Few hydrological studies have been conducted in Greenland, other than on glacial
hydrology associated with the ice sheet. Understanding permafrost hydrology
and hydroclimatic change and variability, however, provides key information
for understanding climate change effects and feedbacks in the Arctic
landscape. This paper presents a new, extensive, and detailed hydrological and
meteorological open access data set, with high temporal resolution from a
1.56 km
Future climate change is expected to be most pronounced in the Arctic region (Kattsov et al., 2005), and the terrestrial and aquatic ecosystems in the Arctic are predicted to undergo fundamental changes in the coming century (Vaughan et al., 2013). To enable predictions of the ecological response to changes in the hydrological and biogeochemical cycles, the understanding and prediction of these responses on the landscape scale have to be improved (Rowland et al., 2010). Hydrology is the key driver for transport of matter within and between ecosystems. To describe the potential future impact on periglacial areas, related to changes in the hydrological cycle in the Arctic, the understanding of present conditions needs to be improved. Woo (2012) summarizes the present knowledge of permafrost hydrology, and in Kane and Yang (2004), water balances from 39 Arctic catchments are presented. However, only few hydrological data sets from Arctic regions exist that contain necessary information on both surface and subsurface waters within the same catchment. This is due to the inaccessibility of these remote areas and the harsh climatic conditions there, but it is also the effect of the relatively short period of the year when unfrozen water is present and can be monitored.
In this work, we present hydrological data obtained from a detailed site investigation performed within the framework of the Greenland Analogue Surface Project (GRASP). The study area, here referred to as the “Two Boat Lake” (TBL) catchment, is located close to the Greenland ice sheet in the Kangerlussuaq region, western Greenland (Fig. 1). The investigations were initiated in 2010, and have resulted in a wealth of information on hydrology, meteorology, geometry (surface topography, lake bathymetry, and regolith and active layer depths), properties and spatial distributions of vegetation and soils, hydrochemistry, soil temperature, and geochemical processes and properties of the limnic and terrestrial ecosystems within the catchment. The aim of GRASP is twofold. The first part is focused on how ecosystems develop and react in a long-term climate change perspective during an entire glacial cycle (Lindborg et al., 2013). The other part is aimed at improving the understanding of water exchanges between surface water and groundwater in a periglacial environment (Bosson et al., 2012, 2013).
The regional area of Kangerlussuaq and the TBL catchment, installations, and sampling points.
The objective of the present paper is to describe the hydrological and meteorological installations and measurements in the TBL catchment, and to make the data set available to the scientific community. The aim of the hydrological field studies presented here was to identify and quantify the main hydrological processes in a periglacial lake catchment, thereby providing input to conceptual and mathematical modeling. In coming studies of the TBL site, this hydrological and meteorological data set will be used as a basis for the modeling of hydrological, biogeochemical, and ecological processes in the catchment.
The Kangerlussuaq region, comprising a hilly tundra landscape with numerous
lakes, is the most extensive ice-free part of Greenland. Continuous
permafrost, interrupted by through-taliks under larger lakes, covers the area.
A permafrost depth greater than 300 m has been observed from temperature
measurements in deep bedrock boreholes,
The TBL catchment, with an area of 1.56 km
The lake is underlain by a through-talik, an observation based on
temperature data from a 225 m long borehole drilled in a 60
Besides the scientific questions to be answered, the measurement program was, to a large extent, determined by what was possible, given the hydrologic, climatic, and logistic conditions at the site. Due to its remote location, but also depending on the harsh climate, the site has only been manned during relatively short periods. This limited the possibilities for long-term or continuous observations of some parameters that could not be measured automatically (e.g., surface water inflow to the lake or groundwater monitoring in the active layer). Lack of infrastructure for electrical power and telecommunications has also been a limiting factor.
Hydrological processes and conditions not considered in temperate climate regions may be of great importance in periglacial areas. Hydrological responses in these cold areas differ from catchments in boreal and temperate regions in fundamental aspects (Kane and Yan, 2004). Most importantly, the hydrology in periglacial environments is intimately connected to the presence of permafrost (White et al., 2007) and active layer dynamics. Snow-related processes (wind drift of snow and sublimation) have also been shown to be of great importance for the annual water balance (Reba et al., 2012; MacDonald et al., 2010).
