Climatology and firn processes in the lower accumulation area of the Greenland ice sheet

Full text

(1)Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1372. Climatology and firn processes in the lower accumulation area of the Greenland ice sheet CHARALAMPOS CHARALAMPIDIS. ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2016. ISSN 1651-6214 ISBN 978-91-554-9571-8 urn:nbn:se:uu:diva-284365.

(2) Dissertation presented at Uppsala University to be publicly examined in Hambergsalen, Geocentrum, Villavägen 16, Uppsala, Friday, 10 June 2016 at 10:15 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Professor Douglas Benn (University of St Andrews). Abstract Charalampidis, C. 2016. Climatology and firn processes in the lower accumulation area of the Greenland ice sheet. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1372. 81 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-9571-8. The Greenland ice sheet is the largest Northern Hemisphere store of fresh water, and it is responding rapidly to the warming climate. In situ observations document the changing ice sheet properties in the lower accumulation area, Southwest Greenland. Firn densities from 1840 meters above sea level retrieved in May 2012 revealed the existence of a 5.5-meterthick, near-surface ice layer in response to the recent increased melt and refreezing in firn. As a consequence, vertical meltwater percolation in the extreme summer 2012 was inefficient, resulting in surface runoff. Meltwater percolated and refroze at six meters depth only after the end of the melt season. This prolonged autumn refreezing under the newly accumulated snowpack resulted in unprecedented firn warming with temperature at ten meters depth increased by more than four degrees Celsius. Simulations confirm that meltwater reached nine meters depth at most. The refrozen meltwater was estimated at 0.23 meters water equivalent, amounting to 25 % of the total 2012 ablation. A surface energy balance model was used to evaluate the seasonal and interannual variability of all surface energy fluxes at that elevation in the years 2009 to 2013. Due to the meltwater presence at the surface in 2012, the summer-averaged albedo was significantly reduced (0.71 in 2012; typically 0.78). A sensitivity analysis revealed that 71 % of the subsequent additional solar radiation in 2012 was used for melt, corresponding to 36 % of the total 2012 surface lowering. This interplay between melt and firn properties highlights that the lower accumulation area of the Greenland ice sheet will be responding rapidly in a warming climate. Keywords: climate change, Greenland ice sheet, accumulation area, automatic weather stations, surface energy balance, melt–albedo feedback, surface mass budget, snow, firn, meltwater, percolation, refreezing, runoff Charalampos Charalampidis, , Department of Earth Sciences, LUVAL, Villav. 16, Uppsala University, SE-75236 Uppsala, Sweden. © Charalampos Charalampidis 2016 ISSN 1651-6214 ISBN 978-91-554-9571-8 urn:nbn:se:uu:diva-284365 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-284365).

(3)

(4)

(5) Have you seen my Queen? She’s a divine child. And have you seen Her ways, how they’re driving me wild? When She dances like a lion, setting the treetops on fire, may Her eyes (ice) never tell me no lies, because I love Her. Samsara Blues Experiment (2010).

(6)

(7) List of papers. This thesis is based on the following papers, which are referred to in the text by their Roman numerals. I. Citterio, M., D. van As, A. P. Ahlstrøm, M. L. Andersen, S. B. Andersen, J. E. Box, C. Charalampidis, W. T. Colgan, R. S. Fausto, S. Nielsen, and M. Veicherts (2015). Automatic weather stations for basic and applied glaciological research. Geol. Surv. Denmark Greenland Bull. 33, 69–72.. II. Charalampidis, C. and D. van As (2015). Observed melt-season snow-pack evolution on the Greenland ice sheet. Geol. Surv. Denmark Greenland Bull. 33, 65–68.. III. Charalampidis, C., D. van As, J. E. Box, M. R. van den Broeke, W. T. Colgan, S. H. Doyle, A. L. Hubbard, M. MacFerrin, H. Machguth, and C. J. P. P. Smeets (2015). Changing surface–atmosphere energy exchange and refreezing capacity of the lower accumulation area, West Greenland. The Cryosphere 9(6), 2163–2181.. IV. Machguth, H., M. MacFerrin, D. van As, J. E. Box, C. Charalampidis, W. T. Colgan, R. S. Fausto, H. A. J. Meijer, E. Mosley-Thompson, and R. S. W. van de Wal (2016). Greenland meltwater storage in firn limited by near-surface ice formation. Nat. Clim. Change 6(4), 390–393.. V. VI. Charalampidis, C., D. van As, W. T. Colgan, R. S. Fausto, M. MacFerrin, and H. Machguth (2016). Thermal tracing of retained meltwater in the lower accumulation area of the Southwestern Greenland ice sheet. Ann. Glaciol. 57(72), n/a–n/a. Charalampidis, C., D. van As, P. L. Langen, R. S. Fausto, B. Vandecrux, and J. E. Box (in review). Regional climate model performance in Greenland firn derived from in situ observations. Geol. Surv. Denmark Greenland Bull.. Reprints were made with permission from the publishers..

(8) c De Nationale Geologiske Undersøgelser for Copyrights. Paper I, II, VI: Danmark og Grønland (GEUS) 2015–2016. c Authors 2015 (Creative Commons Attribution 3.0 License). Paper III: c Nature Publishing Group 2016. Paper IV: c Paper V: Authors 2016 (Creative Commons Attribution 4.0 License). Co-authorships. The contribution to Paper I was the involvement in the development, maintenance and establishment of automatic weather stations on the Greenland ice sheet, as well as the discussion and assistance during the composition of the manuscript. The contribution to Paper IV was the participation in the field campaigns that collected the data, the performance of part of the data analysis, the continuous discussion of the results, and further development of the analysis. Support. The outlined research was conducted at the Geological Survey of Denmark and Greenland (GEUS) in Copenhagen, Denmark, and is a contribution to the Nordic Centre of Excellence SVALI, “Stability and Variations of Arctic Land Ice”, funded by the Nordic Top-level Research Initiative (TRI). This research builds upon data gathered and results obtained within the Greenland Analogue Project (GAP). The measurements were collected during the Snow Processes in the Lower Accumulation Zone (SPLAZ) campaign in 2012, and the Arctic Circle Traverse campaigns in 2013 and 2015 (ACT-13, ACT-15). This work is within the framework of the Programme for Monitoring of the Greenland Ice Sheet (PROMICE), launched and funded by the Danish Energy Agency (Energistyrelsen) under the Danish Ministry of Energy, Utilities and Climate, and within the Danish Cooperation for Environment in the Arctic (DANCEA).. Relevant papers Contributions related to this work, but not part of the thesis are referred to in the text by their Greek numerals: α Van As, D., R. S. Fausto, W. T. Colgan, J. E. Box, and the PROMICE project team (incl. C. Charalampidis) (2013). Darkening of the Greenland ice sheet due to the melt–albedo feedback observed at the PROMICE weather stations. Geol. Surv. Denmark Greenland Bull. 28, 69–72. β Van As, D., R. S. Fausto, K. Steffen, and the PROMICE project team (incl. C. Charalampidis) (2014). Katabatic winds and piteraq storms:.

(9) observations from the Greenland ice sheet. Geol. Surv. Denmark Greenland Bull. 31, 83–86. γ Fausto, R. S., D. van As, J. A. Antoft, J. E. Box, W. T. Colgan, and the PROMICE project team (incl. C. Charalampidis) (2015). Greenland ice sheet melt area from MODIS (2000–2014). Geol. Surv. Denmark Greenland Bull. 33, 57–60. δ Van As, D., R. S. Fausto, J. Cappelen, H. Machguth, R. S. W. van de Wal, R. J. Braithwaite, and the PROMICE project team (incl. C. Charalampidis) (in press). Placing Greenland ice sheet ablation observations in a multi-decadal context. Geol. Surv. Denmark Greenland Bull..

(10) Cover of printed version. Combination of LandsatLook imagery (“Natural Color” for ice sheet and water, and “Thermal” for land surface) of the wider area of Maniitsoq (Sukkertoppen), about 100 km south of Kangerlussuaq (Søndre Strømfjord). The data were acquired by Landsat 7 on 14 August 1999 (scene identifier: LE70060141999226EDC00), and are available from the U. S. Geological Survey. Inuit portrait. Scan of original art archived in the GEUS glaciology library..

