By now, the general public is well aware of greenhouse gases in Earth’s atmosphere and the overall warming effect that burning fossil fuels has on the world’s climate. Less publicized is the effect of a warming climate on the permanently frozen ground—permafrost—in the far northern regions of the world. And still less is known about how gases released by thawing permafrost soils could accelerate the very climate change that caused them to thaw in the first place.
Permafrost soils contain twice as much carbon as is currently present in the atmosphere, and the emission of just a fraction of permafrost carbon would strongly amplify the ongoing warming of the planet. Despite this worrying possibility, we presently lack the ability to detect changes in carbon emissions across the Arctic and boreal region in a timely manner.
A new initiative led by the Permafrost Carbon Network aims to resolve this shortcoming. This synthesis activity, supported in part by the Arctic Data Center, will develop a comprehensive repository for net carbon dioxide (CO2) flux data collected across the northern permafrost zone.
This project builds on the efforts of countless scientists, across decades, who weathered swarms of mosquitoes and harsh weather to make precise measurements of the carbon exchange of treeless tundra, boreal forests, and permafrost soils. Although these measurements tell us a great deal about single sites, the larger picture remains much more elusive.
Previous efforts that synthesized this wealth of data disagree on whether the Arctic is presently a net sink or source of CO2 [Belshe et al., 2013; McGuire et al., 2012]. Although valuable insight has been gained, this basic question about the carbon balance of the permafrost zone remains unresolved.
The direction in which net CO2 exchange will develop with future warming is equally, if not more, uncertain. In general, warmer soils respire more CO2, and thawing permafrost releases carbon that can add to this effect. At the same time, warming also enhances the uptake of CO2 by plants, which at least partly compensates for the loss of soil carbon. Large interannual variations in these two opposing processes make it challenging to identify shifts in net CO2 exchange. Decadal time series, constructed from as many sites as possible, are necessary to determine trends for the entire permafrost zone.
A Sink of Data, a Source of Knowledge
To tackle these issues, we welcome flux observations from Arctic and boreal ecosystems in our data repository. For an essential, comparative perspective, we include data from ecosystems within and outside the permafrost zone. The evolving database will include measurements spanning 3 decades, with the intention to incorporate new observations as near to real time as possible. The fast response time makes it possible to report whether deviations from the historical baseline are occurring.
Such a database is long overdue. The scientific community is currently unable to communicate regional changes in permafrost carbon in the same way as for other components of Earth’s cryosphere. For example, the retreat of glaciers, melting of ice sheets, shrinking of sea ice, and increase in permafrost temperature are all well documented and regularly reported [e.g., Box et al., 2019]. Part of the reason for this difference in up-to-date information is an inability to detect permafrost carbon at the circumpolar scale using remote sensing technologies, and syntheses of site data occur sporadically rather than systematically.
The synthesis activity in our project differs in its scope from existing efforts such as FLUXNET, which has been highly effective at integrating measurements made with the eddy covariance technique. Eddy covariance tower data are considered the current gold standard because of their high temporal and landscape-scale coverage, but this technique will always be limited to fewer locations because it is cost intensive.
In contrast, flux chambers are used to observe carbon exchange in many more locations but are temporally limited. The data they collect are not systematically archived in a central database. The goal of the new database is to compile biweekly, monthly, or seasonal aggregates of net CO2 flux data collected with both eddy covariance and flux chambers, which collectively allow a picture of the whole region to emerge, at timescales relevant to climate and model projections.
Bridging Time and Space
Initial analysis will reveal valuable insights in decadal trends of net CO2 exchange and the response to years with extreme temperatures or precipitation. To minimize uncertainties in these analyses, a number of hurdles involving such factors as spatial variation, fragmented time series, and the representativeness of the collected data need to be cleared. This is challenging because permafrost landscapes are highly diverse with distinct microtopography: Dry elevated areas and wet depressions alternate across short distances.
