Earth’s terrestrial ecosystems, from the tundra to the tropics, constantly exchange water, energy, and carbon with the atmosphere, playing a critical role in shaping the planet’s climate. Anticipating and effectively planning for changes in the climate system require an accurate, mechanistic understanding of these exchanges, or fluxes. However, our limited ability to observe the fluxes through time and across spatial scales remains a major barrier to this understanding.
Currently, a global view of terrestrial ecosystem fluxes is made possible by scaling up insights from site-level research, typically conducted using tower-mounted instruments that provide coverage of areas spanning tens of meters to a few kilometers wide. Better coupling near-ground, site-level observations with remote sensing data from satellites or aircraft would represent a significant advance toward overcoming observational limitations. How can that coupling be achieved?
This past July, 40 scientists representing a wide range of disciplines, geographies, and career stages gathered at the Linking Optical and Energy Fluxes Workshop to start building an international community connecting remote sensing and ecosystem flux science. The group discussed opportunities, challenges, and next steps with respect to using remote sensing to estimate terrestrial water, energy, and carbon fluxes.
Major outcomes from this workshop, which we summarize here, included identifying opportunities for proximal remote sensing data to be used in conjunction with flux data, highlighting major barriers limiting the use of these data, and issuing a call to action for community collaboration to develop a coordinated network of proximal remote sensing instruments.
Bridging Gaps in Go-To Methods
To study ecosystem fluxes, scientists currently rely on high-frequency measurements of wind velocity and direction, atmospheric gases, radiation, and heat collected at sites where instrumentation is mounted atop tall towers. Applying the eddy covariance technique to these measurements allows us to estimate carbon and water fluxes and enables substantial insight into current and potential future environmental sensitivities of these fluxes.
However, several limitations on eddy covariance measurements exist. They are limited in spatial extent because the instrumented towers used are relatively few and far between and they are typically concentrated around well-populated regions, leaving key ecosystems like the Arctic-boreal zone and tropical forests underrepresented. They also provide only area-averaged estimates across a given tower’s coverage footprint (typically a diameter of ~10–3,000 meters), they suffer from temporal gaps, and they are subject to uncertainties associated with data processing.
Remote sensing measurements from satellite instruments—such as from the Global Ecosystem Dynamics Investigation, the Moderate Resolution Imaging Spectroradiometer, the Orbiting Carbon Observatory, the Ecosystem Spaceborne Thermal Radiometer Experiment on the International Space Station, and many others—expand our understanding from isolated site measurements to greater spatial and temporal scales. They can also provide additional information about vegetation structure and function, improving our ability to interpret eddy covariance data. Yet satellite-based remote sensing faces limitations too: Although it provides broader coverage, its spatial and temporal resolution is inherently lower, and data collection is subject to frequent gaps because of clouds or other interferences. Satellite data may therefore miss, or misrepresent, key ecological processes.
Collecting proximal remote sensing data within the footprints of eddy covariance measurements—by using repeat drone flights or additional tower-mounted instruments equipped with a suite of imaging and optical technologies—bridges spatial and temporal gaps between flux measurements and satellite remote sensing. Proximal remote sensing can help validate satellite measurements and provide new insights into physical and biological processes affecting observed water, energy, and carbon fluxes.
Despite the value of proximal remote sensing data, significant barriers restrict the technique’s application for studying ecosystem exchanges.
Despite the value of proximal remote sensing data, significant barriers—such as high costs, limited funding, and limited awareness—restrict the technique’s application for studying ecosystem exchanges, and a serious need exists to increase access to and standardization of proximal remote sensing data products. At the recent workshop, participants discussed how to overcome these barriers. One way is to increase scientists’ awareness of existing synergies, challenges, and opportunities with respect to using proximal remote sensing. Another is fostering new cross-disciplinary connections between the remote sensing and flux science communities to identify and address knowledge gaps and uncertainties associated with the data they collect. Attendees also recognized that increasing the accessibility, usability, and standardization of proximal remote sensing data will help researchers use them to inform flux data and will bring databases into better alignment with the FAIR (findable, accessible, interoperable, and reusable) data principles.