(11) Contents. 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17. 2. Study area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Ablation area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Accumulation area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 19 19 19 20. 3. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 KAN_U automatic weather station . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Firn observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Surface energy balance model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Meteorology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Surface energy balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3 Surface mass budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4 Subsurface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.5 Initialization and validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 The HIRHAM5 regional climate model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 25 25 28 28 29 31 36 37 39 41. 4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Snowpack evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Radiation budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Melt–albedo feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Firn density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Firn temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 43 43 46 47 48 50. 5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Firn processes and simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Firn and atmospheric warming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2 Surface energy balance model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.3 Regional climate model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.4 Further investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Mechanisms and implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 53 53 53 53 57 59 60. 6. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 6.1 Summary in English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 6.2 Sammanfattning på svenska . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66.

(12) 6.3. Resumé på dansk. ............................................................................. 68. ............................................................................................ 71. ......................................................................................................... 75. Acknowledgments References.

(13) 1. Introduction. 1.1 Background In recent past, the cryospheric component to global average sea level rise was largely due to retreating glaciers and ice caps around the globe (about 70 % of the total cryospheric component for the period 1961–2003, equivalent to 0.5 mm yr−1 ; Solomon et al., 2007). Due to their relatively small size, these ice bodies responded faster to the prevailing warming climate. However, in the Northern Hemisphere, the Greenland ice sheet is by far the largest freshwater reservoir. If completely melted, global average sea level would rise by ∼ 7.4 m (Bamber et al., 2013). In recent decades the Greenland ice sheet has been losing mass. The contribution from the ice sheet to the global average sea level rise was estimated at was 0.09 mm yr−1 for the period 1992–2001, but according to the latest IPCC report it has increased to 0.6 mm yr−1 over the period 2002–2011 (Vaughan et al., 2013). Atmospheric warming impacts the ice sheet primarily by increasing surface melt (Huybrechts and de Wolde, 1999; Huybrechts et al., 2011). The air and surface temperatures of the ice sheet have been steadily increasing in recent years (Van As, 2011; Hall et al., 2013), thus surface mass loss has been accelerating (Shepherd et al., 2012). The increasing surface meltwater runoff (hereafter “runoff”) has led to the reduction of the ice sheet’s surface mass budget (SMB; Ettema et al., 2009, 2010). Estimates suggest that the surface runoff contribution was as high as two-thirds of the total ice sheet mass loss between 2009 and 2012 (Enderlin et al., 2014). The increasing surface melt is also correlated to the amount of absorbed solar radiation. The high-energy, shortwave radiation from the Sun is more effectively absorbed by a darker, less reflective ice sheet surface, with a lower “albedo”. Factors that can influence the ice sheet’s albedo include black carbon, dust depositions (Bøggild et al., 2010; Wientjes and Oerlemans, 2010), as well as the presence of meltwater itself (Paper α ; Box et al., 2012). The atmosphere was exceptionally warm over the ice sheet during the summer of 2012 (Tedesco et al., 2013). The 2012 surface melt extent (Fig. 1.1) was unprecedented during the satellite area due to the persistent, record-warm air temperatures. On 12 July 2012, surface melt detected at 98.6 % of the ice sheet surface, including at Summit (72◦ 35 46.4 N, 38◦ 25 19.1 W), the highest location on the ice sheet. Here, at ∼ 3216 m above sea level (a.s.l.), the near-surface, hourly air temperature reached 0.8 ◦ C (Nghiem et al., 2012; McGrath et al., 2013). Summit remained relatively warm throughout the summer of 2012, with the highest JJA (June–July–August) average air temperature 13.

(14) Figure 1.1. The classic image that was widely circulated in the media, showing the unprecedented surface melt extent on 12 July 2012. The image was originally published online on 24 July 2012 by the National Aeronautics and Space Administration (NASA) in the Feature: “Satellites See Unprecedented Greenland Ice Sheet Surface Melt.” Shortly after, in October 2012, Nghiem et al. (2012) released the detailed study with satellite and in situ observations of the event (credit: Nicolo E. DiGirolamo, SSAI/NASA GSFC, and Jesse Allen, NASA Earth Observatory).. (−11.4 ◦ C) observed since 1994 (Hanna et al., 2014). According to ice core evidence from Summit, melting had not occured at the location during the past century, with the last melt event dating back to 1889 (Clausen et al., 1988; Keegan et al., 2014). This Greenland-wide event of increased near-surface temperatures in 2012 was partially attributed to thin, low-level liquid clouds (Bennartz et al., 2013). These – low and optically thick – clouds enhanced downward longwave radiation. At the same time, they were thin enough to allow for substantial amounts of solar radiation on the ice sheet surface. At Summit, these clouds were present about 30 % of the time during the summer months in 2012. 14.

(15) Figure 1.2. Page 25, Figure 15 from Benson (1962), where he defined the facies (areas) of an ice field, after using for years the Greenland ice sheet as his “laboratory”. The pioneering nature of his work is easily understood, as it is still a key reference in glaciology textbooks, as well as contemporary glaciological research. Introducing his Figure 15 at the section “Glacier facies – A clacification of glaciers”, he wrote: “If the strata could be observed from top to bottom in a glacier with all facies present, we would find that: 1) Below the saturation line, all of the material has been soaked at one time or another; 2) Between the saturation line and the dry-snow line, soaked and non-soaked layers alternate; and 3) Above the dry-snow line, no soaked layers are observed (Fig. 15).”. Unlike the regions of the periphery of the ice sheet, where bare ice is at the surface (i.e. ablation area; Fig 1.2), the high interior (i.e. accumulation area) has fresh snow at the surface accumulated on top of multi-year snow called firn. Snow and firn are porous, and can be up to 80 m thick. Meltwater at these elevations percolates and refreezes in snow and/or firn. In the lowest parts of the accumulation area adjacent to the ablation area, meltwater in snow cannot percolate deeper than the same year’s summer surface, and refreezes as superimposed ice (Cogley et al., 2011). However, in the higher parts of the accumulation area percolates and refreezes into firn. This internal accumulation of meltwater in firn is a buffer to the Greenland ice sheet’s runoff contribution to sea level rise (Pfeffer et al., 1991; Bøggild et al., 2005; Harper 15.

(16) et al., 2012; Humphrey et al., 2012). Estimates from regional climate simulations suggest that 42 ± 4 % of the sum of rain and meltwater generated at the ice sheet surface was retained during recent decades (Van Angelen et al., 2013). However, the Arctic climate is warming relatively fast. This fact will likely lead to more frequent and intense ice sheet melt, and runoff events like in 2012. Increased melt will impact the snow facies (Fig. 1.2) and reduce their surface area, restricting them to the highest elevations of the ice sheet (McGrath et al., 2013). Van Angelen et al. (2013) suggested also that a moderate warming of air temperatures could cause a 50 % reduction in ice-sheet-wide firn retention capacity by 2100. However, the associated uncertainties of the fractional retention of melt, based on surface mass budget reconstructions, are quite large (Vernon et al., 2013). This is not only due to modeling limitations and observational uncertainties, but also due to limited understanding of meltwater percolation and refreezing in firn. The stochastic nature of these subsurface processes can be only indirectly observed by means of firn coring and temperature monitoring. This makes it difficult to simulate them, and thus constrain the mass component of internal accumulation. In addition, there is only an emerging understanding of the role of perennial firn aquifers in buffering the transient response of ice sheet runoff to climate change (Forster et al., 2014; Koenig et al., 2014; Kuipers Munneke et al., 2014). Considering the thermal evolution of firn, heat transfer simulations can provide in-depth knowledge. Heat is dominantly transfered by conduction in the high interior of the ice sheet, which is a simple calculation, and thus firn temperatures can be calculated to a good approximation given surface temperature (Cuffey and Paterson, 2010). These areas experience no or limited meltwater refreezing, and according to field observations the annual air temperature cycle does not penetrate below 20 m depth (Cuffey and Paterson, 2010). At various locations in the historic dry firn area, firn temperatures at 10 m depth were within 0.3 ◦ C of the respective average annual, near-surface air temperature (Benson, 1962). However, the recent increases in melt occuring at even higher elevations, have resulted in substantial warming of this near-surface firn. As the snow surface absorbes energy, its temperature increases. If the snow surface is at freezing temperature (0 ◦ C), excess energy will be used to turn snow into water. This meltwater percolates through the porous layers of firn beneath for as long as the temperature of the firn layers is also at freezing point. If it reaches sub-freezing firn layers at certain depth, it refreezes. While it undergoes phase transition again, it releases the previously absorbed energy, the latent heat, which raises the firn’s temperature. As mass is added to that firn layer, its density also increases. At mid-elevations (1400–2500 m) in Northwest Greenland, the top 10 m of firn was recently reported up to 5.7 ◦ C warmer than in the 1950s (Polashenski et al., 2014). Additionally, the inland expansion of annually-persistent surface hydrologic networks, may potentially contribute to the warming of firn or ice. 16.

