Study sites spread across the continuous permafrost zone in Greenland. Red circles represent the 6 areas where 21 different samples of organic permafrost soils were collected for the analysis. The location of the Disko Bay region is shown as a black square. Figure courtesy of Kent Pørksen, Univ. Copenhagen.
Study and sampling sites.
We used 21 samples that were collected from 16 contrasting organic permafrost soils located in 6 different areas in Greenland that were all within the continuous permafrost zone (Fig. 1). In the model, we focused on two contrasting sites: a kitchen midden at Qajaa in West Greenland and a peatland at Zackenberg in northeast Greenland.
The Qajaa midden, situated 18 km southeast of Ilulissat in the western central part of Greenland, was selected as the main study site for investigating the coupling between climate, soil temperatures, decomposition and heat production. The midden has been known at least since 1871 and is considered the site with the best preserved organic remains from the Palaeo-Eskimo Saqqaq and Dorset cultures in all of Greenland11, 25. It covers an area of approximately 2,900 m2, has a maximum thickness of 3 m and consists of peat as well as rocks from fireplaces, animal bones and wood11. Owing to its great historical value, the midden has been monitored since 2009 to evaluate current and future preservation conditions11, 25, 26. The climate in the area is arctic with a mean annual temperature of −4.5 ± 1.7 °C (1974 to 2004) and a mean annual amount of precipitation of 266 mm (1961 to 1984; ref. 27).
The peatland is situated in the Zackenberg Valley near the Zackenberg Research Station in central northeast Greenland (74° 30′ N; 20° 30′ W). The climate is high arctic, with a mean annual air temperature of −9.1 °C and a mean annual amount of precipitation of 220 mm, with 90% falling as snow and sleet. Zackenberg is located in the continuous permafrost zone, and the permafrost thickness has been estimated to be around 400 m (ref. 28).
Sampling and analyses.
Sampling was based on excavation of pits down to the frost table followed by drilling to obtain intact permafrost cores. Top permafrost cores were collected by motorized hand-drilling equipment consisting of a Stihl drilling engine, an expandable drill string and a 40-cm-long core barrel with a drill head. Sample lengths from 3 to 30 cm were packed in plastic bags and kept frozen. The heat production was measured at 16 °C using a thermal activity monitor (type 2277, Thermometric, Sweden, or C3-analysentechnik, Germany) equipped with ampoule cylinders (4 ml twin, type 2277-201, and 20 ml twin, type 2230). Oxygen consumption rates were measured by monitoring the decrease of headspace O2 concentrations over time using oxygen optodes (PreSens) in three replicates 14. Three replicate sub-samples (1–2 g) of humid soil were transferred to 12.1 ml glass vials flushed with atmospheric air and the vials were sealed carefully using a disc of transparent commercial oxygen barrier film (Escal), a silicone gasket and a screw cap with aperture.
Modelling.
Atmospheric conditions were governed by time-variant meteorological inputs, and the model incorporated interface processes from snow and vegetation cover at the boundary between atmosphere and soil. The model regime consisted of a 20-m-deep profile divided into 82 layers. The upper 55 layers (3 m) were considered to be completely organic except for the upper 20 cm organic-rich top soil layer. The soil below 3 m was considered similar to the entisol/cryosols normally found in these parts of Greenland29. The temperature was kept constant at the lower boundary of the model regime. The thermal conductivity (kh) was calculated as a function of soil solids and soil moisture on the basis of empirical equations adjusted to accommodate observations on the volumetric water content from previous investigations30. The unfrozen and frozen values of kh were adjusted to match measured values (Supplementary Fig. 5). For the layers below 3 m, values of kh were based on default values for mineral soils. To adjust the thermal conductivity individually for each soil layer, a scaling coefficient was applied17.
Total modelled soil respiration R was the sum of three pool-specific respiration rates, each of which was simulated (see Supplementary Methods Equation (1)) as the pool-specific decay rate multiplied by the total initial carbon pool multiplied by a fractionation coefficient that describes the ratio of the carbon pool to the total carbon pool (see Supplementary Methods Equation (2)). The heat production used in the model was the mean value of all measurements (excluding the high values from Disko that were considered as outliers).
Climate change scenarios used in the CoupModel to predict future ground temperatures were based on IPCC RCP4.5 (ref. 18) with the increase in air temperature by 2100 relative to the 1986–2005 mean and an Arctic amplification by a factor of two19. The two scenarios are: a low range with a summer warming of 1.1 °C and winter warming of 3.3 °C; and a high range with a summer warming of 4.2 °C and winter warming of 6.2 °C.