Proceedings of a Symposium to Commemorate the Contributions of Professor J. Ross Mackay (1915-2014) to Science in Canada

7th Canadian Permafrost Conference Quebec City, QC 20-23 September 2015

Edited by C.R. Burn

Proceedings of a Symposium to Commemorate the Contributions of Professor J. Ross Mackay (1915-2014) to Permafrost Science in Canada

7th Canadian Permafrost Conference Quebec City, QC

20-23 September 2015

Edited by C.R. Burn

Introduction

th Professor J. Ross Mackay, who passed away in 2014 shortly before his 99 birthday, was an iconoclastic influence on geocryology in Canada and around the world. A symposium to honour his contributions to th rd our field was held at the 7 Canadian Permafrost Conference in Quebec City, 20 – 23 September 2015. th The Permafrost Conference was embedded in the 68 Canadian Geotechnical Conference. This volume is the record of the symposium. Publication of the conference papers as a full Proceedings will occur, we hope, in 2016. A companion memorial to this volume, a special issue of Permafrost and Periglacial Processes, is currently in preparation.

The majority of contributors invited to the symposium were Canadians, but a few international colleagues also presented papers that were chosen to represent particular dimensions of Ross’s career. The papers in this volume were prepared following the instructions and review procedures of the Canadian Geotechnical Conference. The conference organizers provided a template for style and a limit of 8 pages. The papers were submitted for review by a committee before acceptance and publication at the meeting. I am most grateful to Professor Richard Fortier for overseeing the review and preparation of the papers. The papers were published on flash drives given to every registrant but were not paginated or indexed. The pagination of the papers is unique to this volume. Otherwise the papers are reproduced as presented, except that errors incurred during production have been corrected, such as omission of diagrams submitted in the original copy. Copyright to each paper rests with the authors. I am grateful to all of the authors for allowing their papers to be included in this collection.

The papers have been arranged into four sections, in which they are presented in alphabetical order of the last name of the first author. They represent: (1) reflections on the life and work of Professor Mackay; (2) regional studies on permafrost in Canada; (3) discussion of hydrologic effects; and (4) examinations of problems in geocryology.

In addition to these authors, I am most grateful to Professor Antoni Lewkowicz and Dr Jerry Brown who filled spaces in the program that appeared due to emergencies before and during the symposium.

The obituary for Dr Mackay published in Arctic is reproduced with kind permission of the Arctic Institute of North America and the journal’s editor, Dr Karen McCullough.

Christopher Burn Carleton University Ottawa, ON December 2015

ARCTIC VOL. 68, NO. 1 (MARCH 2015) P. 129 – 131 http://dx.doi.org/10.14430/arctic4464

J. ROSS MACKAY (1915 – 2014)

John Ross Mackay, Canada’s pre-eminent authority on per- mafrost, died peacefully in the early morning of 28 October 2014. He was nearly 99 years old. Throughout the geocry- ological community, Ross was known as an exemplary researcher, an audacious feld scientist, and a loyal and friendly man. For more than three decades from the early 1960s, he was acknowledged as Canada’s pre-eminent Arc- tic scientist. His feld research in the western Arctic began in 1951 and continued without interruption from 1954 to 2011, although he offcially “retired” from the University of (UBC) in 1981. He published 201 scien- tifc papers and two memoirs in toto, all but 13 as sole or senior author. His work is the benchmark on thermal con- traction cracking in permafrost and on pore-water expul- sion during freezing of sands. The permafrost community thus knows him best for his work on ice wedges and , but his expertise on terrain conditions in the western Arctic J. Ross Mackay on the East Channel of the Mackenzie River, July 2010. was perhaps of more immediate material signifcance. He was told in the 1970s that his work had saved industry two equivalent to his academic research. His brother, Leslie, full years of investigations in preparation for hydrocarbon joined the RCAF, and was lost over the English Channel in development in the region. 1942. He mourned his brother, taken prematurely like so Ross was born in Tamsui, Formosa (now Taiwan), to many others, for the rest of his life. Canadian missionary parents. His family is renowned in Ross joined McGill University as an assistant profes- Taiwan, particularly for the extensive medical and educa- sor in 1946 and at the same time started his PhD research, tional work initiated by Ross’s paternal grandfather, George enrolled at the University of Montreal. His “offce” was a Leslie Mackay, who founded the Presbyterian Church desk in the department library. During the regular session there. Ross’s paternal grandmother, Tiuⁿ Chhang-miâ, was he taught large classes, dominated by veterans, using only a Taiwanese and this, along with his time in Tamsui, gave blackboard and chalk. He then joined the summer school at him a lifelong affnity with the country. When Ross visited Stanstead, QC, where he met V. Stefansson and N.E. Odell. Taiwan, he was treated as an icon, with people, particularly His PhD thesis (1949) was on the regional geography of the nurses-in-training, queuing up to be photographed with lower Ottawa valley, but the major paper that he published him. In Tamsui, where he had developed a fondness for ani- from it is primarily a robust reinterpretation of the existing mals and atlases as a child, he was given the key to the city. theory on the origin of the landscape. As early as 1949, he His paper in Economic Geography (1951) declared “geogra- demonstrated a trademark ability to read the landscape in phy of the Far East” as one of his research interests. historical terms and to set straight the published record. He Ross was sent to school at the Canadian Academy in moved to UBC the same year, where he then remained. He Kobe, Japan, and went on to Clark (BA, 1939) and Bos- was in the offce daily from seven to fve, and nine to fve ton (MA, 1941) universities, where he shone. The Second on Saturdays. There he was a reserved man, but an inspira- World War interrupted his academic career; nevertheless, tion and mentor to colleagues with interests quite different he regarded his military service in 1941 – 46 as his greatest from his own, always considering how their work could be contribution. He enlisted as a gunner (private) in the Royal incorporated into his research. Canadian Artillery, and transferred to the Canadian Intel- Ross began his Arctic research in the Darnley Bay area ligence Corps in 1942. He spent the war breaking Japanese of the western Arctic coast, providing interpretive keys that codes, particularly with respect to routing of messages, were urgently needed by the newly formed federal Geo- serving in Ottawa in the same offce as Diamond Jenness. graphical Branch for aerial photographs, given the recent A formal history of his unit described him quite simply availability of stereoscopic coverage. He was hired for his as “brilliant.” In 1945, an American general told him that stellar reputation and as one of the very few Canadians who his work had saved a division (more than 15 000 men). He had advanced training in physical geography at that time. ended his service with the rank of major, as Commanding He walked south and east from Paulatuk with his assis- Offcer No. 1 Discrimination Unit, Directorate of Military tants J. Keith Fraser, from the Geographical Branch, and Intelligence, Ottawa, having commanded the Intelligence Joe Thrasher, from the community, accompanied by six Unit, No. 1 Canadian Special Wireless Group, Darwin, pack dogs. They had no maps or radios. For navigation, they Australia, in 1945. His subsequent career was founded on relied on the aerial photos, a compass, and Joe’s knowledge the reputation he had acquired in the Intelligence Corps. of the terrain. The journey began a life-long friendship with Ross described his activities in cryptanalysis as generically the people of Paulatuk. It also led to Ross’s frst paper on 130 • OBITUARY

Arctic , published in 1952. On the journey north from Yellowknife by foat plane, the pilot, Ernie Boffa, had been concerned that the water bodies were still frozen about 200 miles short of their destination, so he turned back and dropped down on a small lake, left Ross and Keith there, and said he would return in a week’s time. Neither Boffa nor Mackay had a map! In 1960, Ross published observations made during that hiatus. The lake now has a permanent name on the maps of Canada. It is called Stopover Lake. During the 1950s, after a season on Cornwallis Island in 1952, Ross began to accumulate feld experience through- out the western Arctic, traveling by boat along the coast east and west of the Mackenzie delta and walking inland. Most of his journeys were by freighter canoe, driven by J. Ross Mackay recording the deformation of an antisyngenetic at a 5 or 10 hp motor, and by schooner. He ran feld parties Garry Island, June 1992. for the Geographical Branch, four times with John Stager, usually with someone from the region as guide and assis- feld measurements on the dynamics of thermal-contrac- tant. His published record from the period was dominated tion cracking and ice-wedge polygon development. Many by contributions to cartography, although from 1956 on the guests, especially his UBC colleagues who spent a week work in Arctic physical geography became consolidated. or so with him at the cabin, were struck by the warmth of His Anderson River memoir, published by the Geographi- his personality and care for their comfort. They became cal Branch in 1958, represents feld observation at the high- immersed in talk of permafrost and the Arctic, commonly est level, but perhaps it was not as infuential as his paper sustained by smoked oysters and lubricated by Glenlivet. in the Geographical Bulletin proposing a ice-thrust The lights of drill rigs on artifcial islands twinkled in the origin for Herschel Island (1959), which drew on substantial Beaufort Sea, not far away. reading of structural with W.H. (Bill) Mathews, his Ross had no mentor to follow, just Leffngwell’s 1919 friend and collaborator. memoir on the Canning River region in adjacent Alaska, Ross was in the vanguard of applying quantitative meth- travelers’ accounts, such as those by Stefansson, and friends ods in geographic research. His wartime expertise was in in the Arctic Institute of North America, particularly Link quantitative analysis, and the cartographic papers followed and Tahoe Washburn. Throughout he drew maps: maps naturally from this. These articles are penetrating and suc- of pingos, maps of ice-wedge polygons, maps of active cinct. They set the stage for his application of quantitative layer depth, usually by plane table. When new technology methods to permafrost science in the early 1960s, which appeared, he used it imaginatively. In 1969–71, he placed he turned into a pivotal era in Canadian geocryology. His thermistors to measure the temperature below the perma- fundamental paper on pingos published in 1962 and the frost table at 10 sites near the Garry cabin and ran several Mackenzie Delta memoir of 1963, also published by the hundred metres of Z-cable into the building to capture the Geographical Branch, were magisterial quantitative state- data on chart recorders. The charts were converted by hand ments on the form and function of the landscape. The paper into daily values, giving us the frst quantitative assessment on pingos is neglected today because few currently study of the effect of snow accumulation on annual mean ground these features and because the Geographical Bulletin is temperatures at local scale, published in 1974. not online. In contrast, the memoir remains the authorita- Ross’s writing is renowned for its clarity and simplicity. tive resource on the western Arctic and is regularly cited in It was not always evident that the explanations offered in research on the region. We should remind ourselves that the the papers were the result of exhaustive examination of all numerous diagrams and extensive computations presented potential alternatives, but those who challenged the analy- were made before the advent of personal computers or even ses were subsequently left in no doubt as to why such ideas, electronic calculators. Slide rules and log tables were tools after consideration, had been rejected. Some of the clearest of the trade, and all of the maps and diagrams were drawn examples we have of the experimental method implemented by hand. It is astonishing to fnd seven publications, includ- in feld science followed from his method of multiple work- ing that memoir, on Ross’s curriculum vitae for 1963. ing hypotheses when a conclusion required fnal testing. Any retrospective view of Ross’s career must empha- The use of snow fences (1978), puncturing of pingos (1977), size his commitment to long-term investigation of geomor- and drainage of Lake Illisarvik in 1978 are three examples phological processes. His feld cabin at Garry Island was of experiments conducted at landform scale in a manner built in 1964 to enable investigation in winter. He designed that is rarely replicated. Ross devoted major effort to sta- feld installations suffciently robust to handle Arctic con- tistical summary of his observations, but his explanations ditions, simple enough to be read under arduous condi- were almost always analytical. He rarely used numerical tions, and yet critically informative of the processes under techniques, primarily because of his adherence to Occam’s investigation. The cabin allowed him to make year-round razor. He often stated baldly, “the simplest is the best!” OBITUARY • 131

It is remarkable for a feld scientist to present the prime received the Massey (1967), Miller (1978), Vega (1986), and contributions of his career while closing in on retirement. Logan (1991) medals; fve honorary doctorates, including But 1970–81 provided Ross’s classical treatment of distur- one from the University of Helsinki; and he was elected to bance to permafrost terrain (1970); segregation as the ori- eight societies and academies, including the Russian Acad- gin of massive ice (1971); preservation of permafrost and emy of Natural Sciences. He was the frst Canadian to be ground ice beneath the Laurentide ice sheet (1972); the recognized as an honorary fellow by the International Asso- origin of offshore permafrost (1972); the development of ciation of Geomorphologists (1993). The Geological Survey pingos from pore-water expulsion (1973); the infuence of of Canada named its inland waters geophysical research snow cover on ground temperatures (1974); the character- vessel MV J. Ross Mackay. Ross was one of only two peo- istics of thermal contraction cracking (1974); experimen- ple who have been elected president of the Canadian Asso- tal demonstration of growth (1977); identifcation of ciation of Geographers and the Association of American an early-Holocene thaw unconformity (1978); the origin of Geographers. From 1983 to 1993 he served as Secretary- hummocks (1980); and plug-like fow in solifuction (1981). General to the International Permafrost Association in its In the same period a number of his trademark short notes formative years. He was quietly proud of his seven Arctic on key points or new techniques appeared: on aggradational namesakes, all descendants of Arctic feld associates. ice (1971); a design for a frost tube (1973); regional ground Ross had met R.M. Anderson during the war, and at the temperatures in the Mackenzie delta area (1974); relict ice Andersons’ home was introduced to Violet Meekins. They wedges (1975 and 1976); permafrost and climate change were married for 53 years, but, sadly, she passed away in (1975); problems with probing the base of the 1997. Their daughters Anne and Leslie live in British (1977); measurement of the widths of ice wedges in oblique Columbia. Ross and Violet enjoyed the outdoors, especially section (1977); the effects of wild fre on the active layer birding, and they loved the dogs that shared their home. (1977); frost heave below 0˚C (1979); and the frst report on Violet traveled north with Ross many times in summer, Illisarvik (1981). These publications demonstrated a distinct staying at Garry Island and Illisarvik, and once in winter loyalty to the Canadian research presses. In the 1950s and at Paulatuk. She insisted that Ross take many of his UBC 60s, Ross had patronized the Geographical Bulletin, but colleagues to the North so that they might glimpse the real- transferred to the Canadian Journal of Earth Sciences ity of his science. Anne was his principal support after her when the Bulletin ceased publication and to the Geological mother’s death. Survey of Canada’s Reports of Activities (from 1978 called Ross’s life was his work. He did not court publicity Current Research). nor engage in self-promotion. His record is due to single- Ross’s formal retirement was merely a punctuation minded pursuit of excellence, meticulous planning, strate- mark in his research record. He steered investigations of gic selection of observations, and unfettered curiosity. The aggrading permafrost at Illisarvik, the lake he intentionally work was received with excitement across a spectrum of drained on Richards Island in 1978, turning it into the cir- disciplines in northern science, from biology and Quater- cumpolar world’s longest-running feld experiment. From nary studies to physical geography and engineering. His 1982 to 2011 he published 52 papers, concentrating on tun- arguments were precise and exhaustive: they could be with- dra lakes, ice wedges, and pingos. Investigations of the lat- ering if he was challenged in the press, but were always ter two features culminated in two lengthy contributions to considered and collegial in private correspondence. It was the journal Géographie physique et Quaternaire (1998 and his brilliance in interpreting the landscape that shook most 2000). (GpQ was the successor name of Revue canadienne people: imagine an intellect that could make hills bounce de géographie, in which Ross published his frst paper in (Pulsating Pingos, 1977)! We will remember him fondly 1947.) Several of the papers drew on decades of feld meas- and with the utmost respect. urements, presenting a temporal perspective on geomorphic processes that will likely remain unmatched. In 2005, he published observations on wind-abraded rocks at Paulatuk, ACKNOWLEDGEMENTS including data he had collected in 1951, and in 2011 his last paper discussed a pingo near Paulatuk that had originally The author is indebted to several of J. Ross Mackay’s been photographed by Stefansson in 1911. His Arctic feld colleagues at UBC, particularly Michael Church, Margaret research began and ended at Paulatuk. North, Olav Slaymaker, John Stager, and Graeme Wynn, for The distinction that permeated Ross Mackay’s research points of information and context. was recognized through numerous awards and medals. He was the frst recipient of the International Permafrost Asso- C.R. Burn ciation’s (IPA) Lifetime Achievement Award (2010), of Department of Geography and Environmental Studies Canada’s Centenary Medal for Northern Science (1984), of Carleton University the Roger J.E. Brown Award of the Canadian Geotechni- 1125 Colonel By Drive cal Society (1986), and of the Award for Scholarly Distinc- Ottawa, Ontario K1S 5B6, Canada tion of the Canadian Association of Geographers (1972). He [email protected] was made an Offcer of the Order of Canada in 1981 and

TABLE OF CONTENTS

Part 1. Reflections on the life and work of Professor Mackay

The geocryological bibliography of J. Ross Mackay (1915-2014) 1 C.R. Burn

John Ross Mackay – Devoted tutor and best friend of Chinese permafrost research 9 Guodong Cheng, Zhijiu Cui, Baolai Wang, and Huijun Jin

Reconstructing geomorphology: an appreciation of the contributions of J. Ross Mackay 15 Michael Church

JRM: His early accomplishments, 1950s to the mid 1980s, and his results, 1990 to 2005 23 Hugh French

Science to Technology – The importance of understanding the fundamentals of permafrost science for engineers practicing in the north 29 Don W. Hayley

Part 2. Regional studies on permafrost in Canada

Geohazard investigations of permafrost and gas hydrates in the outer shelf and upper slope of the Canadian Beaufort Sea 37 S.R. Dallimore, C.K. Paull, A.E. Taylor, M. Riedel, H.A. MacAulay, M.M. Côté, and Y.K. Jin

Late Wisconsin glaciation of Hadwen and Summer islands, Tuktoyaktuk Coastlands, NWT, Canada 43 Julian B. Murton, Mark D. Bateman, Richard I. Waller, and Colin Whiteman

Holocene lake-level recession, permafrost aggradation and formation in the Yellowknife area, Great Slave Lowland 49 S.A. Wolfe and P.D. Morse

Part 3. Hydrologic effects in permafrost regions

Estimating depth beneath lakes in Arctic Alaska 59 Kenneth M. Hinkel and Christopher Arp

Lakes of the western Canadian Artic: Past Controls and Future Changes 65 Philip Marsh, Tyler de Jong, Lance Lesack, Cuyler Onclin, and Mark Russell

Meteorological and geological influences on icing dynamics in subarctic Northwest Territories, Canada 73 P.D. Morse and S.A. Wolfe

Geometry of oriented lakes in Old Crow Flats, northern Yukon 81 P. Roy-Léveillée and C.R. Burn

Summer and winter flows of the Mackenzie River system 89 Ming-ko Woo and Robin Thorne

Part 4. Problems in geocryology

Subdivision of ice-wedge polygons, western Arctic coast 97 C.R. Burn and H.B. O’Neill

Retrogressive thaw slumps: From slope process to the landscape sensitivity of northwestern Canada 103 S.V. Kokelj, J. Tunnicliffe, D. Lacelle, T.C. Lantz, and R.H. Fraser

Geochemistry of the active layer and permafrost in northwestern Canada: from measurements to Quaternary 111 Denis Lacelle, Marielle Fontaine, and Steve V. Kokelj

Topoclimatic controls on active-layer thickness, Alaskan Coastal Plain 119 Frederick E. Nelson and Melanie A. Schimek

Permafrost degradation adjacent to snow fences along the Dempster Highway, Peel Plateau, NWT 127 H.B. O’Neill and C.R. Burn

The thermo-mechanical behaviour of frost cracks over ice wedges: new data from extensometer measurements 135 Denis Sarrazin and Michel Allard

Part 1 Reflections on the life and work of Professor Mackay

The geocryological bibliography of J. Ross Mackay (1915-2014)

C.R. Burn Department of Geography and Environmental Studies - Carleton University, Ottawa, ON, Canada

ABSTRACT For more than four decades Professor J. Ross Mackay was the Canadian authority in permafrost science, and was internationally recognized for his contributions to geocryology. This paper presents the research bibliography of J.R. Mackay with respect to the permafrost environment, and also his contributions on the hydrology of Mackenzie River and his studies of and snow in southern British Columbia.

RÉSUMÉ Durant plus de quatre décennies le Professeur J. Ross Mackay était l’autorité canadienne en matière de science du pergélisol, et était reconnu internationalement pour ses contributions en géocryologie. Cet article présente la bibliographie scientifique de J.R. Mackay en lien avec les environnements de pergélisol, ainsi que ses travaux sur l’hydrologie de la rivière Mackenzie et ses études relatives aux aiguilles de glace et de neige dans le sud de la Colombie Britannique.

1 INTRODUCTION of pingos from pore-water expulsion (1973); the characteristics of thermal contraction cracking (1974); J. Ross Mackay, Canada’s leading permafrost scientist, experimental demonstration of pingo growth (1977); and died on 28 October 2014. He was nearly 99 years old. the origin of hummocks (1980). Professor Mackay’s Arctic field research began in the After his formal retirement in 1981 he published 52 Paulatuk area, N.W.T., in summer 1951 and continued papers concentrating on lakes, ice wedges, and without interruption from 1954 to 2011. He focussed his pingos. Two of these papers, reporting long-term studies work on the western Canadian Arctic, especially the of pingo growth (1998) and the development of ice-wedge Tuktoyaktuk Coastlands. His Arctic research was wholly polygons (2000), are of monograph length. They were conducted from the Department of Geography at the published in Géographie physique et Quaternaire, the University of British Columbia. The permafrost community successor name for the Revue Canadienne de knows him best for his work on pingos and ice wedges, Géographie, in which he had published his first papers (on but his expertise on terrain conditions in the western the Ottawa River valley in 1947 and 1949). In 2005, he Arctic was wide ranging. He began his field research published observations on wind-abraded rocks at taking a physiographic approach to the landscape, and Paulatuk, including data he had collected in 1951, and in during the 1950s developed a mechanistic view, working 2011 his last paper discussed a pingo near Paulatuk closely with W.H. Mathews on the structural geology of originally photographed by Stefansson in 1911. permafrost sediments. In the 1960s he revolutionized In total, Dr Mackay published 201 research papers Canadian geocryology with his physically based approach and two memoirs. About 150 of the papers focus on the to the study of landforms. permafrost environment. He also published extensively on Mackay’s first major contribution to geocryology cartography. He was the sole or senior author of all but 9 concerned glacier ice-thrusting of permafrost sediments to of the papers concerning permafrost. Here I present account for the deformed beds of Nicholson Peninsula Professor Mackay’s full geocryological bibliography, with (1956) and Herschel Island (1959). He based most of his the exception of contributions published as abstracts. The geocryological publications in the 1950s on field bibliography is divided into several categories, which are observations, but in 1962 he presented an assessment of presented in chronologic order of the first paper. The list talik geometry as part of a seminal paper on pingos. This also includes 13 papers regarding freeze-up, break-up, paper and the Mackenzie Delta memoir (1963) were and mixing of Mackenzie River, and eight papers magisterial quantitative statements on the form and reporting research on snow and needle ice, conducted in function of the landscape. From 1959 onwards he had collaboration with W.H. Mathews in southern BC. used computers to assist his statistical examinations of Throughout his career, Dr Mackay was loyal to the freeze-up and break-up of Mackenzie River, which Canadian research presses. His initial work on the North accompanied his transition into fundamental studies of was published in particular in the Geographical Bulletin. geomorphic processes. His work developed along with After this stopped publishing, the Canadian Journal of federal and industrial interest in the western Arctic, so that Earth Sciences became his principal journal. During the 1970-1980 provided a series of key papers, e.g., the 1970s and 1980s, he presented many short papers in the classical treatment of disturbance to permafrost terrain Geological Survey of Canada’s Report of Activities, after (1970); segregation as the origin of massive ice (1971); 1978 called Current Research. A detailed obituary for Dr the origin of offshore permafrost (1972); the development Mackay was published earlier this year (Burn 2015).

1 Mackay, J.R. 1972. Offshore permafrost and ground ice, southern Beaufort Sea, Canada. Canadian Journal of 2 MONOGRAPHS Earth Sciences, 9: 1550-1561. Mackay, J.R. 1972. Permafrost and ground ice. In Mackay, J.R. 1958. The Anderson River map area, Proceedings, Canadian Northern Pipeline Research N.W.T. Geographical Branch, Department of Mines Conference, 2-4 February, 1972. National Research and Technical Surveys, Ottawa, ON. Memoir 5, 137 p. Council of Canada, Associate Committee on Mackay, J.R. 1963. The Mackenzie Delta area, N.W.T. Geotechnical Research, Ottawa, ON. Technical Geographical Branch, Department of Mines and Memorandum 104: 235-239. Technical Surveys, Ottawa, ON. Memoir 8, 202 p. Mackay, J.R., Rampton, V.N. and Fyles, J.G. 1972. Relic Pleistocene permafrost, western Arctic, Canada. Science, 176: 1321-1323. 3 PHYSIOGRAPHY AND LANDSCAPE HISTORY Mackay, J.R. and Mathews, W.H. 1973. Geomorphology and Quaternary history of the Mackenzie River valley, Mackay, J.R. 1952. Physiography of the Darnley Bay near Fort Good Hope, N.W.T., Canada. Canadian area, N.W.T. The Canadian Geographer, 1(2): 31-34. Journal of Earth Sciences, 10: 26-41. Mackay, J.R. 1953. Post-glacial drainage changes in the Mackay, J.R. 1976. Pleistocene permafrost, Hooper Darnley Bay area, N.W.T., Canada. Association of Island, Northwest Territories. In Report of Activities, Pacific Coast Geographers, Yearbook, 15: 17-22. Part A. Geological Survey of Canada, Ottawa, ON. Mackay, J.R. 1955. Physiography. In Geography of the Paper 76-1A: 17-18. Northlands. Kimble G.H.T. and Good D. (eds). Mackay, J.R. 1978. Freshwater shelled invertebrate American Geographical Society and John Wiley & indicators of paleoclimate in north-western Canada Sons: New York, NY. pp. 11-35. during late glacial times: Discussion. Canadian Mackay, J.R. 1956. Deformation by glacier-ice at Journal of Earth Sciences, 15: 461-462. Nicholson Peninsula, N.W.T., Canada. Arctic, 9: 218- Mackay, J.R. 1978. Quaternary and permafrost features, 228. Mackenzie delta area. In Geological and geographical Mackay, J.R. 1956. Mackenzie deltas Ð a progress report. guide to the Mackenzie Delta area. Young, F.G. (ed.). The Canadian Geographer, 2(7): 1-12. Canadian Society of Petroleum Geologists, Calgary, Mackay, J.R. and Mathews, W.H. 1956. Surficial geology AB. pp. 42-50. of the Firth River archaeological site, Yukon Territory. Mackay, J.R. 1981. Dating the Horton River breakthrough, Arctic, 9: 210-211. District of Mackenzie. In Current Research, Part B. Mackay, J.R. 1958. The valley of the lower Anderson Geological Survey of Canada, Ottawa, ON. Paper 81- River, N.W.T. Geographical Bulletin, No. 11: 37-56. 1B: 129-132. Mackay, J.R. 1959. Glacier ice-thrust features of the Mackay, J.R. 1986. The permafrost record and Yukon coast. Geographical Bulletin, No. 13: 5-21. Quaternary history of northwestern Canada. In Mackay, J.R. 1960. Crevasse fillings and ablation slide Correlation of Quaternary deposits and events around moraines, Stopover Lake area, N.W.T. Geographical the margin of the Beaufort Sea: Contributions from a Bulletin, No. 14: 89-99. joint Canadian-American workshop, April 1984. Mathews, W.H. and Mackay, J.R. 1960. Deformation of Heginbottom, J.A. and Vincent, J-S. (eds). Geological by glacier ice and the influence of pore pressures Survey of Canada, Ottawa, ON. Open File Report and permafrost. Transactions of the Royal Society of 1237: 38-40. Canada, Series 3, 54: 27-36. Mackay, J.R. and Slaymaker, O. 1989. The Horton River Mackay, J.R. and Terasmae, J. 1963. Pollen diagrams in breakthrough and resulting geomorphic changes in a the Mackenzie Delta area, N.W.T. Arctic, 16: 229-238. permafrost environment, western Arctic coast, Mackay, J.R. 1964. Arctic landforms. In The Unbelievable Canada. Geografiska Annaler, 71A: 171-184. Land. Smith, I.N. (ed.). Queen’s Printer, Ottawa, ON. Mathews, W.H., Mackay, J.R. and Rouse, G.E. 1989. pp. 60-62. Pleistocene geology and geomorphology of the Mackay, J.R. and Mathews, W.H. 1964. The role of Smoking Hills Upland and lower Horton River, Arctic permafrost in ice-thrusting. Journal of Geology, 72: coast of mainland Canada. Canadian Journal of Earth 378-380. Sciences, 26: 1677-1687. Mathews, W.H. and Mackay, J.R. 1965. Ice-pushed Mackay, J.R. and Burn, C.R. 2005. A long-term field study ridges, permafrost, and drainage: a reply. Journal of (1951-2003) of ventifacts formed by katabatic winds at Geology, 73: 896. Paulatuk, western Arctic coast, Canada. Canadian Mackay, J.R. 1969. The Mackenzie delta. Canadian Journal of Earth Sciences, 42: 1615-1635. Geographical Journal, 57: 146-155. Mackay, J.R. 1970. Recent and late Pleistocene permafrost in the western Arctic of Canada. In 4 ICE WEDGES Selected Papers, Physical Geography, 21st International Geographical Congress, New Delhi, Mackay, J.R. 1953. Fissures and mud circles on 1968. Chatterjee, S.P. and Gupta S.D. (eds). National Cornwallis Island, N.W.T. The Canadian Geographer, Committee for Geography, Calcutta, India. Vol. 1: 72- 1(3): 31-37. 74.

2 Mackay, J.R. 1972. Some observations on ice-wedges, Mackay, J.R. 1989. Ice-wedge cracks, western Arctic Garry Island, N.W.T. In Mackenzie Delta area coast. The Canadian Geographer, 33: 365-368. monograph, 22nd International Geographical Mackay, J.R. 1990. Some observations on the growth and Congress. Kerfoot, D.E. (ed.). Brock University, St deformation of epigenetic, syngenetic, and Catharines, ON. pp. 131-139. antisyngenetic ice wedges. Permafrost and Periglacial Mackay, J.R. 1973. Winter cracking (1967-1973) of ice- Processes, 1: 15-29. wedges, Garry Island, N.W.T. (107 C). In Report of Mackay, J.R. 1992. The frequency of ice-wedge cracking Activities, Part B. Geological Survey of Canada, (1967-1987) at Garry Island, western Arctic coast, Ottawa, ON. Paper 73-1 Part B: 161-163. Canada. Canadian Journal of Earth Sciences, 29: Mackay, J.R. 1974. Ice-wedge cracks, Garry Island, 236-248. N.W.T. Canadian Journal of Earth Sciences, 11: 1366- Mackay, J.R. 1993. The sound and speed of ice-wedge 1383. cracking, Arctic Canada. Canadian Journal of Earth Mackay, J.R. 1974. The rapidity of tundra polygon growth Sciences, 30: 509-518. and destruction, Tuktoyaktuk Peninsula-Richards Mackay, J.R. 1993. Air temperature, snow cover, creep of Island area, N.W.T. In Report of Activities, Part A. frozen ground, and the time of ice-wedge cracking, Geological Survey of Canada, Ottawa, ON. Paper 74- western Arctic coast. Canadian Journal of Earth 1 Part A: 391-392. Sciences, 30: 1720-1729. Mackay, J.R. 1975. The closing of ice-wedge cracks in Mackay, J.R. 1995. Ice wedges on hillslopes and landform permafrost, Garry Island, N.W.T. Canadian Journal of evolution in the late Quaternary, western Arctic coast, Earth Sciences, 12: 1668-1674. Canada. Canadian Journal of Earth Sciences, 32: Mackay, J.R. 1975. Relict ice wedges, Pelly Island, 1093-1105. N.W.T. (107 C/12) In Report of Activities, Part A. Mackay, J.R. 2000. Thermally induced movements in ice- Geological Survey of Canada, Ottawa, ON. Paper 75- wedge polygons, western Arctic coast: a long-term 1 Part A: 469-470. study. Géographie physique et Quaternaire, 54: 41-68. Mackay, J.R. 1976. Ice wedges as indicators of recent Mackay, J.R., and Burn, C.R. 2002. The first 20 years climatic change, western Arctic coast. In Report of (1978/79 to 1998/99) of ice-wedge growth at the Activities, Part A. Geological Survey of Canada, Illisarvik experimental drained lake site, western Arctic Ottawa, ON. Paper 76-1A: 233-234. coast, Canada. Canadian Journal of Earth Sciences, Mackay, J.R. 1976. The growth of ice wedges (1966- 39: 95-111. 1975), Garry Island, N.W.T., Canada. In Geomorphology and Paleogeography, 23rd International Geographical Congress, Moscow, 1976. 5 TUNDRA LAKES Aseyev, A.A., Velitchko, A.A. and Spasskaya, I.I. (eds). Pergamon, Oxford, UK. Section 1, 180-182. Mackay, J.R. 1956. Notes on oriented lakes of the Mackay, J.R. 1977. The widths of ice wedges. In Report Liverpool Bay area, N.W.T. Revue Canadienne de of Activities, Part A. Geological Survey of Canada, Géographie, 10: 169-173. Ottawa, ON. Paper 77-1A: 43-44. Mackay, J.R. 1957. Les lacs orientés de la regional de la Mackay, J.R. 1978. The use of snow fences to reduce ice- Baie de Liverpool: Discussion. Revue Canadienne de wedge cracking, Garry Island, Northwest Territories. In Géographie, 11: 175-178. Current Research, Part A. Geological Survey of Mackay, J.R. 1972. Lake bottom freezing and periglacial Canada, Ottawa, ON. Paper 78-1A: 523-524. effects in western Canadian Arctic. In Selected Mackay, J.R. 1980. Deformation of ice-wedge polygons, Papers, Regional Geography and Cartography, 21st Garry Island, Northwest Territories. In Current International Geographical Congress, New Delhi, Research, Part A. Geological Survey of Canada, 1968. Chatterjee, S.P. and Gupta S.D. (eds). National Ottawa, ON. Paper 80-1A: 287-291. Committee for Geography, Calcutta, India. Vol. 4: 118- Mackay, J.R. and Matthews, J.V., Jr. 1983. Pleistocene 119. ice and sand wedges, Hooper Island, Northwest Mackay, J.R. 1981. An experiment in lake drainage, Territories. Canadian Journal of Earth Sciences, 20: western Arctic coast. In Current Research, Part A. 1087-1097. Geological Survey of Canada, Ottawa, ON. Paper 81- Mackay, J.R. 1984. The direction of ice-wedge cracking in 1A: 63-68. permafrost: downward or upward? Canadian Journal Mackay, J.R. 1985. Permafrost growth in recently drained of Earth Sciences, 21: 516-524. lakes, western Arctic coast. In Current Research, Part Mackay, J.R. 1986. The first 7 years (1978-1985) of ice B. Geological Survey of Canada, Ottawa, ON. Paper wedge growth, Illisarvik experimental drained lake site, 85-1B: 177-189. western Arctic coast. Canadian Journal of Earth Mackay, J.R. 1988. Catastrophic lake drainage, Sciences, 23: 1782-1795. Tuktoyaktuk Peninsula area, District of Mackenzie. In Mackay, J.R. 1988. Ice-wedge growth in newly aggrading Current Research, Part D. Geological Survey of permafrost, western Arctic coast, Canada. In Canada, Ottawa, ON. Paper 88-1D: 83-90. Permafrost, Fifth International Conference, August 2- Mackay, J.R. 1992. Lake stability in an ice-rich permafrost 5, 1988. Senneset, K (ed.). Tapir, Trondheim. Vol. 1: environment: examples from the western Arctic coast. 809-814. In Arctic ecosystems in semi-arid regions: Implications for resource management. Robarts R.D. and Bothwell

3 M.L. (eds). Environment Canada, Saskatoon, SK. Activities, Part B. Geological Survey of Canada, NHRI Symposium Series 7: 1-25. Ottawa, ON. Paper 77-1B: 273-275. Mackay, J.R. 1997. A full-scale field experiment (1978- Mackay, J.R. 1979. An equilibrium model for hummocks 1995) on the growth of permafrost by means of lake (non-sorted circles), Garry Island, Northwest drainage, western Arctic coast: a discussion of the Territories. In Current Research, Part A. Geological method and some results. Canadian Journal of Earth Survey of Canada, Ottawa, ON. Paper 79-1A: 165- Sciences, 34: 17-33. 167. Mackay, J.R. and Burrous, C. 1979. Uplift of objects by an upfreezing surface. Canadian Geotechnical Journal, 6 THE ACTIVE LAYER AND NEAR-SURFACE 16: 609-613. PERMAFROST Mackay, J.R., Ostrick, J., Lewis, C.P. and MacKay, D.K. 1979. Frost heave below 0¡C ground temperature, Mackay, J.R. 1957. Field observation of . Inuvik, Northwest Territories. In Current Research, Canadian Alpine Journal, 40: 91-96. Part A. Geological Survey of Canada, Ottawa, ON. Mackay, J.R. 1958. A subsurface organic layer associated Paper 79-1A: 403-405. with permafrost in the western Arctic. Geographical Mackay, J.R. 1980. The origin of hummocks, western Branch, Department of Mines and Technical Surveys, Arctic coast, Canada. Canadian Journal of Earth Ottawa, ON. Geographical Paper 18, 21 p. Sciences, 17: 996-1006. Mackay, J.R., Mathews, W.H. and MacNeish, R.S. 1961. Mackay, J.R. 1981. Active layer slope movement in a Geology of the Engigstciak archaeological site, Yukon continuous permafrost environment, Garry Island, Territory. Arctic, 14: 25-52. Northwest Territories, Canada. Canadian Journal of Mackay, J.R. 1965. Gas-domed mounds in permafrost, Earth Sciences, 18: 1666-1680. Kendall Island, N.W.T. Geographical Bulletin, 7: 105- Mackay, J.R. 1982. Active layer growth, Illisarvik 115. experimental drained lake site, Richards Island, Mackay, J.R. 1967. Underwater patterned ground in Northwest Territories. In Current Research, Part A. artificially drained lakes, Garry Island, N.W.T. Geological Survey of Canada, Ottawa, ON. Paper 82- Geographical Bulletin, 9: 33-44. 1A: 123-126. Mackay, J.R. 1969. Geomorphic processes, Mackenzie Mackay, J.R. 1983. Downward water movement into valley-Arctic coast, District of Mackenzie. In Report of frozen ground, western Arctic coast, Canada. Activities, Part A. Geological Survey of Canada: Canadian Journal of Earth Sciences, 20: 120-134. Ottawa, ON, Canada. Paper 69-1 Part A: 197-198. Mackay, J.R. 1984. The frost heave of stones in the active Mackay, J.R. 1970. Disturbances to the tundra and forest layer above permafrost with downward and upward tundra environment of the western Arctic. Canadian freezing. Arctic and Alpine Research, 16: 439-446. Geotechnical Journal, 7: 420-432. Mackay, J.R. 1984. Lake bottom heave in permafrost; Mackay, J.R. 1971. Ground ice in the active layer and the Illisarvik drained lake site, Richards Island, Northwest top portion of permafrost. Proceedings of a Seminar Territories. In Current Research, Part A. Geological on the Permafrost Active Layer, 4 and 5 May, 1971. Survey of Canada, Ottawa, ON. Paper 84-1A: 173- National Research Council of Canada, Associate 177. Committee on Geotechnical Research, Ottawa, ON, Parameswaran, V.R., Johnston, G.H. and Mackay, J.R. Canada. Technical Memorandum 103: 26-30. 1985. Electrical potentials developed during thawing of Mackay, J.R. 1971. Geomorphic processes, Mackenzie frozen ground. In Ground Freezing. Proceedings, valley-Arctic coast, District of Mackenzie. In Report of Fourth International Symposium on Ground Freezing, Activities, Part A. Geological Survey of Canada, Sapporo, Japan, 5-7 August, 1985. Kinosita, S. and Ottawa, ON, Canada. Paper 71-1 Part A: 189-190. Fukuda, M. (eds). Balkema, Rotterdam, Netherlands. Mackay, J.R. 1972. Geomorphic processes, Mackenzie Vol. 2: 9-15. valley-Arctic coast, District of Mackenzie (107 B, C, D). Mackay, J.R. 1986. Frost mounds. The Canadian In Report of Activities, Part A. Geological Survey of Geographer, 30: 363-364. Canada, Ottawa, ON. Paper 72-1 Part A: 192-194. Mackay, J.R. 1995. Active layer changes (1968-1993) Kerfoot, D.E. and Mackay, J.R. 1972. Geomorphological following the forest-tundra fire near Inuvik, N.W.T., process studies, Garry Island, N.W.T. In Mackenzie Canada. Arctic and Alpine Research, 27: 323-336. Delta area monograph, 22nd International Mackay, J.R. 1999. Periglacial features developed on the Geographical Congress. Kerfoot, D.E. (ed.). Brock exposed bottoms of seven lakes that drained rapidly University, St Catharines, ON. pp. 115-130. after 1950, Tuktoyaktuk Peninsula area, western Mackay, J.R. and MacKay, D.K. 1976. Cryostatic Arctic coast, Canada. Permafrost and Periglacial pressures in nonsorted circles (mud hummocks), Processes, 10: 39-63. Inuvik, Northwest Territories. Canadian Journal of Mackay, J.R. 1999. Cold-climate shattering (1974 to Earth Sciences, 13: 889-897. 1993) of 200 glacial erratics on the exposed bottom of Mackay, J.R. 1977. Probing for the bottom of the active a recently drained Arctic lake, western Arctic coast, layer. In Report of Activities, Part A. Geological Survey Canada. Permafrost and Periglacial Processes, 10: of Canada, Ottawa, ON. Paper 77-1A: 327-328. 125-136. Mackay, J.R. 1977. Changes in the active layer from 1968 Mackay, J.R. and Burn, C.R. 2002. The first 20 years to 1976 as a result of the Inuvik fire. In Report of (1978/79 to 1998/99) of active-layer development,