(17) by means of cryo-hydrologic warming (Phillips et al., 2010, 2013; Lüthi et al., 2015). Processes observed on smaller ice fields in the Arctic (Bezeau et al., 2013; Gascon et al., 2013) have recently been reported also on the ice sheet. Relatively high melt years in a sequence, with relatively high corresponding meltwater retention, have caused rapid reduction in the pore volume available in near-surface firn (De la Peña et al., 2015). Near-surface pore volume reduction can in turn hamper percolation while favoring meltwater runoff. Meltwater presence and runoff at the surface can lead to melt–albedo feedback (Paper α ; Box et al., 2012), which promotes ablation in the accumulation area. Thus, the ice sheet due to its vast extent and sheer volume, in combination to the relatively large warming of the polar regions will eventually dominate the contribution to cryospheric global global average sea level rise in the current century.. 1.2 Objectives The aim of this study is to investigate the climatology over the higher parts of the Greenland ice sheet, and identify the feedbacks between atmosphere and ice sheet (sub)surface. The work is based exclusively on in situ observations. It analyses the changes in firn properties, and assesses their extent, in the accumulation area (1840 m a.s.l. and higher) of the Southwestern Greenland ice sheet due to the recent pronounced Arctic warming (Paper IV). The observations include firn densities from various locations on the southwestern ice sheet, and detailed firn temperature, as well as automatic weather station (AWS; Paper I) observations from 1840 m a.s.l. In the lower accumulation area, atmospheric forcing is expected to have the greatest impact on firn. At these elevations, the ice sheet surface–atmosphere energy exchange in recent years (2009–2013) is thouroughly described by utilizing a validated surface energy balance (SEB) model (Paper III), thus extending our understanding of regional ice–atmosphere interaction into the accumulation area. This is framed in the context of atmospheric temperature variability. Due to the fundamentally different properties of surface/subsurface in the accumulation compared to the ablation area (Fig. 1.2), emphasis is not only put on energy forcing on the ice sheet surface (Paper III), but also on meltwater refreezing in firn. The melt–albedo feedback, largely due to saturation of the ice sheet surface, is also quantified with focus on the relative importance between atmospheric warming and surface darkening. Percolation in the lower accumulation area is analysed, both in snow and firn. In case of snow (Paper II), the nature of percolation in summer 2012 is identified, while estimating the refrozen meltwater used to raise the snowpack’s temperature to 0 ◦ C relative to the produced meltwater. The timing and 17.

(18) conditions under which refreezing occurs within the snowpack are identified. The refreezing rates are estimated at various depths at a centimeter scale until the complete ablation of the snowpack. Percolation in firn is also investigated (Paper V), while estimating the depths at which water refroze in the lower accumulation area during the extreme melt year of 2012. Firn temperature variability in the period April 2009–September 2013 is outlined while identifying the factors that preserve and distribute heat in firn. The annual firn warming rate during the observational period is also estimated. Considering the large uncertainties in modeling the fractional retention of melt, the subsurface performance of the validated SEB model, as well as of the regional climate model HIRHAM5 is discussed. Both models are evaluated against the firn temperature observations (Paper VI). At the same time, the comparison of both models with the observed firn temperatures provides insight on the state of the firn, and the percolation depth and water movement within, during the observational years. The comparison of the regional climate with the SEB model reveals the reasons leading to a bias and the way toward improved estimates, and more reliable future projections of water retention in Greenland firn.. 18.

(19) 2. Study area. 2.1 General In situ observations are essential for understanding processes, as well as the impact of the changing climatic conditions on the ice sheet. A relatively dense network of in situ measurements (Van de Wal et al., 1995; Greuell et al., 2001; Van den Broeke et al., 2008; Van As et al., 2012) is located in the Kangerlussuaq region, Southewest Greenland (Fig. 2.1, inset). The instrumentation includes seven automatic weather stations (AWSs; Fig. 2.1) and nine SMB stakes (Fig. 2.2(a)). Between the ice sheet and the ocean in the Kangerlussuaq region lie approximately 150 km of mountainous tundra. The ice sheet has a relatively wide (∼ 100 km) ablation area. The equilibrium line altitude (ELA), where annual accumulation and ablation are equal, averaged over 1990–2003, was estimated to be 1535 m a.s.l. (Fig. 2.2(a); Van de Wal et al., 2005). However, including more recent SMB observations (until 2011), the average ELA has increased to 1553 m a.s.l. (Van de Wal et al., 2012). At the end of every ablation season, superimposed ice is exposed at the ice sheet surface at 1520 m a.s.l., and it extends until about 1750 m a.s.l. (i.e. the “firn line” in Figure 1.2; Van den Broeke et al., 2008). At higher elevations, up to about 2500 m a.s.l., lies the percolation area. Finally, from 2500 m a.s.l. and higher lies the dry snow area.. 2.2 Ablation area The ablation area in the Kangerlussuaq region is well-documented in the literature. Radiation below the ELA was presented by Van den Broeke et al. (2008), based on four years of observations (September 2003–August 2007). The lowest surface albedo was found at the intermediate AWS S6 (1020 m a.s.l.; Fig. 2.2(a)) due to a “dark band” of dust and meltwater at that elevation along the ice sheet slope (Greuell, 2000; Wientjes and Oerlemans, 2010). Melt modeling confirmed the increase in summer melt toward the margin, while revealing decreasing sensible heat transfer with increasing elevation. Additionally, it was shown that shortwave radiation was more dominant over the SEB during melt at higher elevations (Van den Broeke et al., 2008, 2011). The surface roughness was found to follow an annual cycle over a large part of the ablation area (Smeets and van den Broeke, 2008). This explained part of the summertime variability in the turbulent heat fluxes (Van den Broeke 19.

(20) Figure 2.1. Map of the southwestern ice sheet in the Kangerlussuaq region, originally published by Van As et al. (2012). The map indicates the locations and names of all but one automatic weather stations (AWSs) that currently operate in the region by the Institute for Marine and Atmospheric research in Utrecht (IMAU), and the Geological Survey of Denmark and Greelnand (GEUS). In addition, GEUS operates an AWS on the tundra, at 1000 m from the ice sheet margin. The map also indicates the delineation of the catchment that was used by Van As et al. (2012) to relate melt from the ice sheet surface to discharge measurements.. et al., 2009). Additionally, it was shown by Van den Broeke et al. (2009) that the regional “katabatic winds” – dubbed as such since they cold and dense air masses forced by gravity, and thus surface slope, to descent (Paper β ) – are over the variable surface roughness at lower elevations, provided significant year-round turbulent heat transfer in a stable surface layer. It was shown that wind speed increases with elevation. This counter-intuitive result was the combined effect of the larger surface roughness near the margin (Smeets and van den Broeke, 2008), the increasing influence of the large-scale pressure gradient force (Van Angelen et al., 2011), and the opposing pressure gradient force in the boundary layer during winter due to the proximity of pooled cold air over the tundra. Van As et al. (2012) quantified the extreme surface melt in the “Kangerlussuaq catchment”, including Russell glacier (Fig. 2.1), in 2010, validated by river discharge measurements. It was shown that during 2010, the ablation zone higher than 1000 m w.e. experienced the largest melt excess. At these elevations, the higher temperatures and the lower albedo contributed equally to the 2010 melt anomaly, while at lower elevations the melt excess was due to the higher temperatures alone.. 2.3 Accumulation area Above the superimposed ice area and below the dry snow area (i.e. ∼ 1750– 2500 m a.s.l.), considerable melt during summer impacts snow/firn properties, but is not sufficient to reveal bare ice. South and West Greenland have expe20.

