Geological Survey of Denmark and Greenland Bulletin 33, 2015, 65-68 65© 2015 GEUS. Geological Survey of Denmark and Greenland Bulletin 33, 65–68. Open access: www.geus.dk/publications/bull Observed melt-season snowpack evolution on the Greenland ice sheet Charalampos Charalampidis and Dirk van As Due to recent warm and record-warm summers in Green- land (Nghiem et al. 2012), the melt of the ice-sheet surface and the subsequent runoff are increasing (Shepherd et al. 2012). About 84% of the mass loss from the Greenland ice sheet between 2009 and 2012 resulted from increased sur- face runoff (Enderlin et al. 2014). Th e largest melt occurs in the ablation zone, the low marginal area of the ice sheet (Van As et al. 2014), where melt exceeds wintertime accumulation and bare ice is thus exposed during each melt season. In the higher regions of the ice sheet (i.e. the accumulation area), melt is limited and the snow cover persists throughout the year. It is in the vast latter area that models struggle to calcu- late certain mass fl uxes with accuracy. A better understand- ing of processes such as meltwater percolation and refreezing in snow and fi rn is crucial for more accurate Greenland ice- sheet mass-budget estimates (Van Angelen et al. 2013). In May 2012, the fi eld campaign ‘Snow Processes in the Lower Accumulation Zone’ was organised by the Geological Survey of Denmark and Greenland (GEUS) at the KAN_U automatic weather station (67°0´0˝N, 47°1́ 1˝W; 1840 m above sea level), which delivers data to the Programme for Monitoring of the Greenland Ice Sheet (PROMICE; Van As et al. 2013) and is one of the few weather stations located in the lower accumulation area of Greenland (Fig. 1, inset). During the expedition, we installed thermistor strings, fi rn compaction monitors and a snowpack analyser; we drilled fi rn cores, performed fi rn radar measurements, gathered me- teorological data, dug snow pits and performed dye-tracing experiments. One important objective of the campaign was to understand the thermal variability in the snowpack dur- ing the melt season by monitoring with high-precision tem- perature probes (Campbell Scientifi c temperature probe, model 107; accuracy: better than ± 0.4°C over the range –24 to 48°C). Six temperature probes were installed in the snowpack of the previous winter at depths of 0.05, 0.10, 0.20, 0.30, 0.40 and 0.70 m below the surface (Fig. 1). Th e data from the probes were stored at 30-minute intervals on data loggers, which also triggered additional measurements of radiation-shielded air temperature at 1.10 m, surface albedo and surface-height change due to accumulation and ablation. Emitted longwave radiation was also recorded to be able to calculate the surface temperature assuming snow to be a black-body radiator. Th e vertical position of the probes relative to the surface, which changes due to ablation and accumulation, was determined by the sonic ranger measurements. Recorded temperatures aft er the probes surfaced were discarded. Th e relatively shallow snowpack (0.70–0.80 m) was on top of fi rn of density ρ >500 kg m-3 which had accumulat- ed in the previous years (Fig. 1). In the upper fi rn we found ice lenses (ρ >800 kg m–3) several metres thick. Within the snowpack, two thin ice layers were present, one at 0.30 m Green- land Fig. 1. After the installation of the temperature probes in May 2012 (thin black cables) at the location of K AN_U on the Greenland ice sheet. The thick grey cables are thermistor strings drilled into the firn. The inset map shows the location of the study area. 6666 and one at c. 0.50 m below the surface, both about 0.01 m thick. Th e average density of the snow was determined to be roughly 360 kg m–3, yielding an accumulation of 0.25 ± 0.08 m water equivalent (w.e.) since the summer 2011 (Charalam- pidis et al. 2015). Below, we present observations from the period 02 May to 23 July and interpret the atmosphere–sur- face interaction and its impact on the subsurface snow layers, with the goal to quantify refreezing in the Greenland accu- mulation area. Atmosphere–snow interaction Th e observations reveal a strong similarity between the near- surface air temperature and the snow-surface temperature with changes of air temparature lagging on average 30–40 minutes behind. Typically, the air remained warmer than the snow surface (Fig. 2), implying a prevailing stable stratifi ca- tion of the near-surface air. At night, the diff erence was larg- er (1.6–1.9°C) due to the reduced sunlight and subsequent cooling of the surface forced by