stem and chamber measurements are conducted only on a biweekly basis and stem fluxes do not cover the higher part of the tree and also tree branches, which could be an additional CH4 sink. These initial findings still suggest that the oak savanna may overall be a small CH4 sink.
Wei Li, Chiba University, Japan
Monitoring Rapid Surface Changes Using Land Surface Albedo Derived from Himawari-8/9: Validation with Flux Tower Observations
The Advanced Himawari Imager (AHI) onboard Japan’s Himawari-8/9 geostationary satellites provides continuous Earth observation every 10 minutes across multiple spectral bands, effectively reducing cloud-related gaps and capturing diurnal surface changes. Land Surface Albedo (LSA) is essential for understanding Earth’s climate, as it determines the amount of solar radiation reflected back into space. This study estimates LSA using Himawari-8/9 AHI data, applying a kernel-driven bidirectional reflectance distribution function model to the 10-minute AHI surface reflectance dataset to derive daily LSA. Results show strong agreement with ground-based measurements and satellite products from MODIS and VIIRS. Compared to these products, AHI LSA offers a more continuous time series and clearly detects rapid surface changes, such as irrigation and harvesting, at cropland sites. These findings highlight the potential of Himawari-8/9 AHI data for large-scale surface change monitoring, supporting future environmental monitoring and modeling efforts.
Johannes Laubach, Manaaki Whenua – Landcare Research, New Zealand
Contributions of eddy covariance to quantifying the greenhouse gas balances of dairy farm practices
Using eddy covariance, the fluxes of the greenhouse gases CO2, N2O and CH4 can be measured continuously. We demonstrate for pasture-based dairy farming in New Zealand how far these flux data contribute to quantifying the greenhouse gas balances of farming practices. For N2O, eddy fluxes represent emissions from the pasture surface well, and we show that a split-footprint approach validates the emission reductions expected for “regenerative” practice (highly diverse pastures and much lower fertiliser input) compared with “conventional” practice. For CO2, eddy fluxes quantify the activity of the pasture-soil system well, provided that grazing events in the footprint are excluded. However, the measured net CO2 balance does not equal the net carbon balance because the latter is also dependent on the amounts of carbon grazed and returned with excreta. Accounting for these carbon exports and imports, the regenerative dairy operation in our study was more likely to lose carbon over a year than the conventional operation, and the C losses were greater than the avoided N2O emissions (in CO2-equivalent terms). For CH4, we argue that routine eddy covariance in one location is unsuitable to quantify the net emissions of such farm systems. The dominant CH4 fluxes, those from animal emissions, cannot be measured in a representative fashion as herds graze different paddocks every day. The minor CH4 flux contributions, from the pasture-soil system, are too small to be resolved against frequently present advective contributions which originate from cow herds at distances of up to 1 km or even beyond. Thus, to construct a full greenhouse gas balance of pasture-based farming, one requires other methods to measure or model the CH4 contributions.
Sung-Ching Lee, Max Planck Institute for Biogeochemistry, Germany
Process understanding under expanding dryland and shrinking coast
Arid and semi-arid ecosystems are one key contributor to the large interannual variations of global carbon (C) budgets in the atmosphere and their distribution is very likely to increase due to droughts. Additionally, potentially imbalanced nutrient availability led by anthropogenic nitrogen (N) deposition can introduce further variations into ecosystem-level carbon and water fluxes. Coastal blue C ecosystems (e.g., seagrasses, mangroves and tidal marshes) have been found to exceed the capability of terrestrial ecosystems to sequester organic C per unit area. This has led to the development of a promising new strategy for climate change mitigation. But spatial coverage of these ecosystems has been decreasing, losing substantial stored C. Therefore, the Eco-Meteorology group tackle these two critical issues by operating multiple sites and conducting experiments to provide the much needed process understanding under expanding dryland and shrinking coast. First, a large-scale N and phosphorus (P) manipulation experiment with three eddy covariance towers have been conducted in a semi-arid tree-grass ecosystem since 2014 at Majadas de Tiétar, Spain. We have found the importance of non-rainfall water inputs, precipitation repackaging, and litter dynamics in ecosystem-atmosphere fluxes. Second, in the framework of Horizon Europe project, C-BLUES, the team collaborates with scientist to have a new eddy covariance tower in a tidal marsh at Barcelona, Spain. We aim to better quantify the annual C budget for European marshes, identify key flux drivers, and link fluxes with microbial activities. Preliminary results from the blue C side will be shared together with the latest findings from the semi-arid ecosystem.
