Breakout discussions are interactive sessions, that are led by various community members, to focus on a specific topic. Find this year’s topics below. Click here to go back to the meeting agenda.
Find the poster abstract book (pdf) here.
| Breakouts Thursday October 23 | |
|---|---|
| Title: Gap-Filling Flux Data Made Easy with Machine Learning | |
| Conveners: Yujie Liu, Housen Chu, Oscar Zimmermann | Description: Eddy covariance measurements are often subject to gaps in the data record. While methods such as marginal distribution sampling (MDS) are well-established for filling short gaps, addressing longer gaps (weeks or more) remains a challenge. Recent studies have shown that the machine learning model eXtreme Gradient Boosting (XGBoost) can serve as a robust alternative to MDS, providing reliable predictions of carbon fluxes in a temperate deciduous forest across different gap lengths (Liu et al., 2025, doi: https://doi.org/10.1016/j.agrformet.2025.110438). In this breakout session, we will walk participants through a workflow using XGBoost in Python for gap-filling carbon and water fluxes using data from a NEON site (US-xBR). Key steps will include hyperparameter tuning, model training with cross-validation, and making predictions for missing data, and model evaluation. This session will include beginner-friendly, hands-on coding activity designed to allow participants to actively engage with the workflow and apply the techniques. We will provide a ready-to-use coding template via Binder, an open-source platform that makes GitHub repositories interactive—so there’s no need to install any packages beforehand. A draft version of the tutorial is already available on GitHub https://github.com/YujieLiu666/NEON_gapfill_test. If time allows, we will dive into approaches for estimating uncertainty in annual sums, and then engage in an open discussion about exciting opportunities and challenges for applying machine learning and AI tools in flux science. To ensure that participants get the most from this workshop, we recommend reading the papers listed below beforehand. We will also prepare a few slides highlighting relevant literature and provide some prompts for discussion. Liu, Yujie, et al. “”Robust filling of extra-long gaps in eddy covariance CO2 flux measurements from a temperate deciduous forest using eXtreme Gradient Boosting.”” Agricultural and Forest Meteorology 364 (2025): 110438. Uyekawa, Jeffrey, et al. “”Machine Learning-Based Prediction of Ecosystem-Scale CO2 Flux Measurements.”” Land 14.1 (2025): 124. Moffat, Antje M., et al. “”Comprehensive comparison of gap-filling techniques for eddy covariance net carbon fluxes.”” Agricultural and forest meteorology 147.3-4 (2007): 209-232. Irvin, Jeremy, et al. “”Gap-filling eddy covariance methane fluxes: Comparison of machine learning model predictions and uncertainties at FLUXNET-CH4 wetlands.”” Agricultural and Forest Meteorology 308 (2021): 108528. |
| Format: Tutorial | |
| Title: AI/ML applications and tools | |
| Conveners: Margaret Torn, AmeriFlux Management Project | Description: AI technologies have the potential to enhance and transform various parts of our work. We are going to briefly introduce three different areas for potential AI applications. We are interested in sharing new resources and garnering community feedback to guide future directions and investment. Where do you see gaps and opportunities?
|
| Format: Short presentations and practical discussions | |
| Title: Signals in wood: insights on carbon and water fluxes from tree rings | |
| Conveners: Soumaya Belmecheri, David Moore | Description: This breakout session explores the use of tree-ring proxies to quantify carbon and water fluxes across tree, stand, and ecosystem scales. Building on recent advances, we will discuss how tree-ring data can inform estimates of carbon sequestration, long-term growth trends, and ecosystem responses to environmental change. The session will also highlight emerging approaches that integrate tree-ring records with forest inventory data to improve ecological forecasting of tree growth and assess future carbon capture potential. Our goal is to foster interdisciplinary dialogue that connects ecophysiological proxies with broader efforts to understand and constrain the dynamics of past, present, and future terrestrial carbon sink. Candidate discussion themes include: – Linking tree-ring records to carbon and water fluxes at flux tower sites. – Integrating dendrochronology with forest inventory and remote-sensing data. – Using tree-ring informed models to forecast ecosystem dynamics and carbon sequestration potential. – Building community capacity: identifying collaborators who can collect tree-ring samples and exploring the development of a “toolbox” to help site teams collect and process samples effectively. Interactive Format: – A short overview of current applications of tree-ring proxies in carbon and water cycle research (5–10 minutes) led by an invited speaker. – Small-group discussions facilitated by the organizers around focused themes with the group voting on which topics (discussed above) would be most productive. – Groups will record insights on googledocs or similar and the workshop organizers will collate and report back with insights, opportunities, and challenges. Proposed Outcomes: – A set of research priorities for integrating tree-ring data into carbon and water cycle science. – Identification of potential collaborations and field support for tree-ring sampling across AmeriFlux and partner sites. – Community input to guide the creation of a practical “tree-ring toolbox” (protocols, resources, contacts) to enable data collection by site teams. – Broader connections between the dendrochronology, forest inventory, and flux communities to better constrain terrestrial carbon sink dynamics across timescales. |