longwave radiation. During the day, the temperature diff erence was smaller (0.6–1.2°C) primarily due to solar radiation heating the surface and re- ducing atmospheric stability. Understandably, when air temperature exceeds 0°C, the temperature diff erence can be larger since the surface cannot exceed the melting point. On 6 May, overcast conditions and atmospheric stillness caused the increase of air temperature above +6°C (Fig. 3A). Th ese were the highest temperatures during the observation- al period and similar temperatures occurred also on two days in June (3 and 18). With the exception of 6 May, the air tem- perature remained negative until the last week of May. Dur- ing this period, the temperature of the upper 0.20 m of snow followed a pronounced diurnal cycle, which at 0.20 m lagged about 10 hours behind the variations in surface temperature (Fig. 3B), signifying the low thermal conductivity of snow. At 1840 m above sea level, the ice sheet generally experi- ences low melt rates. When melt occurs, it displays a diurnal cycle following air temperature. A diurnal cycle of positive air temperatures occurred fi rst on 27 May, marking the be- ginning of the melt season (Fig. 3A). Th e surface ablated in response to the warm conditions, while the snow tempera- tures revealed the distinct progression of a warming and thus wetting front moving vertically through the snowpack (Fig. 3B). Th e temperature at 0.70 m depth was aff ected by this 42 hours aft er surface melt initiated, i.e. an average warming front progression of only c. 17 mm h–1. Th e entire snowpack became temperate aft er six days of ablation. Th e slow pro- gression of the warming front indicates a concurrent hetero- geneous meltwater infi ltration to the fi rn below (Humphrey et al. 2012). During the period 8–12 June, sub-freezing air tempera- tures occurred again (Fig. 3A) and melting ceased. While the upper part of the snowpack remained close to 0°C, possibly containing liquid water, the deeper levels (0.4 m and below) cooled as heat was conducted downwards into the colder fi rn. Melt resumed on 13 June, aff ecting snow temperatures at 0.7 m aft er 30 hours (Fig. 3B), which is faster than in the previ- ous melt period due to the reduced measurement depth, and changed snow properties. Th ereaft er, the snow remained at the melting point until it ablated completely on 11 July. Between 13 June and 11 July, there were fi ve occasions when the diurnal air/surface temperature cycle was inter- rupted by periods with warm night-time conditions result- ing in enhanced ablation (Fig. 3A). Most notably, during the warm week of 8–14 July when the whole Greenland ice- sheet surface area was reported to melt (Nghiem et al. 2012), the air temperature at KAN_U remained above +2°C for six days and melt was large. At the same time, the ‘Watson River’, which drains this section of the ice sheet, experienced the highest discharge in 56 years, judging from the partial destruction of a 1956 bridge near the town of Kangerlussuaq. Snowpack evolution In May, the area received 0.12 m of fresh snow on top of the existing snowpack (Fig. 4B) and the albedo remained at fresh snow values of 0.8–0.9 (Fig. 4A). On 27 May, the surface be- gan ablating and by the end of the day 0.05 m of the fresh snow had melted away. In the period until 8 June the average ablation rate was 0.02 m day–1, reducing albedo to c. 0.75, primarily due to snow metamorphosis. In principle, the energy needed to make temperate a uni- form snowpack 0.7 m thick at –10°C is equivalent to the energy necessary for melting 0.04 m of snow at 0°C and a density of 360 kg m–3. Th erefore, the generation of 15 mm of meltwater and its refreezing within the snowpack raises its temperature to 0°C. By the beginning of June when the entire snowpack had reached 0°C, the fi rst 0.15 m of snow (i.e. c. 50 mm of meltwater) had ablated. Th is implies that approximately 70% of the meltwater either percolated deeper 1 3 5 7 9 11 13 15 17 19 21 23 Time of the day (h UTC) Period: 2 May to 23 July 2012 (T a − T s ) av e ra ge ( °C ) −1 0 1 2 3 Fig. 2. Average temperature difference between air and surface during the day. The error bars show two standard deviations. Solar zenith time is at 15:08 UTC. 67 into the fi rn or was retained in liquid form in the snow by capillary forces. From the beginning of June onward, by lack of cold content, the snowpack was able to respond immedi- ately to surface forcings (Fig. 4B), and all percolating meltwa- ter was routed toward the underlying fi rn. Th e cold conditions and