(21) (a). (b). Figure 2.2. (a) Cutaway diagram of the ice sheet in the Kangerlussuaq region along the K-transect showing all IMAU SMB stake measurements, and IMAU AWSs marked in red. The average 1997–2002 ELA is also indicated (Van de Wal et al., 2005). The figure was originally published by Smeets and van den Broeke (2008) in their study on the regional ice sheet surface roughness. From August 2010 until April 2016, the S10 AWS operated also at the highest SMB point, at ∼ 50 m distance from the GEUStype KAN_U AWS (Paper III). (b) The latest evolution of S10 (IMAU type-III AWS; 4 April 2015).. rienced relatively high air temperatures in recent years (Van As, 2011; Van As et al., 2012, 2014; Fausto et al., 2016). In a warming climate, melt would occur at even higher elevations. The firn properties in a substantial part of the lower accumulation area would alter. Since firn lies in regions of the ice sheet where the surface slope is small, a relatively small vertical increase in melt extent would affect a large area of the ice sheet (McGrath et al., 2013). Established on 4 April 2009 at about 1840 m a.s.l. (67◦ 0 N, 47◦ 1 W) in the framework of the Greenland Analogue Project (GAP), and located approximately 140 km inland from the ice margin, KAN_U is one of the few AWSs in Greenland located in the lower accumulation area (Fig. 2.4). The 1995–2010 average SMB at the location of KAN_U (S10; Van de Wal et al., 2012) is 0.27 with 0.17 m w.e. standard deviation. Historic observations from the southwestern accumulation area include two firn cores retrieved in May–June 1989 from Site J, ∼ 36.2 km east-southeast from KAN_U (Fig. 2.3) at 2030 m a.s.l. (66◦ 51.9 N, 46◦ 15.9 W; Kameda et al., 1995). The deduced Melt Feature Percentage (MFP) in years earlier than 1946 is not more than ∼ 30 %, implying that the location did not experience significant melt events until the 1950s. Strong melt signals (∼ 40 %) were 21.

(22) 60. 75 ° N 50. 70 ° N MFP (%). 40. KAN_U. 65 ° N. Site J. 30. 20. 60 ° N 55 ° W 50 ° W. 10 °. °. W 45° W 40 W 35 0 1900. 1910. 1920. 1930. 1940 1950 year. 1960. 1970. 1980. 1990. Figure 2.3. Melt percentage data from the top part of a firn core retrieved in 1989 at Site J by the Japanese Arctic Glaciological Expedition (JAGE89; Kameda, T. et al., 2004).. identified from 1946 to the early 1960s, owing to melt in years within that decade. Melt percentage higher than 50 % were reported for the 1976, 1988, and 1989 layers, indicative of more frequent and increasing melt occuring at these elevations during the second half of the past century. In a historic study based on 10 m firn temperature observations prior to 1965, the closest site to KAN_U was ∼ 60 km south-southwest, i.e. at a lower elevation. Mock and Weeks (1966) estimated 10 m firn temperature for the elevation of KAN_U at around −14 ◦ C, suggesting that there was no significant refreezing in that period, and therefore ice content within the firn. The above observations bracket the 1840 m a.s.l. elevation contour as they provide evidence from both higher and lower elevations, and put KAN_U in a historical context. Based on these findings, in the first half of the past century, the lower accumulation area had predominantly dry ice sheet surface.. 22.

(23) Figure 2.4. The KAN_U AWS, established on the ice sheet on 4 April 2009 (credit: D. van As).. 23.

(24) 24.

(25) 3. Methods. 3.1 KAN_U automatic weather station Long-term monitoring of snow depth, ice ablation and all physical parameters controlling the energy balance of a glacial surface is possible by means of AWSs. Manufactured at GEUS, the GEUS-type AWS (Paper I) has a robust design for enduring the harsh conditions in glacier ablation areas (Fig. 3.1; Paper β ) and has evolved at GEUS since the 1980s. The evolution of the current AWS setup coincided with the 2007–2008 initialization of glacier monitoring programs of GEUS. The biggest of the programs, the Programme for Monitoring of the Greenland Ice Sheet (PROMICE), is operating about 16 AWS in seven ablation regions around the periphery of the ice sheet (Ahlstrøm and the PROMICE project team, 2008) and on Mittivakkat glacier, Southeast Greenland (Mernild et al., 2015). GlacioBasis, is operating three stations on an individual Greenland ice cap located at the A. P. Olsen Land, Northeast Greenland (Citterio et al., 2016). Lastly, within GAP, the four KAN AWSs, were established in the Kangerlussuaq region (Fig. 2.1). The GEUS-type AWS provides cost-effective, year-round observations of air pressure, relative humidity and temperature at 2.7 m height above the ice sheet surface, wind speed, and direction at 3.1m height, shortwave and longwave radiation components, snow depth and GPS position (Paper I). The measurements are complemented by aspiration of the radiation-shielded air temperature and humidity sensors and tilt monitoring of the station that allows for tilt correction of the shortwave radiation measurements (Wang et al., 2016). A watertight enclosure contains amongst other items the CR1000 data logger, the Campbell AM16/32A analog multiplexer and all supporting circuitry (Paper I). The data are stored in removable flashcard, suitable for very low temperatures. At the same time, the internal memory of the data logger stores hourly averages of all measurements, holding in excess of one year’s worth of measurements, thereby acting as backup for the flashcard. Measurements are also transmitted via satellite data telemetry providing a near-realtime overview of the conditions on the glacier. The power supply of the GEUS-type AWS is from recheargeable, sealed, lead-acid battaries (totaling 112 Ah at 12 V). A 10 W solar panel charges the battery assembly. During polar night conditions, with no solar power and low temperatures, energy conservation is ensured by regulating SBD to one per day, when daily averages of all measurements are transmitted. In the event 25.

(26) (a). (b). Figure 3.1. GEUS-type AWSs in ablation areas: (a) TAS_U, located on the ice sheet in the region of Tasiilaq (Ammassalik), Southeast Greenland (5 September 2012); (b) KAN_L, located on the ice sheet in the region of Kangerlussuaq, Southwest Greenland (28 April 2015).. of voltage drop below 11.5 V, a low-power mode activates. In this mode, the power-demanding functions aspiration, SBD, and GPS are deactivated, thus the AWS continues to operate almost as normal. Typical monthly consumption of the GEUS-type AWS is 17 Ah in summer, 11 Ah in winter and 1.3 Ah in low-power mode. The frame of the GEUS-type AWS is based on 1 and 1.5 aluminum poles that are stabilized by steel wire in tetrahedral shape, resulting in a stable, free-standing tripod. A horizontal boom at ∼ 2.9 m distance from the feet of the AWS accomodates most instrumentation. The ∼ 50 kg battery box is suspended under the mast, thereby providing stabilty by increasing the AWS weight and by lowering the centre of gravity. The tripod is especially designed for airborn transportation, as it can be folded to fit within the cockpit of a small helicopter (Fig. 3.1b). In addition to all standard meteorological parameters measured by AWSs, the observational suite of the GEUS-type AWS includes ice ablation and subsurface temperature, measured by coustom-built and calibrated GEUS instruments (Paper I). In the ablation area of a glacier, ice ablation is monitored by pressure transducer assembly (Fausto et al., 2012). The assembly consists of a hose filled with antifreeze mixture, inside which a pressure transducer is 26.

