Download the complete article (PDF).


Globally, soils are the largest terrestrial contributor of carbon (C) flux to the atmosphere. Carbon is released from soils mainly in the form of carbon dioxide (CO2) emitted by microbial and plant processes. The production of CO2 in soils is closely connected to physical drivers like temperature and soil water content, which may change in response to global climate change.

Ecosystems like heathlands and grasslands cover large portions of the Northern hemisphere and store large amounts of C. These ecosystems experience harsh winter conditions with multiple freeze-thaw cycles and other climatic events that significantly impact CO2 emissions. While overwinter CO2 fluxes have traditionally been assumed to be negligible, recent research has shown winter emissions can contribute significantly to the annual ecosystem C balance.


At the CLIMAITE climate change research facility in Brandbjerg, Denmark, researchers from the University of Copenhagen are studying C dynamics under present and manipulated climate conditions.

Brandbjerg is located 60 km west of Copenhagen on a former glacial outwash plain overlain by wind deposited sand. The soil is very well-drained and nutrient poor. The site is characterized as a heath- and grassland co-dominated by heather (Calluna vulgaris) and grass (Deschampsia flexuosa). The climate at the site is temperate with a mean annual temperature of 8°C and annual precipitation close to 800 mm with about 5-10% falling as snow.

Starting in 2016, an experimental set-up was introduced with a series of permanent rainout shelters removing 40%, 50% or 66% of the annual precipitation.

The Brandbjerg site is part of the AnaEE Denmark consortium (

The eosFD – standalone soil CO2 flux sensor that uses Forced Diffusion technology


At Brandbjerg the aim is to understand how future extreme drought, a consequence of climate change, impacts ecosystem functioning and the exchange of C. These impacts are studied across an experimental drought gradient to investigate the shape and magnitude of the C exchange response.

In the current study eosFD chambers were used to better understand the soil CO2 fluxes in late autumn and through the winter season. As part of the study, soil-atmosphere CO2 fluxes were observed in an area dominated by Deschampsia flexuosa. The eosFD has been especially helpful with measuring in detail the temporal variability of CO2 fluxes during a period of the year where good estimates of soil respiration are often lacking. These data can help disentangle the soil temperature and moisture control on soil-atmosphere exchange of carbon in cold periods, especially periods with snow cover when other methods of respiration measurement are prone to equipment failure.

eosFD’s measuring CO2 flux at the CLIMAITE facility


From October 30th to January 11th, CO2 fluxes were also measured with an eosAC system connected to an LGR UGGA laser-based greenhouse gas analyzer and the Eco2Flux system developed by University of Copenhagen researchers. The comparison data are presented below in Figure 1.

Figure 1. CO2 fluxes over time for the eosFD chambers (light blue and orange) as well as 3 eosAC automated chambers (red, green, yellow) monitored using the LGR UGGA analyzer. Finally ecosystem fluxes (soil and plant respiration) using the Eco2Flux chamber are shown in black.

The CO2 fluxes from all systems show similar temporal responses to changing environmental conditions as well as a similar magnitude of CO2 flux across the spatially distributed locations. As expected the total ecosystem respiration measured by the Eco2Flux chamber is consistently higher than the soil CO2 emissions measured using both the eosFD and eosAC methods, suggesting that plant respiration is an important contributor to the total CO2 emissions during the late autumn and winter. More details on the eosAC results can be found in the “Experimental Drought Reveals Soil CO2 and CH4 Flux Dynamics under Climate Change” Case Study.

eosAC chambers monitoring the control plot (Oct. 30 – Dec. 12)


CO2 fluxes were measured using eosFD chambers through the winter, until approximately March 26, 2018.

Figure 2 and Figure 3 show the response of CO2 flux to temperature over the whole monitoring period. Mean fluxes from eosFD unit #1 and unit #2 are significantly different over the monitoring period, but this is attributable to spatial heterogeneity at the site (demonstrated in Fig. 1).  Fluxes decrease in tandem with temperature during the measurement period. Periods of temperature increase and decrease are highly correlated with increases and decreases in flux, although temperature effects only explain between 70%-80% of the total variation observed over the monitoring period (Figure 4, following page).

Figure 2. CO2 fluxes and soil temperature for eosFD #1. The dark blue is a running average of eosFD measurements. Periods with a white background on the plot indicate when air temperatures were below zero.

Figure 3. CO2 fluxes and soil temperature for eosFD #2. The dark orange is a running average of eosFD measurements. Periods with a white background on the plot indicate when air temperatures were below zero.

Figure 4. CO2 fluxes from both eosFD #1 (blue) and #2 (orange) against soil temperature (0-5 cm). Dark lines show the best-fit for a Q10 function.

Dr. Jesper Christiansen in his rain gear checking on the eosFD and eosAC chambers located in the control plot.

Moisture changes are likely to explain much of the additional variability in the data, however they covary strongly with temperature (Figure 5) which makes it difficult to separate the effects. Other researchers have noted a strong moisture response in winter conditions, where availability of liquid water is the main controller of respiration processes.

Figure 5. CO2 fluxes from eosFD #1 (blue) alongside soil moisture (0-10 cm, grey) and soil temperature (0-5 cm, green). Strong covariance makes separation of the impacts of moisture and temperature difficult.


Understanding C cycling in the winter is critical for developing an accurate carbon balance for Northern ecosystems. The eosFD’s low power consumption and membrane-based approach to flux measurement make it ideal for use in harsh conditions, including under snow cover. The results here highlight the need for further study of the covarying moisture and temperature controls on soil respiration in the winter months.


Thanks to Dr. Jesper Christiansen and Dr. Klaus Steenberg Larsen for performing the measurements and analyzing the data associated with this study.