Xuanlong Ma, Lanzhou University, China
Bridging satellite productivity, fluxnet measurements, and global biodiversity:
Unveiling insights through dynamic habitat indices
Global biodiversity conservation requires scalable tools to effectively monitor species richness patterns, and satellite remote sensing offers a promising avenue. However, the key is how to summarize satellite data into useful insights that is most relevant to biodiversity conservation. Whenever possible, satellite indices for biodiversity conservation should be validated against in-situ measurements such as those from FLUXNET and field-plot network. This study examines the effectiveness of Dynamic Habitat Indices (DHIs) derived from satellite vegetation products, including gross primary productivity (GPP), fAPAR, LAI, NDVI, EVI, and solar-induced chlorophyll fluorescence (SIF), in capturing global species richness across amphibians, birds, mammals, and reptiles. The DHIs is consisted of three sub-indices that each representing an important productivity-species richness hypotheses, namely: annual cumulative productivity (DHI Cum, available energy hypothesis), annual minimum productivity (DHI Min, environmental stress hypothesis), and coefficient of variation of productivity (DHI CV, environmental stability hypothesis). Results showed that DHIs derived from satellite GPP data explain a large proportion of the variance in species richness globally (R2 = 0.70 for amphibians, R2 = 0.78 for birds, R2 = 0.77 for mammals, R2 = 0.77 for reptiles, and R2 = 0.82 when all taxa combined), outperforms other satellite vegetation products. Validation with in-situ DHIs derived from the flux tower-measured GPP across the 124 FLUXNET sites confirmed the robustness of satellite-GPP based DHIs. Globally, protected areas showed significantly higher DHI Cum and Min, and lower DHI CV, underscoring their superior habitat quality for biodiversity conservation. These findings highlight the potential of DHIs as a powerful and scalable tool for linking satellite observations to global biodiversity patterns, thus bridging the gap between remote sensing, fluxnet, and ecological conservation community.
Xuanlong Ma, Lanzhou University, China
Parameterizing an Ecosystem Light-Use-Efficiency Model for Predicting GPP at half-hourly scale in Asia and Oceania using FY4/AGRI geostationary satellite
Gross primary productivity (GPP) through photosynthesis is a crucial ecosystem function that significantly influences food security, carbon cycle, and climate change. Current remote sensing estimates of GPP rely on polar-orbiting satellites that are prone to cloud-induced gaps and unable to resolve diurnal photosynthesis. Hypertemporal observations from the new-generation geostationary satellites offer the possibility of improved satellite remote sensing of GPP with reduced data gaps and much refined temporal resolution. Here we explored the potential of using a simple yet effective ecosystem light-use efficiency (eLUE) model to predict GPP at half-hourly scale from ChinaFlux/AsiaFlux/OzFlux tower sites to a regional scale. Defined as GPP/PAR, eLUE differs from the traditional LUE (GPP/APAR, or ε) in that eLUE essentially integrates canopy light absorption (fAPAR) and the physiological efficiency of photosynthesis (ε), thus eliminating the need for a separate estimate of ε. eLUE was calibrated as a function of FY4/AGRI Enhanced Vegetation Index (EVI), and then GPP can be modelled directly as eLUE × PAR. To quantify the carbon cycle error budget, we analytically derived GPP uncertainty based on the law of error propagation. Cross-validation against more than 50 ChinaFlux/AsiaFlux/OzFlux sites, encompassing multiple plant functional types (PFTs), demonstrated satisfactory performance of the eLUE model. Meanwhile, we found the cloud-induced gaps within the FY4/AGRI generated GPP was much reduced in comparison to MODIS-based GPP datasets, hence provide more timely estimate of GPP for regional scale. We suggest that our eLUE model, in combination of the hypertemporal FY4/AGRI data, can contribute to a better hypertemporal monitoring of terrestrial primary productivity for the Asia and Oceania region.