4 Illisarvik experimental drained lake site, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 8 PINGOS 39: 1657-1674. Mackay, J.R. 1962. Pingos of the Pleistocene Mackenzie Delta area. Geographical Bulletin, No. 18: 21-63. 7 HYDROLOGICAL STUDIES Mackay, J.R. 1963. Origin of the pingos of the Pleistocene Mackenzie delta area. In Proceedings of the First Mackay, J.R. 1961. A study of freeze-up and break-up at Canadian Conference on Permafrost, 17 and 18 April Fort Good Hope, N.W.T. In Thought from the learned 1962. National Research Council of Canada, societies of Canada, 1960. W.J. Gage, Toronto, ON. Associate Committee on and Snow Mechanics, pp. 65-71. Ottawa, ON. Technical Memorandum 76: 79-82. Mackay, J.R. 1961. Freeze-up and break-up of the lower Mackay, J.R. 1966. Pingos in Canada. In Permafrost Mackenzie River, N.W.T. In Geology of the Arctic. International Conference 1963, Lafayette, Indiana. Raasch, G.O. (ed.). University of Toronto Press, National Academy of Sciences Ð National Research Toronto, ON. pp.1119-1134. Council, Washington, DC. Publication 1287: 71-76 Mackay, J.R. 1963. Progress of break-up and freeze-up Mackay, J.R. and Stager, J.K. 1966. The structure of along the Mackenzie River. Geographical Bulletin, some pingos in the Mackenzie Delta area, N.W.T. No.19: 103-116. Geographical Bulletin, 8: 360-368. MacKay, D.K. and Mackay, J.R. 1965. Historical records Mackay, J.R. 1968. Discussion of the theory of pingo of freeze-up and break-up on Churchill and Hayes formation by water expulsion in a region affected by Rivers. Geographical Bulletin, 7: 7-16. subsidence. Journal of Glaciology, 7: 346-350. Mackay, J.R. 1966. The mixing of the waters of the Liard Mackay, J.R. 1972. Some observations on the growth of and Great Bear rivers with those of the Mackenzie. pingos. In Mackenzie Delta area monograph, 22nd Geographical Bulletin, 8: 166-173. International Geographical Congress. Kerfoot, D.E. Mackay, J.R. 1967. Freeze-up and breakup prediction of (ed.). Brock University, St Catharines, ON. pp.141- the Mackenzie River, N.W.T., Canada. In Quantitative 147. Geography, Part II, Physical and cartographic topics. Mackay, J.R. 1973. The growth of pingos, western Arctic Garrison, W.L. and Marble, D.F. (eds). Department of coast, Canada. Canadian Journal of Earth Sciences, Geography, Northwestern University, Evanston, IL. 10: 979-1004. Studies in Geography No. 14: 25-66. Mackay, J.R. 1976. On the origin of pingos Ð a comment. Mackay, J.R. 1970. Lateral mixing of the Liard and Journal of Hydrology, 30: 295-298. Mackenzie rivers downstream from their confluence. Mackay, J.R. 1976. The age of Ibyuk Pingo, Tuktoyaktuk Canadian Journal of Earth Sciences, 7: 111-124. Peninsula, District of Mackenzie. In Report of Krouse, H.R. and Mackay, J.R. 1971. Application of Activities, Part B. Geological Survey of Canada, 18 16 H2 O/H2 O abundances to the problem of lateral Ottawa, ON. Paper 76-1B: 9-60. mixing in the Liard-Mackenzie river system. Canadian Mackay, J.R. 1977. Pulsating pingos, Tuktoyaktuk Journal of Earth Sciences, 8: 1107-1109. Peninsula, N.W.T. Canadian Journal of Earth Mackay, J.R. 1972. Application of water temperatures to Sciences, 14: 209-222. the problem of lateral mixing in the Great Bear- Mackay, J.R. 1977. Permafrost growth and subpermafrost Mackenzie river system. Canadian Journal of Earth pore water expulsion, Tuktoyaktuk Peninsula, District Sciences, 9: 913-917. of Mackenzie. In Report of Activities, Part A. MacKay, D.K. and Mackay, J.R. 1972. Break-up and ice Geological Survey of Canada, Ottawa, ON. Paper 77- jamming of the Mackenzie River, Northwest 1A: 323-326. Territories. In Mackenzie Delta area monograph, 22nd Mackay, J.R. 1978. Sub-pingo water lenses, Tuktoyaktuk International Geographical Congress. Kerfoot, D.E. Peninsula, Northwest Territories. Canadian Journal of (ed.). Brock University, St Catharines, ON. pp. 87-93. Earth Sciences, 15: 1219-1227. MacKay, D.K. and Mackay, J.R. 1973. Break-up and ice Mackay, J.R. 1978. Contemporary pingos: a discussion. jamming on the Mackenzie River, N.W.T. In , 27: 133-154. Hydrologic aspects of northern pipeline development. Mackay, J.R. 1979. Pingos of the Tuktoyaktuk Peninsula Environmental-Social Committee, Northern Pipelines, area, Northwest Territories. Géographie physique et Task Force on Northern Oil Development, Ottawa, Quaternaire, 23: 3-61. ON. Report 73-3: 223-232. Mackay, J.R. 1981. Aklisuktuk (Growing Fast) Pingo, MacKay, D.K. and Mackay, J.R. 1973. Locations of spring Tuktoyaktuk Peninsula, Northwest Territories, Canada. ice jamming on the Mackenzie River, N.W.T. In Arctic, 34: 270-273. Hydrologic aspects of northern pipeline development. Mackay, J.R. 1983. Pingo growth and subpingo water Environmental-Social Committee, Northern Pipelines, lenses, western Arctic coast, Canada. In Permafrost, Task Force on Northern Oil Development, Ottawa, Fourth International Conference, Proceedings, July ON. Report 73-3: 233-257. 17-22, 1983. National Academy Press, Washington, Mackay, J.R. and MacKay, D.K. 1977. The stability of ice- DC. Vol. 1: 762-766. push features, Mackenzie River, Canada. Canadian Mackay, J.R. 1985. Pingo ice of the western Arctic coast. Journal of Earth Sciences, 14: 2213-2225. Canadian Journal of Earth Sciences, 22: 1452-1464.

5 Mackay, J.R. 1986. Growth of Ibyuk Pingo, western Arctic Contribution, Permafrost Second International coast, Canada, and some implications for Conference, 13-28 July 1973, Yakutsk, USSR. environmental reconstructions. Quaternary Research, National Academy of Sciences, Washington, DC. pp. 26: 68-80. 223-228. Mackay, J.R. 1987. Some mechanical aspects of pingo Mackay, J.R. and Black, R.F. 1973. Origin, composition, growth and failure, western Arctic coast, Canada. and structure of perennially frozen ground and ground Canadian Journal of Earth Sciences, 24: 1108-1119. ice: a review. In North American Contribution, Mackay, J.R. 1987. Pingos of the western Arctic coast, Permafrost Second International Conference, 13-28 Canada. In Regionalgeografi Ymer 1987. Elg, M. (ed.). July 1973, Yakutsk, USSR. National Academy of Svenska Sällskapet för Antropologi och Geografi, Sciences, Washington, DC. pp. 185-192. Stockholm, Sweden. Årgång 107: 155-170. Mackay, J.R. 1974. Reticulate ice veins in permafrost, Mackay, J.R. 1988. Pingo collapse and paleoclimatic northern Canada. Canadian Geotechnical Journal, 11: reconstruction. Canadian Journal of Earth Sciences, 230-237. 25: 495-511. Mackay, J.R. and Lavkulich, L.M. 1974. Ionic and oxygen Mackay, J.R. 1988. The birth and growth of Porsild Pingo, isotopic fractionation in permafrost growth. In Report Tuktoyaktuk Peninsula, District of Mackenzie. Arctic, of Activities, Part B. Geological Survey of Canada, 41: 267-274. Ottawa, ON. Paper 74-1 Part B: 255-256. Mackay, J.R. 1990. Seasonal growth bands in pingo ice. Mackay, J.R. 1975. Reticulate ice veins in permafrost, Canadian Journal of Earth Sciences, 27: 1115-1125. northern Canada: reply. Canadian Geotechnical Mackay, J.R. 1994. Pingos and pingo ice of the western Journal, 12: 163-165. Arctic coast, Canada. Terra, 106: 1-11. Mackay, J.R. 1976. at depth in Parameswaran, V.R. and Mackay, J.R. 1996. Electrical permafrost. In Report of Activities, Part A. Geological freezing potentials measured in a pingo growing in the Survey of Canada, Ottawa, ON. Paper 76-1A: 287- western Canadian Arctic. Cold Regions Science and 288. Technology, 24: 191-203. Mackay, J.R. 1978. The surface temperature of an ice-rich Mackay, J.R. 1998. Pingo growth and collapse, melting permafrost exposure, Garry Island, Northwest Tuktoyaktuk Peninsula area, western Arctic coast, Territories. In Current Research, Part A. Geological Canada: a long-term field study. Géographie physique Survey of Canada, Ottawa, ON. Paper 78-1A: 521- et Quaternaire, 52: 271-323. 522. Mackay, J.R. and Burn, C.R. 2011. A century (1910-2008) Mackay, J.R., Konishchev, V.N. and Popov, A.I. 1979. of change in a collapsing pingo, Parry Peninsula, Geologic controls of the origin, characteristics, and western Arctic coast, Canada. Permafrost and distribution of ground ice. In Proceedings of the Third Periglacial Processes, 22: 266-272. International Conference on Permafrost, July 10-13, Mackay, J.R. and Burn, C.R. 2013. Addendum. A century 1978, Edmonton, Alberta, Canada. National Research (1910-2008) of change in a collapsing pingo, Parry Council of Canada, Ottawa, ON. Vol. 2: 1-18. Peninsula, western Arctic coast, Canada. Permafrost Mackay, J.R. 1983. Oxygen isotope variations in and Periglacial Processes, 24: 261. permafrost, Tuktoyaktuk Peninsula area, Northwest Territories. In Current Research, Part B. Geological Survey of Canada, Ottawa, ON. Paper 83-1B: 67-74. 9 GROUND ICE AND EXPOSURES Mackay, J.R. 1986. Fifty years (1935-1985) of coastal retreat west of Tuktoyaktuk, District of Mackenzie. In Mackay, J.R. 1963. Notes on the shoreline recession Current Research, Part A. Geological Survey of along the coast of the Yukon Territory. Arctic, 16: 195- Canada, Ottawa, ON. Paper 86-1A: 727-735. 197. Mackay, J.R. 1989. Massive ice: some field criteria for the Mackay, J.R. 1966. Segregated epigenetic ice and slumps identification of ice types. In Current Research, Part G. in permafrost, Mackenzie Delta area, N.W.T. Geological Survey of Canada, Ottawa, ON. Paper 89- Geographical Bulletin, 8: 59-80. 1G: 5-11. Mackay, J.R. and Stager, J.K. 1966. Thick tilted beds of Mackay, J.R. and Dallimore, S.R. 1992. Massive ice of the segregated ice, Mackenzie delta area, N.W.T. Biuletyn Tuktoyaktuk area, western Arctic coast, Canada. Peryglacjalny, 15: 39-43. Canadian Journal of Earth Sciences, 29: 1235-1249. Mackay, J.R. 1971. The origin of massive icy beds in permafrost, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 8: 397-422. 10 SNOW AND NEEDLE ICE Rampton, V.N. and Mackay, J.R. 1971. Massive ice and icy sediments throughout the Tuktoyaktuk Peninsula, Mathews, W.H. and Mackay, J.R. 1963. Snow creep Richards Island, and nearby areas, District of studies, Mount Seymour, B.C.: Preliminary field Mackenzie. Geological Survey of Canada, Ottawa, investigations. Geographical Bulletin, No.20: 58-75. ON, Canada. Paper 71-21, 16 p. Mackay, J.R. and Mathews, W.H. 1967. Observations on Mackay, J.R. 1972. The world of underground ice. Annals, pressures exerted by creeping snow, Mount Seymour, Association of American Geographers, 62: 1-22. British Columbia, Canada. In Physics of Snow and Ice. Mackay, J.R. 1973. Problems in the origin of massive icy Proceedings of the International Conference of Low beds, western Arctic, Canada. In North American Temperature Science, 1966. Ôura, H. (ed.). Institute of

6 Low Temperature Science, Hokkaido University, Delta area monograph. 22nd International Sapporo, Japan. Vol. 1: 1185-1197. Geographical Congress. Kerfoot, D.E. (ed.). Brock Brink, V.C, Mackay, J.R., Freyman, S. and Pearce, D.G. University, St Catharines, ON. pp. 107-114. 1967. Needle ice and seedling establishment in Mackay, J.R. 1973. Some aspects of permafrost growth, southwestern BC. Canadian Journal of Plant Science, Mackenzie delta area, N.W.T. (107 B, C, D). In Report 47: 135-139. of Activities, Part A. Geological Survey of Canada, Mackay, J.R. and Mathews, W.H. 1974. Needle ice Ottawa, ON. Paper 73-1 Part A: 232-233. striped ground. Arctic and Alpine Research, 6: 347- Mackay, J.R. 1974. Seismic shot holes and ground 359. temperatures, Mackenzie delta area, N.W.T. In Report Mackay, J.R. and Mathews, W.H. 1974. Movement of of Activities, Part A. Geological Survey of Canada, sorted stripes, The Cinder Cone, Garibaldi Park, B.C., Ottawa, ON. Paper 74-1 Part A: 389-390. Canada. Arctic and Alpine Research, 6: 79-84. Mackay, J.R. and MacKay, D.K. 1974. Snow cover and Mackay, J.R. and Mathews, W.H. 1975. Snow creep: its ground temperatures, Garry Island, N.W.T. Arctic, 27: engineering problems and some techniques and 287-296. results of its investigation. Canadian Geotechnical Mackay, J.R. 1975. Freezing processes at the bottom of Journal, 12: 187-198. permafrost, Tuktoyaktuk Peninsula, District of Mackay, J.R. and Mathews, W.H. 1975. Orientation of soil Mackenzie (107 C). In Report of Activities, Part A. stripes caused by needle ice. Journal of Glaciology, Geological Survey of Canada, Ottawa, ON. Paper 75- 14: 329-331. 1 Part A: 471-474. Mackay, J.R. 1977. Needle ice growth, ice segregation, Mackay, J.R. 1975. The stability of permafrost and recent and frost heave under natural conditions. In Report of climatic change in the Mackenzie Valley, N.W.T. In the Sixth Meeting, British Columbia Soil Science Report of Activities, Part B. Geological Survey of Workshop. British Columbia Ministry of Agriculture: Canada, Ottawa, ON. Paper 75-1 Part B: 173-176. Victoria, BC; 173-178. Mackay, J.R. 1975. Some resistivity surveys of permafrost thickness, Tuktoyaktuk Peninsula, N.W.T. In Report of Activities, Part B. Geological Survey of Canada, 11 TECHNIQUES AND INSTRUMENTATION Ottawa, ON. Paper 75-1 Part B: 177-180. Mackay, J.R. and Scott, W.J. 1977. Reliability of Mackay. J.R. 1965. Glacier flow and analogue simulation. permafrost thickness determination by D.C. resistivity Geographical Bulletin, 7: 1-6. soundings. In Proceedings of a Symposium on Mackay, J.R. 1972. Shot hole logs and ground ice. Arctic Permafrost Geophysics, 12 October 1976. Scott, W.J. Petroleum Operators Association, Calgary, AB. 13 p. and Brown, R.J.E. (eds). National Research Council of Mackay, J.R. 1973. A frost tube for the determination of Canada, Associate Committee on Geotechnical freezing in the active layer above permafrost. Research, Ottawa, ON. Technical Memorandum 119: Canadian Geotechnical Journal, 10: 392-396. 25-39. Mackay, J.R. 1974. Measurement of upward freezing Burn, C.R., Mackay, J.R. and Kokelj, S.V. 2009. The above permafrost with a self-positioning thermistor thermal regime of permafrost and sensitivity to probe. In Report of Activities, Part B. Geological disturbance near Inuvik, N.W.T. Permafrost and Survey of Canada, Ottawa, ON. Paper 74-1 Part B: Periglacial Processes, 20: 221-227. 250-251. Mackay, J.R. 1974. Performance of a heat transfer device, Garry Island, N.W.T. In Report of Activities, ACKNOWLEDGEMENTS Part B. Geological Survey of Canada, Ottawa, ON. Paper 74-1 Part B: 252-254. I would like to thank Michael Church and Ian McKendry, Parameswaran, V.R. and Mackay, J.R. 1983. Field both of the Department of Geography, University of British measurements of electrical freezing potentials in Columbia, for helpful discussions and assistance in permafrost areas. In Permafrost, Fourth International checking some of the citations in this bibliography, and Conference, Proceedings, July 17-22, 1983. National Timothy Oke, Olav Slaymaker, John Stager, and Graeme Academy Press, Washington, DC. Vol. 1: 962-967. Wynn of the same department, for many points of Mackay, J.R. and Leslie, R.V. 1987. A simple probe for information. Fabrice Calmels kindly translated the the measurement of frost heave within frozen ground abstract. in a permafrost environment. In Current Research, Part A. Geological Survey of Canada, Ottawa, ON. Paper 87-1A: 37-41. REFERENCE

Burn, C.R. 2015. J. Ross Mackay (1915 Ð 2014). Arctic, 12 GROUND TEMPERATURES AND PERMAFROST 64: 129-131.

Mackay, J.R. 1967. Permafrost depths, lower Mackenzie valley, Northwest Territories. Arctic, 20: 21-26. Mackay, J.R. and MacKay, D.K. 1972. Ground temperatures at Garry Island, N.W.T. In Mackenzie

7

8 John Ross Mackay - Devoted tutor and best friend of Chinese permafrost research

Guodong Cheng1, Zhijiu Cui2, Baolai Wang3, and Huijun Jin1, 4 1. Cold and Arid Regions Environmental and Engineering Research Institute (CAREERI), Chinese Academy of Sciences (CAS), Lanzhou, China; [email protected] 2. Beijing (Peking) University, China; [email protected] 3. Citizenship and Immigration Canada, Ottawa, Canada: [email protected] 4. Corresponding author: HJ Jin at [email protected]

ABSTRACT The scientific accomplishments of Mackay were of great benefit to China, especially his research on ground ice. Permafrost studies began in China in the 1960s. They were mainly concerned with the impact of near-surface layers of ground ice upon engineering infrastructure. However, Chinese geocryologists had little knowledge of ground-ice genesis and massive ground ice bodies. After 1982, under the influence and generous help of Mackay, Chinese research on permafrost and ground ice became increasingly sophisticated. Mackay was respected and admired not only for his outstanding accomplishments in scientific research, but also for his devotion to science and his innovative methods for permafrost research. His family ties with China were deep.

RESUMÉ Les contributions scientifiques de Mackay furent bénéfiques pour la Chine, en particulier ses recherche sur la glace de sol. Les études sur le pergélisol ont débuté en Chine au cours des années 1960. Elles concernaient principalement l’impact de la présence de glace de sol sur les ouvrages de génie civil et les infrastructures. Ë ce moment, les géocryologues chinois avaient des connaissances limitées en matière de formation de la glace dans le sol et de corps de glace massive. Après 1982, sous l’influence et l’aide généreuse de Mackay la recherche chinoise sur le pergélisol et la glace de sol est devenue de plus en plus élaborée. Mackay était respecté et admiré non seulement pour ses réalisations exceptionnelles en recherche, mais aussi pour son engagement scientifique et ses méthodes innovatrices. Ses liens familiaux avec la Chine étaient profonds.

1 INTRODUCTION At the invitation of the Lanzhou Institute of Glaciology and Geocryology (LIGG), Chinese Academy of Sciences In 1982, Guodong Cheng wrote a paper in Chinese (CAS), Professor Mackay visited China in late October- with the title “Formation processes of the thick-layered early November, 1982. In Lanzhou, Mackay delivered six ground ice”. Subsequently, he became aware that several lectures on permafrost: 1) Field experiments on papers had been written by Mackay over a decade earlier permafrost aggradation; 2) The growth of ice-wedge (Mackay, 1972) in which this ice was called “aggradational polygons; 3) The growth of pingos; 4) The genesis of ice” and that it was syngenetic in nature. Upon learning of massive ice; 5) The active layer and 6) Periglacial this, Guodong Cheng painstakingly translated his own phenomena. Attendees included scholars and students paper into English and sent it to Mackay for advice and from LIGG, Lanzhou University and the Northwest improvement. At that time, most Chinese geocryologists Institute of Railway Science. Mackay was also invited to had studied only Russian in school and his clumsy present lectures at the Department of Geology and English was self-taught, hence his translation was difficult Geography, Lanzhou University. Afterwards, he visited the to understand. However, Mackay meticulously and Tianshan Glaciological Station and inspected permafrost patiently revised the paper, almost word by word, and in the Tianshan Mountains (Figure 1). recommended that it be published in the journal Cold Regions Science and Technology. This paper became the first major paper published by Cheng in an international English journal (Cheng, 1983). It was also the first English-language paper on permafrost in China that was published abroad in a major technical journal. This was an inspiring event not only for Guodong Cheng but also for many of his colleagues in China in the early 1980s.

9 Cheng, 1987). Later, in 1993, Chinese scientists discovered additional well-preserved ice wedges at Wuma (52o45N, 120o45E; elevation at 700 m a. s. l.) in northwestern Da Xing’anling Mountains. As Mackay had suspected, radiocarbon dating indicated the wedges formed between 14-10 ka BP, at the end of the Late Pleistocene (Tong, 1993). Subsequently, more detailed investigations confirm that the ice wedges at both Yituli’he and Wuma were much younger. Using research pioneered by Mackay (1983), stable oxygen and hydrogen isotopes allowed inferences at to the paleo-temperature changes, particularly the mean annual air temperatures and mean monthly air temperatures in the coldest month, when the ice-wedge ice was being formed (Yang et al., 2015). On the basis of new 14C-dating and the analysis of the oxygen and hydrogen isotopes of the Yituli’he wedge ice, three short- term cooling events were inferred, with the peak of their cooling at about 2800, 2300 and 1900 a BP, and with amplitudes of cooling of about 2.1, 1.1 and 1.3C, respectively, when compared with the present (Yang and Jin, 2011; Yang et al., 2015). In other words, these ice

wedges, formed during the Neoglacial period (between Figure 1. Mackay in the Tianshan Mountains in 1982 3300 and 1600 a BP), supported the earlier, highly during his visit to the Tianshan Glaciological Station, CAS, cautious, interpretation of Mackay in 1987. near Urumqi, Xinjiang Uygur Autonomous Region, Northwest China 3 MACKAY’S SCIENTIFIC METHOD 2 MACKAY AND MASSIVE GROUND ICE IN CHINA Mackay had an unquenchable curiosity and desire for learning about natural phenomena and revealing the In 1987, Guodong Cheng wrote a letter to Mackay mysteries of nature. Taking the harsh Arctic as his lab, he informing him that the southernmost ice wedge in the devoted his life and passion to permafrost research. As a northern hemisphere had been discovered at Yituli’he visiting scholar at UBC in 1984-1985, Professor Zhijiu Cui (50o32N, 121o29E; elevation at 730 m a. s. l.) in northern recalls (personal communication to Cheng, 2015) that, he Da Xing’anling Mountains, Northeast China. This was feel very fortunate and honored being able to study soon reported in the Chinese Journal of Glaciology and permafrost and periglacial phenomena under the Geocryology (Jia et al, 1987). The top of the wedge ice guidance of Mackay. He remembers that, whenever was at the same depth as the permafrost table ( 0.9 m), learning of some important viewpoints on the formation and the upper width of the ice wedge was about 1 m and processes of permafrost, Mackay would conduct the height of the wedge was 1.5 m (Figure 2). corresponding field experiments for either their verification or dismissal. All of the papers published by Mackay were based on personal observations and/or the results of his experiments. They were well founded and innovative. Even during the 30 years after his formal retirement from UBC in 1981, Mackay continued his work in the Arctic each year, publishing more than 50 high-quality technical papers on the basis of his meticulous work. In 1988, Mackay came to the Fifth International Conference on Permafrost directly from the field, and directly returned to the field immediately after the Conference. Some said that the Arctic was the Paradise of Mackay. His dogged pursuit of science even in the harshest of environments has been a great inspiration to us, the Chinese geocryologists who have conducted permafrost research on the extreme high-elevation Qinghai-Tibet Plateau for several decades. Figure 2. Inactive ice wedge at Yutuli’he in northern Da Mackay had a talent for finding ways to create and/or Xing’anling Mountains, Northeast China design simple but effective instrumentation, and he never The ground surface showed a polygonal pattern to felt bored per se. Once he told Guodong Cheng, with the ice wedges. In reply, Mackay pointedly asked whether excitement, that “I have devised a simple probe with the ice wedge was still active (personal communication to which I can detect small vertical changes (e.g. 0.5 mm) using small magnets buried in the ground. Now, I am able

10 to detect heave in frozen ground in the thaw period” deposits can effectively “cool” the ground. Thus, in the (personal communication to Cheng from Mackay, 1986). design and construction of the Qinghai-Tibet Railway from Professor Zhijiu Cui also recalls impressions from 1984- Golmud, Qinghai Province, to Lhasa, Tibet Autonomous 85: “There was a log cabin with a kitchen at his field Region, “roadbed cooling” became an extremely “hot”, or station on Garry Island. It took two attempts by the popular, research topic in China. The academic aspects of helicopter to find it because of very thick snow cover. After permafrost cooling were quickly and conveniently applied our arrival, we dug a snow pit about 1 m in depth. Then, to the use of blocky roadway materials, the use of shades the observation devices and instruments that Mackay had and awnings, and the installation of air-ducts, with buried the previous year revealed themselves. He then remarkable success in engineering construction and erected two tele-sensors with electric currents in the ecological benefits. Therefore, the division between extending direction of the polygonal frost crack on each theoretical and applied research seems overly simplified side of the crack. The sensors were directly connected to (Cheng, 2004). a recording device. When the crack extended forward, the electrode copper wire would disconnect. The device 5 MACKAY AND THE INTERNATIONAL would then record the exact time, to the hour, minute and PERMAFROST ASSOCIATION seconds, and the date of the break. These devices were based on his formulation and were made by professional In 1982, the late Professor Shi Yafeng, the founding father engineers, and thus unique in the world” (personal of glaciology in China, met Mackay in Lanzhou (Figure 3). communication to Cheng from Professor Zhijiu Cui, 2015). After this historical meeting, Professor Shi attended the It was because of these unique instruments and Fourth International Conference on Permafrost, held in devices that Mackay was able to decipher the secret Fairbanks, Alaska, in 1983. At this conference, that “Big codes of permafrost science that eluded most other Four” (the Union of Soviet Socialist Republics (USSR), people. Many of Mackay’s experimental designs were Canada, the People's Republic of China (PRC) and the creative and innovative. In particular, his full-scale United States of America (USA)) created the International experiment on permafrost growth by the draining of a Permafrost Association (IPA). This was a shared initiative tundra lake (‘the Illisarvik experiment’) has been one of a and Mackay was asked to become its first Secretary- kind. Continuously and for many decades, he relentlessly General. In 1993, after serving 10 years in that office, observed and studied the processes associated with Mackay stepped down and was greatly honored (Figure permafrost formation and pingo growth in the Mackenzie 4). Under Mackay’s early guidance at the IPA, China’s Delta region, indelibly contributing to geocryology in an involvement in international permafrost studies increased impeccable way. rapidly in the following years (Figure 5). Professor Zhijiu Cui emotionally recalls “There were many Chinese visiting scholars abroad. However, I was extremely lucky that I could learn many new thoughts and methods from Ross. Later, I and my students deployed many observational sites in the Tianshan Mountains in West China, in the Andes Mountains, at the Great Wall Station in Antarctica, on the Qinghai-Tibet Plateau, and in the Changbai Mountains in Northeast China, … We were able to accomplish prolifically on the basis of those large amount of observational data…” (personal communications to Cheng from Professor Zhijiu Cui, 2015). It was his large amount of long-term field experiments, innovative experimental designs, and analytical interpretations for process mechanisms that were the foundations for Mackay becoming the international guru on permafrost and ground-ice research. Figure 3. A centennial toasting between the founding father of glaciology in China (Academician Yafeng Shi) 4 MACKAY AND APPLIED RESEARCH IN CHINA and the founding father of geocryology in Canada (Professor Ross Mackay), in Lanzhou, Gansu Province, Mackay focused upon the basic and theoretical China, in 1982 aspects of perennially-frozen ground. However, with increasing discoveries of petroleum and natural gas reserves in arctic regions, people suddenly needed to learn more about permafrost and ground ice. As a result, Mackay’s academic research immediately became applied research, with great economic and environmental benefits. Chinese scientists and scholars felt the same way. Studies on permafrost distribution had demonstrated that local factors, such as peatlands, ice caves, and talus

11 abstracts for the Journal of Glaciology and Geocryology (in Chinese), and provided constructive advice which greatly increased the international impact of the journal. He also suggested that the Qinghai-Tibet Highway and Tianshan Glaciological Station be listed as field trips for the post-conference field excursions of the Sixth International Conference on Permafrost in Beijing in 1993. These greatly added to the success of the Conference. In Taibei (Taipei), Southeast China, there is a well- known Mackay Memorial Hospital. This is named after Mackay’s grandfather who was a missionary. Although not as famous as Dr. Norman Bethune (who served in China during the WWII and almost known to everyone in China), Grandfather Mackay made great contributions to Taiwan. As a missionary, Grandfather Mackay came to Taiwan in Figure 4. A special salute and gift to Mackay was 1872, and married Miss Cunming Zhang (Acun), a local delivered by incoming IPA President Guodong Cheng at Chinese girl, the grandmother of JRM, in 1878. The father the Sixth International Conference on Permafrost, held in of JRM was born in Taiwan but completed his schooling in Beijing, China, in 1993 Canada only to return to Taiwan, again as a missionary, and devoted himself to the education of children. J. R. Mackay was born at Danshui, Taiwan, in 1915 and spent his childhood there. Ross was honored with an honorary citizen of Danshui town in 2001. Danjiang Middle (High) School decided to erect a monument for J. R. Mackay at the Mackay Cemetery (Lin, 2014). Perhaps it is because of his Chinese blood that Ross Mackay felt a close kinship to Chinese colleagues and to China. Dr. Baolai Wang recalled that Mackay once confided in him that he preferred Chinese cuisine. Professor Zhijiu Cui also has a related memory: “In winter 1984, we went to work in the Canadian arctic. It was decided that we should have a Chinese dinner. Ross ushered us to the supermarket, and he set a rule that I (Cui) should prepare the Chinese food for dinner and he (Ross) would prepare breakfast and a lunch sandwich… Figure 5. In 1998, at the Seventh International Conference Possibly, and because of my satisfactory Chinese on Permafrost, held in Yellowknife, NWT, Canada, the cooking, he invited me to prepare a Chinese dinner for his President of the IPA, Guodong Cheng, paid tribute to wife and daughter in his home upon our return to three pioneering geocryologists. At the podium: Vancouver, and with pleasure I accepted the Academician Guodong Cheng. Standing (left to right): invitation…The more interesting was, he had overheard Professors T. L. Péwé, J. R. Mackay, and A. L. the songs of Miss Lijun Deng (Theresa Teng) that I was Washburn) listening to during our field work. He requested that I buy some cassettes (albums then) of her songs upon my 6 MACKAY, CHINA AND FAMILY TIES return to Vancouver” (personal communications with Cheng from Professor Zhijiu Cui, 2015). Miss Theresa Teng was very popular at the time and was widely known Mackay’s help to China was multifaceted. For example, in in China. It is our belief that Ross understood and order to facilitate permafrost research in China, Professor appreciated his Chinese heritage and feelings in Teng’s Ross Mackay and Professor Olav Slaymaker, then Head songs. of the Department of Geography at the University of At age 95, Mackay sent me (Guodong Cheng) a British Columbia, actively assisted China in applying for Christmas card, adding, “I look back, with great pleasure, funding from the International Development Research to my trip to China.” We can deeply feel Mackay’s missing Centre (IDRC) under the theme of “Sino-Canada Joint of China during his senior years. We also deeply miss him Research on Permafrost, Ground Ice and Alpine and will always cherish his sincere friendship and Geomorphic Processes in the Kunlun Mountains, China”. generous help. Although the application was unsuccessful, China will Mackay was not only the best tutor of Chinese never forget the good will and generous efforts of permafrost scholars, but also a devoted friend to us. Professors Mackay and Slaymaker. Mackay served as an honorary professor of the ACKNOWLEDGEMENTS LIGG, CAS, and was an honorary member of the Branch of Glaciology and Geocryology, Geographical Society of Many people have provided memories and pictures to China. He volunteered to reviewing and editing English assist in the preparation of this paper. They include

12 Professor Yongping Shen with the CAREERI, CAS; Yang S Z, Cao X Y, Jin H J, 2015. Validation of ice-wedge Senior Engineer Emeritus Xiaoming Huang with the isotopes at Yituli’he, northeastern China as climate Lanzhou Institute (now Academy) of Railway Science; Mr. proxy. Boreas, 44(3): 502-510. Jindong Wang with the CAREERI, CAS; Ms. Hongxian Yin with the CAREERI, CAS; Professor Tingjun Zhang with the Lanzhou University, and Ms. Xiaojing Jia and Professor Fujun Niu with the CAREERI, CAS. Their contributions to the paper are hereby acknowledged. Professor Hugh French (University of Ottawa and University of Victoria) assisted in manuscript revision and Professor Michel Allard (Université Laval) provided the French abstract.

REFERENCES

Cheng G D, 1983. The mechanism of repeated- segregation for the formation of thick layered ground ice. Cold Regions Science and Technology, 8(1): 57– 66 Cheng G D, 2004. Influences of local factors on permafrost occurrence and their implications for Qinghai-Xizang Railway design. Science in China Series D Earth Sciences, 47(8): 704- 709 French H M, Harry, D G, 1990. Observations on buried glacier ice and massive segregated ice, Western Arctic Coast, Canada. Permafrost and Periglacial Processes, 1: 31-43. Harry, D G, French, H M, Pollard, W H, 1988. Massive ground ice and ice-cored terrain near Sabine Point, Yukon coastal plain. Canadian Journal of Earth Sciences, 25: 1846-1856. Jia M C, Yuan F, Cheng G D, 1987. First discovery of ice wedges in Northeast China. Journal of Glaciology and Geocryology, 9: 257-260 (in Chinese with English abstract). Lin J J, 2014. Report on “In memorial of the late Dr. John Mackay from Taiwan and Toronto churches”. 《TÂI- OÂN KËU-HŌE KONG-Pñ》 (Taiwan Church News), Issue No. 3274, 24-30 November, 2014, p.3 Mackay J R, 1972. The world of underground ice. Annals of the Association of American Geographers, 63: 1-22. Mackay J R, 1983. Oxygen isotope variations in permafrost, Tuktoyaktuk Peninsula area, Northwest Territories. Current Research, Part B-Geological Survey of Canada, Paper 81: 67–74 Mackay J R, 1989. Massive ice: some field criteria for the identification of ice types. Current Research, Part G, Geological Survey of Canada, Paper 89-1G, 5-11. Mackay, J R, Dallimore, S R, 1992. Massive ice of the Tuktoyaktuk area, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 29: 1235-1249. Tong B L, 1993. Ice wedge in Northeast China. Journal of Glaciology and Geocryology, 15(1): 41-46 (in Chinese with English abstract). Wang B L, 1990. Massive ice in . Journal of Glaciology and Geocryology, 12(3): 209-218 (in Chinese with English abstract). Yang S Z, Jin H J, 2010. δ18O and δD records of inactive ice wedge in Yituli’he, Northeastern China and their paleoclimatic implications. Science China Series D Earth Sciences, 54(1): 119-126.

13

14 Reconstructing geomorphology: an appreciation of the contributions of J. Ross Mackay (1915-2014)

Michael Church Department of Geography, The University of British Columbia, Vancouver, British Columbia, Canada

ABSTRACT Based as they are entirely on his own field experiences, Ross Mackay’s geomorphological contributions are commonly regarded as the fruit of solo inspiration. In fact, his work is decidedly the product of its time. At the outset of his career, geomorphology was entering a radical transformation from interpretive study of landscape history toward quantitative study of landscape-forming processes. Accordingly, Mackay’s earliest works are accounts of regional geography. From 1960, however, his work, following the new perspective, expresses its dominant character: quantitative measurement in the field and on maps, and application of physical theory for interpreting the observations. By 1965 his mature style was firmly established. The outstanding aspects of Mackay’s contributions are his genius for making critical field observations and his ability to use them to test geophysical theory. But his early regard for landscape history remained an important aspect of his insight throughout his career.

RÉSUMÉ Fondées sur des travaux originaux sur le terrain, les contributions de Ross Mackay en géomorphologie sont généralement considérées comme le fruit d’une inspiration unique. Néanmoins, son parcours scientifique reflète aussi l’évolution de la discipline durant l’après-guerre. Rappelons-nous qu’au début de sa carrière, la géomorphologie entrait tout juste dans une transformation radicale qui substituera aux études interprétatives portant sur l’histoire des paysages, des investigations quantitatives des processus de formation des paysages. D’ailleurs, les premières publications de Mackay dans les années 50 sont des comptes rendus de géographie régionale reflétant les paradigmes de l’époque. Néanmoins, dès 1960, ses travaux s’engagent dans la nouvelle perspective qui dominera le reste de son Ïuvre : celle-ci est fondée sur les mesures quantitatives sur le terrain et sur les cartes, et l’application des théories physiques dans l’interprétation des observations. Dès 1965, ce Ç style Mackay È était déjà bien établi. Les aspects marquants des contributions de Mackay sont son aptitude à obtenir des observations critiques sur le terrain, puis à les employer pour mettre à l’épreuve des théories géophysiques. Son intérêt initial pour l’histoire des paysages demeurera aussi un aspect important de sa vision de la géomorphologie durant toute sa carrière.