Five Types of Proximal Sensing to Focus On
Workshop participants identified five types of proximal remote sensing data that combined with flux data, are particularly useful for addressing ecosystem science questions. These data types include light detection and ranging (lidar), spectral reflectance, solar-induced chlorophyll fluorescence (SIF), thermal radiance, and microwave backscatter.
Lidar is a form of active remote sensing that can provide 3D views of vegetation canopy structure by emitting laser pulses and measuring the travel times of pulses reflected and returned to the sensor. Lidar is useful to scientists studying ecosystem fluxes because it offers more accurate estimates of total leaf surface area and leaf angle distribution (i.e., the distribution of leaf orientations—whether they spread out flat or hang vertically, e.g.) within a measured area. These traits affect the amount of light absorbed by the canopy and are thus relevant for understanding plants’ role in observed fluxes; they also inform the fate of observed fluxes (e.g., how much absorbed carbon becomes plant tissue).
Spectral reflectance sensors measure broad or narrow bands of visible to infrared light reflected from the canopy. This information, and derived vegetation indices, can be used to estimate plant traits that regulate ecosystem carbon and water fluxes, such as leaf pigment concentrations (e.g., chlorophyll), leaf area, and canopy water content. Collocated spectral reflectance and flux data further provide valuable opportunities to detect spectral signatures of plant stress and photosynthetic activity.
Thermal measurements of forest canopies can inform estimates of evapotranspiration and help scientists understand the effects of temperature and water stress on carbon fluxes.
SIF takes advantage of the fact that sunlight causes green plants to fluoresce. That is, solar energy promotes electrons in chlorophyll molecules into an excited energy state, and when these electrons fall back to their ground state, they release the excess energy in the form of red to far-red light. SIF is sensitive to both the amount of light absorbed by chlorophyll and the amount of photosynthesis occurring in the plant canopy. Collecting SIF measurements within eddy covariance footprints can help constrain the sensitivity of SIF to plant carbon uptake, which subsequently improves estimates of ecosystem productivity over broad spatial scales.
Thermal infrared radiation measurements are used as a proxy for latent heat flux in many remote sensing applications. In terrestrial ecosystem settings, these measurements can be used to determine leaf surface temperatures, which are directly sensitive to latent heat flux (i.e., evapotranspiration). The rates of biochemical reactions in leaves—for example, reactions controlling photosynthesis—are also sensitive to leaf temperature. Thermal measurements of canopies can thus inform estimates of evapotranspiration and help scientists understand the effects of temperature and water stress on carbon fluxes.
Microwave remote sensing methods can be active or passive. Active microwave sensors like radar emit radio signals and measure properties of the backscattered radiation, whereas passive sensors measure the amount of microwave energy emitted by Earth. The extent to which vegetation transmits microwave energy is controlled by factors such as temperature, water content, and canopy structure. Thus, microwave measurements can be used to monitor changes in aboveground biomass and relative water content that are associated with water, energy, and carbon fluxes.
Developing coordinated networks of proximal remote sensing instrumentation and data collection will expand insights into terrestrial ecosystems’ functioning and provide benchmarks to evaluate models and satellite data products.
Individually, each of these types of proximal remote sensing data provides a useful complement to flux data. However, combining them allows for evaluation of an additional suite of ecological parameters, including plant responses to temperature and water stress; carbon sink potential; and vegetation’s capability to use light, water, and nutrients efficiently.
These techniques can be applied across landscape scales to estimate plant pigments, plant stress, and other aspects of plant function relevant to estimating fluxes. Developing coordinated networks of proximal remote sensing instrumentation and data collection at towers equipped for eddy covariance measurements will expand insights into terrestrial ecosystems’ functioning and provide benchmarks to evaluate models and satellite data products.
Toward a Global Proximal Remote Sensing Network
Widespread use of proximal remote sensing techniques is limited by a shortage of funding available for installing new instrumentation, the lack of standardization of data collection across existing instrumented sites, and limited access to existing data streams. At present, intercomparisons of different proximal remote sensing data products are specifically hindered by differences in the types and fields of view of instruments used and the timing and averaging of measurements, as well as by a lack of standardization in calibration practices. Overcoming these challenges first requires collaboration and coordination among remote sensing and ecosystem flux scientists.
