Seeking Closure to Better Understand Earth’s Energy Balance and Water Resources

Earth’s surface and atmosphere constantly exchange energy and gases such as water vapor and carbon dioxide through evapotranspiration (ET), photosynthesis, and respiration. Measuring these exchanges, which regulate fundamental Earth system processes affecting weather and climate and the cycling of carbon and other critical ecosystem components, has challenged scientists for centuries. Until the 1980s, for example, technologies for measuring ET (a term sometimes used synonymously with “evaporation,” as some scientists do not distinguish between plant transpiration and other sources of evaporation) lagged behind those for other meteorological measurements. In his 1979 presidential address to the Royal Meteorological Society, John Monteith recalled how 19th century British meteorologist G. J. Symons referred to evaporation as “the most desperate branch of this desperate science” of meteorology. Monteith then explained, “One of the main reasons for despair was the lack of techniques for measuring evaporation over natural surfaces.” Addressing urgent socioenvironmental issues, including population growth, increasing demands for food, and changing patterns of land use and water availability, demands reliable, direct measurements and robust models characterizing surface-atmosphere exchanges. Today, such measurement and modeling approaches are increasingly applied to evaluate agricultural water consumption, assess the effects of land management practices on carbon stocks, monitor and predict drought and wildfire events, and quantify the impacts of these phenomena on human and ecosystem health. Despite theoretical and technical advances in recent decades, researchers have been plagued by the energy balance closure problem. Since the 1980s, eddy covariance (EC) systems—comprising collocated wind and gas sensors mounted on towers—have emerged as the preferred means for measuring field-scale ET and exchanges of trace gases and heat. Yet despite theoretical and technical advances in recent decades, researchers have been plagued by the energy balance closure (EBC) problem [Hicks and Baldocchi, 2020; Mauder et al., 2020, 2024]. The crux of this problem is that the sum of surface-atmosphere fluxes measured by EC often fails to match the energy available and stored within the ecosystem, violating the fundamental energy conservation principle. Currently, the community of EC practitioners does not coordinate research efforts and applications investigating the EBC problem. This lack of coordination has limited the advancement of techniques to address uncertainty in flux measurements, conveying doubt to nontechnical audiences and potentially influencing management and policy decisions governing agricultural and energy production, water allocations, ecosystem services, and weather adaptation strategies. The accelerating adoption of EC approaches for scientific purposes and regulatory applications is compelling scientists to address the EBC problem collectively. As part of this effort, members of the EC community convened at an AGU Chapman Conference in 2025—outcomes of which inform this article—to discuss long-standing questions about EBC and to spur development of clear, coordinated best practices for collecting, correcting, analyzing, and applying EC measurements and datasets. The Books Aren’t Balanced Eddy covariance measures the sensible and latent heat fluxes in the surface energy balance (Figure 1) [e.g., Lee et al., 2005; Aubinet et al., 2012; Foken, 2017; Burba, 2022; Lalic et al., 2026]. These fluxes contribute, respectively, to changes in the temperature and the evaporation of water. Fig. 1. The basic components of an EC system used for surface energy balance measurements are illustrated. In principle, the sum of sensible (H) and latent (LE) heat fluxes (i.e., the “turbulent flux”) should be the same as the available energy in the near-surface environment (i.e., the net radiation, Rn, minus the soil heat flux, G). Often, heat (energy) storage in the air and canopy layer below the EC system are neglected, as is advection (horizontal transport) of energy from the surrounding landscape. EC sensors measure vertical wind velocities (updrafts and downdrafts), air temperatures, and water vapor concentrations at sampling frequencies of 10–20 hertz (10–20 observations per second). Sensible and latent heat fluxes are calculated on the basis of covariances of vertical wind speed and either temperature or water vapor concentration, respectively. In other words, the method quantifies the extent to which fluctuations in temperature and water vapor contribute to the vertical transport of sensible and latent heat. These covariances are typically evaluated over 30-minute or 1-hour averaging periods. In principle, the sum of sensible and latent heat fluxes should equal the available energy in the near-surface environment. In principle, the sum of sensible and latent heat fluxes should equal the available energy in the near-surface environment. The available energy is normally simplified as the net radiation—itself calculated as the difference between incoming and outgoing (reflected and emitted) shortwave (i.e., solar) and longwave radiation—minus the flux of heat conducted through soil, while other storage terms are often neglected (Figure 1). In practice, however, the measured combined flux rarely balances the available energy, calling into question the accuracy and validity of flux measurements. Mismatches between these measurements and model calculations are more noteworthy when applied across large spatial scales, heterogeneous landscapes, and short durations. Perceived inaccuracies influence how a wide range of data users, such as water resource managers, farmers, ranchers, and modelers, choose to interpret EC measurements and raise unanswered questions about the application of these data: Is our theoretical understanding of EC accurate? Should measured sensible and latent heat fluxes be adjusted