1 INTRODUCTION their drainage basins: hydrophysical approach to quantitative morphology”, and then, in 1950, by Ross Mackay began his scientific career at the close “Equilibrium theory of erosional slopes approached by of the nineteenth century. That is because, in frequency distribution analysis”, written by geomorphology (and in much of earth science), the A.N.Strahler. In 1952, this same author provided the nineteenth century continued until about 1950. The programmatic statement in “Dynamic basis of great scientific project of nineteenth century earth geomorphology”. The new paradigm would seek science was to understand the history of Earth: to quantitative expression of landscape-forming comprehend the significance of the rock column as processes rationalized in terms of Newtonian evidence for that history and to understand the mechanics. Landscape history would be made to historical development of modern landscapes. Work conform with plausible scenarios of physical process. involved mapping and rich description of rocks and Time scales would be those of ordinary human topography, and the construction of historical perception. While focusing mostly on more local narratives of particular landscapes (Church, 2010). features of the landscape that become directly Time scales were geological ones Ð much too long to involved in human works, engineers had been playing admit any detailed consideration of landscape-forming this game for centuries, and so many of the methods processes. For geographers, the narrative could of study and early results under the new paradigm extend to Ð or be largely preoccupied with Ð the were directly imported into geomorphology from people and their means of livelihood in the landscape. engineering. With the appearance, in 1953, of Change began while the world was preoccupied Leopold and Maddock’s The hydraulic geometry of with war. R.A.Bagnold’s book, The physics of blown stream channels and some physiographic sand and desert dunes, based on 1930s explorations implications, a work based squarely on engineering in North Africa and published in 1941, announced Ð regime theory of unlined canals, the mechanistic right from its title Ð a radically different way to paradigm was firmly established. consider landscape. It was followed in 1945 by There had, of course, already been R.E.Horton’s “Erosional development of streams and geomorphologists who followed a process-focused

15 paradigm, perhaps most prominently the American The publications from his early Arctic seasons are geologist G.K.Gilbert (1843-1918). Their lead was not brief and mainly descriptive reports of field activity. followed simply, I expect, because of the dominance However, they reveal a rapid development in the in the earlier period of the great historical questions. focus of his interests and his method of investigation. Ross Mackay’s training and early career spanned the Some of his reports, most notably his two major period of the revolution in modes of landscape study memoirs, written in and immediately following these and the evolution of his style of work reflects the years (Mackay, 1958; 1963), retain the organization of transition. It very much reflects the times in which he a traditional regional geography, with sections on worked. geology, physiography, climate, vegetation and human history. The papers, however, begin to focus more narrowly on describing specific landforms and Ð 2 EARLY WORK (1947-1960) increasingly exclusively Ð landforms of permafrost. Thus fissures and mud circles on Cornwallis Island Ross Mackay graduated from Clark University (1939) (1953), glacier ice-thrust deformation at Nicholson and Boston University (1941), where he would have Peninsula (1956) and on the Yukon coast (1959), received a solid geographical training in the older, oriented lakes at Liverpool Bay (1956), and ‘earth history’ tradition. After war service he took up topography associated with disintegrating glacial ice Ph.D. studies at the Université de Montréal, where he (1960). (The latter paper was based on notes made studied aspects of the late Quaternary history of the during Mackay’s first journey north in 1951 when ice Ottawa valley. This issued in two papers in the newly conditions forced a week’s layover at an anonymous established Revue Canadienne de Géographie, “The lake in the middle of nowhere Ð probably the closest north shore of the Ottawa River” (1947), and he ever came to having an ‘adventure’.) Mackay also “Physiography of the lower Ottawa valley” (1949). The revealed the wide range of his interests in the Arctic former is a traditional regional description covering landscape in this period, reporting twice (1957; 1960) both physical and human geographies of the region; on the condition of small boat harbours along the the latter is the core of his Ph.D. thesis. In that work, western arctic coast, a topic of considerable interest the principal contribution is a demolition of the view to the local people and to boat-borne ‘outsiders’. put forward by the French geographer Raoul Mackay’s regard for historical work is also evident Blanchard (who founded the Institut de géographie in in his early publications, with references to the Université de Montréal) that terraces in the lower observations and mapping by explorers as diverse as valley and around Montréal are depositional John Franklin (the successful, 1825-27 overland landforms. By their landform association with expedition to the Mackenzie Delta), Roald Amundsen erosional scarps, by the nonconformable stratigraphy, (the Northwest Passage expedition of 1903-06); and by consistency with the postglacial history of Father Émile Petitot (Oblate missionary and degradation in the valley, Mackay showed without geographer whose late 19th century travels included doubt that they are erosional features. This work Anderson River); the Canadian Arctic Expedition already displayed qualities of his mature (1913-18, led by Vilhjalmur Stefansson, whom contributions: remarkably keen powers of observation; Mackay subsequently met); and E. deK. Leffingwell, a a determination to consider all plausible hypotheses pioneer American periglacial geologist whose 1919 to explain a phenomenon; a penchant to choose the USGS monograph Mackay extensively quoted. most straightforward hypothesis; a determination to defend the chosen hypothesis by as many 3 TRANSFORMATION (1960-1965) independent lines of evidence as possible; and a concern to use historical evidence to the fullest extent From about 1960, the character of Mackay’s work possible Ð habits likely acquired, if not earlier, during changed radically. Focusing fully on the analysis of his wartime service in Intelligence. But the work also periglacial landforms, his work quite suddenly began exhibited the character of his training in the earth to reflect the hallmark concerns for quantitative history tradition. Focused on late Quaternary history measurement in the field and analysis in terms of of a particular landscape, it invokes the historical- physical theory of the new, process-oriented explanatory framework of the period, including the geomorphology. The change in approach might be anthropomorphized youth-maturity-old age stages of demarcated from a paper published in 1960 (Mathews landform and landscape development. and Mackay, 1960; on glacial ice thrusting) or the first Upon completion of his Ph.D. Mackay took up his pingo paper (Mackay, 1962). It is definitively evident position in the University of British Columbia where he in the content of the Mackenzie Delta memoir was immediately recruited by the Geographical (Mackay, 1963). While the change appears relatively Branch of the Canadian government to join a small abruptly, it seems to have been gestating in Mackay’s group of geographers engaged in exploration of the mind throughout the latter 1950s as he read the Canadian Arctic territories Ð with the advent of the evolving literature and considered how best to Cold War a rapidly escalating strategic imperative. approach the analysis of the periglacial landforms he This catalyzed his love of the western Arctic was describing. landscape and people that set the course for the rest Mackay never discussed the methodological or of his career and life. philosophical motivations for his work so, to get

16 beyond invoking the spirit of the times, one must Characteristically, Mackay conducted extensive speculate on the wellsprings of his conversion. One numerical exercises to confirm the cross-wind probable encouragement was work with his UBC hypothesis, and he methodically considered and colleague William H. Mathews, lead author of the discarded alternate explanations that had been put 1960 paper. Mathews, one of the last professional forward, mostly by invoking his own meticulous field geologists of renaissance proclivities, was trained in observations. physics and geological engineering: he was well The occurrence and growth of massive sheets of prepared to place glacial ice thrusting Ð described by ground ice were analyzed by application of Stefan’s Mackay in earlier papers Ð on a theoretical and solution for freezing of ice lenses, put forward in the quantitative . The 1960 paper is one of the context of freeze and thaw by the American very few of Mackay’s papers in which he is not the geotechnical engineer K. Terzaghi (1952) and lead or sole author, attesting to Mathews’s influence detailed in engineering textbooks of heat conduction on this work. It surely demonstrated to Mackay the consulted by Mackay. But he first used his own field route to quantitative analysis of landforms. The two observations to dismiss the possibility that massive friends subsequently collaborated in a range of ice, including ice exhibiting deformed bedding, was studies of snow creep, needle ice occurrence and relict glacier ice or buried sea ice. patterned ground phenomena in the mountains near The analysis of the development of ‘closed Vancouver. Their joint work may never have come system’ pingos is the most elaborate theoretical about but for the circumstance that, before 1959, exercise that Mackay conducted. The basic concept Geography and Geology were practiced in a single of upward water expulsion from a freezing thaw bulb department at UBC. beneath a shoaling lake (or in a former lake basin) An influence of similar character may have been was advanced by A.E. Porsild (1938) and elaborated the pioneering textbook by S.W.Muller (1945; 1947), by F. Müller (1959). Mackay developed the theory Permafrost or Permanently Frozen Ground and mathematically by adapting analysis by the American Related Engineering Problems. This book, of which a geophysicist A.H. Lachenbruch (1957) for heat copy was held in the UBC Geology library, conduction in permafrost beneath heated buildings. summarized extensive Russian experience of (Unlike most geomorphologists of the period, Mackay permafrost along with the more sporadic (to that time) was competent in applied mathematics.) A lake that European and North American work. It provided does not freeze to the bottom is analogous to a Mackay with an overview of the history of technical building placed on frozen ground. Once the lake is studies of the subject, mostly conducted by engineers removed (or shoaled to the point that it does freeze to in a process-focused perspective. It particularly the bottom) frost begins to prograde into the emphasized the response of permafrost to ground subjacent thaw bulb (talik). The lateral margins being surface disturbances, such as the construction of frozen and water escape denied below by frost, buildings and roadways. It also gave extensive tables saturation or impermeable strata, the remaining water of the geotechnical properties of frozen ground is subjected to increasing pressure as the saturated developed by Russian engineers. Reference to the ground in the shrinking thaw bulb expands upon book appears in Mackay’s papers of the period, freezing. Water expulsion then occurs upward notably the 1960 ice-thrust paper and the 1963 towards the thinnest confining layer, which is Mackenzie Delta monograph. deformed in response to the hydraulic pressure. The The primary influence prompting the change in escaping water freezes under the deforming soil cap Mackay’s style, however, was his reading of the to form the pingo. All this was laid out in full literature and his own quick intelligence. He not only mathematical detail by, in effect, inverting picked up the spirit of change, he mined the literature Lachenbruch’s analysis. The process and predicted he was reading as a source of ideas for application to patterns of pingo growth were abundantly confirmed understand permafrost and its attendant phenomena. by Mackay through often ingenious observations In the 1950s it was still possible to keep track of the during much of the rest of his career. literature of the entire field of geomorphology and, Perhaps the most startling example of Mackay’s accordingly, Mackay found inspiration in a surprisingly ability to borrow and repurpose prior analyses lies in wide range of sources. This is best illustrated by a his analysis of Mackenzie Delta river channels Ð reading of the Mackenzie Delta memoir (Mackay, startling because he did not before or subsequently 1963), in which he repeated the pingo analysis of dwell on the topic, and it is not intrinsically a 1962 and introduced a number of new, theoretically periglacial topic. Yet it seemed called for in an grounded analyses. analysis of the physiography of the Mackenzie Delta. He returned to the oriented lakes he had In the absence (in 1960) of any hydrological or described in 1956, borrowing theory on equilibrium hydrographic measurements of the river, he shoreline form from the American coastal engineer P. considered aspects of the Delta channels that could Bruun (1953), developed and applied to oriented be measured from maps, including channel widths, lakes on the Alaskan north coast by R.W. Rex (1961) junction angles, and meander-form. He adopted to show that the lakes were shaped by winds ‘expected’ relations amongst these variables from dominantly blowing across (perpendicular to) the long work by C.C. Inglis (1949) and by L.B. Leopold and axis of the lakes (a formerly controversial issue). M.G. Wolman (1960). Inglis was a British civil

17 engineer engaged in the design and management of the ice sheets but, so far as I know, they have never unlined ‘regime’ canals and long-time Director of the been followed up. Poona hydraulics research station in British India; Leopold, an American engineer and geologist who 4.1 Scope of work was Chief Hydrologist and Director of the Water Resources Branch of the United States Geological The range of phenomena that attracted Mackay’s Survey. Following their lead, Mackay explored various interest continued to expand in the early mature morphometric relations amongst elements of the years. He initiated work on permafrost depth, on the channels’ geometry and found that the Delta channels annual cycle of freeze and thaw in the active layer, are generally relatively narrower than channels and on frost features in the active layer. He made the elsewhere. He supposed that the contrasting summer most detailed studies of massive ground ice and of and winter (sub-ice) flow regimes may be a reason for ice wedges and published an authoritative that. classification of ground ice (Mackay, 1972). He made Two aspects stand out from the foregoing studies of freeze-up and breakup along Mackenzie accounts of Mackay’s inspirations and analyses: first, River and of the mixing of Mackenzie River waters they were all firmly grounded in his own field and map with those of its tributaries (mixing of waters was a observations and, second, they all drew inspiration heavily researched topic at the time), principally Liard from engineering analyses or, at least, analyses by River, for which purpose he twice canoed the length engineers of analogous situations Ð a process of Mackenzie River. With W.H. Mathews he continued described by his colleague, J.K. Stager of ‘re-ordering to investigate the Quaternary history of the Mackenzie of the known to unravel the unknown’. These are region. All the while he continued his investigations of exactly the hallmarks of the ‘new’ geomorphology of pingos. A hallmark of his papers is that each one the 1950s and 1960s. Ross Mackay led in permafrost focuses on a single landform or a single aspect of studies the development of a style of work, designed environmental process or environmental history. Yet to reveal the details of landform genesis, that his in the field he was simultaneously pursuing a number contemporaries were leading in other branches of of projects. geomorphology. There is often a resemblance in modus operandi amongst these geomorphologists. A 4.2 Observation striking example arises in a comparison of the methods employed by Mackay and those of L.B. The principal basis for Mackay’s success was his Leopold who, as chief of the USGS Water Resources remarkable power of observation. But he was group, led the early reinvention of fluvial ingenious in finding ways to extend his observations geomorphology. Both gave absolute primacy to field (Mathews, 1985). A remarkable example was his use observations; both were keen observers; both of cheap, battery-powered watches that display the borrowed physical theory Ð usually theory adapted for day and week to determine the time of ice wedge engineering purposes Ð to analyze and provide cracking. He rewired the battery via a breaking wire rational explanation of their findings. So far as I know, across the crack position. The combination of date they never met. and day when the watch stopped is unrepeated for seven months, giving him an unequivocal and precise time of cracking. Several such installations positioned 4 MATURE WORK (from 1965) along a crack yielded the speed of lateral crack propagation. Multiple breaking wires buried in the Following the emergence of his propensity to view active layer gave evidence of the origin and direction landforms in genetic and theory-based terms, of vertical propagation of the crack. Mackay’s work assumed its mature form: To determine the geometry of frost cracks Mackay painstakingly detailed and often ingeniously arranged attached spheres of known diameter to a stiff field observations directed toward providing critical measuring tape and inserted them into the crack to evidence to decide amongst alternative, theory-based refusal. He thereby learned the width of the crack at hypotheses for the origin and development of the the known depth. He used these technical innovations landform. Essentially all of Mackay’s papers have to test the theoretical prediction by Lachenbruch their basis in field observation or, in a few papers, (1962) that cracks originate at the top of the ice laboratory investigations of frost heave. wedge (and base of the active layer). Mackay (1974) The single exceptional paper, however, surmised from evidence of ‘multistorey ice wedges’ represents perhaps the most outstanding example of that Lachenbruch was correct; ten years later his ingenuity. That is ‘Glacier flow and analogue (Mackay, 1984a) he presented direct evidence from simulation’ (1965). In that paper he appropriated an his breaking wires that upward cracking occurs about analogue electrical field plotter to model the flow two-fifths of the time (and can propagate into the patterns of the North American Pleistocene ice sheets overlying snow). on the basis that both phenomena are driven by flow down a potential gradient. The results are entirely 4.3 Collaboration plausible as a reflection of the large-scale features of

18 Ross Mackay was by nature inclined to pursue his conventional experimental mode, he studied frost research independently. He did not normally heave by upfreezing from below (Mackay, 1984b). collaborate directly with his graduate students and, His most ambitious experiment, however, was the aside from his colleague, Mathews, he made few joint artificial draining of a lake, “Illisarvik”, to study the investigations. Early in his career he published jointly progradation of permafrost into the lake bottom and with his field assistant (and later colleague) J.K. the development of ground ice features (Mackay, Stager, and he pursued freeze-up/breakup studies 1997). In this endeavor he was repeating an event with his sometime student D.K. Mackay. He also that has happened countless times along the sought and obtained collaboration with specialist retreating coast of the western Arctic. He had long scientists when he needed particular results. Hence before noted the correlation of pingo occurrences with he worked with the isotope chemist H.R. Krause to shallow depressions and reasoned that drained lakes, obtain isotope analyses of river water and ice; he formerly sufficiently deep not to freeze to the bottom, collaborated with the palynologist J. Terasmae to provide the ideal conditions for pingo growth. No examine Holocene pollen, with the palaeontologist doubt he hoped to grow a pingo, and now the site F.J.E. Wagner for mollusc identifications, and with indeed has done so (C.R. Burn, personal various geophysical colleagues at the Geological communication; May, 2015). Survey of Canada for remote sensing and coring of ground conditions. He frequently took academic 4.5 Publication colleagues to the Delta on his winter trips. Finally, late in his career he collaborated with his former postdoc, Ross Mackay was universally recognized within the C.R. Burn Ð his only postdoctoral student. permafrost and periglacial community as the premier investigator of his time. But he was not nearly so well 4.4 Experimentation known in the larger scientific, or even earth science, community as his achievements merit. A reason for Experimental evidence is considered to be the acme that is his loyalty to his own supporters. Throughout of scientific endeavor. It leads to unequivocally his Arctic career he was supported by the interpretable results. So it was natural that Geographical Branch and then the Geological Survey experimentation should attract Mackay’s attention. of Canada, both units of Canada’s national earth However, experimentation in field science is not easy: resources ministry (which has had several names it is not superficially obvious that environmental over the course of Mackay’s career). Accordingly, he conditions can be sufficiently well controlled for published in the Geographical Bulletin (house journal experimental purposes. One may argue, however, of the Geographical Branch), in Current Research that control of boundary and initial conditions is (the research-in-progress serial of the Geological sufficient, that in fact the normal continuation of Survey), and in the Canadian Journal of Earth forcing conditions (weather) is appropriate in field Sciences, Canada’s national earth science journal. experimentation (Church, 2011). They are all good journals, but not major international Mackay may have had experimental possibilities journals. He published but once in Science, and very on his mind from his earliest Arctic visit for, late in his occasionally in other international journals. But he did career, he published a study of ventifacts at Paulatuk not deliberately attract attention to his own that extended from his first season (1951) to 2003 achievements. (Mackay and Burn, 2005). He attempted to discern changes over 52 years to 158 natural rocks of varying lithology. To the extent that this was ‘experimental’ 5 LEGACY work it was an inadvertent experiment. Another inadvertent experiment began with a fire near Inuvik Ross Mackay has left behind a towering set of results in 1968. Mackay quickly established sites to study the in periglacial geomorphology. Always working from response of the active layer, contrasting burned sites field observation and using his observations to test with control sites beyond the perimeter of the fire the application of analytical theories on heat and (Mackay, 1995). water transfer under temperature gradients and at Mackay’s first deliberate experiment was his most freezing fronts, he constructed a body of fundamental spectacular one. Observing seasonal inflation and results that will remain current for decades. At the deflation of pingos on the order of 10-30 cm by means same time, he introduced periglacial geomorphology of survey, he hypothesized that the expulsion of water to quantitative and theory-based discipline. from the freezing bulb beneath the pingo accounts for We have considered how a revolution in the the inflation. He then drilled into one pingo to release approach to earth science in the 1950s appears to the expected water pressure, obtaining a satisfactorily have set Mackay on his course. A second revolution artesian water spout to confirm the hypothesis has occurred within the length of his career. (Mackay, 1977). Beginning in the 1970s, and gathering strength from In the field he manipulated snow cover using the 1980s on, earth science has increasingly come to fences to study the effect on ground temperature and be recognized as a ‘system science’, an endeavor to on ice wedge cracking. In the laboratory, in a more understand earth processes as the result of interaction amongst numerous subsystems. Nowhere

19 is this more evident than at Earth’s surface, with its Physiographic Implications, United States intersection of climate, hydrology, biotic processes, Geological Survey, Professional Paper 252, environmental chemistry and physical processes. The Reston, VA.: 57p. means to analyze the interactions amongst these Leopold, L.B. and Wolman, M.G. 1960. River elements is increasingly by numerical modeling Ð Meanders. Geological Society of America Bulletin, modelling that characteristically mixes analytical and 71: 769-794. empirical results. The model increasingly replaces Mackay, J.R. 1947. The North Shore of the Ottawa closed form theory as the summary of our River, Revue Canadienne de Géographie, 1(2-3): understanding of the environment. Important 3-8. characteristics of Mackay’s work, however, are that Mackay, J.R. 1949. Physiography of the Lower he focused his attention on one phenomenon at a Ottawa Valley, Revue Canadienne de time, and he always approached analysis of his Géographie, 3: 53-96. observations using classical analytic theory, Mackay, J.R. 1953. Fissures and Mud Circles on supported by statistical analyses to encompass the Cornwallis Island, The Canadian Geographer,1(3): range of natural variability in the phenomena that he 31-37. measured. He took no part in the system perspective. Mackay, J.R. 1956. Deformation by Glacier-ice at He had found a successful way to understand a part Nicholson Peninsula, N.W.T., Canada, Arctic, 9: of nature that held endless fascination for him, and he 218-228. remained loyal to that way to the end. Mackay, J.R. 1956. Notes on Oriented Lakes of the Liverpool Bay Area, N.W.T., Revue Canadienne de Géographie, 10: 169-173. ACKNOWLEDGMENTS Mackay, J.R. 1957. Notes on Small Boat Harbours, N.W.T., Geographical Branch, Canada Thanks to Michel Lapointe, Olav Slaymaker and John Department of Mines and Technical Surveys, Stager for reviews and corrections of the paper, and Geographical Paper, 13: 11p. especially to Chris Burn for a thorough review and for Mackay, J.R. 1958. The Anderson River Map Area, insights into Ross Mackay’s career. N.W.T. Geographical Branch, Canada Department of Mines and Technical Surveys, Memoir, 5: 137p. Mackay, J.R. 1959. Glacier Ice-thrust Features of the REFERENCES Yukon Coast. Geographical Bulletin, 13: 5-21. Mackay, J.R. 1960. Crevasse Fillings and Ablation Bagnold, R.A. 1941.The Physics of Blown Sand and Slide Moraines, Stopover Lake area, N.W.T.. Desert Dunes. Methuen, London. Geographical Bulletin, 14: 89-99. Bruun, P. 1953. Forms of equilibrium of coasts with a Mackay, J.R. 1960. Notes on Small Boat harbours of littoral drift. Berkeley, University of California, the Yukon Coast, Geographical Bulletin, 15: 19- Institute of Engineering Research, Waves 30. Research Laboratory, Reports, ser. 3, no. 347: 7p. Mackay, J.R. 1962. Pingos of the Pleistocene Church, M. 2010. The trajectory of geomorphology, Mackenzie Delta Area. Geographical Bulletin, 18: Progress in Physical Geography, 34: 265-286. doi: 21-63 10.1177/0309133310363992. Mackay, J.R. 1963. The Mackenzie Delta Area, Church, M. 2011. Observations and experiments. Ch. N.W.T., Geographical Branch, Canada 14 in Gregory, K.J. and Goudie, A.S., editors, Department of Mines and Technical Surveys, Handbook of Geomorphology, London, Sage: Memoir, 8: 202p.. 121-141. Mackay. J.R. 1965. Glacier flow and analogue Horton, R.E. 1945. Erosional Development of simulation. Geographical Bulletin, 7: 1-6. Streams and their Drainage Basins: Hydrophysical Mackay, J.R. 1972. The world of underground ice. Approach to Quantitative Morphology, Geological Annals, Association of American Geographers, Society of America Bulletin, 56: 275Ð370. 62: 1-22. Inglis, C.C. 1949. The behaviour and control of rivers Mackay, J.R. 1974. Ice-wedge Cracks, Garry Island, and canals. Poona, India, Central Waterpower, Northwest Territories. Canadian Journal of Earth Irrigation and Navigation Research Station, Sciences, 11: 1366-1383. Publication, 13, 2 vols.: 486p. Mackay, J.R. 1977. Pulsating Pingos, Tuktoyaktuk Lachenbruch, A.H. 1957. Three-dimensional Heat Peninsula, N.W.T. Canadian Journal of Earth Conduction in Permafrost beneath Heated Sciences, 14: 209-222. Buildings. United States Geological Survey, Mackay, J.R. 1984a. The Direction of Ice-wedge Bulletin, 1052B: 51-70. Cracking in Permafrost: Downward or Upward? Lachenbruch, A.H. 1962. Mechanicss of Thermal Canadian Journal of Earth Sciences, 21: 516-524. Contraction Cracks and Ice-wedge Polygons in Mackay, J.R. 1984b. The Frost Heave of Stones in Permafrost. Geological Society of America, the Active Layer above Permafrost with Downward Special Paper, 70: 69p. and Upward Freezing. Arctic and Alpine Leopold, L.B. and Maddock, T. 1953. The Hydraulic Research, 16: 439-446. Geometry of Stream Channels and some

20 Mackay, J.R. 1995. Active Layer Changes (1968- 1993) following the Forest-tundra Fire near Inuvik, N.W.T., Canada, Arctic and Alpine Research, 27: 323-336. Mackay, J.R. 1997. A Full-scale Field Experiment (1978-1995) on the Growth of Permafrost by means of Lake Drainage, Western Arctic Coast: a Discussion of the Method and some Results. Canadian Journal of Earth Sciences, 34: 17-33. Mackay, J.R. and Burn, C.R. 2005. A Long-term Field Study (1951-2003) of Ventifacts formed by Katabatic Winds at Paulatuk, Western Arctic Coast, Canada. Canadian Journal of Earth Sciences, 42: 1615-1635. Mathews, W.H. 1985. On the Scientific Method of J. Ross Mackay, In Church, M. and Slaymaker, O., editors, Field and Theory: Lectures in Geocryology, Vancouver, University of British Columbia Press: 1-16. Mathews, W.H. and Mackay, J.R. 1960. Deformation of Soils by Glacier Ice and the Influence of Pore Pressures and Permafrost, Royal Society of Canada, Transactions, Series 3, 54: 27-26. Muller, S.W. 1945; reprinted 1947), Permafrost or Permanently Frozen Ground and Related Engineering Problems, U.S. Army Military Intelligence Division, Office of the Chief of Engineers, Special Report: Strategic Engineering Study, 62: 231p. (1947 edition published at Ann Arbor, MI, by J.W. Edwards, Inc.) Müller, F. 1959. Beobachtungen über Pingos. Meddelelser om Gr¿nland, 153: 127p. Porsild, A.E. 1938. Earth Mounds in Unglaciated Arctic Northwestern America. Geographical Review, 28: 46-58. Rex, R.W. 1961. Hydrodynamic Analysis of Circulation and Orientation of Lakes in Northern Alaska. In Raasch, G.O., editor, Geology of the Arctic, Toronto, University of Toronto Press, v.2: 1021-1043. Strahler, A.N. 1950. Equilibrium Theory of Erosional Slopes Approached by Frequency Distribution Analysis, American Journal of Science, 248: 673Ð 96, 800Ð14. Strahler, A.N. 1952. Dynamic Basis of Geomorphology. Geological Society of America Bulletin, 63: 923Ð37. Terzaghi, K. 1952. Permafrost. Journal of the Boston Society of Civil Engineers, 39(1): 1-50.

21

22 JRM: His early accomplishments, 1950s to the mid 1980s, and his results, 1990 to 2005.

Hugh French1 2 1. Departments of Geography and Earth Sciences, University of Ottawa, Ontario, Canada

2. Department of Geography, University of Victoria, British Columbia, Canada

ABSTRACT Ross Mackay’s main scientific accomplishments were as a permafrost scientist working in the Western Canadian Arctic. Starting in 1951, he would spend approximately 3 months in the field each year for more than 40 years. He described many geographical features including the closed-system pingos of the Mackenzie Delta. In 1968, when the PCSP Western Arctic logistics base moved to Tuktoyaktuk, he commenced long-term permafrost studies in the delta region. His research added significantly to our general understanding of permafrost terrain, the nature of ground ice, the controls over permafrost distribution, the dynamics of the active layer and the effects of natural and man-induced disturbances to the tundra and arctic ecosystems.

RESUME Les principaux accomplissements scientifiques de Ross Mackay ont été en tant que chercheur du pergélisol travaillant dans l’Arctique de l’Ouest Canadian. Dès 1951, il passa environ 3 mois par an sur le terrain durant plus de 40 ans. Il décrivit de nombreuses formes géomorphologiques dont les pingos de système fermé du Delta du Mackenzie. En 1968, lorsque la base logistique d’Arctique de l’Ouest du PPCP fut déménagée à Tuktoyaktuk, il commença des études à long terme sur le pergélisol dans la région du delta. Ses travaux ont contribué de façon majeure à notre compréhension générale des terrains de pergélisol, de la nature de la glace de sol, des facteurs contrôlant la distribution du pergélisol, de la dynamique de la couche active et des effets des perturbations naturelles et anthropiques sur la toundra et les écosystèmes arctiques.

1 INTRODUCTION

Ross Mackay graduated as a physical geographer in 1949 and first visited the Canadian Arctic in 1951 and again in 1952 (Figure 1). Subsequently, he spent over 4 decades conducting field research in the Mackenzie Delta region of the Western Canadian Arctic. His research reflected the emergence in the 1950s of a quantitative terrain science led by individuals such as A. N. Strahler, L. B. Leopold , R. J. Chorley and A. Rapp that emphasized the study of geomorphic processes. Mackay’s first seminal contributions to permafrost science and periglacial geomorphology were regional physical geography monographs on the Anderson River, NWT and the Mackenzie Delta, NWT, both published by the Geographical Branch (Mackay, 1958; 1963a). Beginning in the mid 1960s, Mackay commenced long-term studies of permafrost and related processes in the delta region of the Mackenzie River. He established a field station on Garry Island in the outer part of the delta and used the PCSP station at Tuktoyaktuk as his base. By the late 1980s, Mackay had completed more than 35 field seasons in the Western Arctic and had become the foremost authority on permafrost in Canada, and probably the world. He continued field research for at least another decade and research papers were to appear for 20 more years, the last being in 2005. Others may appear posthumously. Figure 1. JRM about to depart for the field at Resolute, Cornwallis Island, 1952. (Photo supplied by JRM, 1999)

23 2 THE AGGRADATION OF PERMAFROST underlying ice wedge. A second monograph-length paper summarizes his field observations (Mackay, 2000). The pingos of the Pleistocene Mackenzie Delta attracted Perhaps his most elegant, but simple, attempt to Mackay’s interest not only because of their striking understand the formation of permafrost was the Illisarvik appearance but also because they reflected the experiment. This involved the artificial drainage of a aggradation (growth) of permafrost. This was to become a tundra lake in the summer of 1979 and the subsequent central theme of Mackay’s permafrost research. monitoring of the growth of permafrost on the drained Mackay documented the growth of pingos mainly lake-basin floor (Mackay 1997). In 2002, the first 20 years through the precise and repeated survey of growing of active-layer development and ice-wedge growth were pingos, aerial photographs, the drilling of pingos and the summarized (Mackay and Burn, 2002a; 2002b). After the stratigraphic and crystallographic analyses of their internal initial years, it appears that the growth of tundra structures. For example, he was able to show that the vegetation on the lake basin floor trapped snow, the birth of a small pingo (‘Porsild Pingo’) took place between thickness of the active layer increased and the rate at 1920 and 1930 (Mackay, 1988). Another, named which permafrost formed slowed. At the same time, the Aklisuktuk (‘Growing Fast’), was found to be subsiding formation of thermal-contraction cracking on the drained and Mackay speculated that its name, dating back 200 lake-basin floor was monitored (Figure 3). By 2002, no years, probably resulted from early rapid growth due to sign of an embryonic pingo could be detected. the accumulation of a large sub-pingo water lens (Mackay 1981a). A third example is provided by a section he was permitted to excavate through a pingo in July 1986 that illustrated the annual growth bands in the ice that typically comprise the core of a hydrostatic pingo (Figure 2) (Mackay 1990a). A monograph-length paper summarizes Mackay’s 40 years of pingo research (Mackay 1998).

Figure 3. Thermal-contraction cracks developed upon the Illisarvik drained lake floor, July 1985. JRM is talking with J. A. Heginbottom. Photo: H. M. French.

3. THE NATURE OF GROUND ICE

The nature and origin of ground ice is primarily the concern of geocryology. In the late 1950s, Russian scientists such as P.A. Shumskii and A. I. Popov were Figure 2. The JRM Pingo #20, August 1988. The ice core world leaders in this field. Ground ice allows inferences to consists of intrusive ice associated with the freezing of a be made as to the Quaternary history of permafrost. In sub-pingo water lens. Seasonal growth bands, consisting addition, the thaw of ice-rich permafrost gives rise to the of alternating clear and bubble-rich layers, can be seen in complex of processes generally termed ‘’. the ice. Photo: H. M. French Ground ice also presents many geotechnical and engineering problems. The widespread nature of thermal-contraction-crack Mackay was intensely interested in the origin of the polygons in the Mackenzie Delta also attracted Mackay’s massive ground-ice bodies that were exposed in coastal attention. He was able to demonstrate the thermal sections in the immediate vicinity of Tuktoyaktuk. He conditions under which cracking occurred and the utilized the Russian literature and took advantage of the frequency, magnitude, speed and sound of the process. It knowledge of visiting Russian scientists such as N. N. became clear that thermal-contraction cracking was Romanovskii, V. N. Konishchev and V. I. Solomatin. He intimately associated with the formation and growth of simplified the complex Russian classifications of ground permafrost. He was also able to explain the distinctive ice and proposed seven mutually-exclusive ground ice nature of the many low-centred (‘fortress’) polygons by types based upon the source of water prior to freezing demonstrating that the enclosing ramparts reflected the and the principal transfer process by which water moves lateral, thermally-induced movement of the active layer. to the freezing plane (Mackay 1972a). Initially, he Previously, it had been widely assumed that the ridges on concluded that the massive ground ice near Tuktoyaktuk either side of the crack reflected the growth of the

24 consisted of bodies of segregated-injection ice (Mackay, 1971) (Figure 4).

Figure 4. Massive ground ice exposed in summer 1973 at Peninsula Point, near Tuktoyaktuk. A. W. Gell (then a PhD student of JRM) and J. Veillette inspect the ice exposure. Photo: H. M. French.

In subsequent years, as massive icy bodies were described from elsewhere in the Western Arctic and Western Siberia, he referred to the Tuktoyaktuk ice bodies as being ‘intrasedimental’ in nature (Mackay and Dallimore, 1992). Others preferred to interpret their stratigraphy, ice crystallography and internal deformation as reflecting glaciotectonics and the icy bodies themselves as basal glacier ice. During this time, Mackay identified field criteria useful for differentiating between Figure 5. JRM and typical frost-crack instrumentation, massive icy bodies of segregation-injection or buried- Garry Island, 1985. The black tubes with white caps glacier-ice origin (Mackay, 1989). He was also one of the contain free-floating electromagnetic sensors to measure first to use oxygen isotopes to characterize ground ice of crack separation. Photo: H. M. French different types (1983a). the widely-assumed role of the cryostatic pressure 4. THE DYNAMICS OF THE ACTIVE LAYER thought to develop in the active layer during freezing. Instead, he highlighted its desiccated nature. This enabled Mackay developed many innovative techniques to study Mackay to develop his ‘equilibrium’ theory of hummock movements within the active layer. Many were perfected (non-sorted circle) formation (Mackay, 1980) (Figure 6). at his field station on Garry Island (Figure 5). Closely related was his understanding of slope To study the frequency, speed and direction of frost movement in the active layer. On Garry Island, where cracking, he initially used simple breaking cables and hummocks occur on sloping terrain, semi-flexible plastic telescoping hoops. In later years he progressed to tubes had indicated that movement was by plug-like flow electronic devices (see Figure 5). At some sites he used in late summer. Mackay (1981b) attributed this movement snow fences to Influence ground temperature and hence to creep associated with the thaw of the ice-rich layer that to reduce the frequency of cracking. In association with D. typically forms at the active layer-permafrost interface and K. Mackay, he studied the ameliorating effect of snow caused by the migration of unfrozen water upwards (in upon the ground thermal regime (Mackay and MacKay, winter) and downwards (in summer).This mechanism of 1974) and concluded that in areas such as the modern movement was at variance with the more Mackenzie Delta, the effect of snow upon permafrost traditional laminar flow mechanisms suggested by others. distribution can be considerable Mackay (1999a) also described the structures that As regards , Mackay developed a frost develop in the active layer by studying a number of tube to measure freezing levels. He also undertook shallow lakes that drained rapidly in the early 1950s. laboratory experiments to demonstrate the upfreezing of These sites were visited at intervals between 1969 and frost-susceptible stones (Mackay, 1984). He was able to 1997 and provided a unique opportunity to understand the demonstrate the two-sided nature of freezing in the active development of certain periglacial features. Not only did layer and the downward movement of water into frozen ice-wedge growth cease within several decades following ground (Mackay, 1983b). This allowed him to minimize lake drainage but the troughs associated with reactivated

25 syngenetic ice wedges (Mackay, 1990b) (Figure 7) and the manner in which hilltop epigenetic wedges evolve into hillslope anti-syngenetic wedges by prolonged slope denudation (Mackay, 1995a) has yet to be fully appreciated by the wider periglacial community. He concluded that anti-syngenetic wedges, many of which are reactivated Late Wisconsinan wedges, are probably more abundant in the hilly topography of the Western Arctic than either epigenetic or syngenetic wedges. The widespread slope denudation that this implies suggests that permafrost landscapes are uniquely dynamic and continually evolving.

Figure 6. Hummocks in the northern boreal forest south of Inuvik, August 1972. Hummocks (non-sorted circles) are the most common form of patterned ground in many arctic and subarctic regions. Photo: H. M. French ice-wedge polygons became ridges and experienced frost heave and . Collapse pits and cavities formed due to differential heave between silts and sands. These field observations mirror similar structures that can be produced under laboratory conditions. Mackay concluded that the origins of the structures he observed would be extremely difficult to interpret if he had commenced field observations in 2000, approximately 40- 50 years after drainage. His observations may help Figure 7. An anti-syngenetic ice-wedge polygon on a explain the many problematic structures reported in the steep 25⁰ slope, Horton River area, NWT. (Photo supplied Pleistocene literature. by JRM in 1993)

5. , AZONAL PROCESSES AND COLD- Another early process study, initiated in 1951, involved CLIMATE LANDSCAPE EVOLUTION the photography of wind-abraded rocks (ventifacts) that occur near Paulatuk, a small settlement in the Darnley Although permafrost became Mackay’s central research Bay area of the NWT. The same rocks were re-examined interest, he never lost interest in the manner by which in 1968 and again in 2003 (Figure 7). The conclusion was landscape is fashioned by cold-climate processes. that wind abrasion was negligible over the 50 year period In the early years of his research career, Mackay of study (Mackay and Burn, 2005b), This is possibly the travelled each spring by motorized canoe down the longest time period over which the rate of wind abrasion in Mackenzie River from Great Slave Lake to the Beaufort cold regions has been observed. Sea, a distance of over 1500 km. The 1950s and early 1960s was a time when fluvial process studies began to dominant English-speaking geomorphology. It was not unexpected, therefore, that Mackay observed, from first- hand experience, the distinctive cold-climate characteristics of an extremely large river, notably its freeze-up and break-up regimes, the characteristic ice- push features and the mixing of waters at its confluence with major tributaries (Mackay 1961; 1963b; 1970b, 1972b; Mackay and MacKay, 1977). These observations contrasted with the many small and medium scale catchment studies that were being undertaken in mid- latitudes by American and British geomorphologists. Figure 8. Wind-abraded JRM rock #2 at Paulatuk, Later, Mackay adopted a more traditional fluvial and photographed in 1951 (left) and 2003 (right). (Photos geomorphic approach when he described the role of supplied by JRM in 1995 and 2006 respectively). fluvial processes in permafrost landscape evolution and the breakthrough of the Horton River (Mackay and Equally revealing of Mackay’s long-term approach and Slaymaker, 1989). attention to detail was his 1974 to 1993 study of the The manner in which slopes evolve was never a weathering of 200 glacial erratics that lay on the exposed central concern of Mackay. Yet his recognition of anti- bottom of a drained Arctic lake (Mackay, 1999b). During

26 the 19 year period of study, he observed that only approximately 5% of rocks shattered. The majority were those that facilitated the entry of water thus leading to ice segregation. In contrast, impervious granitic rocks hydrofractured from the freezing of water in semi-closed systems and dolomitic rocks shattered either explosively or from thermal shock. These simple observations confirm not only the well-known importance of lithology in rock weathering but also summarize the complexity of frost- shattering and of cold-climate weathering.

6. NATURAL AND MAN-INDUCED DISTURBANCES TO THE FOREST-TUNDRA AND TUNDRA ECOSYSTEMS

Mackay was always aware of the many geotechnical and terrain problems presented by the increasing pace of economic development in the Western Arctic and by changing climate. For example, he noted how the northern (boreal) and the forest and tundra ecosystems had changed over the years in which he had worked in the Western Arctic (Figure 8). He also appreciated the thermal impact of natural climate variability and anthropogenic disturbance to permafrost terrain. He served on many advisory panels and committees and acted as an unpaid consultant to oil companies prospecting in the delta region in the 1970s and early 1980s. He was one of the first to describe the effects of surface disturbances to tundra terrain and clearly explained the nature of thermokarst subsidence (Mackay 1970b). Following the forest-tundra fire near Inuvik in 1968, he monitored active layer changes for 25 years (Mackay 1995b).

7. CONCLUSIONS Figure 9. Vegetation changes near tree line, Reindeer Hills, east side of Mackenzie Delta, 1954-2006. Note the By the mid 1980s, Mackay had completed over 35 years increase in spruce on the ridges. Tundra vegetation is on of fieldwork in the Western Arctic. In addition to spending the hilltop. (Photos supplied by JRM in 2007). 2-3 months in the field during the summer, he had made numerous visits to his field sites in the winter months. His experimental studies and monitoring programs were REFERENCES (JRM) largely in place and ongoing Around this time, and following his retirement from Mackay, J. R. 1958. The Anderson River map-area, NWT. UBC in 1981, he began to contemplate his scientific end Geographical Branch Memoir 6. game. In1985, he invited Alan Heginbottom (GSC) and Mackay, J. R., 1961. Freeze-up and break-up of the the writer to tour his many sites. He asked if we were Lower Mackenzie River, Northwest Territories. In: interested in taking over a number of his projects. For Geology of the Arctic, G. O. Raasch, ed., University of different reasons, we both declined. He fully understood. Toronto Press: 1119-1134, He continued with his fieldwork for another 15 years, Mackay, J. R. 1963a. The Mackenzie Delta area, NWT. started to collaborate with C. R. Burn in the early 1990s, Geographical Branch Memoir 8. and systematically began preparation of the appropriate Mackay, J. R., 1963b. Progress of break-up and freeze-up summary publications, all cited in this paper. along the Mackenzie River. Geographical Bulletin, 19: Mackay’s contribution to our understanding of the 103-116. terrain conditions of the Western Arctic, and of permafrost Mackay, J. R., 1970a. Lateral mixing of the Liard and in general, cannot be over-stated Mackenzie rivers downstream from their confluence. Canadian Journal of Earth Sciences, 7: 111-124. ACKNOWLEDGEMENTS Mackay, J. R. 1970b. Disturbances to the tundra and forest tundra environment of the Western Arctic. Although the writer never formally collaborated with JRM, Canadian Geotechnical Journal, 7: 420-432. he wishes to acknowledge the numerous personal Mackay, J. R. 1971. The origin of massive icy beds in communications over the years in which JRM generously permafrost, western Arctic coast, Canada. Canadian shared his observations, ideas and experience. Journal of Earth Sciences, 8: 397-422.

