Continuous CO2 fluxes in the Cassiope heath

This blog post was co-authored by Dr. Magnus Lund, Gillian Simpson and Zhila Rashidi (Eosense)


This summer, researchers from the GeoBasis team at Aarhus University (Denmark) took six eosFD soil CO2 flux sensors to the Zackenberg Research Station in north-eastern Greenland to conduct continuous measurements of soil CO2 fluxes in the Cassiope heath. The following is our latest update from the team…


Prior to the start of the field-season, we conducted analyses of the historical eddy covariance (EC) footprint. These analyses showed that during the summertime the majority of fluxes measured by our EC setup are typically sourced from a distance of around 30-50 m south-east of the tower; an area dominated by the Cassiope heath plant community type. This vegetation type (prostrate/hemiprostrate dwarf shrub tundra) covers approximately 10% of the high and middle Arctic areas. In order to adequately sample this area, we planned to install two stationary chambers where the majority of the flux measured by our EC tower is sourced, with the remaining four chambers being moved to new locations on a bi-weekly basis. The locations of our ten measurement sites and the dates of each placement period are indicated in Figure 1 and Table 1 respectively.


Figure 1: Map of the area around the eddy covariance tower showing the vegetation classification and indicating our pre-selected sites at which the eosFD chambers were deployed.


Table 1: List of placement periods and site location of each chamber during our field-campaign. Note that some problems were experienced with Chamber 4 during Placement 4, meaning that no measurements were made here. The last column provides average chamber flux per setting period. Note these averages were calculated using only 15-minute measurement intervals where three or more chambers were operating.



eosFD measurements

Positive daytime temperatures were recorded at Zackenberg from early June, however it was not until the 25th June 2017 that daily mean temperatures rose above zero. Collars were inserted into the soil on the 27th June, and field-measurements commenced a week later on the 2nd July 2017, when the area was free of snow and the ground was dry enough for chamber placement. Mean daily temperature here was 1.9 °C, and soils were thawed to around 20 cm depth. During the first two weeks of our field campaign (Placement 1), soil CO2 efflux measured by the chambers was very small (mean: 0.05 ± 0.07 µmol  m-2 s-1) and fluctuated around the zero line. The smallest mean soil respiration flux for this period was measured by Chamber 1 (-0.005 ± 0.07 µmol CO2 m-2 s-1), whereas the largest fluxes were measured by Chamber 4 (mean: 0.14 ± 0.06 µmol CO2 m-2 s-1). On the 14th July chambers were relocated. Chambers 4 and 5 were inserted at Site 2 and were to remain stationary for the rest of the field campaign, whereas the other four chambers would continue to be moved on a bi-weekly basis. Soil CO2 flux rates during placements 2-5 were considerably larger in magnitude than observed for Placement 1 (see Table 1) and were predominantly positive in sign, which indicates that soils in our measurement plots were actively respiring CO2 to the atmosphere. On average, Placement 4 (10th – 21st August) was characterised by the largest average soil CO2 efflux, with rates of up to 0.39 µmol  m-2 s-1 recorded on the 15th August. A sudden reduction in flux rate to around zero was observed on 18th August, before rates increased and then gradually decreased over the fifth and final placement period.

Looking at Figure 2, it is clear that soil respiration rates measured by the chambers displayed strong spatial and temporal variability. In terms of the latter, anomalously high flux values were observed for chambers during the initial 24 hours of each new placement period. We expect this is a result of site disturbance after collar insertion, as no such pattern is seen for chambers 4 and 5 which were quasi stationary. Maximum fluxes outside of these collar insertion periods were seen on the 30th July, where soil respiration rates measured by Chamber 6 peaked at 0.86 µmol CO2 m-2 s-1.

In terms of spatial variability, during Placement 3 we saw a strong contrast in the magnitude of fluxes measured at sites 5 and 6 (see Figure 2). The average chamber flux measured at Site 5 (mean: 0.38 µmol CO2 m-2 s-1) was a factor of six greater than measured at Site 6 (mean: 0.06 µmol CO2 m‑2 s‑1). This is likely due to microsite-scale variability in soil characteristics within the EC footprint, with soils in Site 6 being more clayey than those found at Site 5 (see Table 2). Our soil respiration measurements also show considerable spatial variability within sites. For example, at Site X4 fluxes measured by Chamber 3 were on average larger and more variable (mean: 0.37 ± 0.22 µmol CO2 m-2 s-1) than fluxes measured where Chamber 2 was installed (mean: 0.12 ± 0.08 µmol CO2 m-2 s-1).