(27) (a). (b). Figure 3.2. GEUS-type AWSs in accumulation areas: (a) KAN_U, located on the ice sheet in the region of Kangerlussuaq, Southwest Greenland (28 April 2013; credit: D. van As); (b) ZAK_T, located on the ice cap in A. P. Olsen Land, Northeast Greenland (23 April 2014).. fixed at the bottom. In case of surface melt, the decrease of bottom hydraulic pressure (corrected for atmospheric pressure) is a measure of the concurrent surface lowering. This surface lowering is a direct indicator of ice loss. A valuable parameter for accumulation area monitoring as in the case of KAN_U (Fig. 3.2a) is the temperature of the subsurface. When melt occurs at the year-round, snow-covered high elevations of a glacier, the 8-level GEUS thermistor string is able to thermally trace the percolation and refreezing of meltwater in firn until 10 m depth. Essentially, the thermistor string measures the total energy input in the firn during summer, and the energy loss during winter, while indicating major refreezing events, prevailing percolation depth and timing of latent heat release. Firn temperature monitoring is also essential for validating SEB models applied in accumulation areas. While in ablation areas melt estimates are geometrically validated against the observed ice ablation every year, in the accumulation area ablation monitoring by pressure transducer assembly is not possible as a positive annual SMB will result in increasing surface height. Additionally, surface melt and percolation result in temperate snow and firn (Paper II), within which objects standing freely on the surface tend to sink. The resulting vertical motion of the sonic rangers on the AWS or on the ablation stake assembly (Fig. 3.1b) relative to the surface renders both of their signals unsuitable for SEB model validation. The thermistor string provides signals (i.e. firn temperature measurements) that are measured at fixed depths with respect to the surface during AWS establishment, and thus reference for validating SEB simulations. 27.

(28) (a). (b). Figure 3.3. (a) Drilling and (b) logging teamwork at KAN_U on 27 April 2013.. 3.2 Firn observations The Snow Processes in the Lower Accumulation Zone (SPLAZ) campaign was conducted at KAN_U during the first week of May 2012. On 1 May, firn stratigraphy and density were determined from a 10.68 m long core. Firn densities and stratigraphy were subsequently re-assessed at KAN_U on 27 April 2013, from a 19.12 m long firn core drilled in the framework of the Arctic Circle Traverse 2013 (ACT-13) campaign (Fig. 3.3), and on 5 May 2015 from a 14.4 m long firn core in the framework of the Arctic Circle Traverse 2015 (ACT-15) campaign. The ACT campaigns asssesed the firn density also at several other locations on the ice sheet (Paper IV). At each location, firn stratigraphy was assessed at 1 cm vertical resolution by visual inspection. The cores were sampled for density at 10 cm vertical resolution. The SPLAZ campaign established also on 2 May 2012 the SPLAZ station for detailed firn temperature monitoring to compliment KAN_U, at approximately 50 m distance from the AWS (Paper V). It was equiped by three high-resolution thermistor strings recording temperature down to 15 m depth. Snowpack temperature monitoring was by means of six high-precision temperature probes (Paper II). Additional parameters included radiation-shielded air temperature, surface height, and surface temperature.. 3.3 Surface energy balance model The utilized SEB model (also SEMMIS – Surface Energy and Mass budget Model for Ice Sheets) has been succesfully used in various studies accross the ablation area of the ice sheet (Van As, 2011; Van As et al., 2012, 2014). It uses AWS observations as input, and computes the turbulent energy, subsurface conductive energy, and melt fluxes. In the following, temperature unit is 28.

(29) Figure 3.4. The SPLAZ station on 4 May 2012 (credit: D. van As). Kelvin, while the time step is ten minutes to maintain stable subsurface calculations at 0.1 m spatial resolution (see Sec. 3.3.4).. 3.3.1 Meteorology Before the initiation of the SEB calculation, the preliminary calculation of several meteorological parameters takes place. These parameters are essential for accurate turbulent energy flux calculation. Specific humidity The kinematic viscocity ν is given by: ν=. μ ρa. (3.1). The dynamic viscocity μ is calculated by Sutherland’s formula: 3 Tre f +C Ta 2 μ = μre f · · Ta +C Tre f. (3.2). For air, the reference viscocity and temperature are μre f = 18.27 · 10−6 Pa s and Tre f = 291.15 K. Sutherland’s constant C for air is equal to 120 K. 29.

(30) The atmospheric density ρa is determined from the ideal gas law: ρa = 100 ·. pa Rs,d · Ta. (3.3). The air pressure is denoted by pa and Rs,d is the specific gas constant for dry air (287.058 J kg−1 K−1 ). The factor 100 is introduced to convert air pressure units from Pa to hPa. The saturation vapor pressure for frozen (si ) and liquid (sw ) water is found respectively for below- and above-freezing conditions as: si. T T −9.09718· T0a −1 −3.56654·log T0a +0.876793· 1− TTa +log(s0 ) 0 = 10 . sw = 10 ·. −7.90298·. . T100 Ta −1. +5.02808·log. ⎛. . T100 Ta. ⎛. . . T −3.49149· 100 Ta −1 +8.1328·10−3 ·⎝10. · ·. . ⎞ 11.344· 1− TTa −7 100 ⎝ −1.3816·10 · 10 −1⎠. 10 10 10log(s100 ). (3.4). · ·⎞. (3.5). −1⎠. ·. In the above equations, T0 and T100 are the temperatures at which water undergoes phase transition (273.15 K and 373.15 K), while s0 and s100 are the respective saturation vapor pressures (6.1071 hPa and 1013.246 hPa). The specific humidity at saturation qsi/w is calculated for atmospheric conditions below or above T0 respectively as: qsi/w = εm ·. si/w pa − si/w · (1 − εm ). (3.6). The molecular weight ratio of water vapour and dry air εm is equal to 0.622. The specific humidity q in units kg kg−1 is derived from qsi/w and relative humidity φ as: qsi/w (3.7) q=φ· 100 It should be noted that the instruments on the weather stations are incapable of measuring supersaturation, therefore relative humidity measurements that exceed 100 % are assumed inaccurate and set at saturation. Precipitation Snowfall occurence during the cold period of the year is prescribed in the model based on the smoothed signal from the sonic ranger mounted on the stake assembly, assuming a constant snow density ρi min . The snow temperature is calculated in the surface energy balance. 30.

(31) Rainfall occurence during the warm period of the year is assumed when Ta > T0 , and incoming longwave radiation exceeds black-body radiation using Ta (i.e. EL↓ > σ Ta4 , where σ is the Stefan-Boltzmann’s constant equal to 5.67 · 10−8 W m−2 K−4 ). Based on the snowfall estimate during the cold period, a constant precipitation rate r˙ in units m w.e. h−1 is assumed for each hydrological year. Rain is expected to be limited at the higher elevations of the ice sheet, and its temperature is not expected to deviate much from the melting point. It is therefore assumed equal to T0 , thus rain adds a small amount of mass to the SMB, and energy to the SEB. Potential temperature The measurement heights of wind speed, air temperature and humidity (respectively zUw and zTa ) are updated in every time step accounting for accumulation and ablation. The potential temperature θ is then calculated as: zT · g θ = Ta + a (3.8) ca,d The specific humidity of dry air ca,d is equal to 1005 J kg−1 K−1 , while the gravitational acceleration at sea level, and at 60◦ N is 9.82 m s2 .. 3.3.2 Surface energy balance The energy balance at the atmosphere–ice sheet interface is EM = ER + EH + EE + EP + EG,z=1. (3.9). where EH , EE , and EP are the turbulent sensible energy flux, turbulent latent energy flux, and rain-induced energy fluxes respectively. The subsurface conductive energy flux between the surface and the first subsurface level (i.e. z = 1) is Ts − Ti,z=1 (3.10) EG,z=1 = −ki,z=1 · Δz with Ts the surface temperature, Ti,z=1 the temperature at the first subsurface level in the previous time step, and Δz is the thickness of the subsurface levels. The radiation budget ER at the ice sheet surface is given by the sum of incoming and reflected solar/shortwave (ES↓ , ES↑ ) and downward and emitted terrestrial/longwave (EL↓ , EL↑ ) radiative energy fluxes: ER = ES↓ + ES↑ + EL↓ + EL↑ = ESNet + ELNet. (3.11). Fluxes are positive when directed toward the ice sheet surface. The broadband albedo is the fraction of the incoming shortwave radiation reflected at the ice sheet surface:. E↑. (3.12) α = S↓. E. S 31.