27 Mackay, J. R. 1972a. The world of underground ice. Mackay, J. R. 1999b. Cold-climate shattering (1974 to Annals, American Association of Geographers, 62: 1- 1993) of 200 glacial erratics on the exposed bottom of 22. a recently drained Arctic lake, western Arctic coast, Mackay, J. R., 1972b. Application of water temperatures Canada. Permafrost and Periglacial Processes, 10: to the problem of lateral mixing in the Great Bear- 125-136. Mackenzie River systems. Canadian Journal of earth Mackay, J. R. 2000. Thermally-induced movements in ice- Sciences, 9: 913-917. wedge polygons, western Arctic coast. Géographie Mackay, J. R. 1980. The origin of hummocks, western physique et Quaternaire, 54: 41-68. Arctic coast, Canada. Canadian Journal of Earth Mackay, J. R. and Burn, C. R. 2002a. The first twenty Sciences, 17: 996-1006. years (1978-1979 to 1998-1999) of ice-wedge growth Mackay, J. R. 1981a. Aklisuktuk (Growing Fast) Pingo, at Illisarvik experimental drained-lake site, western Tuktoyaktuk Peninsula, Northwest Territories, Canada. Arctic coast, Canada. Canadian Journal of Earth Arctic, 34: 270-273. Sciences, 39: 95-111. Mackay, J. R., 1981b. Active-layer slope movement in ac Mackay, J. R and Burn, C. R. 2002b. The first twenty ontinuous permafrost environment, Garry island, years (1978-1979 to 1998-1999) of active layer Northwest territories, Canada. Canadian Journal of development, Illisarvik experimental drained lake- Earth Sciences, 18: 1666-1680. site, western Arctic coast, Canada. Canadian Journal Mackay, J. R. 1983a. OxygenÐisotope variations in of Earth Science, 39: 1657-1674. permafrost, Tuktoyaktuk Peninsula area, Northwest Mackay, J. R. and Burn, C.R. 2005. A long-term field Territories. Geological Survey of Canada Paper 83- study (1951-2003) of ventifacts formed by katabatic 1B: 7-74. winds at Paulatuk, western Arctic coast, Canada. Mackay, J. R. 1983b. Downward water movement into Canadian Journal of Earth Sciences, 42: 1615-1635. frozen ground, western Arctic coast, Canada. Mackay, J. R. and Dallimore, S. R. 1992. Massive ice of Canadian Journal of Earth Sciences, 20: 120-134. the Tuktoyaktuk area , western Arctic coast, Canada. Mackay, J. R. 1984. The frost heave of stones in the Canadian Journal of Earth Sciences, 29: 1235-1249. active layer above permafrost with downward and Mackay, J. R. and MacKay, D. K. 1974. Snow cover and upward freezing. Arctic and Alpine Research, 16: ground temperatures, Garry Island, NWT. Arctic, 27: 430-446. 287-296. Mackay, J. R. 1988. The birth and growth of Porsild Mackay, J. R. and MacKay, D. K., 1977. The stability of Pingo. Tuktoyaktuk Peninsula, District of Mackenzie. ice-push features, Mackenzie River, Canada. Arctic, 14: 267-284. Canadian Journal of Earth Sciences, 14: 2213-2225. Mackay, J. R. 1989. Massive ice: some field criteria for Mackay, J. R. and Slaymaker, O. 1989. The Horton River the identification of ice types. Geological Survey of breakthrough and resulting geomorphic changes in a Canada Paper 89-1G: 5-11. permafrost environment, Western Arctic coast, Mackay, J. R. 1990a. Seasonal growth bands in pingo ice. Canada. Geografiska Annaler, 71A: 171-184. Canadian Journal of Earth Sciences, 27, 1115-1125. Mackay, J. R. 1990b. Some observations on the growth and deformation of epigenetic, syngenetic and anti- syngenetic ice wedges. Permafrost and Periglacial Processes, 1, 15-29. Mackay, J. R. 1995a. Ice wedges on hillslopes and landform evolution in the late Quaternary, western Arctic coast. Canadian Journal of Earth Sciences, 32: 1093-1105. Mackay, J. R. 1995b. Active layer changes (1868 to 1993) following the forest-tundra fire near Inuvik, NWT, Canada. Arctic and Alpine Research, 27: 323-336. Mackay, J. R. 1997. A full-scale field experiment (1978- 1995) on the growth of permafrost by means of lake drainage, western Arctic coast: a discussion of the method and some results. Canadian Journal of Earth Sciences, 34: 17-33. Mackay, J. R. 1998b. Pingo growth and collapse, Tuktoyaktuk Peninsula area, western Arctic coast, Canada: a long-term field study. Géographie physique et Quaternaire, 52: 271-323. Mackay, J. R. 1999a. Periglacial features developed on the exposed lake bottoms of seven lakes that drained rapidly after 1950, Tuktoyaktuk peninsula area, western Arctic coast, Canada. Permafrost and Periglacial Processes, 10: 39-63.

28 Science to Technology—The importance of understanding the fundamentals of permafrost science for engineers practicing in the north

Don W. Hayley, Kelowna, British Columbia, Canada

ABSTRACT The author met J. Ross Mackay in 1972 after requesting his assistance to prepare for a pipeline data collection program in the Mackenzie Valley. Mackay provided selflessly of his time to transfer basic knowledge of permafrost and ground ice with advice on what to look for in the field. That relationship carried on to the end of his life with exchange of information useful for engineering purposes. He was outstanding in his dedication to his research and to those who worked with him and then went on to carve out their own niche in this important region of Canada. The paper looks at the importance of understanding the genesis of ground ice and its importance to engineering. Polygonal ground formed by ice wedges is identified as a high-risk terrain unit. The paper relates this knowledge to decisions that had to be made when routing a new highway from Inuvik to Tuktoyaktuk. The route crosses 140 km of some of the most difficult terrain for road building in Canada.

RÉSUMÉ L’auteur a rencontré J. Ross Mackay en 1972 lors des études du projet d'un couloir de gazoduc dans la vallée du Mackenzie. Mackay lui a transmis ses connaissances en pergélisol et en glace de sol ainsi que ce qu’il faudrait chercher sur le terrain. Cette relation a continué tout au long de sa vie avec des échanges d’informations utiles en ingénierie. C’était un chercheur dédié à ses travaux et à ceux qui travaillaient avec lui et qui ont ensuite fait carrière dans cette importante région du Canada. Cet article démontre l’intérêt de comprendre la naissance des glaces de sol et leur importance en ingénierie. La structure polygonale formée par les coins de glace est identifiée comme très dangereuse. Cette information est ensuite liée aux décisions prises lors de la création de l’itinéraire de la nouvelle autoroute entre Inuvik et Tuktoyaktuk, route qui traverse 140 km de terrain le plus difficile pour la construction d’une route au Canada.

1 INTRODUCTION that are indicative of an active ground ice environment. From that time forward, we became friends and shared It is an unfortunate reality today that civil engineers experiences through his many published papers and at graduating from Canada’s Universities learn little about the International Permafrost Conferences. His dedication the permafrost that underlies fifty percent of our country. to the science and genuine desire to help others was an Even graduate students in civil/geotechnical are fortunate enormous benefit to all those who chose to practice if they can get more than a lecture or two on how to engineering in regions where permafrost is present. recognize and accommodate the unique challenges posed by design and construction for permafrost regions. Those that choose to practice engineering in Canada’s 2 GEOTECHNICAL ENGINEERING PRACTICE North must learn the tools by working with a mentor and EVOLVES FROM GOOD SCIENCE by self-study. That self-study component must include a grasp of the fundamental properties of permafrost that There is nothing more effective at bringing science to affect its distribution and its ability to support the engineering practice than a natural impediment to infrastructure that northern Canada needs. We frequently proposed new infrastructure to which there seems to be turn to an active scientific group of permafrost no practical solution. In the early 1970’s, following geographers currently working in Canada, many of whom discovery of oil at Atkinson Point on Tuktoyaktuk can link their mentorship and motivation to the work of Peninsula, it was recognized that permafrost was a Professor J. Ross Mackay. serious impediment to either a warm oil pipeline south This paper examines how Mackay’s work helped from the Mackenzie Delta or a cold gas pipeline in shape engineering practice in the North through the discontinuous permafrost of the upper Mackenzie Valley. 1970’s and 80’s. In 1972, the author was assigned a task There was a clear need to put the engineers who were of planning and executing a helicopter supported, looking for information together with the scientists who geotechnical drilling program that was required for early had been studying permafrost in order to look for feasibility assessment of an Arctic gas pipeline (Gas solutions. That process was initiated at a conference in Arctic Systems). That program covered the lower Ottawa in 1972, organized by the National Research Mackenzie Valley and northern Yukon to the Alaska Council and referred to as the Canadian Northern Pipeline border. Dr. Mackay took an entire day to educate this Research Conference. As a recent graduate with a strong neophyte engineer on how to recognize terrain features desire to better understand the hidden perils of northern

29 terrain, I had the opportunity to attend that conference. the modern Mackenzie Delta was formed by “open Ross Mackay presented a lengthy paper simply entitled system”, downward freezing of unfrozen soils. Mackay “Permafrost and Ground Ice” (Mackay, 1972) in which he refers to this common form of ground ice as; wisely told the engineers that: “aggradational segregated ice” or epigenetic ice, as the soil formation predates the ground ice that is now found in “The most important engineering element in permafrost it. This ice type gives rise to extensive and sometimes construction is the distribution and behaviour of ground near continuous ground ice just below the current active ice in permafrost. If such high ice content permafrost can layer in Mackenzie Valley permafrost. This is in contrast to be mapped and thus avoided or designed for, the older permafrost soils of the Tuktoyaktuk Peninsula and engineering problems in permafrost will be substantially Yukon Coast that have a much longer permafrost history minimized.” with periods of closed system freezing following interglacial periods that produces massive icy beds This seemingly simple message was right for the time as exposed on the coast of the Tuktoyaktuk Peninsula as it helped to establish a solid user group for Ross’s shown in Figure 1 (Mackay 1971a, 1971b). research into the world of underground ice. At the same The distribution of ground ice can be even more conference, Dr. Jack Mollard introduced us to the new complex in regions of central Yukon where relict science of terrain classification and mapping using permafrost, believed to be as old as 700,000 years, has airphotos and Dr R.M. Hardy, then Dean of Engineering at been found at substantial depths in placer mine University of Alberta provided an overview of the exposures (Froese et al., 2008). Ground ice, believed to engineering challenges. We went away from that have formed in early glacial times, was also encountered conference with at least two new tools to work with; the in a deep clay horizon shown in Figure 2. These importance of identifying terrain with a prevalence of unexpected events prove to be challenging for engineers ground ice and stereo airphotos that can provide a tool for working on resource projects where stability of pit walls identification of surface indicators along any proposed and waste material embankments often control route. operations.

3 IMPORTANCE OF GROUND ICE

The realization that ground ice should be treated as a stratum within our geotechnical logs brought about a need to understand how it got there. Not unlike classifying a sand or gravel as alluvium or moraine, we needed to know more about ground ice origins. Ross Mackay’s work on the origin of ground ice in key papers published in the early 1970’s, such as: “The Origin of Massive Icy Beds in Permafrost , Western Arctic Coast, Canada” (Mackay, 1971a), “The World of Underground Ice” (Mackay, 1972b) and “The Growth of Pingos, Western Arctic Coast, Canada” (Mackay, 1973) gave us the basics we needed. Engineers practicing in permafrost regions learned to use that information to make rational judgements on the probable nature of ground ice from terrain features. The Figure 1. Exposure of massive Icy beds, near effective use of stereo airphoto interpretation with Tuktoyaktuk, June 2012 appropriate ground reconnaissance quickly became an essential component of northern geotechnical practice. The continuing possibility that some massive ice . features may be remnants from past that have been preserved by granular cover has been a source of 3.1 Quaternary History debate among permafrost scientists in Canada for some time. Mackay was not a strong proponent of this potential It became clear from Mackay’s early work in the ground ice source because he was able to otherwise Mackenzie Delta Region that some knowledge of rationalize ground ice origin within his research sphere of Quaternary history for the particular terrain provides a the Mackenzie Delta environs. Ground ice features found valuable insight from which to evaluate ground ice at the base of large eskers within the Canadian Shield distribution and risk. He reminded us that the age of Region of Northwest Territories and Nunavut do not fit any permafrost is directly related to regional glacial episodes. of Mackay’s ground ice description. A typical exposure of We think of permafrost in the Western Canadian Arctic as clear blue ice, shown in Figure 3, exposed at the contact Late Pleistocene in age, some 13,000 years old. Mackay between a dense impermeable till sheet below and a showed us the significance of understanding the different clean sandy gravel esker above are common occurrences phases of Laurentide glaciation that left behind permafrost in that region. Inspection of an ice exposure at Ekati in the Mackenzie Delta region with ages ranging from Diamond Mine by a group of permafrost scientists is 44,000 to 13,000 years. The most recent permafrost of shown in Figure 3. That site visit conducted in

30 conjunction with the Seventh International Conference on permafrost. The near surface ice can become a seepage Permafrost, Yellowknife, 1998, resulted in consensus that corridor that will erode the core of a linear ice wedge that the ice must be a glacial remnant. Buried glacial ice, is supporting an embankment resulting in near embedded in moraines and below eskers, has now been instantaneous sinkhole formation. Road collapse caused accepted as ground ice features that also pose risk for by thermal erosion of ice wedges has been a direct cause roads and foundation infrastructure. of at least one traffic fatality on the Dempster Highway (Hayley et al, 1987). Route selection tries to avoid or at least minimize regions of active ice wedge polygons. Not all ice wedges are directly visible as polygonal patterned ground as they also occur on slopes where soilifluction processes can obliterate the surface expression we look for with aerial photography (Mackay 1995). Where ice wedge polygon fields can’t be avoided then the risk of thermal erosion followed by road collapse must be assessed and the embankment design modified to protect the shoulders from pond formation and retrogressive slumping. This usually goes along with a clear expectation that maintenance activities will escalate where ice wedges are included in the foundation soils. Ross Mackay devoted much of his research effort toward understanding the freezing and cracking that initiates ice wedges, how they develop and change over time (Mackay, 2000). His long-term research on Garry Island has provided the source of knowledge that engineers need to evaluate the risk whenever ice wedge Figure 2. Ice lenses up to 1 m thick exposed in a deep polygon fields must be crossed. clay horizon believed to be relict permafrost, Minto Copper Mine, Yukon 4 THE INUVIK TO TUKTOYAKTUK HIGHWAY (ITH)

An all-weather road to Tuktoyaktuk, NT that is under construction in 2015 incorporates many of the practical applications of lessons learned by engineers who have benefitted from the research of Ross Mackay. There were initiatives dating back to the 1960’s to construct an overland highway that would connect Inuvik to Tuktoyaktuk on the Beaufort Sea. That would be a logical completion of the Dempster highway system that would provide road access to an Arctic seaport and service the community of Tuktoyaktuk. A route, approximately 140 km in length, has been studied and refined a number of times but was always considered too difficult because of the nature of the permafrost terrain and therefore too costly. The Federal Government completed construction of the Dempster Highway in 1978 but there was a long follow-up period of contract litigation that showed a need Figure 3. Massive ice exposure at the base of an esker, for better terrain knowledge. Design and contracting Ekati Diamond Mine, 1998 methods also needed fine tuning to the Arctic environment before another major highway project was initiated over 3.2 Ice Wedges and Frost Effects difficult permafrost terrain. Permafrost conditions on the Dempster highway have been described elsewhere in the Near surface ice wedges that form polygonal ground are proceedings by Burn et al (2015). evident to most who work and build in permafrost regions. The most troublesome terrain on the Dempster The explanation for ice wedge polygon formation is highway has been where it crosses the Richardson usually attributed to Lachenbruch (1962) based on his Mountains at the border between Yukon and Northwest assessment of the massive polygon fields on the Alaska Territories. The eastern foothills of the mountains, North Slope. Ground surface cracking at the onset of commonly known as the Peel Plateau, have clearly winter in response to rapidly cooled ground initiates ice demonstrated to highway engineers the risks that shallow wedge growth. The presence of these near surface linear ground ice, both massive ice and wedge ice, poses to ground ice features poses a substantial risk to all forms of embankment stability. The site of the embankment transportation infrastructure in regions of continuous

31 collapse at NT-km 8.5 in 1985, shown in a current • Recognition that a particularly ice rich zone photograph, Figure 4, remains a persistent challenge. occurs virtually everywhere below the active layer and can extend to depths of several metres. This concentration of ground ice is so prevalent that it must be assumed to be continuous • The embankment fill must be continuous with a minimum thickness that can vary somewhat with active layer properties and ground ice type. • The tundra surface is highly sensitive to disturbance everywhere therefore the construction plan must preclude any operation over tundra that was not protected by a fill structure or winter road. All fill placement must occur when the active layer is fully frozen • Material sources are scarce and usually contained excess ice. It was recognized that winter fill placement followed by grading, shaping and compaction of the embankment in summer Figure 4. Long-term embankment instability over shallow would be a requirement. Techniques for material ice wedges, Dempster Highway, NT-km 8.5, June 2015 sorting to minimize ice content would have to be photo developed within the individual borrow pits.

The Project Description Report for the new highway was submitted in 2010 (Kiggiak-EBA 2010), and the environmental hearings were complete in spring of 2013 with a positive endorsement of the project. Construction began during the 2013-2014 winter season after a cost- sharing agreement was signed between the federal and territorial governments. It was anticipated that construction would occur over four successive winter seasons, two of which have passed at the time of this conference. Embankment construction is on schedule and about half complete as of June, 2015.

Figure 6. Thermokarst lake with active thaw-flow slide a few km south of ITH route near Inuvik. June, 2012 photo.

Figure 5. The ITH road alignment had segments with a challenging choice between fields of ice wedge polygons and unstable thermokarst lakeshores. June 2015, photo

The technical challenges that had to be dealt with for this project included: • Route optimization to avoid obvious terrain Figure 7. Material source access road over ice-wedge features indicative of ice wedge polygon fields polygons just south of Tuktoyaktuk. and thaw flow slides (Figures 6 and 7).

32 A number of options were examined at the final design REFERENCES stage to maintain the original active layer beneath the embankment in a fully frozen condition. Embankment Burn, C.R., Moore, J.L., O’Neil H.B., Hayley, D.W., thicknesses were optimized with 2D geothermal analyses Trimble J.R., Calmels, F., Orban, S.N., and Idrees, M. based on four general terrain types. Options that included 2015. Permafrost characterization of the Dempster insulation with sideslope berms for high risk terrain, such Highway, Yukon and Northwest Territories. This as ice wedge polygon fields, were developed as an Preceedings Volume alternative to simply increasing embankment fill thickness. De Guzman, E.M., Piamsalee, A., Alfaro, M., Arenson, However, the design-build contractor had a strong L.U., Doré, G., and Hayley, D.W. 2015. Initial preference for thicker fills rather than address the monitoring of instrumented test sections along the complexities associated with multi-layered embankment Inuvik-Tuktoyaktuk Highway. This preceedings volume systems. There was a strong but understandable pull to Froese, D.G., Westgate, J.A., Reyes, A.V., Enkin, R.J. keep construction simple given the winter conditions. and Preece, S.J. 2008. Ancient permafrost and a The behavior of winter-constructed thick fills required future, warmer arctic, Science, V321,1648/1157525 to achieve design vertical grades were a concern as Hayley, D.W. and Bowen, J.A. 1987. Ground ice in experience has shown that the sideslopes will creep on an continuous permafrost, a hazard to safe highway ice-rich foundation resulting in longitudinal cracking. A operation, Western Association of Canadian Highway test section was constructed with sideslopes that included Officials. Annual Conference, Yellowknife, June 2 geotextile layers as reinforcement in April 2015. A Kiggiak-EBA Consulting Ltd. 2010. Project description description of the test site and how it was constructed is report for construction of the Inuvik to Tuktoyaktuk included in these proceedings (De Gusman et al, 2015) highway. Hamlet of Tuktoyaktuk, Town of Inuvik and Government of Northwest Territories. Available on line 5 CONCLUSIONS at:http://www.dot.gov.nt.ca/_live/documents/cont ent/Project%20Description%20Report%20Inuvik The legacy of Ross Mackay has been permanently %20to%20Tuktoyaktuk%20Highway.pdf embedded in permafrost science in Canada and has also Lachenbruch, A.H. 1962. Mechanics of thermal shaped the evolution of engineering practice for our contraction cracks and ice-wedge polygons in northern environment. Engineers learn by experience permafrost. Geological Society of America, New York, from each project they undertake. In order for it not to be NY, USA. Special Paper 70. 69 p. the same experience repeated again and again, we must Mackay, J.R. 1971a. The origin of massive icy beds in collectively build upon an accumulation of knowledge that permafrost, western Arctic coast, Canada. Canadian contributes to a better understanding of our natural Journal of Earth Sciences, 8: 397-422. environment. That is particularly true in the North where Mackay, J.R. 1971b. Ground ice in the active layer and permafrost terrain continually challenges us to produce the top portion of permafrost. Proceedings of a reliable yet cost-effective infrastructure. Seminar on the Permafrost Active Layer, 4 and 5 May, I was fortunate to deliver a presentation at a special 1971. National Research Council of Canada, celebration for Ross Mackay’s 90th birthday on February Associate Committee on Geotechnical Research, 17, 2006. Several days later I received mail from Ross Ottawa, ON, Canada. Technical Memorandum 103: thanking me for the presentation. The message was 26-30. hand-written on the back of a photograph taken near Mackay, J.R. 1972a. Permafrost and ground ice. In Inuvik. That photograph showed a snow fence he Proceedings, Canadian Northern Pipeline Research installed in 1978 to study the effects of snow cover on Conference, 2-4 February, 1972. National Research active layer thickness and his message to me included an Council of Canada, Associate Committee on update on what 10 years of enhanced snow cover could Geotechnical Research, Ottawa, ON. Technical do to the active layer. It is truly hard to comprehend the Memorandum 104: 235-239. passion he had for his work and his joy at spreading the Mackay, J.R. 1972b. The world of underground ice. knowledge. May we all take a message of inspiration Association of American Geographers, V 62, No1, 1- from a life so well lived. 22 Mackay, J.R. 1973. The growth of pingos, western Arctic ACKNOWLEDGEMENTS coast, Canada. Canadian Journal of Earth Sciences, 10: 979-1004. The author would like to acknowledge the contribution of Mackay, J.R. 2000. Thermally induced movements in ice- Chris Burn for suggesting this paper for the Mackay wedge polygons, western Arctic coast: a long-term Symposium, for reviewing the draft for inconsistencies study. Géographie physique et Quaternaire, 54: 41-68. and for assisting with the reference list. Mackay, J.R. 1970. Disturbances to the tundra and forest tundra environment of the western Arctic. Canadian Geotechnical Journal, 7: 420-432. Mackay, J.R. 1974. Ice-wedge cracks, Garry Island, N.W.T. Canadian Journal of Earth Sciences, 11: 1366- 1383.

33 Rampton, V.N. and Mackay, J.R. 1971. Massive ice and icy sediments throughout the Tuktoyaktuk Peninsula, Richards Island, and nearby areas, District of Mackenzie. Geological Survey of Canada, Ottawa, ON, Canada. Paper 71-21, 16 p. Mackay, J.R. 1995. Ice wedges on hillslopes and landform evolution in the late Quaternary, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 32: 1093-1105. Mackay, J.R. 1972. Offshore permafrost and ground ice, southern Beaufort Sea, Canada. Canadian Journal of Earth Sciences, 9: 1550-1561. Mackay, J.R. 1963. The Mackenzie Delta area, N.W.T. Geographical Branch, Department of Mines and Technical Surveys, Ottawa, ON. Memoir 8, 202 p.

34

Part 2 Regional studies on permafrost in Canada

35

36 Geohazard investigations of permafrost and gas hydrates in the outer shelf and upper slope of the Canadian Beaufort Sea

S.R. Dallimore1, C.K. Paull 2, A.E. Taylor 3, M. Riedel1, H.A. MacAulay 3, M.M. Côté1, and Y.K. Jin4 1Geological Survey of Canada – Pacific Division, Sidney, BC, Canada 2Monterey Bay Aquarium Research Institute, Moss Landing, CA, USA 3Geological Survey of Canada (Retired) 4Korean Polar Research Institute, Incheon, Korea

ABSTRACT Building on the early work of J. Ross Mackay, this paper reviews the state of knowledge of offshore permafrost and gas hydrates beneath the Beaufort Shelf. A thick and extensive interval of transgressed terrestrial permafrost is present beneath the central shelf, warming seaward and pinching out at 90-100 m water depth near the shelf-slope break. A complex gas hydrate regime is recognised with possible intrapermafrost and subpermafrost gas hydrate beneath the shelf, a zone with no gas hydrates at the shelf-slope transition, and a marine gas hydrate zone where water depths are greater than ~270 m. We believe that changes induced by transgression (formation and thawing of ground ice and gas hydrate) have created unique porous media conditions in this setting that can influence geohazards and active geologic processes. To this end we have undertaken a variety of multidisciplinary field investigations that have included marine geophysics, high resolution sea bed mapping, sediment coring, sea floor moorings and ROV dives.

RÉSUMÉ Se basant sur les premiers travaux de J. Ross Mackay, ce manuscrit fait la synthèse de l’état des connaissances sur le pergélisol sous-marin et sur les hydrates de gaz se situant sous le plateau continental de la mer Beaufort. Une vaste zone épaisse de pergélisol terrestre formée lors d’une transgression marine est présente sous la région centrale du plateau. Vers le large, le pergélisol se réchauffe et s’amincit progressivement jusqu’à disparaitre à des profondeurs d’eau de 90-100 m, près de la transition plateau-pente continentale. Un régime complexe d’hydrates de gaz est documenté avec la présence possible d’hydrates de gaz intra et subpergélisol sous le plateau, une zone sans hydrates de gaz à la transition plateau-pente continentale, et enfin une zone d’hydrates de gaz marins là où les profondeurs d’eau sont supérieures à ~270 m. Nous croyons que les changements résultant de la transgression marine (soit la formation et la fonte de glace de sol et des hydrates de gaz) ont créé les conditions d’un milieu poreux unique qui peuvent influencer les risques géologiques et les processus géologiques actifs. Nous avons donc entrepris un programme multidisciplinaire de recherche sur le terrain qui inclut la géophysique marine, la cartographie de fonds marins à haute résolution, le carottage de sédiments, des systèmes d’amarrage aux fonds marins, et des plongées avec un véhicule sous-marin téléguidé.

1 INTRODUCTION indicators of the occurrence of offshore permafrost were also appraised and industry conducted a number of In his 1972 publication in the Canadian Journal of Earth geotechnical coring studies. Sciences (Mackay 1972), J. Ross Mackay made a The recent sale of offshore leases in the outer milestone scientific contribution that substantially shelf and upper slope has stimulated a new phase of advanced the understanding of the factors controlling the exploration and scientific investigations in the Canadian distribution, origin, and thermal regime of offshore Beaufort Sea. In the past five years, we have had the permafrost beneath Arctic shelves. Mackay proposed that opportunity to undertake research expeditions to two types of permafrost were widespread beneath the investigate permafrost and gas hydrates in the transitional Beaufort Shelf; subsea permafrost that was in equilibrium area from the outer shelf to upper slope. Following the with negative seabottom temperatures, and submerged example of Mackay, and fundamental to our science, we terrestrial permafrost which is in disequilibrium with have developed testable hypotheses to guide our seabottom temperatures. He also pointed out that in fieldwork. To this end we have carried out geothermal certain settings, a thin active layer was likely present in modeling to provide a conceptual model of the present the offshore near the sea bed. The concepts proposed by geothermal setting of the southern Beaufort Sea and Mackay were quickly accepted by industry, which at the consider changes that may have occurred in the time was actively undertaking hydrocarbon exploration in permafrost and gas hydrate regime as a consequence of the area. Mackay’s work helped to foster several decades Holocene marine transgression. This paper briefly of close collaboration between industry, government, and reviews the regional setting of this area and provides an academic scientists. This collaboration led to the first overview of our approach to studying geohazards. geothermal measurements and well log interpretations that documented and mapped a thick body of offshore permafrost beneath the Beaufort Shelf. Geophysical

37 2 PERMAFROST AND GAS HYDRATE SETTING OF in the geothermal setting of the transgressed Beaufort BEAUFORT SHELF AND UPPER SLOPE shelf during the last 125,000 years we applied worldwide sea level trends as well as terrestrial and marine ground Throughout the past two million years, shelf areas of the temperature estimates in a numerical model (Taylor et al. Arctic Ocean have experienced substantial changes in 2013). This scenario is considered a reasonable portrayal surface temperatures. During periods of low sea level of the geologic history of areas of the Beaufort Shelf that associated with continental glaciations, many areas were were not glaciated. As shown in a simplified view on terrestrially exposed and experienced mean annual air Figure 2, the model is particularly useful in defining the temperatures that were often -20¼C, or colder (Brigham geothermal basis for permafrost occurrence beneath the and Miller 1983). These cold temperatures caused shelf and the conditions where permafrost gas hydrate aggradation of thick ice-bonded permafrost occurrences could be stable. A thick and continuous body of and established substantial thicknesses of subsurface permafrost, with near isothermal conditions is predicted sediments with pressure and temperature conditions from the coast to approximately 60 m water depth (~80 conducive for the formation of permafrost gas hydrate. In km offshore). The permafrost body then warms and thins contrast, sea level rise during interglacial periods resulted seaward (~80-120 km offshore), until it pinches out at in marine transgression that inundated coastal areas with about 90-100 m water depth near the shelf-slope break. warm seawater. Mackay (1972) recognised that because The geothermal conditions for gas hydrate stability heat diffuses slowly it can take thousands of years after are complex with a permafrost gas hydrate interval transgression to fully melt the subsurface ice-bonded beneath the shelf pinching out at the shelf break permafrost. Indeed, the processes associated with the coincident with the permafrost boundary. Geothermal Holocene marine transgression continue today with conditions for stable marine gas hydrate are predicted ongoing warming and thawing of thick permafrost and gas further offshore, but a notable gap in gas hydrate stability hydrate occurrences beneath shelf areas of the Arctic is inferred from the shelf break to approximately 270 m Ocean. Figure 1 illustrates the geothermal, permafrost water depth. and gas hydrate regime of a typical transgressed We have chosen not to model a scenario for permafrost setting and the contrasting marine geothermal glacial ice cover on the shelf as ground surface setting of the upper slope. temperatures near freezing may have persisted during glaciation, limiting the formation of a significant ice bonded permafrost interval. While there is no published evidence to suggest the shelf offshore of the Tuktoyaktuk Peninsula was glaciated, recent interpretation of reflection seismic data suggest that two glacial tills may be present in the Mackenzie Trough area (Batchelor et al. 2012).

Figure 1. Plots showing examples of contrasting geothermal setting of shelf vs slope. Left panel depicts permafrost and permafrost gas hydrate where transgression has submerged terrestrial permafrost (note near-isothermal section within the permafrost interval). Right panel depicts marine geothermal setting typical for the Beaufort Sea and marine gas hydrate Figure 2. Cross-section portrayal of ground temperature regime stability as expected in the deep water setting of the upper slope. and distribution of permafrost and gas hydrate realms in the shelf and upper slope of the Beaufort Sea. The scenario presented is The scientific underpinning for our research is based on geothermal modeling presented in Taylor et al. (2013) based on establishing reasonable geologic hypotheses for along the transect shown on Figure 3. The arrows show the intervals where substantial changes in the base of permafrost the geothermal and porous media setting of the shelf and and gas hydrate stability were modelled. the contrasting setting associated with the upper continental slope. Field programs were designed to 3 POROUS MEDIA CONSIDERATIONS characterize the geology and geologic processes as a basis to evaluate these hypotheses. Geothermal modeling 3.1 Ice-bonded Permafrost is a particularly effective way to assess the evolution of ground temperatures during marine transgression (Taylor While permafrost is formally defined on a geothermal et al. 1996; 2005; 2013). To evaluate the lateral changes basis as sediments that have mean annual temperature

38 below 0oC, the critical consideration in terms of geologic sediment pore ice. A substantial sub-permafrost gas processes and engineering properties relates to the hydrate interval is also well defined. Similar to the presence or absence of ice within the sediment matrix permafrost realm, the most significant changes in the gas (commonly referred to as ice-bonding). In a setting like hydrate regime can be expected at the base of hydrate the Beaufort Shelf, where sediments have experienced a stability and the base of permafrost. Smaller but varied ground surface history, both the formation and potentially significant changes may also occur at the top degradation of ice-bonding must be considered. The of permafrost. As discussed in a recent paper by Weaver geothermal modeling has provided us with a basis to (2013) changes in the gas hydrate regime can induce predict some the sediment processes that may have significant changes in the porous media conditions of occurred. Because permafrost aggradation occurred as a affected sediments. These relate gas hydrate consequence of very cold terrestrial ground temperatures, degradation, which dramatically reduces sediment pore ice formed in nearly all of the permafrost interval strength, and the potential for the release of both gas and substantially increasing the sediment strength, impeding water creating sediment over-pressure. This has the conventional sediment consolidation and reducing potential consequence of dramatically reducing effective sediment permeability. Under some circumstances, often stress conditions or inducing potential for water or gas associated with fine-grained sediments, ice in excess of migration. the pore space was likely formed. The main geotechnical In the upper continental slope where water consideration as permafrost warms relates to the thawing depths are greater than ~270 m, geologic conditions of ground ice which substantially weakens the affected conducive for the formation of conventional marine gas sediments as a consequence of loss of particle to particle hydrate can be anticipated. bonding. Where the ground ice volume is substantive, the release of excess water can also increase sediment pore 4 REGIONAL PERMAFROST AND GAS HYDRATE pressure reducing the effective stress. OCCURRENCE While the geothermal model presented in Figure 2 is quite simplified, the inferred porous media response The legacy of hydrocarbon exploration data collected by to transgression is intriguing. Two environments with the industry in the 1980s and 1990s provided researchers potential for significant sediment response can be with a basis for the appraisal of regional permafrost anticipated. As shown by the arrows on Figure 2 an occurrence beneath the Beaufort Shelf. The efforts of upward shift in the base of permafrost is predicted with Bernard Pelletier, a contemporary of J. Ross Mackay, are consequent thawing of ground ice and liberation of excess particularly noteworthy as his 1987 Marine Sciences Atlas pore waters. Less obvious from the model are the effects of the Beaufort Sea (Pelletier 1987) includes many of warming of the permafrost interval from the coast to the substantial compilation efforts (i.e., O’Connor and Blasco shelf break. Within the range of temperatures inferred (- 1987; Judge et al. 1987; Pullan et al. 1987). Figure 3 2.5 to 0¡C) sediments with different grain size or pore provides a summary of some of this work by overlaying water salinity can be expected to respond quite differently. basedata from determinations of permafrost extent from Whereas fine-grained saline sediments can thaw seismic refraction studies (Pullan et al. 1987) and well log completely within this temperature range, fresh water interpretations (Judge et al. 1987) with subsequent sands may stay ice-bonded. This reveals the very revisions by Smith and Judge (1993) and Hu et al. (2013). important consideration of geology, with warming of the permafrost creating the possibility for differential 5 DISCUSSION permafrost response to warming in sand vs silt vs clay. The complexities of these effects are challenging to model 5.1 Geohazards Related to Exploratory Drilling without detailed subsurface information. As reviewed by Collett and Dallimore (2002), operators 3.2 Gas Hydrates undertaking exploratory drilling in the Arctic have encountered numerous problems advancing well bores Methane gas hydrate, a solid form of natural gas, has through the permafrost and permafrost gas hydrate been inferred in many offshore exploration wells primarily interval. Given our interest in assessing porous media on the basis of well log response and indications of conditions in the outer shelf area, we have reviewed the subsurface gas (Judge 1982; Majorowicz and Osadetz well history reports and well log data for 16 wells drilled in 2001; Smith and Judge 1993; Weaver and Stewart 1982; the Beaufort Sea in the mid-shelf area to the outer shelf Weaver 2013). In order for gas hydrate to occur in (water depths from ~35-65 m). While many of the drilling sediments, there needs to be cold formation problems in the upper kilometre were managed temperatures, moderate pressure, and a suitable geologic adequately with the operational procedures in place at the setting (presence of pore water and natural gas as well as time, the Immiugak N-05 well experienced a blowout at suitable porous media nucleation sites). In shelf areas the ~500 m depth which necessitated evacuation of personnel gas hydrate stability field is controlled by the thermal from the drill rig and eventual abandonment of the well regime established by the transgressed permafrost body site (COGLA 1989). (Fig. 2). Intrapermafrost gas hydrate (Dallimore and Our preliminary appraisal of the well history Collett 1995) may occur from ~250 m depth to the reports suggest that the porous media changes thermally defined base of permafrost. Conditions occur in associated with marine transgression of terrestrial this setting for the co-existence of gas hydrate and permafrost and gas hydrates (section 3 of this paper) can

39 result in unique drilling challenges. Not surprisingly, processes generated deeper in the substrate. There is formation gas, as observed as high mud gas mounting evidence of ground water and free gas concentrations or short duration gas kicks, was commonly migration from depth (e.g. Hughes Clarke et al. 2009; encountered. Most gas occurrences were observed Paull et al. 2011). In the central and outer shelf area, sea beneath the base of the ice-bonded permafrost interval or bed instability relating to degassing of permafrost gas deeper, perhaps associated with the base of the hydrates has been proposed as a factor affecting the subpermafrost gas hydrate interval. However, three formation of isolated pingo-like features (PLF) (Paull et al. observations of formation gas were observed within the 2007). Evidence for regional gas release associated with ice-bonded interval suggesting the possibility of possible sea bed deformation has also been identified at intrapermafrost gas hydrates. Drilling problems were often the shelf edge (Hughes Clarke 2009) and in the upper associated with gas occurrences and included tight hole slope distinct mud features are actively degassing conditions, oversized hole and some incidents of loss of (Paull et al. 2011). Large submarine , some of circulation. The incident review of the Immiugak N-05 well which have headwalls near the shelf-slope boundary have (COGLA 1989) revealed that the blowout occurred also been identified (Saint-Ange et al. 2014). The failure beneath the ice-bonded permafrost interval. While the modes of these features may also be influenced by report alluded to gas hydrates being a possible cause, it is deeper geologic processes generated by thawing of shelf more likely in our opinion that the event was a free gas edge permafrost or gas hydrate dissociation. release possibly associated with perturbed gas hydrates at the base of the gas hydrate stability field (see Fig. 2). 5.3 Ongoing Geohazard Field Investigations

5.2 Unique Geologic Processes The portrayals of possible permafrost and gas hydrate distribution and response to transgression shown in Our field research has focussed largely on the shelf-slope Figures 2 and 3 have been particularly important in our transition area where there are a rather striking number of design of recent field investigations. In 2013 and 2014 we sea bed features that may be controlled, in part, by conducted reflection seismic studies onboard the Korean

40 ice breaker RV Araon (Jin et al. 2015; Riedel et al. 2014) Conference on Gas Hydrates, Yokahama, Japan, focusing on the shelf-slope transitional zones near our 47-52. modeled permafrost transect. These studies allowed us to Dallimore, S.R. and Collett, T.S. 1995. Intrapermafrost verify the determinations of the boundaries of the gas hydrates from a deep core in the Mackenzie permafrost zones in the outer shelf mapped by Pullan et Delta, Northwest Territories, Canada, Geology, 23: al. (1987) based on industry seismic data. We also 527-530. observed the abrupt pinch-out of shelf permafrost at 90- Hu, K., Issler, D.R., Chen, Z., and Brent, T.A. 2013. 100 m water depth near the shelf-slope break as Permafrost investigation by well logs, and seismic portrayed in Figure 2. In three recent cruises onboard the velocity and repeated shallow temperature surveys, CCGS Sir Wilfrid Laurier we have focused on studying Beaufort-Mackenzie Basin, Geological Survey of geologic processes at the sea floor including shelf edge Canada, Open File 6956. landslides, evidence of sea floor instability and upper Hughes-Clarke, J.E., Church, I.W., Blasco, S., Muggah, slope mud volcanism. Experiments have been designed J., and Toodesh, R. 2009. Identification of gas to document gas release and core studies have focused seeps on the Beaufort Continental Margin: New on developing a seismostratigraphic model that considers multibeam and water column imaging off of the the strength and geothermal setting of near surface CCGS Amundsen, Proceedings of the Annual sediments. Scientific Meeting of ArcticNet, Victoria, Canada, 2009. 6 CONCLUSIONS Jin, Y.K., Riedel, M., Hong, J.K., Nam, S.I., Jung, J.Y., Ha, S.Y., Lee, J.Y., Kim, Y.-G., Yoo, J., Kim, H.S., In response to a new phase of exploration in the outer Kim, G., Conway, K., Standen, G., Ulmi, M., and shelf and upper slope areas of the Canadian Beaufort Schreck, M. 2015. Overview of field operation Sea, we have undertaken an appraisal of the regional during a 2013 research expedition to the southern permafrost and gas hydrate setting by examining Beaufort Sea on the RV Araon, Geological Survey of historical data, undertaking new geothermal modeling and Canada, Open File 7754. advancing multidisciplinary field studies. As recognised Judge, A.S. 1982. Natural gas hydrates in Canada, by Mackay (1972), a thick and extensive interval of Proceedings, Fourth Canadian Permafrost transgressed terrestrial permafrost is present beneath the Conference, Ottawa, Canada, 320-328. central shelf, warming seaward and pinching out at 90- Judge, A.S., Pelletier, B.R., and Norquay, I. 1987. 100 m water depths near the shelf-slope break. A Permafrost base and distribution of gas hydrates, complex gas hydrate regime is recognised with possible Marine Science Atlas of the Beaufort Sea: Geology intrapermafrost and subpermafrost gas hydrate beneath and Geophysics, Geological Survey of Canada, the shelf, a zone with no gas hydrates at the shelf-slope Miscellaneous Report 40. transition, and a marine gas hydrate interval where water Majorowicz, J.A. and Osadetz, K.G. 2001. Gas hydrate depths are greater than ~270 m. Review of drilling distribution and volume in Canada, AAPG Bulletin, information and well logs for hydrocarbon exploration 85(7): 1211–1230. wells drilled in the central and outer shelf suggest drilling O’Connor, M.J. and Blasco, S.M. 1987. Subsea problems and at least one blowout may be related to the acoustically defined permafrost (APF types), Marine complex porous media environment in this setting. Our Science Atlas of the Beaufort Sea: Geology and field studies have endeavored to establish linkages Geophysics, Geological Survey of Canada, between active geologic processes such as landslides Miscellaneous Report 40. and sea floor instability and deeper processes related to Paull, C.K., Ussler, W., Dallimore, S.R., Blasco, S.M., changes in the permafrost and gas hydrate realm. Lorenson, T.D., Melling, H., Medioli, B.E., Nixon, F.M., and McLaughlin, F.A. 2007. Origin of pingo-like 7 REFERENCES features on the Beaufort Sea shelf and their possible relationship to decomposing methane gas hydrates, Batchelor, C.L., Dowdeswell, J.A., and Pietras, J.T. 2012. Geophysical Research Letters, 34: L01603. Variable history of Quaternary ice-sheet advance Paull, C.K., Dallimore, S., Hughes-Clarke, J., Blasco, S., across the Beaufort Sea margin, Arctic Ocean, Lundsten, E., Ussler, W. III, Graves, D., Sherman, Geology, 41: 131-134. A., Conway, K., Melling, H., Vagle, S., and Collett, T. Brigham, J.K. and Miller, G.H. 1983. Paleotemperature 2011. Tracking the decomposition of permafrost and estimates of the Alaskan Arctic Coastal Plain during gas hydrate under the shelf and slope of the the last 125,000 years, Proceedings, 4th International Beaufort Sea, Proceedings of the 7th International Conference on Permafrost, Fairbanks, Alaska, 80- Conference on Gas Hydrates, Edinburgh, Scotland. 85. Pelletier, B.R. 1987. Marine Science Atlas of the Beaufort COGLA 1989. Report of Investigation of Events Sea: Geology and Geophysics, Geological Survey of Culminating in a Blowout of Gas at Gulf et al. Canada, Miscellaneous Report 40. Immiugak N-05, Canada Oil and Gas Lands Pullan, S., MacAulay, H.A., Hunter, J.A.M., Good, R.I., Administration, Canada. Gagne, R.M., and Burns, R.A. 1987. Permafrost Collett, T.S. and Dallimore, S.R. 2002. Detailed analysis distribution determined from seismic refraction, of gas hydrate induced drilling and production Marine Science Atlas of the Beaufort Sea: Geology hazards, Proceedings of the 4th International and Geophysics, Geological Survey of Canada,

41 Miscellaneous Report 40. Riedel, M., Hong, J.K., Jin, Y.K., and Kim, H.S. 2014. Refraction seismic velocity analyses from multichannel seismic data acquired during Expedition ARA04C on the IBRV Araon in the Beaufort Sea, Geological Survey of Canada, Open File 7618. Saint-Ange, F., Kuus, P. Blasco, S., Piper, D.J.W., Hughes-Clarke, J., and MacKillop, K. 2014. Multiple failure styles related to shallow gas and fluid venting, upper slope Canadian Beaufort Sea, northern Canada, Marine Geology, 355: 136-149. Smith, S.L. and Judge, A.S. 1993. Gas hydrate database for Canadian Arctic and selected East Coast wells, Geological Survey Canada, Open File 2746. Taylor, A.E., Dallimore, S.R., and Outcalt, S.I. 1996. Late Quaternary history of the Mackenzie-Beaufort region, Arctic Canada, from modelling of permafrost temperatures. 1. The onshore-offshore transition, Canadian Journal of Earth Sciences, 33: 52-61. Taylor, A.E., Dallimore, S.R., Hyndman, R.D., and Wright, F. 2005. Comparing the sensitivity of terrestrial and marine gas hydrates to climate warming at the end of the last ice age, Scientific Results From the Mallik 2002 Gas Hydrate Production Research Well Program, Mackenzie Delta, Northwest Territories, Geological Survey Canada, Bulletin 585. Taylor, A.E., Dallimore, S.R., Hill, P.R., Issler, D.R., Blasco, S., and Wright, F. 2013. Numerical model of the geothermal regime on the Beaufort Shelf, arctic Canada since the Last Interglacial, Journal of Geophysical Research-Earth,118: 2365-2379. Weaver, J.S. and Stewart, J.M. 1982. In situ hydrates under the Beaufort Sea Shelf, Fourth Canadian Permafrost Conference, Ottawa, Canada, 312-319. Weaver, J.S. 2013. Prediction of pore pressures in thawing in situ hydrates on Arctic Continental Shelves, Proceedings of the Twenty-third International Offshore and Polar Engineering Conference, Anchorage, Alaska.