Figure 2: Temporal evolution of soil CO2 effluxes from our eosFD chambers, with the different placement periods (1-5) indicated by the vertical dashed-lines. The upper plot shows a time-series for each of the six chambers, with colours indicating different measurement sites, and the lower plot shows an average of all chambers taken at our 15-minute measurement frequency for periods when more than two chambers were operational. Note the power outage towards the end of Placement 1 due to a change-over in researchers at the station, where only chambers 3 and 4 were operational.



Figure 3: The sun never sets and the measurements never stop – it’s midnight at the start of Placement 4 and here’s a pair of eosFD chambers that we installed to the south-west of our eddy covariance tower (Photo: Gillian Simpson)


Table 2: Inserted collars at a selection of measurement sites. Photos highlight the differences in soils between sites 5 and 6 during Placement 3; and within site X4 during Placement 5.


Comparison with manual chamber measurements

In addition to the automatic eosFD chamber measurements, over the course of the field-campaign we also conducted over 80 manual chamber measurements with our EGM-5 CO2 gas analyser (PP-Systems, USA). These additional measurements provide us with valuable small-scale data which can be directly compared with our eosFD measurements. It is important to note that measurements with the two systems were not conducted at the same sites, although they were made in close proximity. A quick comparison of the results is given in Figure 5, which shows measurements made with the eosFD chambers in blue, and those with the EGM chamber in red. Generally our EGM soil respiration estimates are higher than those from the average eosFD chamber, however this may be due to spatial variability, and the majority of our EGM measurements do lie within the range of fluxes estimated by the individual eosFD chambers (light blue).

Figure 4: Our PP-Systems EGM-5 dark chamber (Photo: Magnus Lund)

Figure 5: A comparison of chamber-based soil respiration measurements from inside our EC footprint. Shown are data from our eosFD chambers (blue markers) and our manual EGM-5 measurements (red markers)

Comparison with EC estimates

As our EC system measures the net ecosystem exchange (NEE) – the difference between Gross Primary Production (GPP) and ecosystem respiration (Reco) – we cannot directly compare EC flux estimates with those from our soil respiration chambers. We can however partition our NEE fluxes to provide a rough estimation of Reco, by examining the so-called night-time or dark NEE. Previous studies have found that soil respiration contributes a significant portion of Reco, and for Zackenberg the size of this contribution has been estimated at around 40-60% (e.g. Christiansen et al., 2012).

Our preliminary analysis of dark NEE using light response curves indicates that soil respiration as measured by our eosFD chambers is much smaller than our Reco estimates for the EC footprint. For example, over the period 11th – 20th July we estimate a dark NEE rate of 1.23 µmol CO2 m-2 s-1, whereas the mean soil respiration flux measured by our chambers was only 0.17 ± 0.08 µmol CO2 m-2 s-1. Similarly, for the 21st -30th July and 31st July – 9th August we estimate a dark NEE of around 0.98 µmol CO2 m‑2 s‑1, whereas mean soil CO2 efflux estimates from our eosFD chambers were substantially lower (0.20 ± 0.10 and 0.14 ± 0.08 µmol CO2 m-2 s-1 for the two periods respectively). These preliminary data appear to suggest that respiration from additional sources (e.g. autotrophic respiration from the shoots of the Cassiope heath plant community) likely contributes significantly to Reco inside our EC footprint.  However, it is important to remember that the above is a very basic comparison of two different flux quantities, measured at two different scales – one at the microsite scale spanning only bare soil patches, and one at the ecosystem scale which incorporates both vegetated and non-vegetated areas.

Next steps

We are continuing to analyse the data we collected this summer and plan to use them to support a model of soil respiration inside the EC footprint. These data will be extremely valuable in helping to partition Reco at Zackenberg; allowing for assessment of the permafrost carbon feedback and the refinement of ongoing modelling efforts.


Christiansen, C.T., Svendsen, S.H., Schmidt, N.M. and Michelsen, A., 2012. High arctic heath soil respiration and biogeochemical dynamics during summer and autumn freeze‐in–effects of long‐term enhanced water and nutrient supply. Global Change Biology18(10), pp.3224-3236.


For more information on technical specification of eosFD please download our brochure.

Related Posts Plugin for WordPress, Blogger...