The overarching target when planning the TBL field program was to identify and quantify the main hydrological processes including the interactions between surface water in the lake and in the surrounding catchment, and the role of both supra- and sub-permafrost groundwater. The dry climate and the very limited surface water runoff in the area imply that conventional discharge measurements are not meaningful; such measurements are otherwise a cornerstone in water balance studies (e.g., Johansson, 2008; Bosson et al., 2010). Hydrological processes and major events on different spatial scales have been captured by time-lapse cameras installed in the catchment. Figure 1 shows the hydrologically related installations and sampling points of the GRASP field program, and Table 1 summarizes all data presented in the present paper. An automatic weather station (AWS) constituted the platform for the hydrological investigations together with a monitoring program for soil water content, soil temperatures, groundwater levels in the active layer, and lake surface water levels. Data from all monitoring equipment, except from the AWS, were collected manually during field campaigns, either by manual measurements or by retrieving data from loggers. Typically three field campaigns per year (in April, June, and August–September) have been organized during the period for which data are presented.
(a) List of equipment used, start of monitoring, time resolution, and accuracy for each parameter for all monitoring data.
The automatic weather station (AWS) installed in April 2011.
The AWS, labeled KAN_B, was installed at a 70 m distance
from the lake on 13 April 2011 (Fig. 2). The station is similar to
approximately 20 other AWSs installed on the ice sheet within the framework
of the Programme for Monitoring of the Greenland Ice Sheet (PROMICE) (Van As
et al., 2011), of which another 3 are located in the Kangerlussuaq
region (Van As et al., 2012). The station at TBL measures air temperature
(
An important addition to the station, compared to the PROMICE stations
located on the ice sheet, is a precipitation gauge capturing both snow and
rain. The snow is melted in the gauge and the snow water equivalent (SWE) is
measured. A Geonor T-200B was used
(
Maintenence of the AWS occurred once a year and entailed reading out the high temporal resolution data from the logger, and replacing sensors for recalibration. On every visit, the liquid content of the precipitation gauge was removed before a replacement mixture of ethylene glycol and methanol was added.
Observations in the TBL area show that the snow cover also decreased during
some periods of freezing temperatures, especially during clear windy days in
late winter. This indicates that sublimation could be a process of
importance. Therefore, sublimation measurements were performed at three
sites within the catchment (Fig. 1) during three clear and sunny days in
April 2013. Even though measurements were performed during a short time
period, they are judged to give a good estimation of the potential mass loss
via snow sublimation. For each site, five white pans (0.34 m
The same basic methodology was used when measuring evaporation in August 2013, i.e., the same pans were placed at the same sites during 3 days in August and the mass loss was measured. In this case, the boxes were filled with water and the mass loss due to evaporation was measured every day. No rain or snow was observed during the measuring period, which was characterized by variable clear and cloudy skies.
The snow depth and associated snow water content were manually measured in April 2011. The catchment was subdivided into four domains: (i) areas without snow cover, (ii) intermediate areas with snow cover broken up by spots without snow in uphill areas and on steep slopes, (iii) valleys with snow cover, and (iv) the lake ice. Based on this classification, snow samples were taken along one transect in each domain, except domain type (i) (the transects are shown in Fig. 1). The snow depth was measured every 5 m along all transects. At every fifth depth location, snow samples were collected using a standardized snowtube sampler with an inner diameter of 4.5 cm. The snow samples were weighed in the field using portable scales. The corresponding snow water contents were calculated using the known sample volumes.
Surface water levels were monitored using pressure transducers placed at the lake bottom, both in TBL and in the lake located northwest of TBL (Fig. 1). The northwestern lake, which is located at an elevation of approximately 19 m above the TBL water level, is monitored with the purpose of getting a reference lake level fluctuation in the area. This is also a precipitation-driven lake but belongs to a different surface water system and discharges into a lake situated south of TBL. Details of equipment and measuring period are given in Table 1a. The transducers were placed at a minimum depth of 5 m to avoid disturbance from ice during the winter. A steel wire connected each transducer to the shore where it was firmly attached. Total pressure (hydrostatic and barometric) and temperature were logged every third hour.
The shallow active layer depth, in combination with the dry climate, results in unsaturated conditions in most of the active layer. This makes monitoring of groundwater levels complicated. Fifteen groundwater wells (of type PEH50, with inner diameter 41 mm) were installed in the catchment to enable groundwater level monitoring and water sampling. The groundwater wells were primarily installed in local discharge areas along the valley's floor (Fig. 1) in order to maximize the possibility of getting the required volumes of water needed for chemical analyses.