to force EBC to be consistent with model assumptions, and if so, are adjustments based on theory possible and clearly justified? Or rather, are measured EC fluxes accurate while our knowledge of the full micrometeorological system affecting the surface energy budget is lacking? The Colorado River Conundrum Uncertainties in our understanding of EBC and the accuracy of EC measurements bear on many applications, including, most prominently, water use accounting in agriculture. Globally, irrigated farming accounts for nearly 70% of freshwater withdrawals and produces about 40% of the world’s food. In the United States, irrigation consumes approximately 45% of freshwater withdrawals, and the farming it supports contributes more than 50% of the country’s crop value. With water supplies limited, effective water management is critical to food security and economic stability. EC instrumentation extends from a tower above a semiarid grassland landscape near the Santa Rita Mountains in southern Arizona. (Instrumentation appearing in the photo is shown for illustrative purposes only and does not constitute official endorsement or approval by the U.S. Department of Agriculture or the Agricultural Research Service of the manufacturer’s products or services to the exclusion of others that may be suitable.) Credit: Russ Scott/U.S. Department of Agriculture, Agricultural Research Service, Public Domain EC measurements narrow an information gap critical in agricultural water management by helping to evaluate models of ET. Uncertainty in ET models and measurements undermines confidence in the monitoring of consumptive water use and exacerbates global concerns about water availability. The Colorado River system, with its well-documented water shortages, offers an existential use case in which high-quality, validated evapotranspiration models are needed. The Colorado River system, with its well-documented water shortages, offers an existential use case in which high-quality, validated ET models are needed. A century ago, the 1922 Colorado River Compact established guidelines for managing and allocating the river’s water among Arizona, California, Colorado, Nevada, New Mexico, Utah, and Wyoming. Today, the seven Colorado basin states are renegotiating these allocations while attempting to account for prolonged drought, increased evaporation, higher temperatures, and decreasing snowpacks. This effort requires a reliable method, validated by ground truth observations, for quantifying the use of the river’s water and fairly apportioning the available supply. EC observations provide this validation of remotely sensed ET models, which in turn provide an equitable and uniform means of monitoring ET and water use across large geographic regions. One such effort using this approach is OpenET, which produces daily ET data products across the continental United States. OpenET evaluates model outputs against EC measurements that are typically closed and has found that ensemble methods combining multiple model outputs produce the most reliable agreement with EC. Benchmark evaluations for croplands indicate a monthly mean absolute error of roughly 15% between OpenET’s model-derived estimates of ET and estimates from EC data [Volk et al., 2024]. Evaluations comparing daily model outputs and EC data are closer to 25%, however, and evaluations for other land cover types tend to exhibit even greater errors. Dilemmas in these benchmarking validations include how to evaluate EBC and account for closure in reporting uncertainty, as well as whether to adjust measured ET values to close the theoretical energy balance. EBC also has substantial practical implications for assessing water availability and water budgets. The FLUXNET 2015 effort offered the first globally standardized, quality-controlled, and uncertainty-quantified dataset of ecosystem exchanges of carbon dioxide and latent and sensible heat fluxes [Pastorello et al., 2020]. Analyses of FLUXNET 2015 data indicate that measured sensible and latent heat fluxes over natural ecosystems account for an average of about 80% of the available energy [Mauder et al., 2024]. With this imbalance scaled to major biomes, the missing residual energy contributes to large differences in estimated water use at basin, regional, and continental scales. At such scales, the relative uncertainty associated with incomplete EBC ranges from approximately 10% to 25%, with even greater uncertainties reported in European datasets (Figure 2). Fig. 2. Forcing EBC has a large influence on annual global mean estimates of terrestrial latent (LE) and sensible (H) heat fluxes (top), stated as megajoules per square meter per year. The means of forcing closure represent a large source of uncertainty that is not uniform across continents (bottom). Closure was either not forced (None), applied preserving the Bowen ratio (BR = H/LE), or assuming all the missing or residual energy should be added to either LE or H (RES). For details on how these values were derived, see Jung et al. [2019]. Values in the bottom panel were calculated from the LE (blue) and H (red) estimates from different closure scenarios as a ratio of (RES − None)/None. Credit: Mauder et al. [2024], CC BY 4.0 Standardizing Methods for the Preferred Flux Measurement Technique Despite the challenges of EBC, many scientists regard EC as the preferred approach for measuring land surface heat and gas fluxes. It is the most direct measurement of the turbulent fluxes, the theory supporting EC flux estimates is robust, and the required instrumentation has improved significantly over the past half century. Eddy covariance measurements have been performed over thousands of sites worldwide, covering diverse landscapes and climates. As a result, EC measurements have been performed over thousands of sites worldwide, covering diverse landscapes and climates. Many studies have compared EC to other measurement techniques and found that it is the most reliable method for measuring surface fluxes at the hectare scale and answering