The community built during the recent workshop is taking steps toward making proximal remote sensing data more accessible and developing standards of practice for data collection, calibration, and sharing. We have developed—and welcome contributions to—community spaces on Slack and GitHub to track conversations and document best practices and existing knowledge. We are promoting conversations through meetups at major conferences, such as those of the Ecological Society of America and AGU. We are currently preparing a technical paper that details standards of practice and compiles existing data sources for proximal remote sensing at flux towers. And we are working with existing networks, such as FLUXNET, AmeriFlux, and SpecNet, to integrate proximal remote sensing data into flux databases.
These grassroots efforts have been successful in engaging an international community of scientists—the 40 attendees at the workshop represented 12 countries, for example. Yet the biggest barriers to advancing a proximal remote sensing network involve time and money. The instrumentation is costly to purchase and install. And even when external funding is provided to cover installation and initial operations (typically for 2–3 years), there is often limited or no additional funding available to maintain sites after that initial period ends.
The result of this patchy funding and lack of coordination is an ad hoc array of data collected with varying instruments and strategies over short temporal windows by individual groups who are open to sharing. Coordinating and compiling existing data streams into a standardized network—and then maintaining this network—are a major undertaking that will require additional and sustained funding and support at federal or international levels. Because of this need, our community submitted a letter to the U.S. Department of Agriculture responding to the Federal Strategy to Advance Greenhouse Gas Measurement and Monitoring for the Agriculture and Forest Sectors. The letter outlines specific ways in which support for proximal remote sensing will advance preparedness for climate change.
We hope to continue strengthening the community of scientists working at the intersection of ecosystem flux science and remote sensing science.
Further, we hope that federal initiatives promoting open science, such as NASA’s Transform to Open Science (TOPS), will motivate researchers to make existing and future proximal remote sensing data sets publicly available. Integrating proximal remote sensing products into networks such as AmeriFlux, FLUXNET, the Integrated Carbon Observation System, the Terrestrial Ecosystem Research Network, AsiaFlux, ChinaFLUX, and the National Ecological Observatory Network will also be a major step in improving the accessibility and interpretability of these data.
Looking ahead, we also hope to continue strengthening the community of scientists working at the intersection of ecosystem flux science and remote sensing science. Through greater community collaboration and engagement, we can build standards of practice and make data more accessible to advance vital science that helps us understand the current state of Earth’s climate and prepare for changes to come.
Acknowledgments
The authors of this article were the main organizers of the Linking Optical and Energy Fluxes Workshop. However, this article is the result of the intellectual contributions and discussions held at and following the workshop. Therefore, we acknowledge the following individuals for their contributions to the ideas presented in this article: Abdallah Abdelmajeed, Sergio D. Aguirre-García, Michal Antala, Rachel Badzioch, Mallory Barnes, Julia Bigwood, Adriana Uscanga Castillo, Kerry Cawse-Nicholson, Matthew Dannenberg, Charlie Devine, Jen Diehl, Erica Edwards, John Gamon, Koen Hufkens, Miriam Johnston, Tommaso Julitta, Aram Kalhori, Christopher Kibler, Gerbrand Koren, Yujie Liu, Jess Lyons, Riasad Mahbub, Hannah Mast, Karem Meza, Gilberto Pastorello, Kevin Postma, Will Richardson, Elizabeth Arango Ruda, Ben Runkle, Eleanor Serocki, Bill Smith, Charlie Southwick, Christopher Still, Giulia Tagliabue, Rebeca Campos Valverde, Xinlei Wang, Ben Wiebe, Christopher Wong, Will Woodgate, and Koong Yi.
We also thank the FLUXNET Co-op and the AmeriFlux Year of Remote Sensing for financially supporting the workshop and the Niwot Ridge Mountain Research Station in Nederland, Colo., for hosting us.
Z.P. was supported by an appointment to the NASA Postdoctoral Program at the Jet Propulsion Laboratory. A portion of this research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA © 2023. All rights reserved.
Author Information
Zoe Pierrat ([email protected]), Jet Propulsion Laboratory, California Institute of Technology, Pasadena; Troy Magney, Department of Plant Sciences, University of California, Davis; Xi Yang, Department of Environmental Sciences, University of Virginia, Charlottesville; Anam Khan, Department of Forest and Wildlife Ecology, University of Wisconsin–Madison; and Loren Albert, Department of Forest Ecosystems and Society, Oregon State University, Corvallis