The wells, consisting of a bottom plug and a fully screened pipe section, were installed with the bottom plug reaching the permafrost surface; the bottom plugs typically were at a depth of a few decimeters below ground surface. The ground surface and the top of casing (TOC) levels of each well were determined relative to the lake level (Table 2). Since no exact elevation reference point is present in the area, the reference level for all groundwater wells is the lake level at the start of monitoring in September 2010. The groundwater levels in the wells were manually measured during field visits, and in the summer of 2013 the groundwater levels were automatically monitored by pressure transducers in 10 of the 15 wells (Table 2). The pressure transducers cannot handle frozen water, so the equipment was removed in the autumn.
Due to the very restricted amount of running water, no permanent discharge stations measuring surface water inflows to, or outflows from the lake were installed. Instead, a simple temporary installation was used to measure intermittent flowing surface water into TBL, led through a PVC pipe at the outlet of one of the main subcatchments of the area, where running water has been visible during wet periods. This small brook, shaped by ice wedges, dewatered the main part of the northern valley (Fig. 1). Manual measurements of this surface water inflow to the lake were made when field crew were present at the site in June and August–September 2013. Even though the total inflow to the lake was not captured in these measurements, they provided data supporting the development of the conceptual understanding of the site hydrology. In particular, the response to rain events and the magnitude of base flow and peak flow components could be analyzed. No surface water outflow has occurred in the lake since the project started, i.e., during the period 2010–2013, which means that the collection of such measurements was not possible.
Levels of top of casing (TOC) and ground surface (relative the lake level, 4 September 2010) for each groundwater well relative to the lake surface. Automatic monitoring in the well is marked by an X.
Spatial and temporal variations of soil water content were monitored by use of the time domain reflectometry technique (TDR; Stein and Kane, 1983); details about the equipment are given in Table 1. In September 2011, 43 TDR-sensors were installed in three clusters, here referred to as clusters 1–3, within the catchment (Figs. 1, 3). Each cluster consists of a number of depth profiles where TDR sensors were placed, so that they cover the whole interval from just below the ground surface to the permafrost surface that constituted the bottom of the active layer at the time of installation. Clusters 1 and 2 consist of four depth profiles, where the four sensors in each profile were distributed evenly from a depth of 5–10 cm below ground to a depth of 40–50 cm (Fig. 3d). The sensors in cluster 3 were arranged in three depth profiles, each profile containing four sensors, which were placed along a transect from the lake into the catchment (Fig. 3c). The different clusters were arranged so that they form transects both along and transverse to the dominating flow direction (Fig. 3a), which generally was judged to be perpendicular to the lake shoreline.
Schematic figure
The depth profiles of TDR sensors were installed in pits (one pit per
profile) that were dug down to the distinct permafrost surface. Four TDR
sensors were installed, with more or less constant distance between the
sensors, so that it covered the entire depth in each particular pit. All TDR
sensors in a cluster were connected to a central unit, consisting of a data
logger, central electrical power, and an earth spike. A table describing
the depth and ID code of each TDR sensor is available via
Undisturbed soil samples were collected at four different depths in metal cylinders for analyses of soil water retention, total porosity, and saturated hydraulic conductivity (Fig. 4a, b). Since the hydraulic conductivity can be anisotropic, soil samples for saturated hydraulic conductivity measurements were collected both in the horizontal and the vertical direction. In addition, disturbed loose soil samples were collected for grain size analysis. Soil sampling was performed at four different sites (Table 3, Fig. 1). Samples taken in till areas and at the lake shore were subject to grain size analysis and total porosity measurements only.
Photo of undisturbed soil samples
The infiltration capacity of the soil was measured in August 2012 at five sites with different soil and vegetation cover (Table 3, Fig. 1) by using the double-ring infiltrometer method (Bouwer, 1986). Two cylinders were driven into the soil and water was added to both the inner and outer ring. The falling head method was applied, where the declining water level in the inner ring was measured as a function of time (Fig. 4c). The falling head was observed until steady-state conditions were reached, i.e., until the soil profile below the rings became saturated. At this moment the infiltration capacity (expressed as area-specific flow rate) provides the saturated hydraulic conductivity of the upper soil layer.
The grain size distributions of 20 samples were obtained in two steps. First
the samples were sieved and divided into three groups:
Water retention parameters were measured in the laboratory (at the Swedish University of Agricultural Sciences in Uppsala). Water was added to the cylinders with undisturbed soil samples, described above, until saturated conditions were reached. The suction was increased in the steps from 0.05, 0.3, 0.5, 0.7, 1, 2, 3, and 6 m. At each suction step, the samples were left to drain until a steady state was reached.