42 Late Wisconsin glaciation of Hadwen and Summer islands, Tuktoyaktuk Coastlands, NWT, Canada

Julian B. Murton, Mark D. Bateman, Richard I. Waller Department of Geography, University of Sussex, Brighton, UK Department of Geography, University of Sheffield, Sheffield, UK School of Physical & Geographical Sciences, Keele University, Keele, Staffordshire, ST5 5BG, UK & Colin A. Whiteman School of the Environment and Technology, University of Brighton, Brighton, UK

ABSTRACT The exact timing of the last major advance of the Laurentide Ice Sheet onto the Beaufort Sea coastlands of western Arctic Canada is unclear but significant to our understanding of landscape change and palaeo-ice stream chronology. Optical stimulated luminescence dating of preglacial and postglacial aeolian sand from Hadwen and Summer islands, in the Tuktoyaktuk Coastlands, indicates that glaciation took place between about 17.5 and 15 ka, and most likely between 16.6 and 15.9 ka, coinciding with Heinrich event 1. At this time the Mackenzie Trough palaeo-ice stream advanced into a cold-climate sandy desert, interrupting aeolian activity.

RÉSUMÉ La date exacte de la dernière grande avancée de l'Inlandsis laurentidien dans la mer de Beaufort dans l’ouest de l'Arctique canadien n'est pas clair, mais elle est significative pour notre compréhension des changements de paysage et de la chronologie des paléo-flux de glace. La datation avec la luminescence stimulée optiquement (OSL) de dépôts préglaciaires et de sable éolien postglaciaire, provenant de Hadwen et de Summer Islands près des côtes de Tuktoyaktuk, montre que les glaciations ont eu lieu entre environ 17,5 et 15 ka, et probablement entre 16,6 et 15,9 ka, coïncidant ainsi avec l’évènement de Heinrich 1. À ce moment, le paléo-flux de glace de la fosse Mackenzie s'est avancé dans un désert de sable de climat froid, interrompant ainsi l'activité éolienne.

1 INTRODUCTION 2 STUDY AREA

During the Wisconsin cold stage, the Laurentide Ice Sheet The main study area is located on the north coast of (LIS) advanced onto the present-day Beaufort Sea Hadwen Island (Figure 1), where the lithostratigraphic coastlands and continental shelf of western Arctic Canada relationships between the preglacial and glacial deposits (Figure 1). In the Tuktoyaktuk Coastlands the last major are clear and well exposed. The Pleistocene LIS glaciation occurred during the Toker Point Stadial lithostratigraphy comprises brown sand of the Kittigazuit (Rampton 1988). Despite growing consensus that this Formation (Fm) beneath pebbly clay (diamicton) of the stadial was Late Wisconsin in age, the exact timing is Toker Point Member (Mb) of the Tuktoyaktuk Fm, unclear. It has previously been assigned variously to >35 (Rampton 1988; Terrain Analysis & Mapping Services ka (Early Wisconsin; Rampton 1988), within the period of Limited 1993). A pebble-boulder lag and / or aeolian dune about 34–20 14C ka BP (Dallimore et al. 1997), between and sand-sheet deposits cap many sections along the 14,000 and 13,000 14C years BP (about 17–15 ka cal BP; coastal bluffs (Bateman and Murton 2006). A second site Mackay and Dallimore 1992), 21–12 ka (Bateman and is at Crumbling Point, Summer Island, where massive ice Murton 2006) and 22–16 ka (Murton et al. 2007). interpreted as buried basal ice of the LIS is penetrated by Bateman and Murton (2006, figure 9) plotted optical postglacial sand wedges and underlain by a glacitectonite stimulated luminescence (OSL) ages for sands interpreted and Kittigazuit Fm sand (Murton et al. 2005, table 3). as preglacial and postglacial in the Summer Island and northern Liverpool Bay areas (Figure 1), near the margins 3 METHODS of the Mackenzie Trough and Anderson palaeo-ice streams, respectively (Margold et al. 2015). The aim of 3.1 Logging the present study is to refine this plot and date more precisely the advance of the Mackenzie Trough palaeo- Stratigraphic sections were examined and logged ice stream onto Hadwen and Summer islands, where we sedimentologically in several localities along the coastal now have the best available stratigraphic understanding bluffs of northern Hadwen Island to refine these and sequence of OSL ages for pre- and postglacial observations and interpret the origin of the sediments and sediments. Here we report observations on the sedimentary structures. Section 08-14, located at 69¡ 35' lithostratigraphy and sedimentology and evaluate new and 57.5"N; 134¡ 09' 02.6" W, was selected for dating of the previously published OSL ages. Toker Point Stadial on the basis of cross-cutting relationships between the preglacial Kittigazuit Fm, Toker Point Mb till and a postglacial aeolian sand wedge (Figure

43 2). A clast fabric analysis of the till was carried out using poor growth of OSL with laboratory dose or if recycling methods reported in Murton et al. (2004, 2005). values fell beyond 10% of unity. Once these and outliers were excluded, if replicate De values for samples had a 3.2 OSL Dating low overdispersion and were normally distributed they were deemed to have been fully reset prior to burial and Samples for OSL were collected from freshly exposed final De values were derived using a weighted mean with vertical sections in opaque PVC tubes and prepared in associated standard errors. For samples with OD >25% order to clean and extract quartz at the Sheffield and skewing of De replicates partial resetting was Luminescence laboratory as per Bateman and Catt assumed and final De values were derived using the finite (1996). OSL for all samples was measured using a Ris¿ mixture model (Roberts et al. 2000). Dose rates were reader with optical stimulation provided by blue/green calculated from in situ gamma spectrometry LEDs and luminescence detection was through a Hoya U- measurements with palaeomoisture based on those 340 filter. The single aliquot regeneration (SAR) measured at present and a calculated cosmic dose procedure was used to determine the palaeodose (De) contribution as per Prescott and Hutton (1994). The final (Murray and Wintle 2000, 2003). Between 14 and 24 ages are in calender years before 2005 and 2008, when replicate De measurements per sample were carried out. samples were collected. Data from aliquots were rejected if an aliquot exhibited

Figure 1. Location map of the Beaufort Sea coastlands and continental shelf. Ice flow lines and glacial limits after Rampton (1982, 1988): Buckland glaciation of the Yukon Coastal Plain, Toker Point Stadial glaciation of the Tuktoyaktuk Coastlands, and Franklin Bay Stadial glaciation of Amundsen Gulf. An alternative glacial limit for the Toker Point Stadial crosses the eastern Beaufort Sea Shelf north of the Tuktoyaktuk Pensinsula, indicating uncertainty about the topographic profile of the ice sheet here (Rampton, 1988). Inset map shows study area and field sites. Modified from Murton (2009).

4 RESULTS (Figure 2; Table 1). The diamicton is massive and matrix- supported. Its lower contact sharply overlies a layer of 4.1 Sedimentology and Stratigraphy sand 0.15–0.2 m thick that is massive to faintly stratified, and contains lenses aligned more or less parallel to the The Toker Point Mb diamicton in section 08-14 overlies layer. A clast fabric measured from the diamicton provided the Kittigazuit Fm and is cross-cut by a sand wedge a value of fabric isotropy (I = S3/S1) of 0.050, a value of

44 fabric elongation (E = 1 – (S2/S1)) of 0.931 and a mean oversteepened and folded tops, and some are displaced lineation vector of 042¡/272¡. The high value of fabric by horizontal shears directly beneath the sand layer. The elongation and low value of fabric isotropy indicate a diamicton is cross-cut by a sand wedge about 3 m high strong preferred orientation of elongate clasts. The lower and 1 m wide (Figure 2). contact of the sand layer is sharp, truncating foresets in the underlying Kittigazuit Fm. Some foresets have

Figure 2. Kittigazuit Formation beneath Toker Point Till Member, section 08-14, northwest Hadwen Island. (A) Schematic sketch section showing stratigraphy and OSL ages. (B) Photograph of section. (C) Close-up photograph of oversteepened (towards left) and sheared foresets in Kittigazuit Fm beneath Toker Point Till.

45 Table 1. Lithostratigraphy and sedimentology of section 08-14, northwest Hadwen Island

Unit Description Interpretation (thickness) Sand wedge Fine sand, vertically to subvertically laminated, laminae 1 to Aeolian sand infilling thermal several millimetres thick; wedge about 1 m wide, 3 m high; flat contraction cracks bottom of wedge at same depth as base of adjacent diamicton Postdates till because wedge cross-cuts till Toker Point Mb 5Y 3/1 (very dark grey, moist), massive, sandy to silty clay matrix, Toker Point till (1–4 m) texturally variable; matrix-supported diamicton; pebbles and cobbles up to 8 cm in maximum dimension, rounded to Ice flow towards the northeast subangular, dispersed (not abundant); lower contact sharp and undulating; diamicton crops out more or less continuously for few hundred metres along bluff; 2 subfacies: (A) Clayey facies: dark grey, few centimetres thick along base and forming lenses near base; (B) Sandy facies: grey, dominant facies; pebblier towards surface; Clast fabric measured nearby is shown in Figure 3 Sand layer Grey, massive to faintly stratified sand; faint horizontal to Glacitectonically sheared and (0.15–0.2 m) subhorizontal, gently undulating lenses; sharp lower contact mixed sediments Kittigazuit Fm Brown, fine sand and grey silty sand; well stratified, strata few Preglacial aeolian dune sand (≥ 7 m) millimetres to few centimetres thick, foresets dip at 24¡ towards 070¡; upper 0.1–0.4 m of some foresets are oversteepened or Glacitectonic overturning or sheared, with shears horizontal to subhorizontal and parallel to the shearing of the tops of some lower contact of overlying sand layer foresets

Table 2. OSL-related data for samples from section 08-14, northwest Hadwen Island

Stratigraphic unit Lab. code Depth from Palaeodose, Dose Rate Age (ka) surface (m) De (Gy) (Gy/ka) Sand wedge penetrating till Shfd08149 2.4 19.41 ± 0.45 1.738 ± 0.090 11.8 ± 0.66 Kittigazuit Fm 0.4 m below till Shfd08148 4.45 27.62 ± 0.66 a 1.652 ± 0.085 16.6 ± 0.9 a De based on finite mixture modelling.

4.2 OSL Dating

The Kittigazuit Fm provided an OSL age of 16.6 ± 0.9 ka at a depth of 0.4 m below the Toker Point till in section 08- 14 (Figure 2; Table 2). A sand wedge that cross-cut the till provided an OSL age of 11.8 ± 0.66 ka.

5 DISCUSSION

5.1 Glacial Processes

The strong preferred orientation of elongate clasts within the Toker Point till and their mean lineation vector of 042/272¡ are consistent ice flow towards the northeast. The ice movement glacitectonically deformed some of the uppermost Kittigazuit Fm sand. Deformation oversteepened and / or sheared foresets, and mixed sand and diamicton to form a sand layer at the contact of the diamicton and sand (Figure 2C). Collectively these features represent a glacitectonite (sensu Benn and Evans 1996). The horizontal bands of clayey facies and sandy facies in the basal part of the overlying diamicton

suggest that this part of Toker Point ‘till’ also represents a Figure 3. Schmidt equal-area stereonet of clast fabric from glacitectonite that has been incompletely mixed. Toker Point till, near section 08-14, northwest Hadwen

Island. The three principal eigenvalues are S , S and S . 1 2 3

46 5.2 Age of the Toker Point Stadial Glaciation

The age of the Toker Point Stadial glaciation on Hadwen Island is thought to be between about 17.5 and 15 ka based on OSL ages below till or below an erosion surface attributed to glaciation (Tables 2 and 3; Figure 4). Our best estimate for the maximum age of the glaciation is the OSL age of 16.57 ± 0.9 ka from Kittigazuit Fm sand beneath Toker Point till at section 08-14. In addition, OSL ages of 19.9 ± 1.1 ka and 25.6 ± 1.3 ka were obtained by Bateman and Murton (2006, figure 2) from Kittigazuit Fm sand beneath a prominent erosion surface that we Figure 4. OSL ages on aeolian sand deposited before and attribute to glacial or glaciofluvial erosion during the Toker after the Toker Point Stadial on Hadwen and Summer Point Stadial; this erosion surface is located at section islands. Mean age of the three oldest postglacial ages is 3.1, about 1.5 km east of section 08-14 (Figure 1 inset 15.93 ± 0.31 ka with uncertainty excluding original age map). The minimum age for the glaciation is based on an uncertainty (indicated by dark grey bar) or 15.93 ± 0.96 OSL age of 16.12 ± 0.91 ka (Shfd02044) from Kittigazuit ka including original uncertainty (light grey bar). Fm sand 0.15 m above this erosion surface at section 3.1d. Additionally, a fading-corrected optical age of 16.1 Glaciation of Hadwen Island is attributed to advance of 0.2 ka (SFU-0-160) on 180–250 m potassium feldspar the Mackenzie Trough palaeo-ice stream. Brief glaciation grains from a postglacial sand wedge at Crumbling Point, between 17.5 and 15 ka accounts for the permafrost Summer Island (reported in Murton et al. 2007), provides thickness of about 700 m reported by Pelletier (1987) in a similar minimum age for onset of deglaciation (Table 3). the study area. Heat-flow calculations suggest that the 14C ages of 12,500110 years BP (GSC–3302) and time required to grow 500–600 m of permafrost probably 12,900170 years BP (GSC–1321) on organic material exceeds 50,000 years (Mackay 1979), consistent with from the basal part of lacustrine sediments on the Hadwen and Summer islands having been ice-free and Tuktoyaktuk Peninsula indicate that ice-free conditions subaerially exposed for most of the Wisconsin (Taylor et commenced by about 15,500–14,300 cal years BP al. 1996a, 1996b). (Ritchie 1984). An age between 17.5 and 15 ka for the Toker Point Stadial glaciation supports Mackay and 6 CONCLUSIONS Dallimore’s (1992) suggestion that ice advanced to the Tuktoyaktuk area between about 17 and 15 cal. ka. Glaciation of Hadwen Island during the Toker Point Stadial occurred sometime between about 17.5 and 15 ka Table 3. Previously published OSL ages on preglacial and (incorporating errors), most likely between 16.6 and 15.9 postglacial aeolian sand on Hadwen and Summer islands ka (Figure 4). This glacial advance coincided with Heinrich event 1. For most of the LGM, the study area was ice- Stratigraphic Lab. Age Significance free, allowing thick permafrost to form. The Mackenzie unit code (ka) Trough palaeo-ice stream advanced into a cold-climate Sand wedge in SFU-0- 16.1 ± Postglacial desert characterized by dunes and sand sheets of the basal ice of LIS, 160 a 0.2 aeolian sand Kittigazuit Fm, located in the distant rainshadow of Crumbling Point, mountains and the Cordilleran Ice Sheet to the southwest. Summer Island Aeolian sedimentation resumed in some till-free areas Kittigazuit Fm Shfd 15.57 ± after the advance and persisted until it was abruptly above major 02046 b 0.91 ended at about 12.9 ka by erosion attributed to an erosion surface, Shfd 16.12 ± outburst flood (Murton et al. 2010). section 3.1, 02044 b 0.91 Hadwen Is. ACKNOWLEDGEMENTS Kittigaz. Fm Shfd 19.9 ± Preglacial below major 02045 b 1.1 aeolian sand The Aurora Research Institute (Inuvik) provided logistical eros. surf., sect. Shfd 25.6 ± support. Field assistance was provided by Della Murton, 3.1, Hadwen Is. 02043 b 1.3 Fred Wolki and Enoch Pokiak. a D.J. Huntley in Murton et al. (2007) b Bateman and Murton (2006) REFERENCES

Bateman, M.D. and Murton, J.B. 2006. Late Pleistocene glacial and periglacial aeolian activity in the Tuktoyaktuk Coastlands, NWT, Canada. Quaternary Science Reviews 25: 2552–2568. Benn, D.I. and Evans, D.J.A. 1996. The interpretation and classification of subglacially deformed materials. Quaternary Science Reviews 15: 23–52.

47 Mackay, J.R. 1979. Pingos of the Tuktoyaktuk Peninsula Rampton, V.N. 1988. Quaternary Geology of the area, Northwest Territories. Géographie physique et Tuktoyaktuk Coastlands, Northwest Territories. Quaternaire 23: 3–61. Memoir 423, Geological Survey of Canada, Ottawa. Mackay, J.R. and Dallimore, S.R. 1992. Massive ice of the Ritchie, J.C. 1984. Past and Present Vegetation of the Far Tuktoyaktuk area, western Arctic coast, Canada. Northwest of Canada. University of Toronto Press, Canadian Journal of Earth Sciences 29: 1235–1249. Toronto. Margold, M., Stokes, C.R., Clark, C.D. and Kleman, J. Roberts, R.G., Galbraith, R.F., Yoshida, H., Laslett, G.M. 2015. Ice streams in the Laurentide Ice Sheet: a new and Olley, J.M. 2000. Distinguishing dose populations mapping inventory. Journal of Maps 11: 380–395. in sediment mixtures: a test of single-grain optical DOI.org/10.1080/17445647.2014.912036 dating procedures using mixtures of laboratory-dosed Murray, A.S. and Wintle, A.G. 2000. Luminescence dating quartz. Radiation Measurements 32: 459–465. of quartz using an improved single-aliquot Taylor, A.E., Dallimore, S.R. and Judge, A.S. 1996a. Late regenerative-dose protocol. Radiation Measurements Quaternary history of the Mackenzie-Beaufort region, 32: 57–73. Arctic Canada, from modelling of permafrost Murray, A.S. and Wintle, A.G. 2003. The single aliquot temperatures. 1. The onshore-offshore transition. regenerative dose protocol: potential for improvements Canadian Journal of Earth Sciences 33: 52–61. in reliability. Radiation Measurements 37: 377–381. Taylor, A.E., Dallimore, S.R. and Judge, A.S. 1996b. Late Murton, J.B. 2009. Stratigraphy and paleoenvironments of Quaternary history of the Mackenzie-Beaufort region, Richards Island and the eastern Beaufort Continental Arctic Canada, from modelling of permafrost Shelf during the last glacial-interglacial cycle. temperatures. 2. The Mackenzie Delta - Tuktoyaktuk Permafrost and Periglacial Processes 20: 107–125. Coastlands. Canadian Journal of Earth Sciences 33: Murton, J.B., Bateman, M.D., Dallimore, S.R., Teller, J.T. 62–71. and Yang, Z. 2010. Identification of Younger Dryas Terrain Analysis & Mapping Services Limited. 1993. outburst flood pathway from Lake Agassiz to the Arctic Potential Granular Resources and their Geological Ocean. Nature 464: 740–743. Constraints, Northern Richards Island. Report for the Murton, J.B., Frechen, M. and Maddy, D. 2007. Department of Indian and Northern Affairs. Luminescence dating of Mid- to Late Wisconsinan aeolian sand as a constraint on the last advance of the Laurentide Ice Sheet across the Tuktoyaktuk Coastlands, western Arctic Canada. Canadian Journal of Earth Sciences 44: 857–869. Murton, J.B., French, H.M. and Lamothe, M. 1997. Late Wisconsinan erosion and aeolian deposition, Summer and Hadwen Islands, Mackenzie Delta area, western Canadian Arctic: optical dating and implications for glacial chronology. Canadian Journal of Earth Sciences 34: 190–199. Murton, J.B., Whiteman, C.A., Waller, R.I., Pollard, W.H., Clark, I.D. and Dallimore, S.R. 2005. Basal ice facies and supraglacial melt-out till of the Laurentide Ice Sheet, Tuktoyaktuk Coastlands, western Arctic Canada. Quaternary Science Reviews 24: 681–708. Murton, J.B., Waller, R.I., Hart, J.K., Whiteman, C.A., Pollard, W.H. and Clark, I.D. 2004. Stratigraphy and glaciotectonic structures of permafrost deformed beneath the northwest margin of the Laurentide Ice Sheet, Tuktoyaktuk Coastlands, Canada. Journal of Glaciology 49 (No. 170): 399–412. Pelletier, B.R. (ed.). 1987. Marine Science Atlas of the Beaufort Sea: Geology and Geophysics. Miscellaneous Report 40. Energy, Mines and Resources Canada, Geological Survey of Canada, Ottawa, Canada. Prescott, J.R. and Hutton, J.T. 1994. Cosmic ray contributions to dose rates for luminescence and ESR dating: large depths and long-term time variations. Radiation Measurements 2/3: 497–500. Rampton, V.N. 1982. Quaternary Geology of the Yukon Coastal Plain. Bulletin 317, Geological Survey of Canada, Ottawa.

48 Holocene lake-level recession, permafrost aggradation and lithalsa formation in the Yellowknife area, Great Slave Lowland

S.A. Wolfe and P.D. Morse Geological Survey of Canada, Natural Resources Canada, Ottawa, Ontario, Canada

ABSTRACT

The Great Slave Lowland occupies the north shore of Great Slave Lake. After glaciation, it was inundated by Glacial Lake McConnell and ancestral Great Slave Lake. Holocene lake-level recession around Yellowknife is determined from accelerator mass spectrometer ages of and detrital organics. In the last 8000 years, recession occurred at about 5 mm/year, and permafrost is youngest near the modern shoreline and older at higher elevations. Silty-clay sediments are abundant, and (ice-rich permafrost mounds within mineral soils) occurring within 40 m above the present lake level are less than 6000 years old. They are common on Yellowknife River alluvium deposited within the last 3000 years. Lithalsas on this surface are assumed to have developed as permafrost aggraded into saturated sediments, and ground ice has formed within the last 250 years.

RÉSUMÉ

L’écorégion Ç Great Slave Lowland È se situe sur la rive nord du Grand lac des Esclaves. Suite à la déglaciation, elle a été inondée par le lac glaciaire McConnell, ainsi qu’un précurseur du Grand lac des Esclaves. La récession holocène du niveau du lac près de Yellowknife a été déterminée en datant des tourbes et des débris organiques par spectrométrie de masse par accélérateur (AMS). Pendant les 8000 dernières années, le niveau du lac a baissé d’environ 5 mm par année, le pergélisol le plus jeune se retrouve donc près du littoral moderne, alors que le plus âgé se retrouve sur les terres plus élevées. Les argiles limoneuses abondent et l’on retrouve des lithalses (buttes cryogènes riches en glace) plus jeunes que 6000 ans jusqu’à 40 m au-dessus du niveau du lac moderne. Celles-ci se retrouvent souvent sur les sédiments alluviaux de la rivière Yellowknife qui furent déposés durant les 3000 dernières années. Les lithalses sur ces surfaces se sont probablement développées avec l’expansion du pergélisol dans des dépôts saturés, la glace de sol s’étant formée durant les derniers 250 ans.

1 INTRODUCTION the origin of the ice, and the age and depositional environment of the enclosing sediment. The Great Slave Lowland is a low-relief, Precambrian granitic bedrock plain extending about 250 km along the north shore of Great Slave Lake (Fig. 1), and includes 2 BACKGROUND AND STUDY AREA Yellowknife, the capital of the Northwest Territories, which resides near the present-day shoreline at 62.4¡N latitude The Great Slave Lowland is bordered to the south by and 114.4¡W longitude. Following retreat of the Great Slave Lake, and to the north by the Great Slave Laurentide Ice Sheet, inundation by Glacial Lake Upland of higher relief. As with much of northern Canada, McConnell and ancestral Great Slave Lake resulted in the region was glaciated during latest, Wisconsinan, extensive deposition of silts and clays throughout the glaciation. With the retreat of the Laurentide Ice Sheet, Lowland. In a cold-continental setting, permafrost about 13 000 cal BP, it was inundated by Glacial Lake aggradation into these sediments has given rise to McConnell that formed along the retreating glacial margin widespread discontinuous permafrost with extensive ice- between about 13 000 and 9500 cal BP (Lemmen et al. rich permafrost mounds (i.e. lithalsas) across the region 1994; Smith 1994). At its maximum extent at about 10 700 (Wolfe et al. 2014). cal BP, the lake included the combined basins of Great The primary purpose of this paper is to establish a Bear, Great Slave and Athabasca lakes (Fig. 1). Lake preliminary estimate of the rate of Holocene lake-level McConnell was largely the result of glacio-isostatic recession in the Yellowknife area, to provide a baseline loading (Lemmen et al. 1994). Evidence of the lake exists estimate of regional terrestrial emergence and permafrost in the form of raised deltas, strandlines, beaches and aggradation. A secondary purpose is to examine the wave-washed eskers and drumlins. In the Great Slave distribution of lithalsas in the Yellowknife area in relation basin, maximum lake elevation was between 320 and 350 to elevation and surficial sediments in order to estimate m asl along the East Arm of Great Slave Lake (Smith their age and environmental settings of formation. 1994; Kerr et al. 2014), compared to only about 180 m asl Permafrost within a lithalsa exposed by thaw slumping at Fort Smith (Vanderburgh and Smith 1988), the into the Yellowknife River permits direct assessments of elevation difference due to differential isostatic uplift.

49 locally widespread with nearly 1800 identified in the region west of Yellowknife (Wolfe et al. 2014). Lithalsas in the Yellowknife area are typically circular to elongate, 25 to 50 m wide and up to 300 m long and 8 m high. They are typically vegetated by birch or birch- spruce forests and commonly occur along the margins of small ponds, which in some cases represent thermokarst ponds related to thawing of former lithalsa terrain (Fig. 2A). The raised topography of lithalsas, combined with the typical deciduous (birch) forest cover makes these features readily recognizable in the field and on stereo aerial photographs, which were used by Wolfe and Kerr (2014) to map the lithalsa distribution in the area. Terrestrial emergence accompanied lake-level recession throughout the Great Slave Lake basin. Given the low gradient of the Lowland region, in some areas a few metres of elevation adjustment resulted in considerable land exposure over a short time period (Fig. 2B). Permafrost aggradation followed terrestrial emergence, though the regional climate has varied during the Holocene. Prior to 6000 cal BP, exposed terrain north of the Lowland was dominated by tundra, shrub tundra and spruce forest tundra (Huang et al. 2004), implying

conditions colder than present. Between 6000 and 3500 Fig. 1. Great Slave Lowland, Northwest Territories with cal BP treeline moved northward of its present position in approximate margin of the Laurentide Ice Sheet and response to climate warming (Moser and MacDonald extent of Glacial Lake McConnell at about 10 700 cal BP 1990; MacDonald et al. 1993), and it is uncertain if permafrost was locally sustained within the Lowland at During recession, the lake first separated from the Great that time. The regional climate cooled, but was variable, Bear basin, and remained in existence until about 9500 from about 3000 Cal BP to present (Huang et al. 2004). cal BP, when the remaining two basins separated Local mean annual air temperatures have recently (Lemmen et al. 1994). The resulting ancestral Great Slave warmed at a rate of about 0.3¡C per decade since the Lake continued to decline due to differential uplift, towards 1940s (Riseborough et al. 2013) with an accelerated trend its present elevation of 156 m asl, constrained by the of about 0.6¡C per decade since 1970 (Hoeve et al. Mackenzie River outlet at Fort Providence (Vanderburgh 2004), consistent with a pan-Arctic warming trend and Smith 1988). beginning AD 1966 (IPCC 2013). Glacial Lake McConnell and ancestral Great Slave Smith (1994) reconstructed the rate of lake-level Lake left prominent signatures on the Lowland landscape, lowering of Glacial Lake McConnell and Great Slave Lake with silt and clay covering nearly 70% of the land area from radiocarbon-dated buried wood in raised deltaic (Wolfe et al. 2014). The Lowland area, being submerged sediments on the southern side of Great Slave Lake. to depths of 100 to 150 m, was a deepwater basin for fine- Based primarily on woody debris collected along former grained glaciolacustrine sediment deposition. During delta-front positions of the Slave River delta (Fig. 1), recession of ancestral Great Slave Lake, waves and rivers Vanderburgh and Smith (1988) estimated the Holocene re-worked and re-deposited sediments in nearshore lake-level recession of Great Slave Lake to be about 2 environments (Fig. 2A). Consequently, glaciolacustrine mm per year from 7000 14C yr BP (about 8000 cal BP) to and lacustrine deposits are broadly distributed across the present. Although long-term uplift rates for the Yellowknife region. area are unknown, present-day uplift is reported to be on The area presently experiences a continental subarctic the order of 6.3 +/- 0.06 mm per year based on a time climate, with an annual mean air temperature of -4.1¡C, series of 15 years (Mazzotti et al. 2011). cold winters (-25.6¡C January mean), warm summers Given difference in maximum elevation of Glacial Lake (17.0¡C July mean), and an average of 281 mm of McConnell between the northern and southern sides of precipitation with 40% falling as snow (Environment the basin, we hypothesize that the rate of Holocene lake- Canada, 2015). The area occupies poorly-drained low level recession for the north shore of Great Slave Lake relief terrain characterized by numerous water bodies and the Yellowknife area, exceeded that for the south separated by fens, peatlands, mixed woodlands, white shore of the lake derived from the Slave River Delta by birch (Betula papyrifera) and black spruce (Picea Smith (1994). Therefore, this study derives a preliminary mariana) forests and bedrock outcrops (Fig. 2A and B). estimate of the rate of Holocene lake-level recession for Permafrost is extensively discontinuous (Heginbottom et the Yellowknife area. The implications of lake-level al. 1995) and lithalsas, which are permafrost mounds recession are assessed in the context of local permafrost formed by ice segregation in fine-grained sediment, are landscape evolution by reconstructing shorelines and examining lithalsa distribution and cryostratigraphy in the

50

Fig. 2. Lithalsa and ground ice examples in the Great Slave Lowland. A) Yellowknife River extending southward into Back Bay on Great Slave Lake, with alluvial plain and lithalsas in the foreground; B) Lithalsa terrain at Site 5 along NWT Highway 3 near Boundary Creek at an elevation of about 10 m above the present level of Great Slave Lake; note Great Slave Lake at a distance of about 7 km in background; C) Ground ice slump at Site 6 along the Yellowknife River; D) Slump headwall at Site 6 exposing alluvial and lacustrine sediments with segregated ground ice at depth.

Yellowknife area in relation to elevation and surficial Cores were obtained from peatlands in the Yellowknife sediments in order to estimate their ages and area, ranging in elevation from 205 m asl near the margin environmental settings of formation. with Great Slave Upland to 165 m asl within the Lowland. Elevations were derived from multiple sources, including topographic base maps, GPS and Lidar data (where 3 METHODS available), and were conservatively estimated the nearest 5 metres (± 2 m). Elevations were derived for both the 3.1 Peatland cores and lake-level recession sample sites and the nearby surface water levels. Local surface water levels were used as the estimates of land Accelerator mass spectrometer (AMS) dating of peatland surface elevation at the time of terrestrial emergence, as sites represents a means to establishing the timing of these discounted the effects of heave due to permafrost lake-level recession in the Great Slave Lowland, where aggradation and ground ice formation and accumulation terrestrial organic accumulation occurred following of peat above the actual mineral surface, which is variable emergence. Peatlands are relatively rare in the Lowland and can locally exceed 3 m. representing less than 2% of the total surface terrain In most instances, coring was undertaken with a cover (Wolfe and Kerr 2014). Unlike permafrost areas CRREL corer using a 7.6 cm diameter core barrel, further south, peat plateaus are uncommon in the Great although at one site (Site 4) a peat auger was used to drill Slave Lowland, as the surrounding terrain underlain by multiple boreholes to confirm stratigraphic contacts. Holes unconsolidated sediments typically also contains were cored/augered through the active layer and permafrost (Morse et al. in press). Peatland sites selected underlying frozen peat, and into the underlying sediment. for dating were typically near, but above, ponds and did In all cases, in situ organic samples were collected from not appear to have undergone alterations in hydrology the base of the peat unit and above the mineral sediment that, especially, would have affected peat initiation. for AMS dating by Beta Analytic and AMS ages were converted to calendar years using Calib©7.0.4 (2014). In

51 situ basal ages are considered to represent the minimum- 4 RESULTS AND DISCUSSION limiting ages of peat accumulation, and thus for terrestrial emergence. In instances where two cores were obtained 4.1 Peatland cores and Holocene lake-level recession from the same site (e.g. Sites 3 and 4), the older of the two AMS ages was used to represent the limiting age. At Table 1 summarizes AMS ages, locations, depths, one site, detrital organics (wood charcoal) collected from material and depositional environments of organic within the underlying subaqueous alluvial sediments were samples. Figure 3A illustrates the near-surface also dated to bracket emergence. These detrital ages are stratigraphy and AMS ages derived from seven peatland considered to represent maximum-limiting ages of cores at five sites of varying elevation in the Lowland. subaqueous alluvial deposition in relation to former Great Core depths vary from about 1.5 m at Site 2 to 4.5 m at 5 Slave Lake levels. and peat thicknesses range from 93 to 295 cm. Calibrated AMS ages range from about 8000 to 1100 14C years BP. 3.2 Shoreline reconstruction and landscape evolution A general relation is apparent between basal peat AMS ages and elevations across the area (Fig. 3B). Relations between lake-level history and landscape Sediments immediately underlying peat at these sites evolution were examined using topographic Canadian are typically clayey-silts, with layers of sandy-silt and silty- Digital Elevation Data (CDED) available from GeoGratis clay at greater depths (Fig. 3A). Note that at Site 3, (http://geogratis.cgdi.gc.ca/), combined with published bedrock was encountered beneath a relatively thin layer surficial sediment and lithalsa distributions for the of sediment underlying peat. These fine-grained Yellowknife area (Wolfe and Kerr 2014). Topographic data sediments most likely represent alluvial or lacustrine were used to reconstruct raised shoreline levels. Timing of sediments pre-dating terrestrial emergence. At Site 5, sediment deposition or reworking, including beach sands near Boundary Creek 30 km west of Yellowknife, detrital and alluvial silts, as well as the maximum-limiting ages of charcoal contained within clayey-silt sediments date to lithalsas, were estimated from their elevations and about 1710 (1810 to 1610) and 1460 (1533-1395) Cal BP, discussed in the context of reconstructed lake-level respectively, with an apparent age reversal relative to recession rates. depth of deposition (Fig. 3A). These ages indicate A slump exposing near-surface sediments and ground deposition within a subaqueous environment, possibly ice, situated on an island in the Yellowknife River about 1 either a shallow bay of ancestral Great Slave Lake or a km (Fig. 2C) north of Back Bay, was examined to former alluvial plain of Boundary Creek. In either case, the determine sediment and ground ice origin and age. Debris charcoal was most likely derived from a burned, forested along a portion of the headwall was cleared to expose at surface at a higher elevation. This is significant, as it 2.2 m high by 6 m wide section (Fig. 2D). This was indicates that terrestrial emergence here must have measured, sketched and photo-documented, noting occurred between about 1460 (the age of the younger stratigraphic contacts and ice-textures. Sediment samples detrital organics) and 1170 (the age of the basal peat) Cal were collected for grain size analysis, Atterberg limits, and BP, at a site presently about 10 m above, and more than moisture content performed by the Geological Survey of 6 km inland of, Great Slave Lake (Fig. 2B). Thus, given Canada. Ice samples were collected from the lower the low elevation gradient, considerable terrestrial portion of the section, where visible ice content exceeded emergence occurred over a relatively short time period. 70%, and waters were submitted to the NWT Lab A preliminary Holocene lake-level reconstruction may for isotopic analysis. In situ roots within unfrozen and be derived for ancestral Great Slave in the Yellowknife frozen portions of the exposure were collected for AMS region from the AMS ages obtained beneath the dating by Beta Analytic. peatlands in the region. Figure 3B illustrates the derived lake-level recession rate for the Yellowknife area using calibrated radiocarbon ages. A rate of approximately

Table 1. AMS dates used to infer late Holocene lake-level recession and terrestrial emergence in the Great Slave Lowland. Site Age 2σ range Lat. Long. Elev.1 Depth Material Depositional (14C BP) (Cal BP) (¡N) (¡W) (m asl) (m) environment 1 7130 ± 40 8020-7865 62.5537 114.0172 210 / 205 2.95 Peat terrestrial 2 6730 ± 30 7670-7515 62.4796 114.4013 210 / 205 1.30 Wood terrestrial 3 6580 ± 40 7565-7430 62.4564 114.5311 195 / 195 2.35 Wood terrestrial 3 5880 ± 30 6780-6640 62.4564 114.5311 195 / 195 2.45 Peat terrestrial 4 1870 ± 30 1880-1725 62.5061 114.2759 185 / 175 1.55 Wood terrestrial 4 1590 ± 30 1545-1410 62.5061 114.2759 185 / 175 1.55 Peat terrestrial 5 1240 ± 30 1270-1075 62.5280 114.9606 170 / 165 0.93 basal peat terrestrial 5 1770 ± 30 1810-1610 62.5280 114.9606 170 / 165 2.20 burned wood alluvial 5 1570 ± 30 1533-1395 62.5280 114.9606 170 / 165 2.55 burned wood alluvial 6 modern na 62.5275 114.3128 165 / 156 0.85 in situ root terrestrial 6 80 ± 30 260-25 62.5275 114.3128 165 / 156 1.80 in situ root alluvial 1 Elevations include collection elevation and local surface water (lake/pond/or stream) elevation. Local water level elevations, rather than collection elevations, are used in lake-level reconstructions.