Parkin Maskulrath, Faculty of Environment, Kasetsart University, Thailand
Seasonal Surface Energy Balance and Carbon Dioxide Flux Dynamics of Tropical Urban Green Areas, Thailand
In the reconstruction of the urban green area into an urban forest, the establishment of the Deep Forest zone within the Urban Green area (Forestias project)revealed that the forest is nearing maturity indicated the measurement of the Deep Forest resembles that of the natural forest-like area; however, the intensity of the carbon sink and the Energy balance shows a great deal of influence by the activities that promotes the excess of CO2 and heat being released. The data acquired from the Eddy covariance reveals an increasing of the latent heat flux as lowering of the sensible flux and the reduction on the Energy Balance Ratio closure value, resulting in lower temperatures recorded, while the expanding canopy coverage and complexity processes a direct impacted to an increased CO2 flux sequestration in parallel with the forest growth and Photosynthesis Flux density. The data can be used as a fundamental study towards the recommendation for the impacts of green area development in tropical cities.
Darian Ng, North Carolina State University, United States
Evolving spatial heterogeneity of methane fluxes in a transitional ecosystem from bottomland forest to forested wetland.
Methane gas fluxes (FCH4) from bottomland ecosystems can be profoundly impacted by sea level rise due to changes in microbial, hydrological, and vegetation conditions. These changes add further uncertainty to the highly complex spatiotemporal variability of FCH4 dynamics. In southeastern United States, rising sea levels have raised water table heights and driven tree mortality across coastal forests, bringing large-scale shifts in many ecosystems from bottomland forests to forested wetlands.
This study uses footprint-weighted flux maps derived from eddy covariance measurements to investigate the spatial heterogeneity of FCH4 in a bottomland forested wetland (Timberlake Forest) on the southeastern United States coast. Timberlake comprises a series of transitional ecosystem regions that represent a chronosequence from bottomland forest to wetland. These transitional regions have been identified with remote sensing data to reveal bands of increasingly wet conditions coupled with observations of tree mortality. Following these spatial patterns, the flux maps reveal highly spatially heterogeneous FCH4 throughout the site. Within the tower’s flux footprint exist areas of methane sources neighboring areas of methane sinks, which spatially correspond to the changing land surface conditions.
Ongoing research is being conducted to investigate dominant subsurface processes that are driving this FCH4 spatial heterogeneity, such as the control of salinity on methane production and oxidation, the differences in soil redox potential, and the composition of methane emission pathways.
Findings from this study can benefit our understanding on how coastal ecosystems will change from a carbon budget perspective as sea levels continue to rise, and can also help improve the spatial representation of ecosystem-level FCH4 estimates in land system models.
Gilberto Pastorello, Lawrence Berkeley National Laboratory, USA
ONEFlux development efforts and recent updates
ONEFlux (Open Network-Enabled Flux processing pipeline) is used by regional flux networks to generate the FLUXNET data product for their member sites. It is fundamentally the same codebase applied to generate the widely used FLUXNET2015 dataset, making different generations of data products compatible, and ready for combined usage. ONEFlux is jointly developed and maintained by the AmeriFlux Management Project, the European Fluxes Database, and the ICOS Ecosystem Thematic Centre, with new contributors joining the development efforts for code improvements and new features. In this presentation, we will highlight recent developments for the codebase for ONEFlux, including changes to behavior for some of the processing steps and impacts on the FLUXNET data product. We discuss changes such as allowing for steps to be skipped dynamically during the execution, better handling for sites and site-years with lower data availability, and streamlining how long gaps are reflected in the final data products. We also show how we are reducing the dependency of ONEFlux on software environments that were too heterogeneous, for instance, by including steps that were executed outside of ONEFlux into the core codebase, and translating the implementation of all steps into Python and C, the only two languages that will remain in use for the codebase. We will also discuss plans for new features being introduced in the near future, and the logistics of ONEFlux runs for regional networks in preparation for integration with the FLUXNET Shuttle, which will allow for global datasets to be generated on-demand, pulling up to date data directly from the regional network data portals.