Undisturbed soil samples collected in the metal cylinders, described above, were placed in a device equipped with water pipes that slowly added water to the cylinders. The samples were saturated and had free drainage at the bottom of the cylinders. Water was added at a constant rate and the outflow at steady-state conditions provided estimates of the saturated hydraulic conductivity.
A soil temperature station consisting of seven subsurface temperature sensors distributed over a depth of 2 m, and one sensor measuring air temperature at 1.6 m above ground, was installed in September 2010 (Figs. 1, 4d, and Table 1a). A hole with a diameter of approximately 60 mm was drilled through the active layer into the permafrost. In order to make sure that the sensors were placed at the right depths, the cables and sensors were attached to a PVC pipe, which was placed in the borehole. The sensors were connected to data loggers (Table 1a) that were installed in boxes placed on the ground surface; the boxes and cables were covered by a mound of stones in order to protect them from animals, ice, and snow.
Supporting information on the spatial distribution of the active layer thickness (i.e., the depth to the permafrost surface) was obtained by soil probe measurements along transects within the catchment (Fig. 1). The measurements were performed in late August or September when the active layer had reached its maximum thickness.
Three time-lapse cameras were installed in August 2012 at different locations within the catchment (Fig. 1), with the purpose of monitoring hydrological events during the year and to support interpretations of the quantitative time series obtained from the AWS. The time-lapse camera installations provided very useful information for mapping local temporal variability in ponded water or snow cover, and to validate the snow models developed for the catchment. One camera was placed to give an overview of the whole catchment, C1 (Fig. 1), and the other two were placed to monitor details of hillslopes, C2 and C3 (Fig. 1), one in the southern part of the catchment (C2), and one in the northern part (C3). Photos were taken every second hour; the camera equipment is described in Table 1a.
Soil sampling sites and types of analysis performed.
* RS is regolith sampling; the coordinates for each sampling site are given
in the data set:
Intervals used in the grain size analysis.
The Geonor precipitation gauge collects the precipitation in a bucket and weighs the liquid with a vibrating load sensor that gives a frequency output. The output vibrating frequency to the logger contains noise that has to be accounted for, and the time series containing data on cumulative precipitation, given in millimeters, have been corrected accordingly. A thin layer of oil impedes evaporation losses, which means that any negative load can be neglected, and the first step of the correction was to delete all negative values from the time series.
Once the precipitation time series was corrected for noise, it had to be corrected for losses due to wind and adhesion. The gauge undercatch is in general larger for solid precipitation, and thus more important to take into account in Arctic environments than in areas where the main part of the precipitation falls as rain. The correction of precipitation is of special interest if the time series is used as input data in hydrological modeling, where an uncorrected precipitation may cause large water balance errors.
The location of the gauge was classified to be in number 6 out of 7 classes with
respect to wind exposure; see Table 5 (Alexandersson, 2003), and the applied
correction factors were in line with other work reported from Greenland
(Hasholt, 1997; Førland et al., 1996). Losses due to adhesion were
corrected for by adding 0.1 mm for every precipitation event in all classes 1–7.
According to Alexandersson (2003), the Geonor precipitation gauges have been
assigned an extra correction factor due to the larger height of the gauge
resulting in a higher wind exposure and higher potential risk for wind
losses; see Table 5. The classes can be determined by studying photos and
relevant spatial data from the site (tree and vegetation cover, topography).
However, the best way to classify the gauge is to visit the station. In all
correction calculations, a threshold temperature of 0
The precipitation data from the station in Kangerlussuaq (Cappelen, 2014) were also corrected for wind and adhesion losses by applying the Alexandersson (2003) corrections from class 4, Table 5. The precipitation bucket used at the DMI station is not a Geonor, which means that the extra correction factor of 7 % was excluded for the Kangerlussuaq data.
Classification of precipitation stations and correction factors based on exposure to wind.
The equipment used for the measurements of groundwater and surface water levels measured total pressure, which means that the pressure data had to be corrected for barometric pressure. Locally measured barometric pressure data were available from the AWS at TBL. In the correction of pressure data, barometric pressure was subtracted from the measurements of total pressure, resulting in time series of groundwater and surface water pressure fluctuations.