52

Fig. 3. Elevation-age relations in the Great Slave Lowland. A) Near-surface stratigraphy and AMS ages beneath terrestrial peatlands at varying elevations. B) Calibrated AMS ages in relation to elevation. Dashed line represents approximation of lake-level recession and terrestrial emergence.

5 mm/year is inferred. This is roughly two and half times that of the southern side of Great Slave Lake (Smith and Fig. 4. Shoreline reconstructions based on elevation in the Vanderburgh 1988). This may be attributed to either a true Yellowknife area, with surficial sediment and lithalsa difference in the rates of Holocene post-glacial isostatic distributions. A) Present-day shoreline, also showing + 20, adjustment as hypothesized, or to differences in + 40 and >50 m elevations. B) Shoreline configuration at methodology (i.e. in situ terrestrial peatlands dated in this + 40 m, and showing > 50 m elevation relative to present study versus allochthonous buried logs used on the Slave day lake level. TR = Tartan Rapids River Delta). Although Smith (1994) also estimated the rate prior to 8000 Cal BP (18 mm/year) no estimate is Back Bay on Great Slave Lake where alluvial plains also made for Yellowknife, though it may be significantly occur at the north end of the bay (Fig. 4A). In contrast, greater. with lake levels at + 40 m relative to present-day elevation this was part of the larger Yellowknife Bay (Fig. 4B), and 4.2 Shoreline reconstruction and landscape evolution the river entered Great Slave Lake several kilometres tothe north (not shown in figure). At that time, lake levels Figure 4 depicts the present-day (Fig. 4A), and + 40 m were locally near the dividing limit between the Great reconstructed (Fig. 4B), shorelines in the Yellowknife Slave Upland and Lowland, and the surface of sandy area. At present, the Yellowknife River flows along a 6 km outwash deposits in the Yellowknife area were reworked channel from Tartan Rapids (TR in Fig. 4A), with broad as beaches (Fig. 4B). With lake-level recession, the alluvial plains and two main islands along the river, into Yellowknife River became confined to the present-day

53 channel, with alluvium deposited in shallows along the margins of the river. Based on the local relief, this occurred when the lake was less than + 20 m relative to present-day level (Fig. 4A). Utilizing the Holocene lake-level recession of about 5 mm per year over the last 8000 years, the + 40 m shoreline relative to present-day is estimated at about 8000 Cal BP (Fig. 4B). Thus, most of the Great Slave Lowland was terrestrially exposed within the last 8000 years. More specifically, in the local Yellowknife area, Back Bay formed about 4000 years ago when lake levels were about + 20 m relative to present. The Yellowknife River was not confined to its present channel below Tartan Rapids until about 3000 years ago, when lake level was about + 15 m. Tartan Rapids itself is relatively new, as Prosperous Lake above the rapids is less than + 2 m relative to Great Slave Lake. Similarly, most of the alluvium along the Yellowknife River was deposited within the last 3000 years, and probably more recently. Lithalsa distribution in relation to elevation and alluvial deposits in the Yellowknife area provides further insight into potential timing of permafrost aggradation. Figure 5 shows the frequency distribution of lithalsas mapped in Figure 4, relative to their elevation and, based on lake- level recession rates, their maximum age. Whereas most lithalsas reside within + 30 m of present lake level, and are thus no older than 6000 years, more than half reside with + 15 m, and thus have formed potentially within the last 3000 years. Notably, nearly all lithalsas formed within + 10 m of present lake level elevation (the last 2000 years) have developed on alluvium of the Yellowknife River and Back Bay. This suggests that these saturated clayey-silt deposits, in close proximity to a water source, are conducive to lithalsa formation during permafrost aggradation. Noting the location of other lithalsas in Figure 4, it is likely that these have similarly formed where alluvium or nearshore lacustrine sediments were deposited.

Fig. 6. Cryofacies description, analytical results (including AMS age and δ18O values) and environmental interpretation of ground ice exposure in slump at Site 6 on the Yellowknife River.

was examined to assess the origin of the ice, and the age and depositional environment of the enclosing sediment. The slump has regressed landward from the river by about 40 m (Fig. 2 C) and the elevation of the ground

surface at the headwall (Fig. 2D) is about 8 m above river Fig. 5. Lithalsa occurrence by elevation in the Yellowknife elevation. Figure 6 shows the stratigraphic and analytical area also showing occurrence of lithalsas on Yellowknife results, and interpretation of the section and Table 1 River alluvium and maximum-limiting age approximations includes results from two AMS ages at this site. An based on Holocene lake-level recession rates. organic cover (Unit 1), comprised primarily of un-

decomposed moss, extended to 20 cm depth, below 4.3 Recent ground ice formation which an unfrozen clayey-silt (Unit 2) with a blocky texture

extended to a maximum depth of about 90 cm, with Regarding the abundance of lithalsas along the abundant woody roots in the upper 20 to 45 cm. This unit Yellowknife River, a retrogressive thaw slump (Site 6) on contained nearly equal amounts of clay (< 4 µm) and silt an island in the river, 1 km north of Back Bay (Fig. 4A), (< 63 µm to 4 µm), and a small fraction of fine-grained

54 sand, and was a low-plasticity clay in which the existing The nature of the ground ice from 120 cm to the base water content of 28% dry weight was slightly above the of the exposure is indicative of segregated ice that formed plastic limit of 23%. A thawed, undulating clayey silt and epigenetically as permafrost aggraded into alluvial sandy silt unit (3A and 3B) extended across the section sediments following terrestrial exposure. The significant from about 83 to 100 cm depth. The clayey-silt sub-unit increase in volume of ice below 160 cm depth suggests contained 70% silt, 27% clay and minor sand, and was a freezing rates were slow with unlimited available water. low plasticity clay with a water content of 19% at about the Significantly, the in situ roots within these the sediments plastic limit. An AMS age from a root contained within the indicate the ground was unfrozen until about 80 ± 30 14C sediments was modern. The sandy-silt sub-unit contained BP. This age equates to between 253 and 25 calendar 71% silt, 10% clay and 19% fine sand, with a moisture years ago (Calib.©7.0.4, 2014), suggesting that content of 16%. A silty-clay (Unit 4), with an erosional permafrost aggradation into the sediment likely occurred upper contact with Unit 3, extended to the base of the within the last two hundred years at most. Furthermore, as exposure. This was divided into two sub-units based on the ground surface of the headwall is presently 8 m above the thaw depth at time of observation. The upper thawed river level, this suggests that as much as 8 m of ground sub-unit (4a) extended from 100 to 120 cm depth, and ice has formed within this time period, heaving the ground contained 51% clay, 46% silt and 3% sand, and was an surface. intermediate plasticity clay with a moisture content of Atterberg limits indicate clay-like properties of these 30%, which was well above the plastic limit of 20%, and fine-grained lacustrine and alluvial sediments, being approaching the liquid limit of 38%. The lower frozen sub- conducive to both ice segregation and sensitivity to thaw. unit (4b) extended from 120 cm to the base of the Field water contents, exceeding the plastic limit at the exposure at 220 cm. It contained 58 to 64% clay, 35 to base of the active layer and within underlying frozen 40% silt and a small fraction of fine-grained sand, and an sediments, enable ongoing thaw slumping by promoting intermediate plasticity clay. Ice content within the frozen sliding and flowing of liquefied sediments. It is possible sub-unit increased substantively from 40 to 70% visible that slumping is further promoted by the increase in slope ice by volume, and ranged from wavy lenticular to layered from surface heave through ice segregation in permafrost sub-horizontal to inclined lenses to reticulate ice with at depth, facilitating soil failure at the base of the active suspended ataxtic soil clasts up to 20 cm in diameter. layer. Moisture content of the combined soil and ice in the upper The ground ice exposure on the Yellowknife River portion of the sub-unit was 45%, equivalent to the liquid (Site 6 shown in Fig. 4B) provides an example of limit. In contrast, the moisture content of the soil comparatively recent historical permafrost aggradation component (without excess ice) obtained from a clay clast into alluvial sediments, accompanied by epigenetic in the lower portion of the sub-unit was 36%, still ground ice formation, with ongoing lake-level recession. exceeding the plastic limit of 23%. Oxygen isotopic values This suggests that the processes of permafrost of the excess ice from between 160 and 200 cm depth aggradation and ground ice (i.e. lithalsa) formation have ranged from -16.0 to -17.5‰ and the δ18O value of a continued under historically modern climate conditions. single ice crystal was -16.6‰. Co-isotopic ratios of δ18O/ δD fall within the range of natural lakes in the area (Gaanderse 2015), confirming a modern water source for 5 CONCLUSIONS the ice. An AMS age of an in situ root along a sediment- ice margin in the section was 80 ± 30 14C BP. The Great Slave Lowland was initially inundated by Underlying silty-clay sediments from Unit 4 are Glacial Lake McConnell, and subsequently by ancestral interpreted as lacustrine deposits, based on their high Great Slave Lake at about 8000 cal BP. Concerning combined silt and clay content. The clay content within regional terrestrial emergence and permafrost this unit is lower, however, than that typical of Glacial aggradation, the main conclusions of this study are: Lake McConnell sediments, which usually exceed 70% 1. In the Yellowknife area, Holocene lake-level (Gaanderse 2015), suggesting that these are not recession of ancestral Great Slave Lake occurred at a glaciolacustrine sediments. More likely, they represent rate of about 5 mm/year over the last 8000 years. ancestral Great Slave Lake deposits, likely in relation to 2. As lake-level recession exposed land within the the expanded Yellowknife Bay as depicted in Fig. 4B or Great Slave Lowland, permafrost aggraded into fine- are more recent. The overlying sandy-silt and clayey-silt grained lacustrine, alluvial and underlying glaciolacustrine Unit 3, with an erosional unconformity at its base, is sediments following terrestrial emergence. Ice-rich interpreted as fluvial channel deposits, with the sandy-silt permafrost mounds (lithalsas) formed in some areas as representing Yellowknife River channel deposits and the permafrost aggraded into saturated sediments. clayey-silt representing side deposits in shallower water. 3. In the local Yellowknife area, most lithalsas This would correspond to a time when the Yellowknife formed within the last 6000 years, and many are locally River was confined below Tartan Rapids and entering into associated with alluvial deposits along the Yellowknife the north end of Back Bay as it does today. Unit 2 River, which were deposited within the last 3000 years as sediments are interpreted as alluvial plain deposits, the river became confined to a channel below Tartan representing a shallow-water or seasonally-flooded Rapids. environment as presently observed along the Yellowknife 4. The process of permafrost aggradation and River shoreline. ground ice formation has continued in recent historical

55 times along alluvial floodplains the present-day shoreline continent North America. Geophysical Research of the Yellowknife River. Letters, 38: L24311. Moser, K.A. and MacDonald, G.M. 1990. Holocene vegetation change at treeline Northwest Territories, ACKNOWLEDGEMENTS Canada. Quaternary Research 34: 227-239. Morse, P.D., Wolfe, S.A., Kokelj, S.V. and Gaanderse, This paper is a contribution to the Climate Change A.J.R. In Press. The occurrence and thermal Geoscience Program of Natural Resource Canada. We disequilibrium state of permafrost in forest ecotopes of gratefully acknowledge the field assistance of Adrian the Great Slave region, Northwest Territories, Canada. Gaanderse and Mike Palmer and review of a draft version Permafrost and Periglacial Processes. of the paper by Phil Bonnaventure. Riseborough, D.W., Wolfe, S.A. and Duchesne, C. 2013. Permafrost modelling in northern Slave region Northwest Territories, Phase 1: Climate data REFERENCES evaluation and 1-d sensitivity analysis; Geological Survey of Canada, Open File 7333, 50pp. Calib.©7.0.4. 2014. Calib Manual. Stuiver, M., Reimer, Smith, D.G. 1994. Glacial Lake McConnell P.J. and Reimer, R.. Quaternary Isotope Lab, paleogeography, age, duration, and associate river University of Washington. deltas, Mackenzie River Basin, western Canada. Environment Canada. 2015. Online Climate Data. Quaternary Science Reviews 13: 829-843. http://climate.weather.gc.ca/ [accessed 14 May 2015]. Vanderburgh, S. and Smith, D.G. 1988. Slave River delta: Gaanderse, A.J. 2015. Geomorphic origin of a lithalsa in geomorphology, sedimentology, and Holocene the Great Slave Lowlands, Northwest Territories, reconstruction. Canadian Journal of Earth Sciences Canada. MSc. Thesis. Department of Geography and 25: 1990-2004. Environmental Studies. Carleton University. 155 pp. Wolfe, S.A., Stevens, C.W., Gaanderse, A.J. and Heginbottom, J.A., Dubreuil, M.A. and Harker, P.A. 1995. Oldenborger, G. 2014. Lithalsa distribution, Canada-Permafrost. In National Atlas of Canada, 5th morphology and landscape associations in the Great Edition. National Atlas Information Service: Natural Slave Lowland, Northwest Territories, Canada. Resources Canada, Ottawa, ON; Plate 2.1. MCR Geomorphology 204: 302-313. 4177. Wolfe, S.A. and Kerr, D.E. 2014. Surficial geology, Hoeve, T.E., Seto, J.T.C. and Hayley, D.P. 2004. Yellowknife area, Northwest Territories, parts of NTS Permafrost response following reconstruction of the 85-J/7, NTS 85-J/8, NTS 85-J/9 and NTS 85-J/10, Yellowknife Highway. International Conference on Geological Survey of Canada, CGM 183. Cold Regions Engineering. Edmonton. Huang, C., MacDonald G.M. and Cwynar, L.C. 2004. Holocene landscape development and climate change in the Low Arctic, Northwest Territories, Canada. Palaeogeography Palaeoclimatology Palaeoecology 205:221Ð234. IPCC. 2013. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (eds.). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1535 pp. Kerr, D.E., Knight, R.D., Sharpe, D.R., Cummings, D.I. and Kjarsgaard, B.A. 2014. Surficial Geology, Snowdrift, NTS 75-L, Northwest Territories; Geological Survey of Canada, Canadian Geoscience Map-137, (preliminary), scale 1:125 000. Lemmen, D.S., Duk-Rodkin, A. and Bednarski, J.M. 1994. Late glacial drainage systems along the northwest margin of the Laurentide ice sheet. Quaternary Science Reviews 13: 805-828. MacDonald G.M., Edwards, T.W.D., Moser, K.A., Pienitz, R. and Smol, J.P. 1993. Rapid response of treeline vegetation and lakes to past climate warming. Nature 361:243Ð246. Mazzotti, S., Lambert, A., Henton, J., James, T.S., and Courtier, N. 2011. Absolute gravity calibration of GPS velocities and glacial isostatic adjustment in mid-

56

Part 3 Hydrologic effects in permafrost regions

57

58 Estimating talik depth beneath lakes in Arctic Alaska

Kenneth M. Hinkel University of Cincinnati, Cincinnati, OH, USA Christopher Arp University of Alaska, Fairbanks, AK, USA

ABSTRACT Water temperature and morphometric data were collected from 28 lakes in Arctic Alaska of various size, geometry, depth and genesis. Using methods pioneered by J. Ross Mackay, calculations of temperature beneath the lake indicate that 20 of these have through . We estimate that over 2100 lakes in the study area have through taliks, suggesting that the continuous permafrost is riddled with taliks that may connect surface lake water with sub-permafrost water.

RÉSUMÉ Des données de température de l'eau et des données morphométriques ont été recueillies pour 28 lacs de tailles, de géométries, de profondeurs et de genèses variées en Alaska arctique. En utilisant des méthodes mises au point par J. Ross Mackay, le calcul de la température sous le lac indique que 20 d'entre eux sont associés à des taliks traversant. Nous estimons que plus de 2100 lacs dans la zone d'étude ont des taliks traversant, suggérant que le pergélisol continu est criblé de taliks qui peuvent connecter l'eau de surface d’un lac avec l'eau sous le pergélisol.

1 INTRODUCTION 2 STUDY AREA

The development of a lake above permafrost introduces a The North Slope of Arctic Alaska is a region of tundra thermal disturbance that causes warming of the frozen underlain by continuous permafrost with a thickness of up ground beneath the lake bed. Generally, if the lake does to 600 m. Elevation gradually declines northward from not freeze to the bed in winter, there is a residual pool of 1500 m in the foothills of the east-west trending Brooks water and a local depression of the permafrost table to Range. The area is characterized by thousands of lakes form a thaw bulb or talik. Beneath larger and deeper that are oriented in a north-south direction. These lakes lakes, the thermal perturbation may propagate into the are developed on a surface largely composed of frozen permafrost to create a through talik that extends to the unconsolidated sediments sitting above Cretaceous base of the permafrost body. Borehole data and sedimentary bedrock. geophysical surveys (e.g., Nolan et al. 2009) ) have The NSF-sponsored Circumarctic Lakes Observation demonstrated the existence of taliks beneath large lakes Network (CALON) project is designed to collect and and rivers, but these data are expensive to acquire and assess basic physical, biological and geochemical spatially restricted. information from a representative sample of lakes A second approach is to model the effects of the collected along north-south geomorphic and climatic thermal disturbance of the lake (e.g., Ling and Zhang gradients. The climate generally warms southward across 2003, West and Plug 2008). Adapting a three-dimensional the 2 degrees of latitude, so interior lakes tend to become theoretical heat conduction model developed by Arthur ice-free earlier in the summer and have a longer ice-free Lachenbruch (1957, 1959), J. Ross Mackay estimated the season. Lakes along the Arctic coast become ice-free 2-4 thermal condition beneath lakes to understand pingo weeks later due to cooler and cloudy maritime conditions. formation and dynamics in his classic 1962 paper Pingos One consequence is an increase in the summer water of the Pleistocene Mackenzie Delta Area. Michael Smith temperature of interior lakes. (1976) and Chris Burn (2002), both students of Mackay, Near the Arctic Ocean, the Outer Coastal Plain is a subsequently modified and expanded his approach to flat, low relief region with large oriented thermokarst lakes account for elongated water bodies, and applied these developed in marine silt and sand. This region grades methods using field data (Burn 2002). southward into the Inner Coastal Plain; a higher, hillier The analysis presented here is an extension of the terrain with aeolian sand at the surface. Lakes here are thermal equilibrium approach pioneered by Mackay, Smith numerous and oriented, but not as large as those near the and Burn. The requisite morphometric and temperature coast and tend to be deeper. Further inland and at higher data have been collected for 28 lakes across a large elevations is the silt belt of ice-rich deposits expanse of Arctic Alaska. The objective is to assess the (), a higher relief terrain with fewer lakes. At the nature of the talik beneath these lakes given variations in southern part of the study area is the Arctic Foothills at lake size, depth, geometry and average water and ground elevations above 120-200 m, which is largely covered by temperature. glacial till in which are found kettle lakes.

59 3 METHODOLOGY sensor string was returned to the original position near the lake center. Data from these sensor strings are analyzed Two north-south transects were used to monitor lakes, in this paper as they provide a continuous record for the and are separated by several hundred kilometers (Figure lake center over the period including late summer, 1). Ten nodes were established along these transects, freezeback, winter ice thickening, and meltout. with nearby terrestrial meteorological stations deployed at Characteristics of these lakes are given in Table 1 and are each node to collect regional data. A total of six shown in Figure 1. representative lakes from each node have been A detailed bathymetric survey was conducted on many monitored since 2012. These lakes were selected to cover of the lakes using a sonar unit. Using a nested sampling a range of lake size, depth and origin. scheme, soundings were first made parallel to the The meteorological stations collect hourly shoreline or littoral shelf, then progressively shifted toward measurements of air temperature, relative humidity, the center of the lake (Hinkel et al. 2012). Using this rainfall, solar radiation, wind speed and direction using method, many thousands of water depth readings were Onset Computer sensors. obtained. The inflatable boat was restricted to use in Lake instrumentation was based on a hierarchical water depths greater than 0.5 m, so the littoral terraces strategy. A rope with an anchor and float was deployed in were avoided. Following data processing, the shoreline April through the ice near the center of all lakes. Above was digitized, assigned a depth of 0.0 m, and attached to the anchor was a water level sensor (Onset U20-001, the spreadsheet of latitude, longitude and depth ±0.2¡C) collecting hourly readings of water pressure and measurements. In ArcMap, these data were interpolated temperature, and just below the float was a high-precision onto a 5 m grid using a natural neighbour algorithm. sensor (Onset U12-015, ±0.1¡C) to measure near-surface water temperature. A 25-cm hole was augured through the ice near the center of each lake. After measuring ice 4 BACKGROUND thickness, the sensor string was lowered into the hole, and the float pushed sideways beneath the ice. In spring In deeper lakes, where the ice does not freeze to the lake during breakup, the float remains beneath the moving ice bed in the central pool, the temperature at the base of the slab, and gradually rises toward the lake surface as the water column and in the lake bed sediments is above zero ice melts and thins. Though not foolproof, this strategy throughout the year to create a talik. Lachenbruch (1957) partially alleviates the problem of float movement and presented a three-dimensional theoretical study of heat possible destruction during spring break-up, but the near- conduction in permafrost beneath a heated building, surface sensor measurement is only valid for the post- which is analogous to a surface thermal disturbance melting period. All floats were retrieved in August, induced by lake formation. Mackay (1962) tailored this serviced, and redeployed the following April. approach to estimate the depth of the talik beneath deeper lakes on the Mackenzie delta area (Mackay 1963); in his one-dimensional approach, latent heat effects are neglected since steady-state conditions are assumed, and the thermal properties of the ground have no influence. Building on work by Smith (1976), Burn (2002) applied it to lakes on Richards Island in western Canadian Arctic coast that have a deeper central pool and shallow littoral terrace underlain by permafrost. Assuming steady-state conditions, the temperature (T) at any depth Z can be estimated as (Mackay 1962):

T(Z) = Tg + (Z/I) [1]

where Tg is the mean annual ground surface temperature and I is the local geothermal gradient (m C-1). The formation of a lake at the ground surface introduces a positive thermal disturbance that is superimposed on the ground thermal profile. If the mean annual water temperature in the central lake pool is above freezing, as is the case for lakes that do not regularly Figure 1. Study area in Arctic Alaska freeze to the bottom in winter, the thermal perturbation beneath the center of a circular lake is estimated by In half the lakes, more intensive instrumentation was Mackay (1962, eq. 7) as: used. Sensor strings with the same configuration were 2 2 0.5 deployed in the lake center in August and left throughout T(Z) = Tg + (Z/I) + {(Tl-Tg)*(1-[Z/(Z +R l) ])} [2] the year. Because the floats became embedded within the ice cover in winter, they tended to drift and move during where Tg is the mean annual ground temperature of the spring break-up. Often, they would be carried many tundra, Tl is the mean annual lake water temperature, and hundreds of meters, and sometimes ended up near shore Rl is the radius of a lake. The thermal impact is or on the littoral terraces. Once retrieved and serviced, the

60 maximized beneath the geometric center of the lake. deep boreholes from the DOI/GTN-P array (Deming et al. Given enough time, the surface warming can ultimately 1996; Clow 2014) are found in and around the study area, lead to basal warming of the permafrost body and the and are shown in Figure 1. The long-term ground surface upward displacement of the lower permafrost boundary. temperature for each borehole location was estimated by Often, lakes are shallower near the shore with bed-fast best-fit extrapolation of the temperature profile to the ice in winter, and many lakes develop a shallow littoral surface, and is used here as an estimate of Tg. The terrace. In either case, the mean annual temperature is borehole temperature measurements in the upper 400 m warmer than the undisturbed tundra, though not as warm were also used to calculate the borehole-specific as the lake center. Thus, there is a central inner pool of geothermal gradient (I), which varied from 21 to 40 m C-1. deep water with radius Rp and temperature Tp, and an The geothermal gradient and ground temperature at each annulus of shallow water with bedfast ice in winter. The lake was estimated by spatially interpolating the borehole radius of the lake (Rl) is equal to Rp+t, where the subscript data across the study area. “t” refers to the terrace of temperature Tt, such that Tp > Tt > Tg. The primary disturbance at depth Z beneath the Table 1. CALON lake characteristics; those highlighted lake center is the sum of the thermal impact of the in grey do not have a through talik. unfrozen central pool and that of the surrounding terrace, which introduces an additional term to Eq. 2, as shown in Area Depth Eq. 3 (Mackay 1962, eq. 8; Smith 1976): Lake Lat (Dd) Long (Dd) (ha) (m) Tp C 2 2 0.5 T(Z) = Tg + (Z/I) + {(Tp-Tg)*(1-[Z/(Z +R p) ])} + Ikp-002 70.815 -154.424 14.8 0.5 -0.68

2 2 0.5 2 2 0.5 Ikp-003 70.793 -154.517 11.2 2.7 1.84 {(Tt-Tg)*(( [Z/(Z +R p) ] - [Z/(Z +R p+t) ]))} [3] Tes-002 70.789 -153.470 265.3 0.8 -2.34 Assuming the lake has been stable for a long duration Tes-003 70.868 -153.773 356.7 0.9 -2.86 and the one-dimensional substrate temperature is in equilibrium, Eq. 3 can be used to estimate the Tes-005 70.752 -153.869 34.0 0.6 -3.99 temperature at any depth beneath the center of a circular FC-M9925 70.247 -151.478 87.4 0.4 -2.77 lake with littoral terraces, and to estimate the depth to the Ini-002 70.000 -153.037 1.3 0.8 -1.76 top and base of the permafrost layer where Tg = 0¼C. Lake geometry also has an influence. Many lakes in Ini-005 70.018 -153.186 4.9 2.1 3.60 the study area are elongated and oriented in a north-south BRW-100 71.242 -156.774 183.8 2.3 2.41 direction. For elongated lakes with terraces, Smith (1976, eq. 7) and later Burn (2002) modified the above equations BRW-103 71.123 -156.317 179.8 1.9 2.25 so that T(Z) can be estimated as Eq. 4: BRW-107 71.274 -156.497 125.0 1.9 1.54

-1 Ikp-001 70.790 -154.450 68.7 2.7 2.20 T(Z) = Tg + (Z/I) + {[(Tp-Tt)/ π]*(2tan (Hp /Z))} + LMR-400 70.754 -156.720 252.3 1.5 2.53 -1 {[(Tt-Tg)/ π]*(2tan (Hp+t /Z))} [4] LMR-402 70.728 -156.843 356.0 1.0 1.92

Where Hp is the half-width of the deep central pool and Tes-001 70.766 -153.562 979.9 2.5 3.23 Hp+t is the half-width of the lake (pool plus terrace). In this Tes-006 70.706 -153.924 110.1 2.2 3.07 approach, the last two terms reflect the thermal impact of the unfrozen pool and terrace, respectively, on the ATQ-200 70.455 -156.948 271.3 2.5 4.58 thermal profile. An elongated lake is defined as one where ATQ-202 70.288 -156.985 148.8 2.4 4.81 the length exceeds the width by at least a factor of two. ATQ-207 70.329 -156.592 353.6 3.5 3.90 Eqs. 2 through 4 are not single-value functions of depth (Z), and cannot be used to directly determine the FC-L9819 70.270 -151.355 100.5 1.8 3.19 upper and lower surfaces of a permafrost body; for this FC-L9820 70.267 -151.386 128.5 1.1 0.77 reason, an iterative approach is used. Application of the method depends on obtaining lake morphometrics, the FC-R0066 70.147 -151.765 104.5 2.4 4.10 local geothermal gradient (I), and mean annual Ini-001 69.996 -153.070 66.4 4.4 3.50 temperature of the lake water (Tp), littoral terrace (Tt), and Ini-003 69.959 -152.951 417.3 2.2 4.50 unaffected nearby tundra (Tg). Given the simplifying assumptions, measured values can be used to determine Ini-006 70.219 -153.172 361.7 4.3 3.91 the existence of a talik , to estimate the depth of the talik RDC-300 69.961 -156.546 63.8 6.0 4.18 beneath the central lake pool, or to ascertain if a through talik penetrates the permafrost. Data collected from the RDC-308 69.986 -156.424 78.7 2.2 4.38 CALON project are used to assess the thermal state RDC-311 69.996 -156.689 76.9 7.0 3.65 beneath 28 lakes where the necessary data have been collected (Table 1). The geothermal gradient (I) and ground temperature Although the water temperature on the terraces (Tt) (Tg) varies across the Arctic North Slope. A total of 13 varies annually with water depth, air temperature, and ice

61 duration, it is typically 3-5¡C warmer than the surrounding winter. Conversely, lakes on the Inner Coastal Plain and tundra (Tg). For that reason, we conservatively estimate Foothills tend to be deeper; lakes here that are smaller Tt as 3¡C warmer than the local Tg. than 66 ha may have a through talik. Twenty eight circular and elongated lakes of varying size have adequate field data to perform the calculations, as shown in Table 1. Eq.3 was applied to the 14 circular lakes and Eq. 4 to the 14 elongated lakes. Lake dimensions were measured on high-resolution satellite imagery or bathymetric maps by establishing a line down the longest axis of the lake, which is typically oriented north-south. The minor axis is perpendicular to the long axis, with axes intersecting in the geometric center of the lake. All measurements of pool and terrace width (R) were made along the minor axis.

5 RESULTS AND DISCUSSION

The results of the analysis estimate that the majority (20) of the 28 lakes may have a through talik (Figure 2). All lakes smaller than 50 ha have no talik or a shallow talik. Three of these lakes (Ini-002, Ikp-002 and Tes-005) have depths of less than 1.0 m and freeze to the bottom in winter, so the mean annual temperature of the central pool is less than 0¡C and there is no talik. By contrast, the other two small lakes (Ini-005 and Ikp-003) have depths of 2.1 and 2.7 m, respectively. These small lakes are deep Figure 2. Talik condition based on lake depth and area; enough to maintain a pool of basal water throughout the note logarithmic scale for area winter, and a mean annual water temperature of 2-4 ¼C. The calculations suggest that these lakes have a talik, There are several limitations to this study. First, it is with thawed sediments to a depth of 12-18 m. unlikely that lake morphology remains static over the Three somewhat larger lakes (FC-9925, Tes-003 and lifetime of a lake. Thermokarst lakes tend to get larger Tes-002) also lack a talik. Although lake area ranges from and become more oriented from thermomechanical bank 87 to 357 ha, lake depth is no greater than 0.9 m and a erosion, and littoral shelves can develop and widen over corresponding mean annual lake bed temperature of time. Additionally, lake water levels change. Many lakes about -2.5¡C. Bedfast ice in winter prohibits basal show evidence of partial drainage, especially in regions of permafrost degradation. higher relief away from the coast. A reduction in lake size To summarize, results suggests that small lakes (<50 from partial drainage would likely result in a narrowing of ha) or very shallow lakes (<1.0 m) do not have an talik the through talik over time as permafrost aggrades through the permafrost. Permafrost that is warmer than laterally, and possibly lead to closure of the vertical the undisturbed tundra exists beneath the lake, and the thawed zone. base of the permafrost body has been locally displaced A second assumption is that lake ice thickness has upward toward the surface. remained constant over time. The water depth limitation All lakes with an area exceeding 66 ha and water of 1.0 m identified in this study is based on contemporary deeper than 1.0 m have a through talik irrespective of measurements, and all lakes with water depths >0.9 m their morphometry (circular or elongated ) or presence of have mean annual water temperatures greater than zero. littoral shelves. Using 66 ha as a very conservative However, evidence from Arctic Alaska suggests that estimate of the minimum lake size needed to develop a regional lake ice thickness has recently declined owing to through talik, we can estimate the number of lakes in the warmer winters, an extended duration of the ice-free region that would have a through talik. Of the 35,383 season, and thicker snowcover (Arp et al. 2012). This lakes in the study area (> 1 ha in size), about 6% (2100) implies that ice thickness in the recent past was greater, have area greater than 66 ha; these are shown in dark so that the ice-free period would have been shorter and blue in Figure 1. These lakes cover at least 12% of the the mean annual water temperature lower. Indeed, study area. This suggests that the permafrost is riddled studies in Barrow report ice thickness of ~ 2 m in the late with through taliks, and that the lake water may be 1950s (Brewer 1958). Many of the shallow lakes would hydrologically connected to sub-permafrost groundwater. have regularly frozen to the lake bed, with consequent This approach ignores the critical impact of water depth impact on Tp and talik development. and thus basal water temperature, and also the effects of Similarly, the permafrost and Tg have been warming lake geometry and radius of the central pool. More such that the current estimates of I, Tg and Tt are not importantly, it neglects the spatial patterns noted earlier. representative of the long-term, steady state condition. Lakes on the Outer Coastal Plain are larger but generally The magnitude of recent warming is often cited as 1-2 ¡C. shallower, and more likely to freeze to the bottom in To test the possible impact of warming, the mean annual

62 ground temperature (Tg), terrace temperature (Tt) and is riddled with taliks that may connect surface lake water water pool temperature (Tp) were reduced by 1.5 ¡C, and with sub-permafrost water. the temperature field beneath each lake was recalculated using the procedure described above. The overall results were the same; no lake changed category from no or ACKNOWLEDGEMENTS shallow talik to through talik. This suggests that the recent observed warming does not have the potential to open a This research was support by funding from the National through talik beneath any lakes. However, for the eight Science Foundation, grants AON-1107607 to KMH and lakes without a through talik, the base of the permafrost AON-1107481 to CDA. The authors thanks Gary Clow was substantially deeper under colder conditions. Further, and Vladimir Romanovsky for providing borehole data the 12-18 m thick talik immediately beneath Ikp-003 and from the DOI/GTN-P network. Ini-003, both deep and small (<11 ha) lakes, was notably thinner. In other words, recent warming has caused some deepening of the thawed zone beneath these small lakes. REFERENCES Over the Holocene, as lakes developed, expanded in size and deepened, taliks began to penetrate the Arp, C.D., Jones, B.M., Lu, Z. and Whitman, M.S. 2012. permafrost body. Episodic or periodic lake drainage Shifting Balance of Lake Ice Regimes on the Arctic would locally counter this overall trend as permafrost Coastal Plain of Northern Alaska. Geophyscial aggraded into the unfrozen zone and eventually closed Research Letters, 39(L16503): 1-5. the talik. Recent warming has likely impacted mean Brewer M.C. 1958. The Thermal Regime of an Arctic annual lake water temperatures, enhancing permafrost Lake. Transactions AGU, 39: 278Ð284. degradation beneath lakes. On a time scale of centuries Burn, C.R. 2002. Tundra Lakes and Permafrost, Richards to millennia, more open taliks would develop, especially Island, Western Arctic coast, Canada. Canadian where permafrost is thinner. At some point, the Journal of Earth Sciences, 39: 1281-1298. continuous permafrost of Arctic Alaska would become Clow, G.D. 2014. Temperature data acquired from the discontinuous permafrost. DOI/GTN-P Deep Borehole Array on the Arctic Slope Imikpuk Lake near Barrow has a radius of 370 m, of Alaska, 1973-2013. Earth Syst. Sci Data. ACADIS depth of 2.8 m, and mean annual basal water temperature repository at http://dx.doi.org/10.5065/D6N014HK. of 1.8¡C. Our calculations suggest that this lake would Deming, D., Sass, J.H., and Lachenbruch, A.H. 1996. have a through talik. However, Max Brewer measured a Heat flow and subsurface temperature, North Slope of talik thickness of 60 m below Imikpuk Lake in the late Alaska. In Thermal Evolution of Sedimentary Basins 1950s. This disparity might be explained by the model in Alaska, USGS Bulletin 2142, Mark J. Johnsson and assumption of temperature equilibrium. If we set Tg and David G. Howell, Editors, US Government Printing of Tp 1.5¡C cooler to reflect the effects of recent warming so fice, Washington: 21-44. that Tp is 0.3¡C, the lake radius is still sufficiently large to Hinkel, K.M., Sheng, Y., Lenters, J.D., Lyons, E.A., Beck, have a through talik. Only if the lake radius is reduced to R.A., Eisner, W.R. and Wang, J. 2012. Thermokarst about 250 m would permafrost exist, with the talik 30 m Lakes on the Arctic Coastal Plain of Alaska: deep. For this reason, the results we present here only Geomorphic Controls on Bathymetry. Permafrost and reflect the potential condition beneath smaller lakes after Periglacial Processes, 23: 218Ð230. DOI: the centuries or millennia that would be required for the 10.1002/ppp.1744. underlying permafrost to reach equilibrium temperature Lachenbruch, A.H. 1957. Three-Dimensional Heat conditions. The implicit model assumption is that climatic, Conduction Beneath Heated Buildings. U.S. geomorphic and lacustrine conditions remain static at the Geological Survey, Bulletin 1052-B, pp. 51Ð69. surface. This overarching assumption is probably Lachenbruch, A.H. 1959. Periodic Heat Flow in a unjustified given the time scales of the processes Stratified Medium With Application to Permafrost involved. Problems. U.S. Geological Survey, Bulletin 1083-A, pp. 1-36. Ling, F. and Zhang, T. 2003. Numerical Simulation of 6 CONCLUSIONS Permafrost Thermal Regime and Talik Development Under Shallow Thaw Lakes on the Alaskan Arctic The results from this study suggest that through taliks Coastal Plain. Journal of Geophysical Research- may exist beneath larger lakes in Arctic Alaska. Atmospheres, 108 (D16): 4511. Temperature and morphometryic data were collected from Mackay, J.R. 1962. Pingos of the Pleistocene Mackenzie 28 lakes of various size, geometry, and depth. Delta area. Geographical Bulletin, 18: 21Ð63. Calculations of temperature beneath the lake indicate that Mackay, J.R. 1963. The Mackenzie Delta Area, N.W.T. 20 of these have through taliks. Ignoring the effects of Geographical Branch, Mines and Technical Surveys, lake depth, lake geometry, and the presence of littoral Otttawa. Memoir 8. 202 pp. shelves, a proxy of 66 ha is used as the minimum lake size necessary to develop a talik through the permafrost. In this case, over 2100 lakes in the study area have through taliks, suggesting that the continuous permafrost

63 Nolan, J.T., Slater, L.D., Parsekian, A., Plug, L.J., Grosse, G. and Walter Anthony, K.M., 2009. Thaw Bulb Dimensions Determined Using Electrical Imaging Across Thermokarst Lakes, Seward Peninsula, Alaska. Eos Trans. AGU, Fall Meeting Supplement, Abstract NS23A-1123. Smith, M.W. 1976. Permafrost in the Mackenzie Delta, Northwest Territories. Geological Survey of Canada, Paper, 75-28. West, J.J. and Plug, L.J., 2008. Time-Dependent Morphology of Thaw Lakes and Taliks in Deep and Shallow Ground Ice. Journal of Geophysical Research 113, F01009, 1-14..

64 Lakes of the western Canadian Arctic: Past Controls and Future Changes

Philip Marsh, Tyler de Jong Wilfrid Laurier University, Waterloo, Ontario, Canada Lance Lesack Simon Fraser University, Burnaby, BC, Canada Cuyler Onclin, Mark Russell Environment Canada, Saskatoon, SK, Canada

ABSTRACT The western Canadian Arctic is lake rich with over 40,000 lakes in the Mackenzie Delta, and thousands more to the east of the Delta in the Tuktoyaktuk Peninsula. To address unknowns concerning these lakes, Professor Mackay published at least 19 papers over the period 1956 to 2013 that dealt with various aspects of lakes in the western Canadian Arctic. These papers outlined many of the controlling processes and considered the complex interactions between lakes and the permafrost landscape. Subsequent research has built on Professor Mackay’s legacy and has gradually developed a better understanding of the hydrology and ecology of these lakes, as well considered the interactions between climate, permafrost and hydrology, and developed predictive models.

RÉSUMÉ L'ouest de l'Arctique canadien est une région avec plus de 40 000 lacs dans le delta du Mackenzie, et des milliers d'autres à l'est du delta dans la péninsule de Tuktoyaktuk. Pour répondre aux incertitudes concernant ces lacs, le professeur Mackay a publié au moins 19 articles, entre 1956 et 2013, qui traitent de divers aspects des lacs de cette région. Ces documents présentent la plupart des processus de contrôle et de considération des interactions complexes entre les lacs et le pergélisol. Des recherches ultérieures se sont appuyées sur l'héritage du professeur Mackay et ont progressivement développé une meilleure compréhension de l'hydrologie et de l'écologie de ces lacs. Elles ont ainsi examiné les interactions entre le climat, le pergélisol et l'hydrologie, en plus de développer des modèles prédictifs.

1 INTRODUCTION

Many areas of arctic Canada, Alaska and Eurasia are extremely lake-rich, and Smith et al. (2007) showed that northwards of 45¡N lake abundance and lake area are higher in permafrost regions compared to non-permafrost terrain. This increase in lake occurrence is likely due to thermokarst processes which can form and enlarge lakes during warm periods (Mackay, 1992) and the aggradation of permafrost and ground ice which can both modify the surface topography and limit water drainage due to impermeable permafrost limiting groundwater flow from lakes. Many areas of the Canadian Arctic are lake rich due to other processes. For example, the Canadian Shield is often lake rich due to geologic controls. In contrast to such shield lakes, the western Canadian Arctic is lake rich due to the effect of permafrost. This is especially true for the Tuktoyaktuk Peninsula (TP), Richards Island (RI), and the Mackenzie Delta (MD) of the western Arctic (Mackay, 1992) (Figure 1), a region where Professor J. Ross Mackay focussed much of his research. Professor Mackay greatly enhanced our understanding of lakes of the western Arctic through at least 19 papers over the period 1956 to 2013. Many of these papers considered key aspects of the controlling processes, considered the complex interactions between Figure 1: Lakes (darker bodies) of the Mackenzie Delta, lakes and the permafrost landscape, and set the stage for Richards Island and the Tuktoyaktuk Peninsual immediately to the east of the Delta.