Boonsiri Sawasdchai, Xishuangbanna Tropical Botanical Garden, China
Comparison of Canopy Phenology and Productivity in Two Tropical Forests from Southwestern China and Northern Thailand
Tropical forest ecosystems are vital for climate change mitigation since they have strong carbon uptake ability and high productivity. The phenology of tropical forests in Southeast Asia is affected by large seasonality, thereby influencing productivity through photosynthetic activity. Still, understanding the relationship between ecosystem productivity and phenology is insufficient in this region. To explore the relationship between phenological indices extracted from canopy cameras and eddy covariance flux data, this study compared two years of monitoring data from flux towers in a tropical rainforest of Southwest China (BNS) and a dry dipterocarp forest of Thailand (DPT) (January 2020-April 2022). The canopy color indices extracted from Red, Green, Blue digital numbers (RGBDN) are divided into RGB chromatic coordinates (RGBCC: RCC, GCC, BCC), Green Excess Index (GEI), RGBratio of Vegetation Condition Index (VCI), Green-Red Ratio (GRr), and Red-Blue Ratio (RBr). The results showed that vegetation indices exhibited clear seasonal variations, reflecting the phenological characteristics of tropical forests. At the BNS, vegetation indices were significantly positively correlated with Gross Primary Production (GPP) (GRr: R2=0.91, GEI: R2=0.90, GCC: R2=0.88, VCI: R2=0.87; P<0.05), while the DPT exhibited weaker relationships compared to the BNS (VCI: R2=0.77, GCC: R2=0.77, GEI: R2=0.71, GRr: R2=0.45). The BNS showed higher sensitivity to phenological changes than the DPT, with a higher annual average GPP [5.92–7.79 gC/(m2·day) vs 4.64–5.63 gC/(m2·day)]. Soil temperature may be a key driving factor for phenological changes in the BNS site, while relative humidity may significantly impact phenology in the DPT forest. Long-term digital camera observations provided valuable data for assessing GPP and understanding the response of canopy phenology to climate change. Future research should focus on quantifying the relationship between canopy phenological changes and ecosystem carbon exchange in tropical forest ecosystems.
Yu-Ting Shih, Research Center for Environmental Changes (RCEC), Academia Sinica, Taipei, Taiwan, Taiwan (R.O.C.)
Eddy Covariance Measurements of CO₂ Fluxes in Pennisetum purpureum Fields: Assessing Carbon Capture Potential in Subtropical Taiwan Under Typhoon-Disturbed Conditions
Yu-Ting Shih1*, CHARLES C.-K. Chou1, Yi-Ying Chen1, Tzu-Rung Li2
1Research Center for Environmental Changes (RCEC), Academia Sinica, Taipei, Taiwan
2Taiwan Livestock Research Institute, Ministry of Agriculture, Tainan, Taiwan
+ Presenter, * Corresponding author, [email protected]
Abstract
Pennisetum purpureum, a highly productive C4 grass, is recognized for its efficient carbon fixation, making it a promising carbon sink crop in tropical regions. However, limited research exists on its carbon flux dynamics within subtropical monsoon climates, particularly concerning the impact of typhoon disturbances and seasonal rainfall patterns. This study conducted long-term, in situ observations from 2023 to 2025 at the Taiwan Livestock Research Institute in Tainan, Taiwan, utilizing eddy covariance (EC) techniques to monitor CO₂ flux dynamics in a 1.1-hectare P. purpureum field. The primary aim was to assess its carbon capture potential in Taiwan. Over the 59-week monitoring period, the field exhibited a cumulative net CO₂ uptake of 34.0 tons CO₂/ha, comprising 73.2 tons CO₂/ha absorbed in daylight and 39.2 tons CO₂/ha emitted at night, with a harvested fresh biomass yield of 57.5 tons/ha (approximately 21.9 tons CO₂e/ha). Peak weekly mean noontime CO₂ uptake fluxes reached -60 μmol/m²/s, while nighttime fluxes ranged from 10 to 20 μmol/m²/s. Environmental factor analysis revealed temperature as the primary limiting factor for carbon absorption efficiency, with precipitation as a secondary influence. Consequently, summer conditions, characterized by high temperatures and abundant rainfall, promoted significantly higher growth rates than winter. Notably, the study period included two typhoon events, which resulted in lodging and subsequent reductions in harvesting efficiency. Considering both carbon sequestration efficiency and biomass preservation, a 16-week harvest cycle is recommended, with harvest periods scheduled at the end of February, June, and October to mitigate the adverse effects of the plum rain and typhoon seasons. These findings contribute valuable long-term observational data on the carbon flux dynamics of P. purpureum in a subtropical monsoon climate, elucidating its diurnal and seasonal variations. Furthermore, they provide a scientific basis for optimizing field management strategies to enhance the potential of P. purpureum as a sustainable bioenergy and carbon sequestration crop.