Only corrected pressure data are included in the data set presented in the present paper (see next section). The time series of lake level data (corrected for barometric pressure) was compared to manual measurements of the lake surface level, which was measured by a leveling instrument in August 2010, 2012, and 2013.
In this section, the resulting hydrological and meteorological time series
data and investigation results, achieved up until 31 December 2013, are
presented. All parameters and time series presented in this paper are
accessible in the database PANGEA via
Climate data time series from April 2011 until December
2013.
It is noted that a pronounced annual cycle in air temperature occurred at the
study site (Fig. 5a). Non-freezing temperatures persisted throughout the
summer months, with hourly values up to 15
Wind speed was typically low, with values below 5 m s
For the measurement period (April 2011–December 2013), barometric pressure was
966 hPa on average. It was typically
The air temperature cycle was also evident from the downward long-wave
radiation measurements, which varied between a minimum of
Surface height measurements illustrate that little snow accumulated at the
weather station in winter, likely because it is situated on a small hill. In
2011/12 and 2012/13, roughly 10 cm accumulated (Fig. 5g). However, it
should be noted that the snow cover was redistributed during storms and an
irregular snow depth occurred at the site; for further details, see Sect. 5.3. In
summer, a surface height increase of similar magnitude was recorded; this
was due to vegetation growth as could be determined from the low albedo
values of
The annual corrected precipitation in 2012 was 365 mm and in 2013, it was 269 mm;
see Fig. 5h. The major part of the precipitation fell in April–May and
August–October, and approximately 40 % of the annual precipitation fell as
snow. The measured precipitation at TBL was approximately twice as large as the
precipitation measured in Kangerlussuaq. The precipitation in Kangerlussuaq
is measured by DMI (Cappelen, 2014) and is not included in the data set
presented at
Time series with hourly data on corrected precipitation from April 2011 until December 2013 are included in the delivered data set. Hourly, daily, and monthly values are presented in the data set for all other parameters measured by the AWS.
The mean daily sublimation measured in April 2013 ranged from 0.39 to 1.06 mm day
Based on data from the investigated profiles, the average snow depth in the area in April 2011 was estimated to be 17 cm, with a corresponding snow water equivalent (SWE) of 20 mm. The largest snow depth was found in the valleys, and the smallest along the hillsides. At the time of measurement, the mean snow depth on the lake was similar to that of the whole catchment. However, time-lapse photos show that the snow cover on the lake was reduced to zero after certain periods of strong winds; for further details, see Sect. 5.8. Snow depth and corresponding SWE for each transect are presented in the data set.
The uppermost 10 cm of the active layer was characterized by high porosity
and high saturated hydraulic conductivity (
Vertical (
Saturated hydraulic conductivity calculated from grain size analysis and total porosity from undisturbed soil samples.
The average saturated infiltration capacity of all samples was 2.
pF curves and drainable pore space (saturated water
content – water content at field capacity) based on data from 40 undisturbed
soil samples
Water retention curves (pF curves; see Fig. 6a–b) were obtained from
undisturbed soil samples taken at different depths in two different types of
soils: peaty silt and sandy silt. The high porosity in the upper 10 cm of
the active layer was reflected as a high drainable pore volume (–) for both
sample sites. The drainable pore volume decreased with depth and the values
below 20 cm were in the same order of magnitude for both types of sampled
soil. The drainable pore volume is defined as the difference between
saturated water content and the water content at field capacity (denoted by pF 2).
Data from all sampling sites and associated results for hydraulic parameters
(
The annual variation in the lake water level in TBL was between 0.2 and 0.3 m (Fig. 7a). The dry spring and summer in 2011 caused a distinct decrease in lake level. The annual variation in the northwestern lake during the period 2012–2013 was 0.25 m, and there was a clear covariation between the levels in the two lakes. The larger catchment and lake area make the northwestern lake somewhat less sensitive to short-term temporal variations. The corrected time series for both lakes are found in the data set.
Lake level in TBL co-plotted with the lake level of the
northwestern lake
All groundwater wells situated within the catchment showed similar temporal variations, whereas the temporal pattern in wells 13 and 14, situated downstream from the TBL catchment area, diverged from the wells inside the catchment. In Fig. 7b, the groundwater levels along the investigated transect in the northern part of the catchment are co-plotted with precipitation data. Two events with large increases in groundwater levels can be observed, namely the rain events around 26 July and 15 August 2013. At the beginning of the active period, decreasing groundwater levels were observed in well 12, located at the highest point along the monitored transect. Thawing of the active layer allowed water to percolate deeper into the saturated zone, at the same time as previously frozen groundwater was released and transported downstream. Wells 9 and 11 did not experience a falling groundwater level due to active layer thawing. These wells are located in local discharge areas in the lower parts of the hillslope, where thawed groundwater released at higher elevation discharged. The same pattern can be seen for the other transects. Time series for each of the wells (1, 3, 4, 6, 8, 9, 11, 12, 13, and 14) are found in the data set.