(32) By the inclusion of albedo and utilizing the Stefan–Boltzmann law, Eq. (3.11) can be rewritten as ER = (1 − α) · ES↓ + EL↓ − ε · σ · Ts4 − (1 − ε) · EL↓. (3.13). The longwave emissivity ε for snow/firn is assumed equal to 1 (black-body assumption). Rainfall is assumed to be at T0 , and thus EP is non-zero when Ts is below freezing: EP =. ρw · cw · mrain · (T0 − Ts ) Δt. (3.14). where cw is the specific heat of water, equal to 4.21 kJ kg−1 K−1 at T0 and density ρw = 999.84 kg m−3 . The right-hand side of Eq. (3.9) is solved for the one unknown variable Ts which is limited to T0 . In case of imbalance, the excess energy is attributed to melt. For sub-freezing Ts , all other SEB components are in balance and surface melt does not occur. In detail, the SEB is calculated by an iterative method. For every time step Δt, the iterative scheme is initialized by assuming Ts = T0 , and the initial surface temperature iteration step ΔTs in search for balance amongst all energy components is set to 10 ◦ C. If Ts is not equal to T0 (i.e. Ts < T0 ) while the calculation overshoots zero total for all SEB components, ΔTs becomes smaller according to a bisection EM < 0, then ΔTs = 0.5 · ΔTs prev , and SEB = EM ). In this process (i.e. if SEB prev case: • If EM < 0, then Ts = Ts − ΔTs . • If EM > 0, then Ts = Ts + ΔTs . The iteration stops either for melting surface (i.e. Ts = T0 and EM > 0) or for balance amongst all SEB components. Sensible and latent energy fluxes The surface temperature is specifically important to the calculation of the turbulent energy fluxes, occurring in every above-mentioned iteration step. In the calculations it is assumed that measured gradients in the near-surface atmosphere can be used to calculate vertical transport of heat, described below. In these calculations one important constant needs to be estimated: The surface roughness length for momentum z0 . At the higher elevations in the accumulation area of the ice sheet can be ∼ 10−4 m (Smeets and van den Broeke, 2008). During summer, due to the effect of melt, the ice sheet surface smoothes while attaining smaller value (∼ 10−5 m). Increased roughness occurs during winter 32.

(33) due to sastrugi. Additionally, drifting snow (Lenaerts et al., 2014) can increase z0 up to 10−3 m. In light of these estimates, it is considered that a constant z0 of 10−4 m is a valid year-round approximation. In every iteration of every time step, the shear velocity u∗ is initially estimated as: Uw (3.15) u∗ = κ · zUw log z0 The Von Kármán constant is set to 0.4 (reported range: 0.35 < κ < 0.42). The Reynolds number follows as: Re = u∗ ·. z0 ν. (3.16). Over aerodynamically smooth surfaces, the roughness lengths for heat zh and moisture zq are estimated according to Andreas (1987): zh = z0 · ech1 +ch2 ·log(Re)+ch3 ·log. 2 (Re). zq = z0 · ecq1 +cq2 ·log(Re)+cq3 ·log. 2 (Re). (3.17). Depending on Re, the polynomial coefficients are equal to: • Re ≤ 0.135 ch1 = 1.25 ch2 = 0 cq1 = 1.61 cq2 = 0. ch3 = 0 cq3 = 0. • 0.135 < Re < 2.5 ch1 = 0.149 ch2 = −0.550 cq1 = 0.351 cq2 = −0.628. ch3 = 0 cq3 = 0. • Re ≥ 2.5 ch1 = 0.317 cq1 = 0.396. ch3 = −0.183 cq3 = −0.180. ch2 = −0.565 cq2 = −0.512. In cases of atmospheric stillness (i.e. very low wind speeds), turbulunt energy exchange is minimal. The wind speed limit below which turbulent enery exchange is assumed negligible is set to its lowest possible value for which stable calculations occur for the entire calculation period (0.95 m s−1 ). On the ice sheet surface with temperature Ts , the saturation vapor pressure si,s and specific humidity at saturation qs i,s are:. 33.

(34) si,s. T T −9.09718· T0s −1 −3.56654·log T0s +0.876793· 1− TTs +log(s0 ) 0 = 10. qs i,s = εm ·. si,s pa − si,s · (1 − εm ). (3.18). (3.19). The calculation of the turbulent energy fluxes follows and iterative scheme based on the bulk method of the integrated flux-profile relations, in which the Obukhov length L is searched in order to find agreement between Eq. (3.15) and Eq. (3.17) that are functions of each other. For stable stratification (θ ≥ Ts ), the vertically integrated stability functions are calculated by utilizing the flux profile correction constants a = 0.7, b = 0.75, c = 5 and d = 0.35 following Holtslag and de Bruin (1988):.

(35). z0 Ψm1 = − a · zL0 + b · zL0 − dc · e−d· L + b · dc zUw

(36) z z Ψm2 = − a · ULw + b · ULw − dc · e−d· L + b · dc. 34. Ψh1.

(37). zh = − a · zLh + b · zLh − dc · e−d· L + b · dc. Ψh2. zTa

(38) z z = − a · LTa + b · LTa − dc · e−d· L + b · dc. Ψq1.

(39) z zq z = − a · Lq + b · Lq − dc · e−d· L + b · dc. Ψq2. zTa z

(40) z = − a · LTa + b · LTa − dc · e−d· L + b · dc. (3.20).

(41) For unstable stratification (θ < Ts ), the vertically integrated stability functions take the form (Paulson, 1970): 2 1+X12 1+X1 · 2 Ψm1 = log − 2 · arctan X1 + π2 2 Ψm2 = log Ψh1 Ψh2 Ψq1 Ψq2. = log = log = log = log. . 1+X2 2. 1+Y1 2. 1+Y2 2. 2. ·. 1+X22 2. . − 2 · arctan X2 + π2. 2 2 . 1+Yq1 2. 1+Yq2 2. (3.21). 2 2 . The parameters in the above are equal to: X1. =. X2. =. Y1. =. Y2. =. Yq1 = Yq2 =.

(42) 1. zUw

(43) 14 L.

(44) 1. zTa

(45) 12 L. zq

(46) 12 L.

(47) 1. 1 − γ · zL0 1−γ ·. 1 − γ · zLh 1−γ · 1−γ ·. 1 − γ · zLT. 4. 2. (3.22). 2. So, for stable and unstable (non-neutral) conditions, the calculation of Eq. (3.15) Ψm1 and Ψm2 : u∗ = κ ·. Uw zUw − Ψm2 + Ψm1 log z0 . (3.23). 35.

(48) The analogue scales for temperature and moisture scales θ∗ and q∗ for the surface layer are given by: θ∗ = κ ·. z θ−Ts log zTa −Ψh2 +Ψh1 h. q∗ = κ ·. q−qs i,s log zTqa −Ψq2 +Ψq1. (3.24). z. Finally, EH and EE are found from: EH = ρa · ca,d · u∗ · θ∗ = ρa · Ls/e · u∗ · q∗. EE. (3.25). The parameter Ls/e represents the latent heat of sublimation or evaporation of water. If the calculation is performed for Ts = T0 , then it is set to evaporation/condensation (Le = 2.50 · 106 J kg−1 ), else to sublimation/deposition (Ls = 2.83 · 106 J kg−1 ). The Obukhov length L is calculated from the expression L=. m 1 + q · 1−ε u2∗ · θ εm · m g · κ · θ∗ 1 + q∗ · 1−ε ε. (3.26). m. The iteration scheme for the turbulent energy fluxes stops when:. L previous − L. L previous < Llimit. (3.27). The limit of relative change in L between iterations is set to 0.01 m.. 3.3.3 Surface mass budget Once the energy balance at the surface has been determined, the mass components are calculated. In case of Ts = T0 , surface melt in physical distance is calculated as: ΔHm = −. EM · Δt L f · ρi,z=0. In m w.e., it equals: Δmm = −ΔHm ·. ρi,z=0 ρw. (3.28). (3.29). The latent heat of fusion of water L f is equal to 3.335 · 105 J kg−1 . The density at the surface ρi,z=0 is between ρi min (i.e. fresh snow) and ρi max , where the latter value is assumed to be the maximum ice density. 36.