65 future research to build upon Mackay’s legacy. However, next improvement in understanding MD lakes, the one of Mackay’s early achievements was simply to raise development of the first predictive model of MD channel awareness of the importance of these lakes to the water levels using a hydraulic model of the main MD ecology, hydrology, and permafrost of the region, and to channels and water levels of the Beaufort Sea (Nafziger the communities of the MD region (Ft. McPherson, et al., 2009; personal communication, F. Hicks and W. Aklavik, Tuktoyaktuk, and Inuvik). The MD, located where Perrie). the Mackenzie River enters into the Beaufort Sea (Figure Predictive tools are essential for understanding 1), is approximately 12,000 km2 in area and is the largest ongoing impacts of climate change on the MD (Lesack delta in Canada and one of the largest Arctic Deltas. and Marsh, 2007) and for understanding the impacts of Unlike many deltas, the MD is extremely lake rich, with developing the vast natural gas fields that lie below the over 40,000 lakes (Emmerton et al, 2007). The formation MD and which would be the source fields for the process of these lakes has been discussed for many proposed, and approved, Mackenzie Gas Project. years, and are believed to have formed due to a Professor Mackay’s early efforts were of great importance combination of sedimentation, permafrost aggradation, in setting the stage for our understanding of the delta thermokarst, and channel migration. hydrology and played a key role in carrying out the Mackay (1963) realized that the water level regime of environmental assessment of the Mackenzie Gas Project. MD lakes depended on the elevation that they were RI and the TP, located to the east of the Mackenzie perched above the main channels and the connection Delta, are of similar size to that of the MD and as Mackay between the lake and the channel. The controlling (1992) noted, is “a land of lakes where water covers about elevation along the connecting channel is often termed 15 to 50% of the total surface area”. Mackay (1992) the lake sill elevation. Based on this Mackay defined lake showed that many of these lakes are thermokarst lakes classes as: no-, low-, and high-closure lakes. No-closure that formed during an earlier warm period over 8 000 lakes are those with a low sill elevation and “open out to, years ago. These thaw lakes are an important and very are joined with, or are crossed by channels”. Low-closure dynamic component of regions such as RI and the TP and lakes are those that “tend to flood annually” and high the Alaskan Coastal Plain for example (Frohn et al., closure lakes having a high sill elevation and that “flood 2005). For these thaw lakes, lake area may change infrequently”. Marsh and Hey (1989) used observations of dramatically over time due to erosion and melting of the channel water levels and dates of lake flooding in the shoreline by thermokarst processes; a change in the lake vicinity of East Channel near Inuvik to quantify Mackay’s water balance and therefore lake level; or increased definitions by the flood return period. No-closure lakes groundwater outflow as permafrost is melted. Such were then defined as lakes that are connected to the main changes have been described for numerous regions of channels of the delta for much of the summer, with the the arctic. For example, Yoshikawa and Hinzman (2003) lake sill elevation lower than the one-year return period for and Smith et al. (2005) found that lakes in discontinuous summer low water levels. Low-closure lakes have a sill permafrost areas of Alaska and Siberia were elevation such that the lake is cut off from the main disappearing, and hypothesized that this is caused by channels for some of the summer but flooded by high increased sub-surface drainage due to permafrost water during the spring breakup at least annually. melting. Smith et al. (2005) also suggested that Specifically they have a sill elevation greater than the one- thermokarst processes resulted in an increase in lake year return period for low water levels and less than the numbers in the continuous permafrost areas of Siberia. one-year return period for spring water levels. High- For the RI and TP regions, Mackay (1992) showed closure lakes are then those that flood less than annually that thaw lakes have been draining, and removed from the during spring breakup and never flood in the summer. landscape over the last 8000 years, resulting in drained Marsh and Hey (1989) and Marsh et al. (1999) used aerial thaw lake basins (DTLBs). These DTLBs have numerous observations and photography to map flooding levels effects. For example, pingos, a major research topic of across larger areas of the delta in order to further extend Mackay over his entire research career (e,g. Mackay, our understanding of the spatial variability of lake sill 1979) only form within these drained lake basins, while elevations and lake hydrology. Hopkinson et al. (2011) van Huisstedent et al. (2011) has outlined the effect of reported on an International Polar Year (IPY) project that these lakes on methane emmissions. used airborne LiDAR to accurately map lake elevations Mackay (1992) documented the rapid drainage of across three broad areas of the MD, further extending our thaw lakes due to melting of an outlet channel through understanding of the spatial variability of lake types. Such ice-rich permafrost. Such rapid drainage has been termed data provides essential information needed to understand catastrophic drainage by Mackay (1992). The ongoing the impact of changes in ice jam flooding and storm removal of these thaw lakes from the landscape across surges (Lesack and Marsh, 2007; Lesack et al., 2013) on the arctic is extremely important as it affects permafrost, the hydrology and ecology of these lakes. This hydrology and the ecology of these regions. However, the combination of studies has resulted on a greatly improved details still poorly understood. The following sections of understanding of the spatial variability of lake hydrology, this paper will outline Mackay’s contribution to our and the links to aquatic ecology (Emmerton et al., 2007; understanding of such lakes in the western Canadian Squires et al., 2009 and carbon fluxes (Tank et al., 2008). Arctic, and consider recent contributions on this topic that However, our predictive ability is still extremely poor. build upon Mackay’s efforts, and we will consider future Building on these advancements, a Beaufort and Region research needs. Environmental Assessment (BREA) project has led to the

66 2. THERMOKARST LAKES OF THE WESTERN occurrence of subsurface flow was clearly demonstrated CANADIAN ARCTIC by the existence of ‘pool ice’ in newly exposed channel banks. 2.1 Lake characteristics Mackay (1992) suggested that a key process leading to lake drainage was the existence of ice wedges and ice Many of the tens of thousands (Figure 1) of lakes wedge thermal contraction cracking over the winter. When across RI and the TP are thermokarst lakes that such cracks intersected lake outlets, then lake drainage developed during a postglacial warm period between 13 could occur the following spring or summer when water 000 BP and 8 000 BP (Mackay, 1992). At this time, active was able to flow through the crack. Mackay (1992) layer depths increased, resulting in the thawing of the suggested that lake drainage is initiated during periods of upper layers of ice-rich permafrost (Mackay, 1992), high water in the spring runoff period, after periods of ground surface settlement, and ponding of water that lead heavy rain, or in the spring due to the existence of a snow to pond and lake development. Most of the existing lakes dam at the normal lake outlet, when water was able to in this region are less than 5 m in depth, with a few having flow through ice wedge thermal contraction cracks. The depths exceeding 13 m (Burn, 2002). Although this area is importance of ice wedge terrain to lake drainage is in the continuous permafrost zone, Mackay (1992) suggested by a comparison of all drained thaw lake estimated that thaw basins exist beneath those portions of basins and pingos across the TP (Figure 4) and the lakes where water depth exceeds approximately two- distribution of ice wedge polygonal terrain across the thirds of the maximum winter ice thickness (up to 2 m in same area (Figure 5) as mapped by Kokelj et al., (2014). thickness). In addition, Mackay (1992) estimated that if These figures show a remarkable similarity between the the lake diameter considerably exceeds the thickness of distribution of ice-wedge polygonal terrain and the permafrost, the thaw basin would eventually extend distribution of drained thaw lake basins and pingos. throughout the depth of permafrost. Burn (2002) carried Ongoing research is required in order to consider this out a detailed analysis of lake size and thaw bulb relationship in greater detail. configuration under equilibrium conditions, and suggested that about 25% of the lakes in this area have taliks that penetrate the permafrost. As will be discussed in the next section, these thaw lakes are prone to rapid drainage due to the melting of ice rich permafrost that surrounds these lakes.

2.2 Drainage of thermokarst lakes

Following a cooling trend that started approximately 4 5000 BP, thousands of thaw lakes across the TP have been lost from the landscape due to erosion of drainage channels through the ice-rich permafrost (Mackay, 1992), resulting in complete or nearly complete emptying of the lake (Figure 2). As Mackay (1979) has shown, pingos in the TP only form in DTLBs as permafrost aggrades in the former lake bed. A lake removal rate from the TP landscape of one to two per year over the last few thousand years has been estimated by Mackay (1988, 1992) from mapping both pingos and lake drainage between 1950 and 1985. A common feature of DTLBs is the existence of large drainage channels (Figure 3) and large volumes of ground ice in the walls of the drainage channel (Mackay, 1992). In the vicinity of the drained lake sites discussed by Mackay (1992), Pollard and French (1980) showed that ice volume in excess of the pore volume was common, with ground ice volume in excess of 60% in the upper 2 m and with ice wedges averaging 1.5 m in width, having an average depth to width ratio of 3:1, and constituting as much as 50% of earth materials in the upper 1 m of permafrost. Most of the examples of lake drainage described by Mackay (1992) include initial lake outflow over a relatively steep and narrow drainage divide, with flow either through ice-wedge thermal contraction cracks or over areas of ice- wedge polygons. In many cases the drainage channel Figure 2: Example of a nearly completely drained lake does not coincide with the original lake outlet (Marsh and (top) and partially drained lake (bottom). Neumann, 2001). Mackay (1992) suggested that the

67

Figure 3: DTLB channel formed by a combination of ground ice melting and sediment erosion. Note person in the drainage channel for scale. It is common for the walls of these drainage channels to expose large volumes of ground ice. Such channels often follow the path of the previous lake outlet channel, and case simply deepens and widens the existing channel. In other cases, the channel does not follow the existing outlet channel.

Figure 5: Density of polygonal terrain as mapped by Kokelj et al. (2014) and ice wedge characteristics for a portion of the area mapped in Figure 4. Note that areas with higher numbers of DLTBs in Figure 4, appear to coincide with areas of higher density of polygonal terrain. 2.4)

2.3 Discharge rates during lake drainage

Although Mackay (1992) outlined possible processes of lake drainage, few details were known at that time. Although the initial stages of lake drainage have not been well documented from field observations, observations after drainage (Mackay, 1992) suggest that there is typically a period of rapid or catastrophic outflow with near-complete lake drainage occurring within a few hours (Marsh and Neumann, 2001). Such a period of rapid drainage was also described for the experimental drainage of Lake Illisarvik on Richards Island in 1978 (Mackay, 1981). Although it seemed likely that lake drainage occurred due to melting of ground ice, the relative importance of thermal versus mechanical erosion during the period of rapid channel enlargement were not known. However, the importance of the melting of ground ice was suggested by the observation that ‘remarkably little sediment was deposited by the floodwater at the terminus of the gully and along the river, considering the Figure 4: All drained lake thaw basins (DLTB) (shown by size of the gully excavated during lake drainage’ (Brewer triangles) and pingos (closed circles) as mapped by et al., 1993) and by ‘signs of thaw slumping of ice-rich Marsh et al. (2009). Dashed red line outlines the area of permafrost’ in a recently formed outlet drainage channel mapped polygonal terrain shown in Figure 5. (Mackay, 1988). Marsh and Neumann (2001) suggested that if melting of ice is important, and possibly the

68 dominant process leading to lake drainage, then as a first vegetation, such as the expansion of shrubs across the approximation it could be assumed that the processes tundra. Ongoing research is also required to consider controlling the rapid enlargement of the drainage channel these changes. controlling the rate of drainage are similar to that of glacier overtopping or tunnelling controlling the sudden release of water from glacier-dammed lakes where the role of 3) DISCUSSION thermal enlargement has been well documented. Fortier et al. (2007) and Godin et al. (2012) described similar J. Ross Mackay made a significant contribution to our thermo erosional processes in the eastern Canadian understanding of lakes and permafrost across the western Arctic. Canadian Arctic, and highlighted the variety of processes Applying a glacial lake drainage model to two controlling lake water level regime and changes in lake examples of lake drainage where discharge was size and depth. For lakes of the Mackenzie Delta, he was available, Marsh and Neumann (2001) showed that the first to outline the role of lake sill elevation and main modelled discharge was in fact similar to measured, and channel water level in controlling the water level regime of as a result, melting of ground ice was likely the dominant the lakes. Ongoing research has further quantified lake process controlling rapid lake drainage. Marsh and water level regime based on extreme low and high Neumann (2001) and Marsh et al. (2008) modelled channel water level frequency. Building on this work are discharge during lake drainage events for a number of efforts to develop a hydraulic model of the Mackenzie lake sizes, and groups of lakes, reporting that lakes could Delta Channels, including the effect of off channel water in fact drain completely in less than a day and that peak storage in delta lakes, and linking lake hydrology to lake discharge could be orders of magnitude larger than would ecology. Such efforts are essential to understanding the occur during spring snowmelt. ongoing impact of a changing climate (including Beaufort Marsh et al. (2009) mapped all thaw lakes and ponds Sea level and storm surges, and discharge from the (Figure 4) in the study area using National Topographic Mackenzie River basin) and future natural gas extraction Data Base (NTDB) map sheets based on 1950 aerial on the hydrology and ecology of the Mackenzie Delta. photographs and estimated the area of each water body. Mackay also outlined the key aspects of rapid Mackay (1992) also used this data source for his drainage of lakes in the western Arctic. Such rapid estimates of 1950 lake boundaries in the Tuktoyaktuk drainage is a natural occurrence in this region, and has Coastlands. DTLBs were identified from a series of aerial been shown to be sensitive to climate. Recent research photographs taken in August 2000, including partially has shown that the rate of lake drainage is declining, drained lakes. Mapped DTLBs were confirmed via field however it is not clear if this trend will continue. In fact it visits. DTLBs that drained since 1950 were separated seems likely that lake drainage will increase in the future from all other lakes and DTLBs, and lake drainage was as thermokarst melting increases, and results in higher then determined for the periods 1950 to 1973, 1973 to rates of lake drainage. It is expected that such a lake 1985 and 1985 to 2000 using available aerial drainage event will occur during the summer or fall of photographs. 2015 in an area west of Ft. McPherson in the Mackenzie The overall drainage rate for the period 1950 to 1985, Delta region (S. Kokelj, person communication). This the time period used by Mackay (1992) was similar to the event will occur due to a massive slumping of ice rich 1 or 2 lakes per year found in Mackay’s study. However terrain that will encroach on a small lake. It seems likely the full data set showed a decreasing rate of lake such lake drainage events will occur more often in this drainage as follows: for the period 1950 to 1973, the region as the climate warms over the coming decades. annual drainage rate was 1.13/yr, decreasing to 0.83/yr Ongoing changes to permafrost affected lakes, either for 1973 to 1985, and only 0.33/yr for the period 1985 to increasing or decreasing lake drainage or expansion of 2000. There is a need to consider the drainage rate from lake area due to thermokarst processes or changes in 2000 to present and determine if it continues to decline. lake water balance, will have significant ecological The processes controlling this decline in lake drainage implications and may impact infrastructure built in this rates are not well understood. Pohl et al. (2009) modelled area. Drained lakes obviously result in a significant lake levels to see if a changing climate over the 1975 to change in habitat, with a change from lakes used by 2006 period could be responsible for a decline in lake waterfowl to DTLBs being a location of enhanced shrub drainage. For this 32 year period, they found no trends in growth and therefore moose habitat for example. Future lake levels. Suggesting that a change in hydrology is likely changes are also important for understanding the impact not responsible for the declining lake drainage rates. on linear infrastructure such as pipelines or roads such as Since ice wedge thermal contraction cracking plays a very the Inuvik-Tuktoyaktuk Highway that is currently under important role in lake drainage, a decrease in cracking is construction. Rapid lake drainage poses a significant risk a possible driver of the decrease in lake drainage. Kokelj to such infrastructure as discharge from these lakes is et al. (2014) outlined some of the factors controlling ice typically not planned for in the design of pipelines or wedge cracking across the TP. Although cracking is roads, but is orders of magnitude larger than snow or dependent on ground temperature, the impact of factors rainfall generated runoff. Lake drainage would have major controlling ground temperature and therefore cracking are implications to downstream roads or pipelines. It is not well known. These could include winter air important to know if this risk will decrease across the TP, temperature and snow depth, with snow depth being as it has done over the past 65 years, or will it increase controlled by snowfall, blowing snow and changes in with a warming climate

69 4) REFERENCES Mackay, J.R. 1988. Catastrophic lake drainage, Tuktoyaktuk Peninsula area, District of Mackenzie. In Brewer MC, Carter LD, Glenn R. 1993. Sudden drainage Current Research, Part D. Geological Survey of of a thaw lake on the Alaskan Arctic coastal plain. Canada, Ottawa, ON. Paper 88-1D: 83-90. Sixth International Conference on Permafrost, Beijing, Mackay, J.R. 1992. Lake stability in an ice-rich permafrost China. Proceedings Vol. 1. South China University of environment: examples from the western Arctic coast. Technology Press: Guangzhou, China; 48Ð53. In Arctic ecosystems in semi-arid regions: Implications Burn, C. R. (2002). Tundra lakes and permafrost, for resource management. Robarts R.D., Bothwell Richards Island, western Arctic coast, Canada. M.L. (eds). Environment Canada: Saskatoon, Sask.; Canadian Journal of Earth Sciences, 39(8), 1281Ð NHRI Symposium Series 7, 1-25. 1298. Marsh, P. and Hey, M. (1989). The flooding hydrology of Emmerton, C. A., Lesack, L. F. W., Marsh, P., L.F.W. Mackenzie Delta lakes near Inuvik, N. W. T., Canada. Lesack, and P. Marsh. (2007). Lake abundance, Arctic, 42(1), 41Ð49. potential water storage, and habitat distribution in the Marsh, P., Lesack, L. F. W., and Roberts, A. (1999). Lake Mackenzie River Delta, western Canadian Arctic. sedimentation in the Mackenzie Delta, NWT. Water Resources Research, 43, doi:10.1029/ Hydrological Processes, 13(16), 2519Ð2536. 2006WRoo5139. doi:10.1002/(SICI)1099- Fortier, D., Allard, M., & Shur, Y. (2007). Observation of 1085(199911)13:16<2519::AID-HYP935>3.0.CO;2-T rapid drainage system development by thermal Marsh, P., and Neumann, N. N. (2001). Processes erosion of ice wedges on Bylot Island, Canadian Arctic controlling the rapid drainage of two ice-rich Archipelago. Permafrost and Periglacial Processes, permafrost-dammed lakes in NW Canada. 18(3), 229Ð243. doi:10.1002/ppp.595 Hydrological Processes, 15(18), 3433Ð3446. Frohn, R. C., K.M. Hinkel, and W.R. Eisner. 2005. doi:10.1002/hyp.1035 Satellite remote sensing classification of thaw lakes Marsh, P., M. Russell, C. Onclin, & H. Haywood. (2008). and drained thaw lake basins on the North Slope of Modelling discharge during the rapid drainage of thaw Alaska. Remote Sensing of Environment. 97, 116-126. lakes in the Western Canadian Arctic. In D. L. Kane & Godin, E., Fortier, D., & Burn, C. R. (2012). K.M. Hinkel (Eds.), Ninth International Conference on Geomorphology of a thermo-erosion gully, Bylot Permafrost (Vol. Vol 2, pp. 1143Ð1147). Fairbanks, Island, Nunavut, Canada. Canadian Journal of Earth Alaska: University of Alaska Fairbanks. Sciences, 49(8), 979Ð986. doi:10.1139/e2012-015 Marsh, P., Russell, M., Pohl, S., Haywood, H., & Onclin, Hopkinson, C., Crasto, N., Marsh, P., Forbes, D., & C. (2009). Changes in thaw lake drainage in the Lesack, L. (2011). Investigating the spatial distribution Western Canadian Arctic from 1950 to 2000. of water levels in the Mackenzie Delta using airborne Hydrological Processes, 23(17), 145Ð158. LiDAR. Hydrological Processes, 3011(June), 2995Ð doi:10.1002/hyp.7179 3011. doi:10.1002/hyp.8167 Nafziger, J., F. Hicks, R. Andrishak, P. Marsh, & L. Kokelj, S., T. Lantz, S. Wolfe, J. Kanigan, P. Morse, R. Lesack. (2009). Hyraulic model of river flow and Coutts, M. Molina-Giraldo and C. Burn. 2014. storage effects in the Mackenzie Delta, Canada. In Distribution and activity of ice wedges across the 17th International Northern Research Basins forest-tundra transition, western Arctic Canada. Symposium and Workshop (pp. 237Ð248). Iqualuit- Journal of Geophysical Research: Earth Surface. Pangnirtung-Kuujjuaq, Canada. 10.1002/2014JF003085 Pohl, S., P. Marsh, C. Onclin, & M. Russell. (2009). The Lesack, L. F. W., and Marsh, P. (2007). Lengthening plus summer hydrology of a small upland tundra thaw lake: shortening of river-to-lake connection times in the implications to lake drainage. Hydrological Processes, Mackenzie River Delta respectively via two global 10.1002/hy. change mechanisms along the arctic coast. Pollard WH, French HM. 1980. A first approximation of the Geophysical Research Letters, 34. doi:10.1029/ volume of ground ice, Richards Island, Pleistocene 2007GL031656 Mackenzie delta, Northwest Territories, Canada. Lesack, L., Marsh, P., Forbes, D., & Hicks, F. (2013). Canadian Geotechnical Journal 17: 509Ð516. Timing, duration, and magnitude of peak annual water Smith LC, Sheng Y, MacDonald GM, Hinzman LD. 2005. levels during ice breakup in the Mackenzie Delta and Disappearing Arctic Lakes. Science 308: 1429. the role of river discharge. Water Resour. Res, 1Ð16. Smith LC, Sheng Y, MacDonald GM. 2007. A first pan- doi:10.1002/2012WR013198 arctic assessment of the influence of glaciation, Mackay, J.R. 1963. The Mackenzie delta area, N.W.T. permafrost, topography and peatlands on northern Geographical Branch, Memoir 8, Department of Mines hemisphere lake distribution. Permafrost and and Technical Surveys, Ottawa. 202 p. Periglacial Processes 18: 201Ð208. Mackay, J.R. 1979. Pingos of the Tuktoyaktuk Peninsula Squires, M. M., Lesack, L. F. W., Hecky, R. E., Guildford, area, Northwest Territories. Géographie physique et S. J., Ramlal, P., & Higgins, S. N. (2009). Primary Quaternaire, 23: 3-61. Production and Carbon Dioxide Metabolic Balance of a Mackay, J.R. 1981. An experiment in lake drainage, Lake-Rich Arctic River Floodplain: Partitioning of western Arctic coast. In Current Research, Part A. Phytoplankton, Epipelon, Macrophyte, and Epiphyton Geological Survey of Canada, Ottawa, ON. Paper 81- Production Among Lakes on the Mackenzie Delta. 1A: 63-68

70 Ecosystems, 12(5), 853Ð872. doi:10.1007/s10021- 009-9263-3 Tank, S. E., Lesack, L. F. W., & Hesslein, R. H. (2008). Northern Delta Lakes as Summertime CO2 Absorbers Within the Arctic Landscape. Ecosystems, 12(1), 144Ð 157. doi:10.1007/s10021-008-9213-5 Yoshikawa, K. and L. Hinzman. 2003. Shrinking thermokarst ponds and groundwater dynamics in discontinuous permafrost near council, Alaska. Permafrost and Periglacial Processes. 14, 2, 151-160. Van Huissteden, J., Berrittella, C., Parmentier, F. J. W., Mi, Y., Maximov, T. C., & Dolman, A. J. (2011). Methane emissions from permafrost thaw lakes limited by lake drainage. Nature Climate Change, 1(2), 119Ð 123. doi:10.1038/nclimate1101

71

72 Meteorological and geological influences on icing dynamics in subarctic Northwest Territories, Canada

P.D. Morse1 & S.A. Wolfe1,2 1Geological Survey of Canada, Natural Resources Canada, Ottawa, Ontario, Canada 2Department of Geography and Environmental StudiesÐ Carleton University, Ottawa, Ontario, Canada

ABSTRACT In this study, we test the significance of identified meteorological forcing variables against a long-term dataset of icing dynamics and distribution developed for the Yellowknife region, Northwest Territories. Overall, 28% of icing density interannual variation is explained by winter warming periods (≥5¡C) and autumn rainfall. Interannual icing density variation and significant meteorological forcing variables differ among ecoregions where varied geological permafrost conditions influence the hydrological regime.

RÉSUMÉ Dans cette étude, un ensemble de données à long terme qui représente la distribution et la dynamique de formation de couches de glace dans la région près de Yellowknife, Territoires du Nord-Ouest, a été utilisé pour tester l’influence de variables de conditionnement météorologique préalablement identifiées. En général, 28% de la variation interannuelle de la densité de couches de glace est attribuée aux réchauffement hivernaux (>5¡C) et aux précipitations automnales. Les variations interannuelles de la densité des couches de glace et les variables significatives de conditionnement météorologique varient d’une écorégion à l’autre, où différents contextes géologiques et de pergélisol influencent le régime hydrologique.

1 INTRODUCTION

Icings (a.k.a. aufeis, naled) are sheet-like masses of layered ice that form during the winter by freezing of successive flows of water on the ground surface, or on top of river or lake ice (ACGR 1988). Common throughout the northern hemisphere (Yoshikawa et al. 2007), icing development has been linked to periodic increases in winter air temperature (winter warming periods) (Kane 1981), high autumn precipitation (Hall and Roswell 1981), and low early-winter snow fall (Carey 1973). Icings are a geohazard, negatively impacting the performance of infrastructure where overflow encroaches on roadways or buildings, or damaging infrastructure if extensive springtime flooding is induced (Carey 1973). Controlled by local and regional factors, icings are spatially recurrent, but not annually, nor to the same extent. Thus, long-term datasets of icing dynamics and distribution are required to improve process studies and identify potential terrain hazards. In this paper we assess how identified local and regional factors affect long-term icing dynamics in the subarctic. We hypothesize that icing area varies over time according to meteorological forcing variables, with spatial Figure 1. Ecoregions within the Northwest Territories variation due to hydrological setting moderated by study area showing shaded relief. Figure 2 image extent geological conditions. We test this hypothesis by is outlined. examining the influence of autumn rainfall, early-winter snowfall and air temperature, total winter snowfall, and frequency of sporadic winter warming periods on 2 BACKGROUND interannual variation of icing area. Divided into four ecologically and geologically distinct ecoregions (Fig. 1) 2.1 Icings (ECG 2007; 2008), hydrological conditions in the study area are representative of much of the subarctic. Icing occurrence and process dynamics are influenced by climate, hydrology, geology, permafrost, and topography

73 (Carey 1973; Kane 1981). Conditions most favourable for outcrops and in bedrock fractures (Kerr and Wilson 2000), icing formation are severe continental climate, cold where wetlands are common. Elevations range from winters with low snow cover, the presence of groundwater about 160 to 200 m asl, with the northern boundary springs or unfrozen, water-bearing taliks, and highly coinciding with the limit of extensive lacustrine sediments permeable materials such as sand or gravel deposits, (Stevens et al. 2012). Permafrost, associated with those organic matter, or karst terrain (Romanovskii et al. 1996). sediments, is more extensive in LHB than the other Interannual variation of icing activity reportedly relates to ecoregions (Morse et al. in press). A wide array of forest the autumn rainfall (Carey 1973; Hall and Roswell 1981), types occur, ranging from black spruce (Picea mariana) and the extent of icing development may vary inversely dominant to pure paper birch stands (Betula papyrifera). with snow cover (Carey 1973; Yoshikawa et al. 1999). Jack pine and aspen (Populus tremuloides) forests occur Overflow events are driven by a rise in hydrostatic on well drained areas. head associated with winter warming periods (Kane In contrast, UHB (8215 km2) is a gently southwest- 1981). Yoshikawa et al. (1999) noted the frequency of sloping Precambrian sedimentary bedrock plain with few winter warming periods of 15¡C d-1 seemed related to local promontories, and is dominantly overlain with thin icing development at Caribou-Poker Creeks Research discontinuous till veneers and scattered outwash deposits. Watershed, AK, where Kane (1981) identified the pore Glacial Lake McConnell reached a maximum elevation of water pressureÐtemperature relation. The process about 290 m asl, covering most of UHB, where elevations underpinning this relation is unknown, but repeated range from at least 200 to 300 m asl (Kerr and Wilson overflow events throughout the winter increase the ice 2000). Consequently, thin, discontinuous deposits of thickness, and by the end of the cold season the channel wave-washed tills, glaciolacustrine sediments, and and adjacent floodplain may fill completely with ice. glaciofluvial materials occur in rock fractures and between outcrops, where the substrate is sufficiently thick for 2.2 Study Area discontinuous black spruce and jack pine forests. More rugged than the UHB, ULS (4211 km2) ranges The study area, delimited by the extent of one Landsat from gently rolling terrain, to rugged local hill systems, satellite scene (WRS-2 Path 47, Row 16; 21887 km2 land with several large lakes and rivers. Southwest sloping, area), is in subarctic Northwest Territories and includes this Precambrian granitoid bedrock upland with two broad Yellowknife (Fig. 1). The regional climate is strongly plateaus ranges in elevation from about 275 to 450 m asl. continental, with cold winters (-25.6¡C January mean Frost shattered bedrock dominate the temperature) and relatively warm summers (17.0¡C July landscape, with discontinuous till and sparse spruce and mean temperature). Nearly 76% of the annual birch forests. Locally extensive outwash deposits occur, precipitation (289 mm) falls as rain from May to but are mainly to the east and north, whereas organic November, with scant snowfall over the winter months. deposits are numerous but have a limited extent. Due to higher autumn rainfall since 1997, there has been an increase in the frequency of autumn runoff events and increased winter streamflow (Spence et al. 2011, 2014). 3 DATA AND METHODS Beginning in late-September, snow accumulates steadily throughout the winter until April when the normal month- 3.1 Icing Mapping end snow depth is 0.37 m. Snowmelt is largely complete by the end of April, with total ablation by the end of May. Morse and Wolfe (2014) developed a new semi- The study area, located in the extensive discontinuous automated remote sensing technique to map icings in the permafrost zone (Heginbottom et al. 1995), is divided by subarctic, which is summarized as follows: four ecoregions of the Great Slave geological district Icing distribution was mapped with Landsat satellite (Fig. 1) (ECG 2007, 2008), which from south to north are: image data (TM, ETM+, and OLI sensors), which are (i) the Plain High Boreal (PHB); (ii) the Lowland High advantageous because icings are clearly evident in Boreal (LHB); (iii) the Upland High Boreal (UHB); and (iv) images acquired following the snowmelt season (Fig. 2a) the Upland Low Subarctic (ULS). (Hall and Roswell 1981; Yoshikawa et al. 2007), the PHB (5418 km2) is characterized by a gentle, dome- images cover large areas and are easy to obtain shaped topography, underlain by horizontally bedded (http://glovis.usgs.gov/), and the 30-year archive enables dolomite, limestone and sandstone of Cambrian to long-term dataset generation. Late-spring images of Devonian age, with an average elevation of 200 m asl sufficiently low cloud cover were available for 24 years. Frequently burned, this dry, well drained plain, capped by Following radiometric and geometric corrections, snow washed till through the centre of the region, is forested by and ice were discriminated from soil, rock, and cloud extensive young jack pine (Pinus banksiana) stands and cover with a Normalized Difference Snow Index (NDSI) white spruce (Picea glauca) that grow adjacent to small (Hall et al. 1995) (Fig. 2b): streams and wetlands. The karst topography with numerous shallow marl ponds is unique in the Northwest NDSI = (G Ð SWIR1) / (G + SWIR1) (1) Territories (ECG 2007). LHB (4042 km2) is a nearly-level Precambrian where G and SWIR1 are green and shortwave infrared granitoid bedrock plain with extensive, discontinuous portions of the electromagnetic spectrum, respectively. wave-washed tills, glaciolacustrine and lacustrine Following Hall et al. (1995), 0.4 was selected as the deposits, and glaciofluvial materials occurring between threshold value for snow and ice (Fig. 2c); however, this

74

Figure 2. Digital images of processing steps: (a) A true colour composite of a typical late-spring Landsat image (23 May 2013) with several bright white icings visible (b) Normalized Difference Snow Index (NDSI); (c) Threshold NDSI values shown in blue; (d) Maximum Difference Ice Index (MDII); (e) Threshold MDII values shown in yellow; (f) The true color composite overlain by the total extent of icings mapped using the 24-year dataset, with the return frequency of each mapped pixel indicated, and lake and river ice removed by a black water mask. threshold value did not exclude wet marl and shallow 3.2 Assessment of Meteorological Forcing Variables marl-bottomed lakes. Consequently, ice was extracted with an exaggerated basic difference index, the new Daily average meteorological data available for Maximum Difference Ice Index (MDII) (Fig. 2d): Yellowknife airport from 1943 to 2014 (Environment Canada http://weather.gc.ca/) were compiled in order to MDII = (G2 Ð SWIR12). (2) determine the relative influence of autumn rainfall, early- winter snowfall and air temperature, and frequency of MDII threshold values of 0.144 and 0.040 discriminate ice winter warming periods on icing dynamics, and to in scenes with and without snow, respectively (Fig. 2e). determine historical trends. Autumn rainfall was the sum Following ice extraction, a water mask derived from a of September-to-October events, early-winter snowfall number of summer images was applied to remove river was the October-to-December sum, and early-winter air and lake ice, conservatively leaving only land-fast ice temperature was the October-to-December mean. (Fig. 2f). Warming period magnitudes were determined from the Icing activity was determined from the count of sum of consecutive daily increased air temperature. As instances (years) when ice was present within the the most influential magnitude of warming on icing maximum aggregate extent of ice (Fig. 2f). Icings were variation is unknown, the frequency of warming periods classified according to return frequency as infrequent (i.e. the frequency of events with a ≥5¡C total increase in (1-6), intermittent (7-18), or frequent (19-24) on the 24- air temperature) was calculated for a range of magnitudes year dataset. (1-15¡C). Forward stepwise multiple-linear regressions were used to determine the influences of meteorological forcing variables on annual icing density.

75 4 RESULTS extensive icing development at LHB also occurred in 2004 (Fig. 5). 4.1 Mapped Icings UHB has the highest icing count (1175) and total icing area of 19 km2 of the shield bedrock ecoregions, yielding A total of 5500 icings are mapped in the study area (Fig. a 0.23% density (Figure 4). Annual icing densities range 3). Satellite image date and cloud cover have little overall from a minimum 3 × 10-5 km2 km-2 in 2012 to 7 × 10-4 km2 influence on interannual icing variation (Fig. 4a). It was km-2 in 2004 (Fig. 5). Icing return frequencies are up to impossible to quantitatively assess the mapping error 88% (Fig. 4c), but 91% of icings are infrequent (Table 1). against an independent measurement of icing area, as At ULS the count (836) and area (17 km2) are less aerial photographs of the region are taken in mid-summer, than at UHB, but the icings are the most densely as are most very-high resolution satellite images that also distributed (0.40% overall density) and active in the Taiga have limited coverage. Consequently, icing mapping Shield ecoregions (Figs. 3 and 5). In stark contrast to results were assessed visually (though many of the icings other ecoregions, about 15% of the icings at ULS are mapped in 2013 and 2014 were confirmed via helicopter); intermittent, and 2.4% re-develop almost every year the results were overlaid on image data where icings are (Table 1). The average icing size is also much larger than visually distinguishable (Fig. 2a). at other ecoregions north of Great Slave Lake (Fig. 4b). Visual inspection of all scenes with snow indicated icings are underestimated at the expense of 4.3 Meteorological Forcings on Interannual Icing overestimating snow. However, snow free images with a Variation lower MDII threshold are not an overall improvement, because the increased ablation means that smaller icings Regression statistics for significant meteorological may no longer be evident and the margins of larger icings forcings on icing dynamics are shown in Table 2. Overall, have likely retreated. Because of the nature of icing winter warming periods (≥5¡C) and autumn rainfall explain dynamics and the timing of image acquisition each spring about 28% of interannual variation of icing density, with season, the relative number and area of icings is more warming periods making a somewhat greater contribution. important to this study than absolute values for Like interannual variations of icing density (Fig. 5), comparative purposes. The broad agreement between significant meteorological forcing variables are not icings mapped, observed, and visually assessed suggests coincident amongst ecoregions (Table 2). At PHB only icing dynamics were adequately captured for the purpose warming periods (≥6¡C) enter the regression, explaining intended. 14% of interannual variation of icing density. Winter warming periods (≥8¡C) are also the only significant 4.2 Icing Distribution and Dynamics forcing at LHB, but they significantly explain 24% of the interannual variation. At UHB warming periods (≥5¡C) are Variation of total icing extent (1985 to 2014) by ecoregion also dominant with autumn rainfall as a secondary is shown in Figure 3, icing dynamics are statistically determinant, together explaining 31% of icing variation. summarized in Table 1 and shown in Figure 4, and Further north at ULS, autumn rainfall and an inverse variation of annual icing extent by ecoregion is shown in 2 relation to the early-winter mean air temperature explain Figure 5. The maximum extent of icings (86 km ) indicates 34% of interannual icing density variation. they affect 0.39% of the land (Fig. 3). Interannual variation -4 2 of icing density ranges from a maximum (13 × 10 km -2 -5 2 -2 km ) in 2004 to a minimum (6 × 10 km km ) just two 5 DISCUSSION years later (Fig. 5). Discrete icings range in size from 2 2 1800 m to 4 km , (Fig. 4b). The majority (93%) of icings 5.1 Geological Influences on Icing Dynamics are infrequent and return frequency generally increases with icing size (Fig. 4c, Table 1). Icings vary substantially among ecoregions in relation to Icing distribution patterns, densities, and dynamics geological setting (Figs. 3, 4, and 5). Limestone and vary by ecoregion (Figs. 3, 4, and 5). The highest dolomite bedrock with karst topography and hydrology, cumulative total icing count (3214), density (0.88%) and beneficial to icing development (Romanovskii et al. 1996), 2 area (48 km ) of any ecoregion occurred at PHB (Fig. 3). defines PHB. Consequently, many icings there develop in The high density can be attributed mainly to extensive the vicinity of springs that feed the numerous marl ponds. icing development in 2002 and 2004 (Fig. 5). However, The flat terrain allows for potentially large icings because -5 2 -2 annual icing density ranged to as low as 1 × 10 km km overflow may not be confined to a valley. Accordingly, (Fig. 5). Icings are small but numerous (Fig. 4), with even infrequent icings may be extensive, diminishing the intermittent or frequent ice cumulatively constituting only strength of the relation between size and return 3.5% of icings and 2.3% of the total icing area. frequency, as compared to icings that develop within the At LHB, 98% of the icings are infrequent, none are Precambrian Shield (Fig. 4c). However, there are distinct frequent, and consequently LHB has the lowest differences within the Shield bedrock region. Low relief, 2 cumulative total icing area (3 km ) and total icing density thick fined-grained deposits, and extensive permafrost at (0.07%) of any ecoregion with only 275 icings mapped LHB likely limit the occurrence and discharge from (Figs. 3 and 4, Table 1). The degree of interannual springs, and thus icing activity. Icings are likely more variation is subdued compared with PHB, though the most numerous at UHB than at LHB because surficial deposits are thin and more discontinuous, and permafrost is less

76 Figure 3. Total distribution of infrequent (1-6 returns), intermittent (7-18 returns), and frequent (19-24 returns) in the Great Slave region digitally mapped from Landsat data (1985 to 2014) for Lowland High Boreal (LHB), Upland High Boreal (UHB), Upland Low Subarctic (ULS), and Plains High Boreal (PHB) ecoregions, with an inset summarizing ecoregion area, total icing area, and total icing density. Figure 2 image extent is indicated.

Table 1. Percentage of icings distributed among Infrequent (1-6), Intermittent (7-18), and Frequent (19-24) return frequency classes for Lowland High Boreal (LHB), Upland High Boreal (UHB), Upland Low Subarctic (ULS), and Plains High Boreal (PHB) ecoregions (counts indicated in brackets).

Return Frequency Overall (5500) PHB (3214) LHB (275) UHB (1175) ULS (836) Infrequent 93.3 (5131) 96.5 (3102) 97.8 (269) 91.1 (1071) 82.4 (689) Intermittent 6.2 (340) 3.2 (104) 2.2 (6) 8.8 (103) 15.2 (127) Frequent 0.5 (29) 0.3 (8) - 0.1 (1) 2.4 (20)

77

Figure 4. Icing dynamics for Lowland High Boreal (LHB), Upland High Boreal (UHB), Upland Low Subarctic (ULS), and Plains High Boreal (PHB) ecoregions: (a) Icing density versus satellite image acquisition date and percent cloud cover; (b) Cumulative discrete icing area versus discrete icing size. (c) Relations between maximum extent discrete icings (n = count) versus return frequency (m is the slope of the linear regression and r is the correlation coefficient).

Figure 5. Interannual variation of icing density, 1985 to 2014, for Lowland High Boreal (LHB), Upland High Boreal (UHB), Upland Low Subarctic (ULS), and Plains High Boreal (PHB) ecoregions, determined from Landsat satellite images.