Naoya Takeda, Queensland Univeristy of Technology, Australia
Integrating Tropical Grassland EC Flux Data to Improve Grassland GPP Estimates
Eddy Covariance (EC) flux towers allow direct, high-frequency measurements of land-atmosphere carbon (C) exchange, providing a unique opportunity to evaluate C cycle processes in terrestrial ecosystems. In grasslands — one of the most expansive and productive biomes covering over 40% of the Earth’s land surface yet highly heterogeneous — EC observations offer valuable insight into gross primary productivity (GPP) and its variability across climatic gradients. However, global EC flux datasets such as FLUXNET2015 remain heavily weighted toward temperate regions, leaving tropical grasslands underrepresented and introducing uncertainty into global C estimates.
This study synthesised EC flux data at ~20 grassland sites predominantly in the tropical regions in Australia with publicly available grassland EC flux datasets including FLUXNET 2015, AMERIFLUX and ICOS, together with meteorological and remote sensing data.
Compared to the existing satellite-based products such as MODIS GPP, the synthesis of EC flux tower network in this study improved the GPP estimates across the temperate and tropical grasslands.
Our findings provide GPP estimates spatio-temporally scalable to a broader range of global grasslands and highlight the importance of regional coverage of the flux tower network to account for complex climatic and edaphic impacts on C cycling. Further establishment and inclusion of EC flux towers into the tower networks will enable improvement of C cycle models, refinement of satellite-based productivity estimates and uncertainty reduction in global C cycle assessments.
Qiufeng Wang, Institute of Geographic Sciences and Natural Resources Research, CAS, China
Declining Atmospheric Phosphorus Wet Deposition in China
Atmospheric phosphorus (P) deposition has become a significant external P source for terrestrial and aquatic ecosystems, influencing functions like productivity by altering P bioavailability. However, systematic quantification of atmospheric P deposition in China remains lacking. Based on China Wet Deposition Observation Network (ChinaWD) data from 2014 to 2022, we explored the wet deposition fluxes, spatiotemporal patterns, and influencing factors of various atmospheric P components. The annual average wet deposition fluxes of total phosphorus (TP), dissolved total phosphorus (DTP), and total particulate phosphorus (TPP) in China’s atmosphere were 0.75 ± 0.53, 0.40 ± 0.28 and 0.36 ± 0.28 kg P ha-1 yr-1, respectively, with total deposition amounts of 0.72, 0.38 and 0.34 Tg P yr-1. Deposition fluxes of TP and DTP increased from northwest to southeast. Although no significant trend in annual changes was observed over the 9 years, the average P wet deposition flux during 2019 – 2022 declined by 26.72% – 40.25% compared to 2014 – 2018, likely due to the ecological restoration and atmospheric environmental governance measures China has implemented in recent years. This study provides a systematic examination of the dynamics and mechanisms of atmospheric P wet deposition in China, offering valuable data for understanding its ecological impacts.
Ruci Wang, Chiba University, Japan
Multi-Model Comparison of Gross Primary Production Estimates across Machine Learning, Remote Sensing, and Process-Based Approaches
Gross Primary Production (GPP) is a key indicator of terrestrial ecosystem carbon uptake, and its accurate estimation is crucial for understanding carbon cycle dynamics. Various methods have been developed for GPP estimation, including Machine Learning (ML)-based models, Remote Sensing (RS)-based models, and process-based models (e.g., Trendy). However, significant discrepancies exist among these approaches due to differences in data sources, assumptions, and model sensitivities to environmental drivers.
This study compares GPP estimates from three approaches: Machine Learning (ML), Remote Sensing (RS), and process-based models (e.g., Trendy) at a spatial resolution of 0.25°. By analyzing seasonal variations, spatial distributions, and interannual trends, we assess model consistency and uncertainty. Correlation coefficients, root mean square error (RMSE), and bias are computed to evaluate their performance.
Results indicate that the variation range of GPP estimates from the Trendy models is relatively large, with some models producing significantly higher values than ML and RS models. Compared to ML and RS, Trendy models exhibit stronger fluctuations, particularly in high-latitude regions (e.g., Siberia and Northern Canada), where cloud cover and satellite data limitations contribute to estimation discrepancies. In tropical rainforest areas (e.g., the Amazon), some models yield lower GPP estimates, likely due to vegetation canopy cover affecting ML and RS model accuracy. Among the evaluated models, NIES estimates are close to the mean value, whereas E3SM produces higher GPP estimates than other models.