critical questions about water and carbon fluxes [e.g., Pastorello et al., 2020]. Paradoxically, the scientific community has not yet developed and agreed on standards for applying the EC method. Several observation networks involving national and international collaborations, such as AmeriFlux, the Integrated Carbon Observation System, the National Ecological Observatory Network, and others, have contributed to emerging standards by providing instrument deployment guidelines, data processing protocols, and standardized datasets. Yet most scientists experienced in micrometeorology struggle to commit to specific standards when deploying EC instrumentation. At the same time, growing demand for near-real-time data is pushing scientists to deploy EC systems rapidly and with diminished emphasis on methodological nuances and sensor uncertainties. While the scientific community acknowledges the need for better surface flux measurements for addressing societal challenges, methodological developments are not yet able to fully meet those needs. This dilemma has motivated a new community effort to establish accepted EC standards and progress toward solving the EBC problem. Aims for this effort include improving understanding of errors in surface energy balance terms, reducing systematic EC biases contributing to EBC gaps, cataloging the influence of ecosystem type and land cover on closure and measurement uncertainties, and describing corrections from theory that can compensate for measurement biases. Those aims serve as objectives for a forthcoming special collection of studies emerging from last year’s Chapman conference. The collection will explore the design of EC observational networks and the effects of land surface heterogeneity and data processing on flux estimates and EBC (Figure 3). The collection will also consider how to define EC data quality related to the use of best measurement practices and EBC metrics for practical management applications and socioeconomic decisionmaking, as well as the implications of EBC for model validation. Fig. 3. This schematic offers a qualitative interpretation of the magnitude of EBC as a function of measurement timescale and landscape homogeneity based on a plethora of EC observations and discussions at the 2025 AGU Chapman Conference. Terms in the definition of EBC beside the color scale bar are defined in Figure 1. Higher closure values, which indicate better agreement between total heat fluxes measured by EC and the available energy, typically occur when longer timescales are considered and when landscape cover is more homogeneous. Ultimately, this community effort will identify critical research priorities and fill knowledge gaps after nearly half a century of EC research. It also looks to further develop the technique for real-world needs, from quantifying water consumption and resource availability to illuminating how changing landscapes are affecting weather and human and ecosystem health. Acknowledgments More information about the special collection, which will be published across several AGU journals, including Geophysical Research Letters, Journal of Advances in Modeling Earth Systems, Journal of Geophysical Research: Atmospheres, Journal of Geophysical Research: Biogeosciences, and Water Resources Research, is available in the “Call for Papers.” Additional contributions to the collection are encouraged. References Aubinet, M., T. Vesala, and D. Papale (Eds.) (2012), Eddy Covariance: A Practical Guide to Measurement and Data Analysis, Springer, Dordrecht, Netherlands, https://doi.org/10.1007/978-94-007-2351-1. Burba, G. (2022), Eddy Covariance Method for Scientific, Regulatory, and Commercial Applications, LI-COR, Lincoln, Neb., www.licor.com/resources/books/ec-book. Foken, T. (2017), Micrometeorology, 2nd ed., Springer, Berlin, https://doi.org/10.1007/978-3-642-25440-6. Hicks, B. B., and D. D. Baldocchi (2020), Measurement of fluxes over land: Capabilities, origins, and remaining challenges, Boundary Layer Meteorol., 177, 365–394, https://doi.org/10.1007/s10546-020-00531-y. Jung, M., et al. (2019), The FLUXCOM ensemble of global land-atmosphere energy fluxes, Sci. Data, 6, 1–14, https://doi.org/10.1038/s41597-019-0076-8. Lalic, B., et al. (Eds.) (2026), Micrometeorological Measurements: An Introduction for Beginners, Springer, Cham, Switzerland, https://doi.org/10.1007/978-3-032-03884-5. Lee, X., W. Massman, and B. Law (Eds.) (2005), Handbook of Micrometeorology: A Guide for Surface Flux Measurement and Analysis, Atmos. Oceanogr. Sci. Library, vol. 29, Springer, Dordrecht, Netherlands, https://doi.org/10.1007/1-4020-2265-4. Mauder, M., T. Foken, and J. Cuxart (2020), Surface energy balance closure over land: A review, Boundary Layer Meteorol., 177, 395–426, https://doi.org/10.1007/s10546-020-00529-6. Mauder, M., et al. (2024), Energy balance closure at FLUXNET sites revisited, Agric. For. Meteorol., 358, 110235, https://doi.org/10.1016/j.agrformet.2024.110235. Pastorello, G., et al. (2020), The FLUXNET2015 dataset and the ONEFlux processing pipeline for eddy covariance data, Sci. Data, 7, 225, https://doi.org/10.1038/s41597-020-0534-3. Volk, J. M., et al. (2024), Assessing the accuracy of OpenET satellite-based data to support water resource and land management applications, Nat. Water, 2, 193–205, https://doi.org/10.1038/s44221-023-00181-7. Author Information William P. Kustas ([email protected]), Agricultural Research Service, U.S. Department of Agriculture, Beltsville, Md.; Jeffrey Wood, University of Missouri, Columbia; Jason Kelley, University of Idaho, Moscow; Nicolas Bambach, University of California, Davis; and Jose D. Fuentes, Pennsylvania State University, University Park Citation: Kustas, W. P., J. Wood, J. Kelley, N. Bambach, and J. D. Fuentes (2026), Seeking closure to better understand Earth’s energy balance and water resources, Eos, 107, https://doi.org/10.1029/2026EO260281. Published on 11 September 2026. Text not subject to copyright.Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.