(49) Evaporation/condensation for Ts = T0 is quantified in m w.e. as: Δme =. EE · Δt Le · ρw. (3.30). In case of Ts < T0 , sublimation/deposition in physical distance is quantified as: EE · Δt (3.31) ΔHs = Ls · ρi,z=0 The total surface height change in physical distance is calculated as: ΔHtotal = ΔHm + ΔHs + ΔHsnow. (3.32). The total amount of surface water is calculated in m w.e. as: Δmtotal = Δmm + Δme + Δmrain. (3.33). It should be noted that in case of melt and positive-calculated EM , ΔHm is calculated negative while Δmm positive. Also, in case of sublimation (evaporation), and thus negative EE , ΔHs < 0 (Δme < 0).. 3.3.4 Subsurface The subsurface part of the model consists of 200 vertical levels, each of thickness Δz equal to 0.1 m (20 m depth total), and is stable for time step of 10 minutes. It functions with Dirichlet boundary conditions, as the surface and 20 m temperatures are prescribed in every time step. Level properties Each level z is initialized with temperature Ti,z and density ρi,z . The specific heat of ice at each level ci,z is dependent on Ti,z , and is calculated according to Yen (1981): ci,z = 152.2 + 7.122 · Ti,z. (3.34). The effective conductivity of ice at each level ki,z is dependent on ρi,z , and is calculated according to Sturm et al. (1997): ki,z = 0.138 − 1.01 · 10−3 · ρi,z + 3.23 · 10−6 · (ρi,z )2. (3.35). Both ci,z and ki,z are updated in every time step. Refreezing scheme At a specific time step, after the SEB and SMB calculations, and in case of Ts = T0 and available Δmtotal at the surface, percolation and refreezing are estimated according to the maximum refreezing that can occur in each level 37.

(50) (Illangasekare et al., 1990). Liquid water retention is not considered. As the surface level (z = 0; first 0.1 m) is assumed to be at Ts , i.e. with no available cold content for refreezing, the process runs for z = 1 until 199. For every level z with Ti,z < T0 and ρi,z < ρi max , , i.e. cold content and pore space available, the maximum refreezing per m2 due to the available cold content, and due to the available pore volume are respectively: mr,z maxT mr,z maxV. = ci,z · =. T0 −Ti,z Lf. ρi max −ρi,z ρw. ·. ρi,z ρw. · Δz (3.36). · Δz. The comparison of the two quantities above leads to three cases, according to which Ti,z and ρi,z are updated: • The available pore volume is the limiting factor: (mr,z maxT > mr,z maxV , while Δmtotal ≥ mr,z maxV ) mr,z. = mr,z maxV. Ti,z-new. f = Ti,z-old + ci,z-old ·. L. ρi max −ρi,z-old ρi,z-old. (3.37). ρi,z-new = ρi max • The available cold content is the limiting factor: (mr,z maxT < mr,z maxV , while Δmtotal ≥ mr,z maxT ) mr,z. = mr,z maxT. c ρi,z-new = ρi,z-old · 1 + i,z-old · (T − T ) 0 i,z-old Lf Ti,z-new. (3.38). = T0. • The amount of surface water is the limiting factor: (Δmtotal < mr,z maxV , and Δmtotal < mr,z maxT ) mr,z. = Δmtotal. Ti,z-new. f = Ti,z-old + Δmtotal · ρw · ρi,z-old ·ci,z-old ·Δz. L. (3.39). ρi,z-new = ρi,z-old + Δmtotal · ρΔzw Whichever of the three cases applies at a given level, the amount of available surface water after refreezing at that depth is recalculated before the examination of the next level: Δmtotal,new = Δmtotal,old − mr,z 38. (3.40).

(51) Estimated from field observations, runoff occurs if the percolating surface water reaches a six-meter thick ice layer (i.e. 60 adjacent levels, each at ρi max ) Vertical shift and conduction scheme The shift of the layers occurs by means of interpolation of the density and temperature profiles along the column, accounting for the calculated ΔHtotal . The temperature of the first layer is updated according to the Ts from the SEB calculation, while in case of snowfall surface density is set to ρi min . The one-dimensional, subsurface conductive energy flux is explicitly calculated between adjacent levels with z > 1 by a central-difference, Euler numerical scheme as: t t Ti,z−1 − Ti,z t+Δt = −kti,z · (3.41) EG,z Δz The temperature at each level is updated as: t+Δt t Ti,z = Ti,z −. Δt ρt. i,z+ 12. · cti,z. ·. t+Δt t+Δt EG,z − EG,z+1. Δz. (3.42). The density ρi,z+ 1 is the average density between adjacent levels: 2. t ρi,z+ 1 2. =. t + ρt ρi,z i,z+1. 2. (3.43). t+Δt t+Δt The boundary condition at the bottom of the column is Ti,z=z = Ti,z=z , max max −1 thus allowing for (small) temperature change.. 3.3.5 Initialization and validation The first 4.5 years of KAN_U observations were used to calculate the SEB components. There was good agreement between observed (determined from longwave radiation emissions) and simulated Ts (i.e. 10−4 m; Paper III), indicative of accurate measurements and calculations. The geometrical validation of calculated ablation (accumulation input was derived from measurements), which is more relevant to the subsurface calculations, was based on a series of realistic assumptions. In principle, for the same amounts of melt energy per unit surface, larger surface ablation is observed for less dense snow/firn. The average observed snow density at KAN_U in spring 2012 and 2013 was 360 kg m−3 . However, for the purposes of the simulation, ρi min was rounded up to 400 kg m−3 in order to better capture the measured surface ablation in the beginning of all melt seasons, thus tuning this model constant to improve the calculations. Accordingly, the average observed density of ice was set to the common value of ρi max = 900 kg m−3 . 39.

(52) Figure 3.5. Initialization density profile for the SEB model on 4 April 2009 (Paper III), based on the observed firn densities on 2 May 2012 at KAN_U (Paper IV, V).. While the thermal initialization of the subsurface scheme was based on observed firn temperatures from April 2009, the density initialization, in lack of deep 2009 firn density profiles, was based on one of the firn cores from May 2012 (Fig. 3.5; Paper IV, V). Preliminary simulations based on the snow desnity and accumulation assumptions, indicated that from 4 April 2009 to 2 May 2012 there was lowering of the surface (i.e. negative SMB). So, on top of the 2012 firn core, accumulation had to be added, equal to the simulated surface height change (i.e. Htotal ) between 4 April 2009 and 2 May 2012. Testing the SEB model with various initialization density profiles resulted in a suitable density profile shown in Figure 3.5 (Paper III). In this optimum profile, 0.70 m of accumulation was added at the top of the 2012 firn core. Additionally, part of the upper 2012 high density firn, presumably the result of refreezing in the years before, was removed in order to achieve approximately equal surface lowering (i.e. ∼ −0.72 m). Sensitivity tests showed that this density modification of the initial core had negligible effect on model output. It is also understood that firn was de facto less dense in 2009 than in 2012. The resulting firn profile for April 2009 (Fig. 3.5) was in accordance to shallow (∼ 3 m) firn stratigraphy retrieved on site in 2009 (Paper IV, Suplementary information). The surface height simulation was validated by means of additional firn temperature observations. The KAN_U thermistor string was installed on 4 April 2009 with lowest recording thermistor at 7.41 m depth (according to the spring 2009 field report). The depths of the KAN_U firn temperatures adapted according to the simulated surface height (Fig. 3.6), were compared to the 40.

(53) Figure 3.6. SEB model validation: observed and simulated relative surface height (Paper III), and observed firn temperature for the period of observations (Paper V).. firn temperatures measured by the SPLAZ station for the SPLAZ period, from 2 May 2012 to 1 February 2013 (e.g. Fig. 3.7; Paper V). The comparison revealed good agreement, while the simulated surface height uncertainty was estimated less than 0.7 m in total for 2009 until 2012 (Paper V).. 3.4 The HIRHAM5 regional climate model Regional climate models are essential in conducting large-scale (Langen et al., 2015) or ice-sheet-wide (Fettweis, 2007; Ettema et al., 2009) mass budget investigations in Greenland. The accuracy of the estimates depends strongly amongst other factors on the physics and the parameterizations implimented in the utilized climate model. The performance of the HIRHAM5 regional climate model (Christensen et al., 2007) is discussed, based on the firn temperature observations from KAN_U and the overall conclusions about the changing properties of the ice sheet. The climate model was forced at the lateral boundaries by 6-hour interval ERA-Interim weather reanalysis data (Dee et al., 2011), while solar radiation absorbtion at the surface was regulated at the surface by daily-smoothed, MODIS-based surface albedo (Box et al., 2012). The subsurface scheme (version 7.11) of HIRHAM5 consists of 25 layers with total depth of 70 m w.e., which at the location of KAN_U, and for the period 2009–2013, is equivalent to ∼ 78 m of physical distance. The scheme accounts for heat diffusion, vertical water transport, and refreezing. The scheme 41.