Inflow to the lake was measured during a week in June and from mid-August to
the end of September 2013. There is a clear correlation between precipitation
and surface water flow; see Fig. 7c. The small storage volume available for
the groundwater, reflected in the rapid response in groundwater levels to
precipitation, was further manifested by the quick response in surface water
flows. When the second measurement period, illustrated in Fig. 7b, started
in August, the discharge station was dry. Almost 90 mm of rain fell during
the second measurement campaign, and the surface water discharge increased
from 0 to approximately 0.2 L s
Soil water content (%) at different depths co-plotted
with precipitation (mm) data from the AWS, during the active period in 2012
The user interface of the time-lapse film.
Time series data on soil water content, given as volume of water per total volume of soil, are available from late August 2011 until 31 December 2013. Data from all 43 sensors in the three TDR clusters (Fig. 1) are available in the data set. Figure 8a and c show examples of soil water contents at different depths as functions of time and precipitation during the period from August 2011 to August 2012, and in the summer of 2012. The depth trend in porosity (Sect. 5.4) is reflected as decreasing soil water contents with increasing depth; there is also an increasing delay in the response to rain events with increasing depth. The relatively low precipitation in July 2012 kept the soil water content in the upper 10 cm close to saturation, without affecting the deeper soil layers. The heavier rains in August quickly saturated the uppermost part (10 cm), and percolating water caused increased soil water content at the lower levels (i.e., at depths of 30, 45, and 55 cm).
The quick changes in apparent soil water content during autumn and late spring were actually due to a phase change from water to ice or from ice to water in the soil matrix. Figure 8c–d illustrate how the active layer froze and thawed from the ground surface downwards, with the uppermost part of the active layer freezing and thawing approximately 2 months before the deepest parts.
At the location for the soil temperature measurements, the transition from
permafrost to the active layer occurred at a depth of approximately 0.9 m
(Fig. 8a, b). The temperature sensor at 0.75 m depth measured temperatures
above 0
Photos from all three cameras, together with time series data from the AWS, the soil temperature station, and the lake level monitoring, were put together in an interactive graphic interface (Fig. 9), hereafter called “the time-lapse film”. The time-lapse film facilitates the analysis of hydrological events and their spatial variations within the catchment. In Arctic landscapes, observations by time-lapse cameras have been shown to be very useful in order to explain and analyze monitoring data, since conventional hydrological investigation techniques are not always applicable. The film has been used to analyze snow and lake ice dynamics, lake level response to precipitation and snow melt, and the occurrence of ponding water on the ground surface. The time-lapse film is available in the data set.
We present a new, extensive, and detailed data set with high temporal
resolution for key hydrological parameters in an Arctic periglacial lake
catchment. The high spatial resolution of the data, in this relatively small
catchment area, makes the data set suitable for detailed hydrological and
ecological catchment-scale studies. The data set also includes time series of
photographs taken by time-lapse cameras, which have been arranged in an
interface where photos, together with time series data from the catchment, can
be studied in a film. This is a useful tool to test interpretations of the
quantitative data. The data set is open access and available at
Emma Johansson prepared the manuscript with contributions from all co-authors and she was involved in all field work and data processing related to the data presented in the present paper. Emma Johansson (hydrology and meteorology) and Tobias Lindborg (ecology and chemistry) are responsible for the GRASP field program. Tobias Lindborg, Sten Berglund, Johannes Petrone, Lars-Göran Gustafsson, Jens-Ove Näslund, and Hjalmar Laudon were involved in the field work regarding the different measurement campaigns, as well as installations and maintenance of the equipment. Dirk van As was responsible for installation and maintenance of the automatic weather station as well as post-processing of the climate data.
Thanks to John Cappelen, Danish Meteorological Institute, for providing climate data from the DMI station in Kangerlussuaq, and Christina Öhman, Swedish University of Agricultural Sciences, for help with the analysis of soil samples.Edited by: A. Gelfan