melt pause from 8 to 12 June were accompanied by snowfall resulting in 0.05 m of fresh snow accumulation (Fig. 4B), thereby increasing the albedo above 0.8 (Fig. 4A). Melt resumed on 15 June with an average abla- tion rate of 0.03 m day–1 and with albedo dropping as low as 0.7, indicative of wet snow. Small snowfall events also oc- curred in the beginning of July. During the warm days of 9 and 10 July the ablation rate exceeded 0.05 m day–1, remov- ing the last of the 2011–2012 winter snowpack and reveal- ing the underlying, water-saturated fi rn. Consequently, the albedo dropped below 0.7, enhancing melt through the melt- albedo feedback (Box et al. 2012). Simulated refreezing rates A combination of temperature measurements and thermal conductivity simulations reveals the amount of refrozen wa- ter in the snow. A heat-conduction model was used to simu- late the evolution of subsurface temperatures, using meas- ured surface temperature and surface height change as input. Th e model was run at temporal and spatial resolutions of 10 minutes and 0.10 m, respectively, and was re-initialised each day by measured temperature profi les at 00:00 UTC. Th e ef- fective conductivity of the snow is a function of snow density (Sturm et al. 1997) and the specifi c heat of snow depends on temperature (Yen 1981). Density profi les were initialised based on snow pit density measurements at the installation of the probes and were updated throughout the run taking refreezing into account. Th e diff erence between the simu- lated and measured temperatures at the end of the day is a measure of the added latent heat during the day, and thus of the daily refreezing rates. Th e simulation reveals that during the fi rst week of melt starting 27 May, refreezing occurred at all measurement depths within the snowpack (Fig. 5A). Th e peak refreezing occurred on 30 May and at depths below 0.50 m. During this period the total refreezing rate in the snowpack was compa- rable to the average melt rate of that fi rst period of melt (6 kg m–2 day–1; Fig. 5B). Th e subsequent cold content reduction and thinning snowpack resulted in low refreezing values be- low 0.4 kg m–2 day–1 at all depths. Note that refreezing rates during the sub-freezing early period of our simulation are non-zero and large near the surface. It is possible that short- wave penetration in the snow plays a role or that our conduc- tion model is fl awed in conditions of large temperature gra- dients in well-ventilated, low-density snow, which is valid for the start of the simulation period. However, in terms of total refreezing the early results add up to small values (Fig. 5B). 0 4 2012 A 06 May 20 May 03 Jun 17 Jun 01 Jul 15 Jul −30 −25 −20 −15 −10 −5 0 T e m p e ra tu re ( °C ) B air (1.10 m) surface 0.05 m 0.10 m 0.20 m 0.30 m 0.40 m 0.70 m T e m p . ( °C ) Fig. 3. Observed temperatures of the near-surface atmosphere (A) and (sub)surface (B). 0.7 0.9 A lb e d o 2012 A In it ia l d e p th f ro m s u rf ac e ( m ) B 06 May 20 May 03 Jun 17 Jun 01 Jul 0.7 0.4 0.3 0.2 0.1 0.0 −28 −24 −20 −16 −12 −8 −4 0 (°C) Fig. 4. Observed surface albedo (A) and thermal evolution (B) of the snowpack. The dark red contour signifies 0°C. Fig. 5. Calculated refreezing rates in the snow at 0.1 m spacial resolution (A) and combined (B). x 10−3 x 10−3 0.10 m 0.20 m 0.30 m 0.40 m 0.50 m 0.60 m 0.70 m 0 0.2 0.4 0.6 0.8 1 2012 A 06 May 20 May 03 Jun 17 Jun 01 Jul 15 Jul 0 2 4 6 R e fr . r at e B total R e fr e e zi n g ra te ( 1 0 3 k g m − 2 d ay − 1 ) 6868 Th e non-zero values of roughly 0.1 kg m–2 day–1 at greater depth are considered the uncertainty for the entire simula- tion period. During the cold period in June, the refreezing rates in- creased again to 3 kg m–2 day–1 (Fig. 5B), which is an indi- cation that liquid water was available, primarily at depths 0.20–0.30 m, while the required cold content was being supplied by the surface. Th is method of refreezing requires liquid water retention in the snow matrix while cold content becomes available, as opposed to the refreezing of meltwater percolating into layers at sub-freezing temperatures. Th e heat between 0.60–0.70 m that was conducted to depths below the seasonal snow layer increased the available cold content, thus when melt occurred again, refreezing was prominent at those depths (15 June; Fig. 5A). As the average melt rate aft er 13 June was c. 8 mm w.e. day–1, the refrozen water in the snowpack was less than 10% of this amount, implying liquid water retention or the routing of meltwater to the lay- ers below. Overall, the