| Format: Guided discussion | |
| Title: Care work in atmospheric and natural sciences | |
| Conveners: Carolina Ramirez, Barbara Antonucci | Description: The objective of the discussion is to share issues related to parenthood and other care work in the context of people working in atmospheric and environmental sciences, given that this is a field that requires fieldwork, data processing, project management, among other complex aspects for balancing work and active parenthood. Initially, the objective of the discussion will be presented, along with some guidelines to ensure its fluency. The topics to be addressed are: 1- The beginning of the journey/pregnancy/postpartum/deadlines; 2- Daily reality and the balance between professional and personal life and fieldwork; 3- Institutional challenges; (1–3 with questions related to the topic to initiate dialogue); 4- Support networks and growth, where experiences, advice, and life stories will be shared. For the closing, we will provide a summary of the main points and suggestions for actions, both at the personal level and in terms of public policies within academia. |
| Format: Guided discussion | |
| Breakouts Friday October 24 | |
|---|---|
| Title: Evapotranspiration partitioning in wetlands | |
| Conveners: Angela Che Ing Tang, Inke Forbrich, Pushpendra Raghav, Patty Oikawa | Description: Understanding how evapotranspiration (ET) is partitioned into evaporation (E) and transpiration (T) is critical for quantifying ecosystem water use and improving models of surface energy and carbon exchange in wetlands. Fluctuating water levels and heterogeneous plant communities alter surface–atmosphere exchanges, making ET partitioning in wetlands difficult to interpret using methods developed for drier, transpiration-dominated systems.
This breakout session will highlight recent advances in ET partitioning in wetland ecosystems, focusing on the application of the Flux Variance Similarity (FVS) approach with high-frequency eddy covariance data. The session will include a hands-on tutorial of the FVS workflow and short presentations illustrating its applications across wetland sites. Participants are encouraged to bring laptops to follow along. The session will conclude with an open discussion inviting participant questions, feedback on technical implementation, and ideas for applying FVS and related partitioning methods across diverse wetland environments.
|
| Format: Guided Discussion, Tutorial | |
| Title: Sustaining the Annual FLUXNET Paper | |
| Conveners: David Moore, Trevor Keenan | Description: This breakout will provide a brief update on the status of the annual FLUXNET paper, followed by a highly interactive “world café” style discussion. The annual paper is intended to serve multiple audiences: (1) researchers engaged in international efforts on carbon, water, and energy (e.g., the Global Carbon Project, IPCC authors), (2) researchers outside the flux community who make use of FLUXNET data, and (3) the FLUXNET community itself, who value a synthesis and “state of the network” report.
The discussion will focus on identifying key themes and priorities for this year’s paper, as well as brainstorming future directions. In line with feedback from past meetings, the session will minimize formal presentations and instead emphasize structured conversation and community input. We especially welcome contributions from data providers and data wranglers, whose expertise and perspectives are essential for shaping the scope and impact of the paper. |
| Format: Interactive Format: A short update on the progress and timeline of the paper (5–10 minutes). Break into rotating small-group conversations, each centered on a theme (e.g., new scientific insights from FLUXNET data, data gaps and challenges, community reporting needs, or emerging opportunities for future syntheses). Groups will record and share highlights, which will be synthesized into a collective set of priorities. | |
| Title: Code Snippet Sharing for Flux Research | |
| Conveners: Xiangmin Sun, Roel Ruzol, Yujie Liu, Angela Che Ing Tang, FLUXNET Early Career Network Committee | Description: In the spirit of collaboration and knowledge sharing, we propose hosting a “Code Snippet Sharing Session” within the AmeriFlux community. A series of short, timed presentations where participants can present their code snippets and explain how they help with common challenges in flux research. This session will serve as a platform for researchers, scientists, and data practitioners to exchange useful code snippets that enhance data management, quality control (QAQC), and analysis in flux research. The focus will be on practical solutions, from formatting raw data for tools like EddyPro to code snippets for data visualization, automating repetitive tasks, and integrating various packages in R and Python. We are inviting community collaborations to develop an online coding textbook that leverages modern tools (R or python)and packages for flux science. The session will explore best practices to maintain quality and consistency via different programming languages. The session will also cover resources and platforms where participants can further their learning on R/Python for flux data analysis, especially for students and early-career flux researchers without coding experiences. |
| Format: Panel discussion, Tutorial | |