(54) KAN_U thermistor string (Paper V) 0. 0.69. −1. 1.69. −2. 2.69. −3 −4. 3.69. −5 −6 −7 −8 −9. 6.69. −10 −11 −12 −13 −14. initial depth from surface (m). depth from simulated 2 May surface (m). thermistor string #2 (Paper V) 0. 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00. −1 −2 −3 −4 −5 −6 −7 −8 −9 −10. 7.50. −11 −12 −13 10.0. −14. −15. −15. −16. −16. −17. −17. −18. −18. −19. −19. −20 Jun Jul. Aug Sep Oct Nov Dec Jan Feb (°C) month in 2012. (a). −20 Jun Jul. Aug Sep Oct Nov Dec Jan Feb (°C) month in 2012. (b). Figure 3.7. Validation of the SEB model (Paper V): (a) KAN_U-measured firn temperatures with simulated surface height against (b) SPLAZ-measured firn temperatures.. also accounts for snow/firn densification based on temperature and pressure parameterizations (Vionnet et al., 2012). Maximum liquid water retention in each layer amounts to 2 % of the available pore volume, with excess water percolating deeper. The “pore close-off” density (830 kg m−3 ; Herron and Langway Jr., 1980) is assumed to be the layer density at which water is prevented from percolating further, running off instead. Before running off over this ice layer, the water lingers to allow for superimposed ice formation.. 42.

(55) 4. Results. 4.1 Snowpack evolution Before and during the extreme 2012 melt season (Fig. 4.1), the ice sheet surface at 1840 m a.s.l. remained colder than the near-surface atmosphere (measurements at ∼ 1.1 m height), suggesting a prevailing stable atmospheric stratification (Paper II). Changes in atmospheric temperature lagged on average 30–40 minutes the changes in surface temperature. The temperature difference between surface and atmosphere was typically between 0.6–1.9 ◦ C, the difference being greatest during evening hours. On 2 May, the snowcover was ∼ 0.7 m thick, a typical winter snow layer thickness for this site (Fig. 4.2; Paper II). The snow temperature at 0.2 m depth lagged the diurnal variations of surface temperature ∼ 10 hours. Between 2–26 May, the area received 0.12 m of snow, thereby further insulating the subsurface and decreasing the magnitude of temperature variations at the upper measurement depths (0.05–0.20 m). On 26 May, right before the beginning of the melt season, the snowpack temperature was at all depths close to −10 ◦ C. Following the onset of melt (27 May), snowpack warming had the form of a slow wetting front. The shallow snowpack reached a uniform 0 ◦ C temperature after six days of melt, at which point 0.15 m of snow had ablated. In theory, the cold content γ of a snowpack of thickness zmax is defined as (Cogley et al., 2011): γ=. zmax 0.

(56) ci (z) · ρi (z) · T0 − Ti (z) dz. (4.1). If the snowpack of mass mi and at Ti = −10 ◦ C, is uniform with certain density ρi , it needs energy Eh to reach 0 ◦ C equal to: Eh = ci · mi · ΔT. (4.2). This energy can be provided to the snowpack by the latent heat release associated with refreezing within its pore volume meltwater (i.e. water at 0 ◦ C) of mass mr : Er = L f · mr. (4.3) 43.

(57) (a). (b). Figure 4.1. The SPLAZ station: (a) Establishment on 2 May 2012 (credit: D. van As); (b) Maintenance visit and data retreival on 26 August 2012 by M. Eijkelboom and H. Snellen from IMAU. The comparison of the two pictures confirms the ∼ 2 m total 2012 surface lowering at 1840 m a.s.l.. Combining the two equations while assuming that the refrozen meltwater originated from thawed part of the snowpack of mass mm (i.e. mr = mm ), results in: mm ci · ΔT = ms Lf. (4.4). The above mass ratio is essentially a volume ratio under a uniform density assumption, and can thus be considered a snow height ratio: Hm ci · ΔT = Hs Lf. (4.5). The above result for ci (Ti ) = 2 · 103 J kg−1 K−1 shows that the thawing and refreezing of ∼ 6.1 % of the initial snowpack can raise its temperature to 0 ◦ C (Paper II). The comparison of the theoretical analysis to the observed snowpack evolution in the first six days of melt suggests that ∼ 30 % of the produced meltwater was used for raising the snowpack’s temperature to 0 ◦ C. This implies that the other ∼ 70 % either flowed heterogeneously toward the underlying firn or was withheld in the, already temperate, upper layers of snow by capillary forces. During a period of subfreezing surface and air conditions (8–12 June), the snowpack remained at 0 ◦ C temperature. The persistence of 0 ◦ C temperature indicated the presence of liquid water within the snow matrix. Heat from the bottom of the snowpack was eventually conducted lower toward the firn, accompanied by a significant temperature drop between 0.4–0.7 m depth with respect to the 2 May surface. The difference of the measured temperature profiles at the end of each day (00:00 UTC) with simulated, conduction-only profiles is a measure of the tem44.

(58) 1 0.95. albedo. 0.9 0.85 0.8 0.75 0.7 0.65 06 May 13 May 20 May 27 May. 03 Jun. 10 Jun 17 Jun week in 2012. 24 Jun. 01 Jul. 08 Jul. 15 Jul. 22 Jul. (a) observed snowpack evolution (Paper II) 0 −1 −2 −3 −4 −5 −6 −7 −8 −9 −10 −11 −12 −13 −14 −15 −16 −17 −18 −19 −20 −21 −22 −23 −24 −25 −26 −27 −27.7924. 0.00. initial depth from surface (m). 0.05 0.10. 0.20. 0.30. 0.40. 0.70. 06 May 13 May 20 May 27 May. 03 Jun 10 Jun week in 2012. 17 Jun. 24 Jun. 01 Jul. 08 Jul. (°C). (b). Figure 4.2. Observed surface albedo (a) and thermal evolution (b) of the snowpack (Paper II). The dark red contour signifies 0 ◦ C.. perature increase due to latent heat release from refreezing meltwater. The comparison verified maximum refreezing within the snowpack after four days of surface melt (∼ 5 kg m−2 day−1 on 30 May; Paper II). This was due to the relatively high available cold content of the snowpack at the onset of melt, that was ∼ 6 MJ m−2 . Increased refreezing rates in the snowpack (∼ 3 kg m−2 day−1 ) were verified also during the cold period of June (Paper II). Temperature decrease due to heat conduction toward the relatively cold surface was counteracted by latent heat release from refreezing of available liquid water. The effect was most prominent at depths 0.2–0.3 m where temperature gradients were largest. 45.

(59) 0.9. albedo. 0.8. 0.7. 2009 2010 2011 2012 2013. 0.6. 1. 2. 3. 4. 5. 6 7 month. 8. 9. 10. 11. 12. Figure 4.3. Seasonal cycle of surface albedo for the years 2009–2013 based on monthly averages (Paper III).. Before ablation began, the surface albedo remained between 0.8–0.9, indicative of a fresh snow cover (Fig. 4.2). From 27 May until 8 June, due to snow metamorphosis in response to melt and increasing temperature, surface albedo dropped to ∼ 0.75. Snowfall on 8 June increased the albedo to above 0.8. However, on 13 June when surface melt resumed, the albedo dropped to as low as 0.7. The shallow 2012 snowpack ablated completely on 11 July (Paper II), exposing the saturated firn underneath and reducing the surface albedo between 0.6–0.7 for the first time in the instrumental record at this elevation (Paper III).. 4.2 Radiation budget Based on the observations from the KAN_U AWS (Paper I) over the period 2009–2013, the pronounced annual cycles of solar radiative energy fluxes, and to a lesser extent also the terrestrial radiative fluxes, were identified (Paper III). Fluctuations on synoptic time scales in both cases are primarily dependent on cloud cover. In case of net solar radiation, daily values can be up to 100 W m−2 . A net solar energy flux of this magnitude is also observed at lower elevations (Van den Broeke et al., 2008). This result implies that the increasing incoming solar radiation with increasing elevation is moderated by increasing albedo with elevation. The surface of the ice sheet experiences radiative cooling throughout the year by emitting longwave radiation back to the atmosphere. The daily mag46.

No results found