This comparative analysis highlights the strengths and limitations of each approach, providing insights into their applicability for carbon cycle research. The findings contribute to improving multi-model GPP assessments and enhancing our understanding of terrestrial carbon fluxes under changing climate conditions.
Baptiste Wijas, Cary Institute of Ecoystem Studies, USA
Contribution of deadwood decomposition to ecosystem level carbon fluxes
Deadwood represents ~8% of global forest carbon (C) stocks and its decay is a major driver of C cycling. Despite deadwood being an important source of C as it decomposes, little is known of its contribution to the total flux of C within ecosystems. Deadwood C stocks constantly change through time and can become especially large after disturbance events such as windstorms, fires or beetle outbreaks. Understanding deadwood decomposition rates through time is crucial to better understand total contribution of deadwood decomposition to C budgets of ecosystems. Towards this goal, I set up a wood decomposition experiment across six climatically diverse TERN SuperSites which are part of the OzFlux Network. This experiment builds off a similar protocol used 8 years ago at all TERN SuperSites and at a total of 133 sites globally, value adding by putting in additional wood pre-treatments simulating fire or flooding exposure. Further, it provides an additional set of measures of deadwood decomposition for six SuperSites. To better understand Australian carbon stores, I also repeated measurements of deadwood C stocks at the six SuperSites to characterise changes in deadwood C stocks through time. Moving forward, I plan to combine temporal dynamics in deadwood decomposition rates and C stocks with flux tower measurements to build a more accurate picture of deadwood contribution to total ecosystem C emissions through time. By combining field experiments with flux tower measurements, I will increase understanding of ecosystem C cycling to improve modelling of deadwood in process-based Earth System models. This knowledge can be used to better predict future C emissions from natural ecosystems.
Xiangming Xiao, University of Oklahoma, USA
Vegetation Photosynthesis Model and prediction of gross primary production over the FLUXNET sites and the globe
Plant photosynthesis is composed of light absorption process and CO2 fixation process (the Calvin Cycle). Plant photosynthesis or gross primary production (GPP) has large temporal (diurnal, pulse, seasonal, and interannual) variation, driven by extreme weather events, climate variation, and atmospheric CO2 concentration. Many models with different levels of complexity have been developed to estimate GPP, but their prediction skill varies substantially. One group of these models, often called as data-driven models or light (energy) use efficiency models, emphasizes accurate characterization of light absorption process and then use light use efficiency to represent CO2 fixation process. In this presentation we will provide brief review of the data-driven models and then introduce latest version of the Vegetation Photosynthesis Model (v3.0). VPM estimates daily GPP (g C/m2/day) as a product of the amount of light absorbed by chlorophyll (APARchl) and light use efficiency. VPM v3.0 has three major improvements. First, it uses leaf traits (broadleaf vs needleleaf) in estimating amount of light absorbed by chlorophyll (APARchl) in broadleaf plants and needleleaf plants. Second, to represent the relationship between GPP and air temperature at individual sites due to temperature acclimation, it uses site-specific optimal air temperature for GPP (Topt-site), instead of biome-specific Topt-biome. Third, it uses two different equations to represent the relationship between GPP and water. We will report the results of VPM (v3.0) simulations over many eddy covariance tower sites as part of our effort for model accuracy and uncertainty evaluation. The comparisons between predicted GPPVPM estimates and GPPEC from the eddy covariance tower sites are carried out. We will also report the global GPP products from VPM simulations with MODIS images and ERA5 climate dataset over the period of 2000-2024. The resultant global GPP data products over 2000-2024 could be used for many applications that have significant societal benefits, including ecosystem goods and services, food and water security, and the United Nations Sustainability Development Goals.