simulated density increase within the snowpack was between 70–80 kg m–3 for most levels. Meltwater refreezing is a positive component in the mass budget (mass storage; Harper et al. 2012), although in a warming climate with more frequent extreme melt condi- tions, the larger meltwater fl uxes in the snow and fi rn may result in rapid reduction of pore volume (Van Angelen et al. 2013). Th e large melt of 2012 at the elevation of KAN_U was a result of both high atmospheric temperatures (Ben- nartz et al. 2013) and a relatively low albedo from the expo- sure of the water-saturated fi rn aft er the early removal of the relatively thin winter snowpack (Charalampidis et al. 2015). Th e high ice content of the fi rn as found during the measure- ment campaign is an indication of intense percolation dur- ing previous years. Th ese snow processes are still quite poorly represented in modelling eff orts, also due to the dependency of horizontal meltwater runoff on the ice layers formed by re- freezing. Our results illustrate that especially the melt-albe- do feedback in relation to pore-volume reduction makes the lower accumulation area of the Greenland ice sheet highly responsive in a warming climate. Acknowledgements Th e data presented in this paper were gathered in close collaboration with the Greenland Analogue Project. We are grateful to our Snow Processes in the Lower Accumulation Zone project partners Horst Machguth, Mike MacFerrin, Andreas Mikkelsen, Rickard Pettersson, Katrin Lindbäck, Alun Hubbard and Sam Doyle. Th is is a publication in the framework of the Programme for Monitoring of the Greenland Ice Sheet (PROMICE) and contribution number 63 of the Nordic Centre of Excellence SVALI, ‘Stability and Variations of Arctic Land Ice’, funded by the Nordic Top- level Research Initiative (TRI). References Bennartz, R., Shupe, M.D., Turner, D.D., Walden, V.P., Steff en, K., Cox, C.J. Kulie, M.S. Miller, N.B. & Pettersen, C. 2013: July 2012 Greenland melt extent enhanced by low-level liquid clouds. Nature 496, 83–86. Box, J.E., Fettweis, X., Stroeve, J.C., Tedesco, M., Hall D.K. & Steff en, K. 2012: Greenland ice sheet albedo feedback: thermodynamics and atmospheric drivers. Th e Cryosphere 6, 821–839. Charalampidis, C. et al. 2015: Changing surface-atmosphere energ y ex- change and refreezing capacity of the lower accumulation area, west Greenland. Th e Cryosphere Discussions 9, 2867–2913. Enderlin, E.M., Howat, I.M., Jeong, S., Noh, M.-J., van Angelen, J.H. & van den Broeke, M.R. 2014: An improved mass budget for the Green- land ice sheet. Geophysical Research Letters 41, 866–872. Harper, J., Humphrey, N., Pfeff er, W.T., Brown, J. & Fettweis, X. 2012: Greenland ice-sheet contribution to sea-level rise buff ered by meltwater storage in firn. Nature 491, 240–243. Humphrey, N.F., Harper, J.T. & Pfeff er, W.T. 2012: Th ermal tracking of meltwater retention in Greenland’s accumulation area. Journal of Geo- physical Research: Earth Surface 117, F01010. Nghiem, S.V., Hall, D.K., Mote, T.L., Tedesco, M., Albert, M.R., Keegan, K., Shuman, C.A., DiGirolamo, N.E. & Neumann, G. 2012: Th e ex- treme melt across the Greenland ice sheet in 2012. Geophysical Re- search Letters 39, L20502. Shepherd, A. et al. 2012: A reconciled estimate of ice-sheet mass balance. Science 338, 1183–1189. Sturm, M., Holmgren, J., König, M. & Morris, K. 1997: Th e thermal con- ductivity of seasonal snow. Journal of Glaciolog y 43, 26–41. Van Angelen, J.H., Lenaerts, J.T.M., van den Broeke, M.R., Fettweis, X. & Meijgaard, E. 2013: Rapid loss of fi rn pore space accelerates 21st century Greenland mass loss. Geophysical Research Letters 40, 2109–2113. Van As, D., Fausto, R.S., Colgan, W.T., Box, J.E. & the PROMICE project team 2013: Darkening of the Greenland ice sheet due to the melt-albedo feedback observed at the PROMICE weather stations. Geological Sur- vey of Denmark and Greenland Bulletin 28, 69–72. Van As, D. et al. 2014: Increasing meltwater discharge from the Nuuk region of the Greenland ice sheet and implications for mass balance (1960–2012). Journal of Glaciolog y 60, 314–322. Yen, Y.C. 1981: Review of thermal properties of snow, ice and sea ice. CR- R EL Report 81–10, 27 pp. Hanover, New Hampshire: US Army Corps of EngineersCold Regions Research and Engineering Laboratory. Authors’ addresses C.C.* & D.v.A., Geological Survey of Denmark and Greenland, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark. E-mail: cc@geus.dk *Also at: Department of Earth Sciences, Uppsala University, Villavägen 16, 752 36 Uppsala, Sweden.