Surface wetness has a strong influence on carbon exchange. Soil respiration rates are generally lower under wet conditions because of the lack of oxygen in the soil, which is why wet ecosystems are often hot spots for CO2 uptake. The availability of water also controls the types of vegetation present: Dwarf shrubs are typically found in dry elevated areas, whereas sedges are common to wet areas. Sphagnum and other moss species also follow moisture gradients. These differences in ecosystem composition lead to varying rates of plant litter input and decomposition. Thus, the spatial distribution of soil carbon is strongly tied to the interplay of geomorphology, surface hydrology, and ecosystem composition.
Capturing the complexity of nature is a colossal task in itself, but the locations and times at which scientists have aimed to do so introduce additional challenges. High-latitude fieldwork has historically been concentrated around a few research stations in Scandinavia and Alaska, which has led to spatial biases where colder environments are poorly sampled [Virkkala et al., 2018]. An analysis of past trends in CO2 exchange, if done incorrectly, may be more representative of the North Slope in Alaska than the entire permafrost zone. This patchiness is clear from a recently launched mapping tool of northern flux stations (Figure 1).
Also, flux measurements have rarely been continuous in time because of financial and logistical constraints. Existing snapshots need to be stitched together while accounting for seasonal biases: Most fieldwork is concentrated in summer, even though the cold season can last up to 9 months. And nongrowing season emissions may be crucial in determining whether the northern permafrost zone is a net source of CO2.
Given these challenges of scale, we need sufficient metadata to place flux measurements from individual locations into a larger context. What was the water level, thaw depth, and soil temperature at the time of measurement? Which types of plants were present, and in what abundance? Furthermore, do the chamber flux measurements cover vegetation types that are representative of the larger region? We propose a consistent framework for metadata, which will make it possible to match flux measurements across space and time and to upscale them to the whole of the permafrost zone. Future steps may also include instituting a more automated procedure to incorporate publicly available metadata.
This database is a first step to further understanding of the permafrost carbon feedback. Methane and nitrous oxide (N2O) emissions and lateral (land-to-water) carbon losses are currently excluded. Measurements over lakes and rivers are also not considered. However, terrestrial CO2 emissions represent the bulk of the potential release of greenhouse gases from permafrost thaw, which is why the monumental challenge of capturing the state of the permafrost carbon pool needs to start there. Other synthesis efforts of methane and N2O are underway, and parallel analyses will provide more comprehensive insights.
Observing the Present to Predict the Future
Beyond our intent to better understand the natural system, this activity will also highlight the economic and societal value of maintaining a flux observation network across the permafrost zone. For example, the Paris Agreement aims to keep warming of the planet this century to well below 2°C, preferably below 1.5°C. Carbon release from permafrost may hamper the feasibility of achieving that goal, even though the likelihood of different warming scenarios is highly uncertain. This uncertainty has consequences for the accuracy of economic projections of the cost of climate change.
By acting as a benchmark, the flux database will lead to more realistic projections that will indicate which future scenarios are unlikely. Focusing on the most likely scenarios lowers costs by reducing the necessity for policy makers to prepare for all possible eventualities. An improved quantification of future permafrost carbon loss can therefore represent an economic benefit.
This database will be a major step in determining how the rapid changes in the Arctic may affect the permafrost carbon feedback. Many in the scientific community have already contributed, and we hope you will join us in this collective effort to deliver the most comprehensive and up-to-date view of the net CO2 exchange of the northern permafrost zone.
We thank the many researchers who have already contributed data and other input to expand our initial data sets. We also thank all workshop participants who helped put together the ideas and initial data for this synthesis. In particular, we thank Brandon Rogers for putting a lot of thought into our initial metadata format and Gerardo Celis for programming and data support. And we thank M. Goeckede, G. Celis, and M. Pallandt for their hard work in developing the carbon flux sites mapping tool shown in Figure 1. We acknowledge the Arctic Data Center for providing workshop funding, as well as the National Science Foundation’s collaborative research award (grant 1331083): Research, Synthesis, and Knowledge Transfer in a Changing Arctic: Science Support for the Study of Environmental Arctic Change. F.-J.W.P. received additional support from the Norwegian Research Council under grant agreement 274711 and from the Swedish Research Council under registration number 2017-05268.