Bai Yang, Campbell Scientific, Inc., USA
Field trials of trace-gas analyzers designed for eddy covariance flux measurements of methane (CH4) and nitrous oxide (N2O)
Scott S. Cornelsen 1, Cody Hatch 1, Ivan Bogoev 1, Bai Yang *1, Michael R. Schuppenhauer 2, Stephen W. Chan 2 and Sebastien C. Biraud 2
1 Campbell Scientific Inc, Logan, Utah, USA
2 Lawrence Berkely National Laboratory, Berkeley, California, USA
The eddy covariance (EC) flux technique is a broadly accepted method to measure the exchange of greenhouse gases between surface ecosystems and the atmosphere. It is widely applied for studies of water vapor (H2O) and carbon dioxide (CO2). The relative abundance of these gases makes them suitable for measurement with non-dispersive infrared (NDIR) gas analyzers. Instruments based on narrow-band tunable infrared lasers are also employed to measure trace gases. NDIR sensors have traditionally been preferred for EC due to their higher frequency response, smaller physical size, lower power, and lower cost. However, as greenhouse gas research has expanded to include the trace gases Methane (CH4) and Nitrous Oxide (N2O), the application of the EC method has been hampered by the lack of sensors with these same advantages. Campbell Scientific Inc has developed the TGA300 series of closed-path trace gas analyzers for CH4 and N2O to specifically meet the requirements of EC flux measurements. These instruments are built on the tunable infrared laser technology of the TGA200 with a long track-record in the field. In this study, we examine the performance of these new instruments during field trials conducted in large-scale agricultural ecosystems in California, USA. The performance will be examined at the sensor level and the system level. The sensor-level analysis considers the frequency response, noise performance, and accuracy of the analyzers. The system-level analysis compares total data availability, power requirements, signal synchronization, maintenance routines, and logistics of a complete EC station. While these instruments offer novel capabilities for EC studies, an effort has been made to compare their performance to similar sensors. The study concludes that these new sensors are a significant advancement in the overall quality and performance of eddy covariance flux measurements of CH4 and N2O.
Liyao Yu, National University of Singapore, Singapore
The shift in resource limitations of ecosystem productivity detected by long-term eddy covariance observations
The gross primary productivity (GPP) of terrestrial ecosystems is constrained by the availability of resources, such as radiation and precipitation. Understanding the primary resource limitation of ecosystems is critical to predicting carbon cycle responses to climate change. Here, we analyzed long-term (>15 years, 1981–2020) eddy covariance data from 55 flux sites and introduced a novel approach to quantify the radiation use efficiency (RUE) and precipitation use efficiency (PUE). We hypothesize that low interannual variability in RUE or PUE implies strong radiation or precipitation limitation on GPP. We found that the difference between PUE and RUE (PUE−RUE) declined and turned negative at many Europe sites (17 out of 34), especially during the severe drought in 2018–2020, indicating an increasing precipitation limitation on GPP. Meanwhile, most sites in North America (14 out of 21) exhibited generally positive or increasing PUE−RUE, suggesting stronger control of solar radiation on GPP. These results highlight the spatial heterogeneity in the resource limitation and the changes in resource limitation for global ecosystems and reveal a potential constraint of ecosystem productivity due to the mutual impacts of both radiation and precipitation availability under future climate.
Ladislav Šigut, Global Change Research Institute of the Czech Academy of Sciences, Czechia
ICOS foliar sampling data set: the concentrations of leaf nutrients relate to water-use efficiency of European semi-natural ecosystems
The ecosystem CO2 and H2O fluxes are typically studied in the context of changing micrometeorological conditions, such as light, water availability, and temperature. However, CO2 and H2O budgets are further modulated by nutrient availability (NA) and air deposition (AD). The influence of changes in NA and AD is observable only over longer periods, thus when evaluating the time series of eddy covariance fluxes it is more practical to compare across different sites with contrasting levels of NA and AD. Water-use efficiency (WUE) is a useful indicator of ecosystem performance and fitness, and it allows us to evaluate changes in ecosystem functioning since it reflects the degree of stomatal regulation of carbon assimilation and water loss.
In this contribution, we explore the ICOS foliar sampling data set available for Class 1 and Class 2 ICOS ecosystem stations. We particularly focus on C, N, Ca and P concentrations in sampled sun-exposed current leaves. Though nutrient content in the soil is not available, leaf samples are expected to be a better measure as they should reflect both soil NA and nutrient accessibility to the plant. The wide selection of semi-natural ecosystems across Europe allows us to evaluate the capability of the ICOS network to capture the impact of NA on WUE. Furthermore, we evaluate how leaf NA relates to AD, especially in the case of nitrogen depositions. For this purpose, EMEP MSC-W modeled air deposition results are used. We show that N/Ca ratio relates with WUE for the evergreen needleleaf forests that are dominant in the data set (and across Europe).