Table 2. Beta coefficient (β), adjusted coefficient of determination (Adjusted R2), and significance (p) statistics for forward stepwise multiple-linear regression models of interannual variation of icing density forced by autumn rainfall, early winter snowfall and mean air temperature, and periodic warming events that remained in the model at the 0.05 level of significance, according to overall study region and Lowland High Boreal (LHB), Upland High Boreal (UHB), Upland Low Subarctic (ULS), and Plains High Boreal (PHB) ecoregions.

Model Components Overall PHB LHB UHB ULS Total rainfall (mm), Sep. To Oct. (β) 0.40 0.43 0.53 Total early-winter snowfall (cm), Oct. to Dec. (β) Mean air temperature (¡C), Oct. to Dec. (β) -0.44 Count of air temperature increases, >=5¡C (β) 0.54 0.56 Count of air temperature increases, >=6¡C (β) 0.42 Count of air temperature increases, >=7¡C (β) Count of air temperature increases, >=8¡C (β) 0.53 Adjusted R2 0.28 0.14 0.25 0.31 0.34 p 0.012 0.041 0.008 0.007 0.005 extensive. The configuration of basins in the exposed valleys in contrast to PHB. Compared to UHB, the shield leads to a hydrology defined by a spill-and-fill landscape at ULS is dominantly exposed, broken bedrock regime (Spence and Woo 2003), with icings confined to with plateaus covered by bouldery till veneers and sandy

78 to gravelly outwash deposits that are highly permeable, basins with larger lakes attenuate the influence of autumn which promotes icing development (Romanovskii et al. rainfall on early-winter baseflow in streams (Spence et al. 1996), and may account for the higher icing density. 2011), the spill-and-fill regime dominates the hydrology Indeed, Veillette and Thomas (1979) observed icings and the rainfall signal is still present. However, within the resulting from groundwater flowing through taliks present ULS portion of the study area there is a relatively high in coarse-grained glaciofluvial deposits in the District of concentration of large rivers that have generally more Keewatin, NT, where permafrost is continuous. winter streamflow, which is indicated by the highest icing return frequencies and icing density compared to UHB 5.2 Meteorological Influences on Icing Dynamics and LHB (Figs. 3, 4, and 5, Table 1). It may be that with high winter streamflow, relatively cold early-winter Our findings (Table 2) confirm field observations made by conditions are important because a more extensive freeze others that icing development is related to winter warming must occur to create a stable, continuous ice cover. It is periods (Kane 1981) and autumn rainfall (Hall and not apparent why winter warming periods were not Roswell 1981), but here we show that the regional air significant in this ecoregion. Perhaps with generally higher temperature regime is the dominant overall driver of icing stream flow and thus pore water pressure, overflow may dynamics, with autumn rainfall of generally less be more continuous, requiring little influence by warming importance. The most significant air temperature effect periods. Alternatively, the ice thickness may be important relates to the frequency of warming periods with a because the relation between pore water pressure and air minimum 5¡C rise in temperature. The winter of 2003- temperature becomes increasingly disassociated as ice 2004 exhibited the highest icing densities of any year (Fig. thickness increases (Kane 1981). Therefore, as the icings 5), and it had second highest count of warming periods at UHB and ULS are both confined to valleys, the larger ≥5¡C and the highest counts of warming periods ≥6, 7, or and hence thicker icings at ULS may not be affected to 8¡C. Icing counts in 2009 were also high in most regions the same degree by periodic warming. (Fig. 5), with an average of 24 warming periods (≥5¡C), It is important to note that meteorological relations but autumn 2008 was the wettest autumn with more than established in this study depend upon data from a single double the 48 mm average autumn rainfall. climate station. The climate is regionally consistent However, under a similar climate, the hydrological (Morse et al. in press), but local-scale factors that may be regime varies throughout the region, likely due to different very important to icing dynamics at a particular location geology, topography, and permafrost conditions that are not captured, such as the influence of local snow modify hydrological conditions. At PHB, the limestone and depth variation due to drifting. dolomite bedrock create a karst hydrology that is substantially different than the spill-and-fill hydrology of shield bedrock (Spence and Woo 2003). This likely 6 SUMMARY AND CONCLUSIONS eliminates the importance of autumn rainfall and reduces the influence of the air thermal regime as the hydrology is Overall, 5500 icings are mapped from 24 late-spring not as physically constrained to surface waters as with Landsat images of the Great Slave region around spill-and-fill hydrology. Autumn rainfall is also not Yellowknife. These data, generated using a new semi- significant at LHB, where the population of icings (springs) automated approach, enable the first ever analysis of is relatively low compared to the other ecoregions. This is regional icing distribution and dynamics produced in likely due to a lesser spill-and-fill hydrological regime Canada, and establish a baseline for winter hydrological resulting from past basin infilling by sediment causing variability in the region against which future icing generally low relief. There may be additional hydrological conditions, under changing climate and precipitation restrictions due to permafrost that is more extensive regimes, may be compared. These data also provide locally than at the other ecoregions, as it occurs within the important geoscience information required to assess ecozone-defining widespread lacustrine deposits (Morse terrain risks to northern transportation infrastructure, and et al. in press). may guide route selection for future transportation At UHB, winter warming periods and autumn rainfall corridors. were both significant determinants of icing dynamics. Icing size is directly related to return frequency. Over Autumn rainfall entered the model likely because this 93% of icings are infrequent, occurring 25% of the time or ecoregion is typified by the spill-and-fill hydrological less. Icing area varies substantially and among regime (Spence and Woo 2003). At UHB, compared with ecoregions because of differing geology, topography, and LHB, numerous lake basins are less infilled by sediment permafrost conditions that constrain the hydrological (Stevens et al. 2012). Because of low precipitation to regime and the extent to which overflow may spread at evaporation ratios (Spence and Rouse 2002), summer the surface. Limestone and dolomite bedrock terrain storage deficits accumulate that prevent a basin runoff promotes numerous icings via karst drainage networks. response until about 100 mm of rainfall has fallen Where bedrock is capped by thick unconsolidated between mid-July and the end of October (Spence et al. deposits with permafrost, springs are likely limited and 2011, 2014). Thus, autumn rainfall is an important driver icings are uncommon. The greatest density of icings in the of early-winter baseflow and therefore icing dynamics. study region north of Great Slave Lake occurs where the Further north at ULS, interannual variation of icing extensively exposed granitoid Precambrian Shield density is related directly to autumn rainfall and inversely bedrock is highly fractured (and may be permafrost free), to the early-winter mean air temperature. Though larger which may permit relatively high hydraulic connectivity

79 between numerous recharge and discharge zones. Natural Resources Canada, Geological Survey of Icing area and count, which varied considerably from Canada, Ottawa, ON. year to year, is related to the frequency of winter warming Morse, P.D. and Wolfe S.A. 2014. Icings in the Great periods and the amount of autumn rainfall preceding Slave Region (1985Ð2014), Northwest Territories, freeze-up. However, icing dynamics and significant Mapped from Landsat Imagery, Geological Survey of meteorological forcing variables are not coincident among Canada, Open File 7720, Natural Resources Canada, ecoregions due to differences in hydrological regimes Geological Survey of Canada Ottawa, ON. among the ecoregions. Morse P.D., Wolfe S.A., Kokelj S.V. and Gaanderse A.J.R. In press. The occurrence and thermal disequilibrium state of permafrost in forest ecotopes of ACKNOWLEDGEMENTS the Great Slave region, Northwest Territories, Canada. Permafrost and Periglacial Processes. Derived annual icing data are available in Morse and Romanovskii, N.N., Gravi, G.F., Melniko, E.S. and Wolfe (2014). This work was conducted under the Great Leibman. M.O. 1996. Periglacial processes as Slave Ð TRACS (Transportation Risk in the Arctic to geoindicators in the cryolithozone, in Geoindicators: Climate Sensitivity) program of the Geological Survey of Assessing Rapid Environmental Changes in Earth Canada, Natural Resources Canada. Thanks to F. Systems, Berger, A.R. and Iams, W.J. (eds.), Hassan for downloading Landsat data. This is ESS Balkema, Rotterdam, Netherlands: 47-67. contribution #20150062. Spence, C. and Rouse, W.R. 2002. The energy budget of subarctic Canadian Shield subarctic terrain and its impact on hillslope hydrological processes, Journal of REFERENCES Hydrometeorology, 3: 208-218 Spence, C. and Woo, M.-K. 2003. Hydrology of subarctic Associate Committee on Geotechnical Research (ACGR). Canadian Shield: soil filled valleys, Journal of 1988. Canadian Glossary of Permafrost and Related Hydrology, 279: 151Ð166. Ground-Ice Terms, Technical Memorandum No. 142, Spence, C., Kokelj, S.V. and Ehsanzadeh, E. 2011. National Research Council of Canada, Ottawa, ON. Precipitation trends contribute to streamflow regime Carey, K.L. 1973. Icings Developed from Surface Water shifts in northern Canada, in Cold Region Hydrology in and Ground Water, Cold Regions Science and a Changing Climate, Proceedings of symposium H02 Engineering Monograph 111-D3, U.S. Army Cold held at Melbourne, Australia, July 2011, Publication Regions Research and Engineering Laboratory, 346, Yang, D., Marsh, P., and Gelfan, A. (eds.), Hanover, NH, USA. International Association of Hydrological Sciences, Ecosystem Classification Group (ECG). 2007. Ecological Oxfordshire, UK: 3-8. Regions for the Northwest Territories Ð Taiga Plains, Spence, C., Kokelj, S.A., Kokel, S.V. and Hedstrom, H. Department of Environment and Natural Resources, 2014. The process of winter streamflow generation in Government of the Northwest Territories, Yellowknife, a subarctic Precambrian Shield catchment, NT. Hydrological Processes, 28: 4179-4190. Ecosystem Classification Group (ECG). 2008. Ecological Stevens, C.S., Kerr, D.E., Wolfe, S.A. and Eagles, S. Regions for the Northwest Territories Ð Taiga Shield, 2012. Predictive surficial materials and geology Department of Environment and Natural Resources, derived from LANDSAT 7, Yellowknife, NTS 85J, Government of the Northwest Territories, Yellowknife, Northwest Territories, Geological Survey of Canada, NT. Open File 7108, Natural Resources Canada, Hall, D.K. and Roswell C. 1981. The origin of water Geological Survey of Canada, Ottawa, ON. feeding icings on the eastern North Slope of Alaska, Veillette, J.J. and Thomas. R.D. 1979. Icings and seepage Polar Record, 20: 433-438. in frozen glaciofluvial deposits, District of Keewatin, Hall, D.K., Riggs, G.A. and Salomonson, V.V. 1995. N.W.T., Canadian Geotechnical Journal, 16: 789-798. Development of methods for mapping global snow Yoshikawa, K., Petrone, K., Hinzman, L.D. and Bolton, cover using Moderate Resolution Imaging W.R. 1999. Aufeis development and stream baseflow Spectroradiometer (MODIS) data, Remote Sensing of hydrology in the discontinuous permafrost region, Environment, 54: 127-140. Caribou Poker Creek Research Watershed, Interior Heginbottom, J.A., Dubreuil, M.A. and Harker, P.A. 1995. Alaska, in Program and Abstracts, 50th Arctic Science Canada-Permafrost, in National Atlas of Canada, 5th Conference, American Association for the ed., MCR 4177, National Atlas Information Service, Advancement of Science, 19-22 September 1999, Natural Resources Canada, Ottawa, ON. Denali National Park and Preserve, Alaska, American Kane, D.L. 1981. Physical mechanics of aufeis growth, Association for the Advancement of Science Arctic Canadian Journal of Civil Engineering, 8: 186-195. Division, Fairbanks, AK, USA: 202-203. Kerr, D.E. and Wilson, P. 2000. Preliminary surficial Yoshikawa, K., Hinzman, L.D. and Kane, D.L. 2007. geology studies and mineral exploration Spring and aufeis (icing) hydrology in Brooks Range, considerations in the Yellowknife area, Northwest Alaska, Journal of Geophysical Research, 112: Territories, Natural Resources Canada, Geological G04S43. Survey of Canada, Current Research 2000-C3,

80 Geometry of oriented lakes in Old Crow Flats, northern Yukon

P. Roy-Léveillée École de dévelopment du nord, Université Laurentienne, Sudbury, Ontario, Canada C.R. Burn Department of Geography, Carleton University, Ottawa, Ontario, Canada

ABSTRACT Old Crow Flats is an interior basin with thousands of thermokarst lakes. These lakes have irregular shapes where they are surrounded by trees and tall shrubs that may remain rooted after bank subsidence and protect the underlying sediment from erosion. In polygonal tundra, the vegetation cover is easily removed and wave action can erode and redistribute bank sediment to form rectilinear shores. The majority of lakes with rectilinear shores are aligned parallel to dominant winds and expand most rapidly in this direction. This is contrary to the oriented lakes of the Arctic coastal plain and is due to the fine texture of glacio-lacustrine deposits in OCF, which contain very little sediment sufficiently coarse to accumulate near-shore along the leeward side of the lake, leaving the bank vulnerable to thermo-mechanical erosion caused by wave action.

RÉSUMÉ La plaine d’Old Crow est un basin intérieur parsemé de milliers de lacs thermokarstiques. Ces lacs sont de forme irrégulière lorsqu’ils sont entourés de fardoche et d’arbres qui peuvent rester enracinés malgré l’affaissement des berges, et qui empêchent ainsi l’érosion des sédiments sous-jacents. Dans la toundra, où le couvert végétal est facilement rompu, les vagues érodent et redistribuent les sédiments pour former des rivages rectilignes. La plupart de ces lacs sont parallèles aux vents dominants et ont une croissance accélérée dans cette direction, ce qui est contraire à la configuration des lacs de la plaine côtière de l’Arctique. Cette différence est due à la granulométrie fine des dépôts glacio-lacustres de la plaine d’Old Crow: très peu de sédiments sont suffisamment grossiers pour s’accumuler près des berges exposées au vent, laissant ces dernières vulnérables à l’action thermo-mecanique érosive des vagues.

1 INTRODUCTION results in the development of an elongated form with sediment accumulation near the centre of the leeward Field evidence indicates that the orientation and shape of shore and maximum sediment transport near the ends of thaw lakes near the western North American Arctic coast the lake, where shoreline curvature is often accentuated reflect the effects of wave action and wind-generated (Rex 1961). Carson and Hussey (1962) provided field currents on shore erosion and sediment transport observations supporting this model, and indicated that (Mackay 1956; Rex 1961; Carson and Hussey 1962; sediment accumulation on the leeward side of the lake Mackay 1963; Côté and Burn 2002). Oriented thaw lakes protects the shore from wave action and is the key are also found in interior basins, such as Old Crow Flats process for the initiation of lake elongation perpendicular (OCF), northern Yukon (Roy-Leveillee and Burn 2010), to the prevailing wind direction. They also suggested that where little is known of the factors controlling their expansion perpendicular to the prevailing winds is development and morphometry (Mackay 1956; Côté and accelerated in large lakes due to wind-driven circulation Burn 2002). In this paper we discuss the distribution and cells that create currents of sufficient strength to erode the characteristics of oriented lakes in OCF. We examine lake ends. Recent remotely sensed images of gyres in variations in lakeshore geometry within the Flats and use oriented lakes of the Alaskan Arctic coastal plain support a combination of field observations and aerial this circulation model, and confirm that the development photographs to discuss relations between patterns of lake of such currents increases with lake size and wind velocity expansion and lake orientation. (Zhan et al. 2014). Oriented lakes can also develop parallel to the 1.1 Background dominant wind direction, as observed in parts of the Lena River delta and in OCF (Morgenstern et al. 2011; Roy- Along the western North American Arctic coast, clusters of Leveillee and Burn 2010), but information regarding the thermokarst lakes are oriented perpendicular to the development of lakes with such an orientation is scarce. dominant wind direction (Carson and Hussey 1962; Mackay 1963). This orientation has been attributed to 1.2 Old Crow Flats longshore sediment drifting and wind-driven circulation patterns. Rex (1961) used hydrodynamic principles to OCF (Fig. 1) is a 5600 km2 wetland containing thousands show that longshore sediment drifting is greatest where of thermokarst lakes and ponds. It is within the forest- the angle between the wave orthogonals and the normal tundra ecotone of northern Yukon, and is in the to the shoreline is approximately 50¼, and least where continuous permafrost zone. The vegetation cover is a waves are parallel to shore. A prevailing wind direction heterogeneous mosaic of woodlands, tall shrubs, low

81

Figure 1 Location map of Old Crow Flats showing (a) northern Yukon with extent of Glacial Lake Old Crow in Bell, Bluefish, and Old Crow basins. The approximate maximum limit of the Laurentide Ice Sheet is shown in light grey (after Zazula et al., 2004, Fig. 1); and (b) a generalized map of land cover structure in OCF where lakes are in black, low shrubs, herbaceous vegetation, bryophytes and barren ground are in beige, and tall shrubs, woodlands and coniferous forest are in green (modified from Turner et al 2014, Fig. 4) shrubs, and herbaceous communities (Turner et al. 2014). OCF was not glaciated during the Wisconsinan Stage but was submerged beneath a 13,000 km2 glacial lake that drained catastrophically 15,000 years ago (Fig. 1a) (Zazula et al. 2004). The glaciolacustrine silts and clays are blanketed with peat. River-bank exposures indicate that excess ice in the upper 40 m of the ground is limited to the glaciolacustrine sediments, which are up to 9 m thick (Matthews et al. 1990). Permafrost temperatures at the depth of zero annual amplitude vary between -5.1¼C and -2.6¼C, depending primarily on snow cover (Roy- Léveillée et al. 2014). The mean annual air temperature at Old Crow, the nearest community, is -8.3¼C (Environment Canada climate data are available at http://climate.weather.gc.ca/, accessed on May 23rd, 2015). The Old Crow wind record is short and incomplete, but wind speed and direction recorded during 1996 – 2014 indicate that winds are primarily from the NE during the open water season, which extends from June to October. Roy-Léveillée and Burn (2010) found a similar wind distribution in a tundra Figure 2. Frequency distribution of wind speed and area near Johnson Creek in June to August 2008-09, with direction during the 2008-09 open water seasons in OCF, 68% of winds over 4 m/s from the NE and ENE (Fig. 2). modified from Roy-Leveille and Burn (2010) and adjusted The lakes of OCF lack littoral shelves and are shallow, to represent windspeed 10 m above the ground following with a mean depth of 1 to 1.5 m. They exhibit key features Resio et al. (2002).

82

Figure 3. Lake geometry and shore erosion in tundra and taiga areas. a) Oriented lakes with rectilinear shorelines in an areas where the vegetation cover is dominated by low shrubs and grasses; b) shore section with overhanging peat curtains and thermo-erosional niche (bank height = 3 m); c) lakes with irregular shorelines in an area where the vegetation cover is dominated by taiga; d) shore section protected from thermo-mechanical erosion by partly submerged trees and tall shrubs (bank height = 2 m). of the thermokarst lake cycle, such as lake expansion by 2 METHODOLOGY thawing of ground ice and catastrophic drainage, followed by permafrost recovery and lake re-initiation. Drained lake 2.1 Field conditions at thermokarst lakeshores basins are abundant throughout the Flats, commonly with deeply incised outlets. In parts of OCF, the lakes and The distribution of lakes with irregular and rectilinear drained basins have strikingly rectilinear shorelines shores was examined on a map of OCF with land cover whereas in other parts lakes tend to have irregular grouped into tundra and taiga (Turner et al. 2014). shapes. Morrell and Dietrich (1993) suggested that the Several examples of lakes with irregular and rectilinear orientation and morphology of OCF lakes may be shores were examined in the field, and bank controlled by underlying geology, but field observations of characteristics were described qualitatively and rapidly receding rectilinear shores suggest that photographed. geometrical control by static underlying features is unlikely Lake-bottom sediment samples were collected in a (Roy-Leveillee and Burn 2010). lake with rectilinear shores at distances of 0, 5, 10, and 20 In order to discuss controls on lake geometry and m from a SW-facing shore with a 1600 m fetch. Twelve orientation in OCF we (1) discuss field observations of samples from the top 2 m of the surrounding area were shoreline conditions associated with different lake used to represent the texture of bank sediment. Sediment geometries; (2) compare the distribution of lake texture between 0.4 and 2000 μm was determined using a orientation to wind patterns during the open water season; Beckman Coulter LS 13 320 laser diffraction analyser and (3) investigate patterns of shore recession in lakes of (Neville et al. 2014). The samples were loaded into the different geometries, orientations, and sizes. machine until an obscuration level of 10 ±3% was reached and statistics were computed using the Fraunhofer

83 diffraction model (Murray 2002; Neville et al. 2014). The polygonal tundra were commonly remnant ponds within mean particle size distributions for lake-bottom samples drained lake basins. Water bodies of ≤0.01 km2 generally and for shore bank samples were used to build had irregular shorelines, particularly where they formed histograms. The littoral cut-off diameter represents the and expanded via degradation of ice wedges. In areas upper limit of particle sizes that are removed from the dominated by tall shrubs and taiga, small lakes had littoral zone by wave action (Limber et al. 2008). It was smoother shores. estimated as D10, the grain size for which 90% of a sample is coarser and 10% is finer (Limber et al. 2008). 3.1.2 Sediment texture

2.2 Lake orientation and shore recession The sand fraction in near shore sediment was more than twenty times that in the bank sediment, and the clay Lake orientation was determined using the ArcGIS fraction in near-shore sediment was reduced to 2% from bounding containers toolbox (http://arcscripts.esri.com/ 14% found in the banks. The mean littoral cut-off diameter details.asp?dbid=14535) to provide a minimum area was 42 μm. Less than 4% of the mineral fraction in shore bounding rectangle and long axis azimuth for all lakes and bank sediment was coarser than 42 μm (Fig. 4). ponds of OCF in the CanVec digital topographic dataset of the National Topographic System, and for two subsamples of 230 lakes each: lakes that were clearly rectangular or triangular in the first subsample, and lakes that did not have rectilinear shores, thus deemed ‘irregular’, in the second sample. Shore recession between 1951 and 1996 was determined for a subsample of 20 lakes using aerial photographs taken in 1951 and 1996. The images were superimposed and co-registered using ice-wedge networks around the lakes. Lakes shores were traced and total area of land eroded was calculated for each lake as the difference between the 1996 and 1951 lake polygons. Mean erosion rate was estimated by dividing the total area of land eroded by the perimeter of the 1951 lake polygon. Shore recession in specific locations was calculated along a normal to the 1951 shoreline. Where a strip of land was eroded due to shore recession on two sides (e.g. during the complete erosion of an island or the merging of lakes), recession rate was calculated based on the width of land eroded divided by two.

Figure 4. Mean particle-size distribution in nearshore lake- 3 RESULTS bottom sediment samples and bank samples. The littoral

cut-off diameter, D , is shown with a vertical dashed line 3.1 Field observations of shore conditions 10 (Limber et al. 2008). The portion of the bank sediment

distribution that is coarser than D is shaded. 3.1.1 Rectilinear and irregular shorelines 10

3.2 Lake orientation In OCF, lakes with rectilinear shores are generally in polygonal tundra, where the vegetation cover is In OCF, the proportion of oriented lakes increases with dominated by low shrubs, grassy tussocks, and mosses 2 lake size (Fig. 5a, b, c, d). Waterbodies ≤0.01 km have (Fig. 1b and 3a). When examined, the vegetation cover 2 no dominant orientation, but those between 0.01 km and was often ruptured at the top of the shore bank (Fig. 3b), 2 1 km are more often oriented NE-SW and ENE-WSW ripped along the bank slope, or peeled back by the action (Fig. 4a and b), i.e., parallel to dominant winds (Fig. 2). of ice push, exposing the underlying unconsolidated 2 Beyond 1 km the proportion of lakes oriented sediment. The few beaches along oriented lakeshores perpendicular to dominant winds increases and the were limited to shore sections sheltered from wave action distribution of lake orientations progressively becomes or formed temporarily along leeward shores during calm 2 bimodal. Lakes larger than 10 km are almost exclusively periods. oriented perpendicular to prevailing winds (Fig. 5d). Only Lakes with irregular shapes were generally expanding 13 water bodies fall in the latter size category, but they in areas where patches of trees and tall shrubs dominated represent over 20% of the total lake area in OCF, making the landscape (Fig. 1b and 3c, d). Standing trees and tall them, and their orientation, noticeable features of the shrubs rooted in submerged ground were found along the Flats, as noted by Mackay (1956). Rectangular and shores. Live trees and shrubs were found closer to shore triangular lakes are aligned either parallel or perpendicular banks and some dead shrubs were occasionally present to dominant winds in 90% of the cases (Fig. 5e). Lakes at the lakeward edge of the submerged vegetation. Water and ponds with irregular shapes do not have a dominant bodies with irregular shorelines in areas dominated by 2 orientation overall (Fig. 5f), but lakes >1 km are

84 Figure 5. Distribution of lake and pond orientation for different size classes in OCF. aligned parallel or perpendicular to dominant winds in More specifically, two conditions must be fulfilled: 1) the approximately 60% of the cases, 10% more frequently thawing of lakeshore banks must yield unconsolidated than if lakes orientations were evenly distributed. The sediment for transport in the littoral zone, which is lakes of OCF are surrounded with paleoshorelines and facilitated if wave action, ice push, or mass-wasting drained basins outlines. Near Johnson Creek (Fig. 1b), processes result in the destruction of the vegetation cover where oriented lakes are abundant, numerous drained to expose underlying sediment to erosion; and 2) long- lake basin outlines are clearly visible. Similar to lakes, shore sediment drifting must lead to sufficient sediment small basins are oriented parallel to prevailing winds and accumulation along leeward shores to impede thermo- larger basins perpendicular. Triangular basins, similar to erosional processes associated with direct contact triangular lakes, are oriented perpendicular to prevailing between waves and the shore bank. These two conditions winds and are at least 5 km2 in area. can be used to explain characteristics of lake geometry and orientation in OCF. 3.3 Shore erosion and lake expansion 4.1 Rectilinear and irregular shorelines During 1951-1996 lakes with rectilinear shores, including large lakes oriented perpendicular to prevailing winds, Rectangular and triangular lakes are found in parts of expanded most rapidly in a NE-SW direction (Fig. 6a, b, c, OCF dominated by low shrub polygonal tundra. There, e). Rapidly eroding shore sections were irregular at the shore-bank vegetation cover is easily broken and local scale (10 to 100 m), largely due to differences in removed by ice push and wave action, exposing erosion rates between ice wedges and polygon centres, unconsolidated sediment for erosion and transport. The but appeared smooth when considered at a larger scale. combination of zones of erosion and accumulation along For this reason, water bodies smaller than 0.01 km2 rectilinear shores indicates that ice push, solar radiation, generally did not have rectilinear shores in polygonal or other processes pertaining only to shore erosion cannot tundra (Fig. 6g). However, water bodies larger than 0.01 be the fundamental mechanisms controlling lake km2 developed increasingly rectilinear shores as they geometry (Mackay 1963). Consistently with descriptions expanded (Fig. 6e). Lakes with irregular shores had no for lakes of the western North American Arctic coastal clear directional trend for expansion, but shore erosion plain, the rectilinear shores of oriented lakes in OCF proceeded fastest for islands and peninsulas, causing appear to result from differences in rates of sediment lakes to become less irregular as they expanded (Fig. 6d). removal and accumulation along shore banks. Prominent shore features are exposed to more aggressive wave 4 DISCUSSION action whereas bays are sheltered and may accumulate sediment, resulting in a natural evening of the shoreline. In order for sediment drifting to lead to the genesis of In parts of OCF where taiga and tall shrubs dominate oriented thaw lakes as described for the North American the landscape, vegetation can remain anchored in the Arctic coastal plain (Rex 1961; Carson and Hussey 1962), sediment after subsidence of the shore banks beneath unconsolidated sediment must be available for water level, and form a barrier protecting the shore from redistribution and accumulation along the leeward shores. wave action and ice push (Fig. 3c and d). With limited

85 removal of slumped sediment and no thermo-erosional action at the bank foot, heat conduction from the lake into the bank becomes the dominant process for permafrost thaw and lake expansion Sediment redistribution along the shore of these lakes is impeded by the persisting vegetation cover, resulting in irregular lake shapes (Fig. 3b and d). Along shore sections where the vegetation barrier is absent, thin, or exposed to very aggressive wave action, processes similar to those prevailing in polygonal tundra affect the shore bank. This distribution of lakes with rectilinear and irregular shores within a forest-tundra ecotone is consistent with observations of thermokarst lake geometry north and south of treeline. The clusters of oval, ellipsoid, triangular, rectangular, and heart shaped oriented thaw lakes of the Alaskan Arctic coast, the Siberian north coast, and the Canadian western Arctic coast are limited to tundra environments (Carson and Hussey 1962; Mackay 1963; Morgenstern et al. 2008) whereas thermokarst lakes expanding in unconsolidated sediment south of treeline have irregular shorelines (e.g. Burn and Smith 1990; Marsh et al. 2009).

4.2 Orientation of lakes with rectilinear shores

4.2.1 Parallel to prevailing winds

The increase in the proportion of oriented lakes with increasing lake size indicates that the processes controlling lake orientation are associated with shore erosion and lake growth. The orientation and expansion of tundra lakes parallel to prevailing winds in OCF is contrary to reports on lakes of similar size on the Alaskan Arctic coastal plain and the Tuktoyaktuk Peninsula, where oriented lakes generally have their long axis approximately perpendicular to the prevailing summer winds (Mackay 1956). However, Rex (1961) noted the importance of an abundant, sandy sediment input to allow sediment accumulation in the leeward littoral zone and, in OCF, sediment input to the littoral zone from eroding lake banks includes only a small fraction of fine sand (Fig. 4). When comparing the texture of lake-bottom sediment near a SW-facing shore to that of bank sediment, the majority of the input sediment is smaller than the littoral cut-off diameter and is apparently removed from the littoral zone by wave action (Limber et al. 2008). This leaves only a very small fraction of sediment available for longshore drifting and accumulation in the littoral zone. Hence, the textural characteristics of the sediment input impede Figure 6. Shore recession patterns between 1951 (dotted accumulation in the near-shore zone and allows waves to line) and 1996 (solid line) in lakes of different sizes, reach the foot of the shore bank. Contact between waves orientation, and geometry. Lakes a), b), c), e), and g) are and the bank foot accelerates erosion by mechanically from areas dominated by low shrub tundra where lakes removing slumped sediment and preventing a thawing and basins generally are oriented and have rectilinear bank from stabilizing. Where there is contact between shores; lakes d), f), and h) are from areas dominated by water and permafrost, heat transfer into the bank is tall shrubs and taiga, where lakes generally had irregular greater than through accumulated sediment. We suggest shapes. The location where the maximum rate of erosion that this accelerated erosion of the leeward shores is was measured is marked with a black triangle. Lake area responsible for the NE-SW orientation of the majority of in 1996, total land loss to erosion and mean and lakes with rectilinear shores in OCF. maximum erosion rates during 1945-96, are indicated for each lake. *Lake area in 1945 is indicated for lake b).

86 4.2.2 Perpendicular to prevailing winds plain and the glacio-lacustrine deposits of OCF, result in contrasting responses to the dominant wind direction. Large lakes and drained basins, particularly those that are >10 km2 are almost exclusively oriented perpendicular to 6 ACKNOWLEDGEMENTS the dominant wind directions. Some of these lakes and basins have a triangular rather than a rectangular shape, The research was supported by the Government of similar to lakes of the Tuktoyaktuk Peninsula. The recent Canada International Polar Year program, the National expansion of these large lakes in a direction opposite to Science and Engineering Research Council of Canada, their orientation (Fig. 6a and b), suggests that they are the Polar Continental Shelf Program, Natural Resources likely not in equilibrium with current conditions. The Canada, and the Northern Scientific Training Program, expansion pattern observed in 1951-1996 could not be Aboriginal Affairs and Northern Development Canada. sustained for thousands of years without resulting in a Essential logistical support was provided by the Vuntut NE-SW orientation. Little information is available on past Gwitchin First Nation Government and the Aurora wind patterns in the area, but Lauriol et al. (2002) Research Institute, Inuvik. We thank T. Patterson and M. examined cliff-top aeolian deposits and report signs of Pisaric for use of laboratory equipment. Several field vigorous summer winds from the SW for several thousand assistants contributed to data collection including A. J. years after drainage of Glacial Lake Old Crow. There is Jarvo, A.L. Frost, B. Brown, D. Charlie, E. Tizya-Tramm, insufficient information to resolve the cause of the NW-SE K. Tetlichi, L. Nagwan, M. Frost Jr., S. Njoutli, S. Frost, orientation of the lakes, but past increases in the intensity and C.Z. Braul. of summer winds may have led to the development of wind-induced circulation cells of sufficient strength to cause shore erosion and lake elongation perpendicular to 7 REFERENCES prevailing winds, as observed on the Alaskan coastal Burn CR, Smith MW (1990) Development of thermokarst plain (Carson and Hussey 1962). Near Barrow and lakes during the Holocene at sites near Mayo, Yukon Tuktoyaktuk, winds greater than 6 m/s are approximately Territory. Permafrost and Periglacial Processes, 1: twice as frequent as in OCF, and recent lake expansion 161–175. patterns indicate that such erosion has not dominated the Carson CE, Hussey KM (1962) The oriented lakes of larger lakes of OCF since the 1950s. Arctic Alaska. Journal of Geology, 70: 417–439. Côté MM, Burn CR (2002) The oriented lakes of 5 CONCLUSIONS Tuktoyaktuk Peninsula, Western Arctic Coast, This paper examined the distribution of lakes with Canada: a GIS-based analysis. Permafrost and irregular and rectilinear shores in relation to vegetation Periglacial Processes, 13: 61–70. structure in OCF, and discussed differences between the Lauriol B, Cabana Y, Cinq-Mars J, Geurts M-A, Grimm oriented lakes of the OCF and lakes of the western North FW (2002) Cliff-top eolian deposits and associated American Arctic coastal plain. Our main findings are that: molluscan assemblages as indicators of Late (1) In areas where trees and tall shrubs surround the Pleistocene and Holocene environments in Beringia. Quaternary International, 87: 59–79. lakes, the vegetation may remain rooted after bank subsidence, protecting the shore from erosion and Limber PW, Patsch KB, Griggs GB (2008) Coastal impeding longshore sediment transport, leading to the Sediment Budgets and the Littoral Cutoff Diameter: A development of lakes with irregular shapes; Grain Size Threshold for Quantifying Active Sediment Inputs. Journal of Coastal Research, 2: 122–133. (2) Lakes with rectilinear shores are concentrated in areas dominated by polygonal tundra, where the vegetation Mackay JR (1963) The Mackenzie Delta area, NWT. cover is easily torn by wave action or ice push to expose Department of Mines and Technical Surveys, Ottawa unconsolidated bank sediment to erosion and Mackay JR (1956) Notes on oriented lakes of the Liverpool Bay area, Northwest Territories. Revue redistribution by wave action; (3) Contrary to the western North American Arctic coastal canadienne de géographie, 10: 169–173. plain, the majority of oriented lakes in OCF are oriented Marsh P, Russell M, Pohl S, Haywood H, Onclin C (2009) parallel to prevailing winds. This orientation develops as Changes in thaw lake drainage in the western the glacio-lacustrine sediment of OCF is too fine to drift Canadian Arctic from 1950 to 2000. Hydrological Processes, 23: 145–158. and accumulate along leeward shores as described by Rex (1961). Rather, the bulk of the sediment is Matthews JV, Schweger CE, Janssens JA (1990) The last suspended and removed from the near shore zone by (Koy-Yukon) interglaciation in the northern Yukon: wave action. evidence from Unit 4 at Ch’ijee’s Bluff, Bluefish Basin. 2 Géographie physique et quaternaire, 44: 341–362. (4) Nearly all lakes > 10 km are oriented perpendicular to dominant winds. Recent patterns of shore recession for Morgenstern A, Grosse G, Günther F, Fedorova I, these lakes indicate that their orientation is not in Schirrmeister L (2011) Spatial analyses of thermokarst equilibrium with current conditions. lakes and basins in Yedoma landscapes of the Lena Delta. The Cryosphere, 5: 849 867. The oriented lakes of Old Crow Flats and those of the – North American Arctic coastal plain are both shaped and Morgenstern A, Grosse G, Schirrmeister L (2008) oriented by wave action. However regional differences in Genetic, morphological, and statistical characterization environmental conditions, primarily textural differences of lakes in the permafrost-dominated Lena Delta. Proceedings of the Ninth International Conference on between the sandy Pleistocene deposits of the coastal

87 Permafrost, 29 June - 3 July, 2008, Fairbanks, Alaska, Vol. 2. Institute of Northern Engineering, University of Alaska Press, Fairbanks, pp 1329–1244. Morrell G, Dietrich JR (1993) Evaluation of the hydrocarbon prospectivity of the Old Crow Flats area of the northern Yukon. Bulletin of Canadian Petroleum Geology, 41: 32–45. Murray MR (2002) Is laser particle size determination possible for carbonate-rich lake sediments? Journal of Paleolimnology, 27: 173–183. Resio D, Bratos S, Thompson E (2002) Meteorology and wave climate. In: Vincent L, Demirbilek Z (eds) Coastal Engineering Manual, Part II, Hydrodynamics, Chapter II-2. US Army Corps of Engineers, Washington D.C., p II.2.1 – II.2.72 Rex RW (1961) Hydrodynamic analysis of circulation and orientation of lakes in northern Alaska. Geology of the Arctic: Proceedings of the First International Symposium on Arctic Geology, Calgary, Alberta, January 11-13, 1960. University of Toronto Press, Toronto, pp 1021–1043. Roy-Leveillee P, Burn CR (2010) Permafrost conditions near shorelines of oriented lakes in Old Crow Flats, Yukon Territory. Proceedings of the Sixth Canadian Permafrost Conference, Calgary, Canada, September 12-16, 2010. Canadian Geotechnical Society, Calgary, pp 1509–1516. Roy-Léveillée P, Burn CR, McDonald ID (2014) Vegetation-Permafrost Relations within the Forest- Tundra Ecotone near Old Crow, Northern Yukon, Canada: Permafrost Temperatures within the Forest- Tundra near Old Crow, YT. Permafrost and Periglacial Processes, 25:127–135. Turner KW, Wolfe BB, Edwards TWD, Lantz TC, Hall RI, Larocque G (2014) Controls on water balance of shallow thermokarst lakes and their relations with catchment characteristics: a multi-year, landscape- scale assessment based on water isotope tracers and remote sensing in Old Crow Flats, Yukon (Canada). Global Change Biology, 20: 1585–1603. Zazula GD, Duk-Rodkin A, Schweger CE, Morlan RE (2004) Late Pleistocene chronology of Glacial Lake Old Crow and the north-west margin of the Laurentide ice sheet. In: J. Ehlers and P.L. Gibbard (ed) Quaternary Glaciations-Extent and Chronology Part II: North America. Elsevier, London, pp 347–362 Zhan S, Beck RA, Hinkel KM, Liu H, Jones BM (2014) Spatio-temporal analysis of gyres in oriented lakes on the Arctic Coastal Plain of Northern Alaska based on remotely sensed images. Remote Sensing, 6: 9170– 9193.

88 Summer and winter flows of the Mackenzie River system

Ming-ko Woo & Robin Thorne School of Geography and Earth Sciences McMaster University, Hamilton, Ontario, Canada

ABSTRACT The Mackenzie River Basin contains major hydro-physiographical regions representative of northern permafrost and non-permafrost zones. The regional environments exert influences, best expressed by the summer and winter hydrographs of headwater rivers. Summer flow is particularly responsive to the varied climatic, topographic and hydrologic settings of the Cordilleran mountains, the low gradient Interior Plains and the bedrock terrain of the Canadian Shield with myriad lakes. In winter, streamflow is sustained by groundwater discharge and by lake outflow. Most rivers in Mackenzie Basin show only weak flow trends in the summer but winter flows in the lower basin exhibit statistically significant rising trends between years, which may be attributed to permafrost degradation and increased autumn rainfall, though these hypotheses warrant further verification.

RÉSUMÉ Le bassin du fleuve Mackenzie contient des régions hydro-physiographiques majeures représentatives des zones nordiques de pergélisol et d’autres sans pergélisol. Les environnements régionaux exercent une influence, mieux exprimée par les hydrogrammes d’été et d’hiver des fleuves nourriciers. L’été, l’écoulement est particulièrement sensible à la variation des paramètres climatiques, topographiques et hydrologiques des montagnes de la Cordillière, ainsi qu’au faible gradient des plaines intérieures et du substratum rocheux du Bouclier canadien avec ses myriades de lacs. En hiver, l’écoulement est maintenu par le débit des eaux souterraines et par la décharge du lac. La plupart des rivières dans le bassin du Mackenzie montrent de faibles tendances d’écoulement en été, tandis qu’en hiver, les écoulements dans le bassin inférieur montrent d’importantes tendances à la hausse d’une année à l’autre. Ces dernières peuvent être attribuées à la dégradation du pergélisol et à une augmentation des précipitations à l’automne. Ces hypothèses méritent cependant d’être vérifiées davantage.

1 INTRODUCTION winter (from November 1 to the end of March). For this study, daily streamflow data are acquired from the HYDAT Mackenzie River is the largest northward-flowing river in database, the National Water Data Archive compiled by Canada. Many northern settlements along the river Water Survey of Canada. course, agriculture in the southern section, hydro-power production and resource development in various parts of the basin draw upon the river system for water supply. 2 SUMMER FLOW AND BASIN ENVIRONMENT Furthermore, barge traffic along the main river requires information on summer flow, while the management of The Mackenzie Basin can be distinguished into four major aquatic ecological well being can benefit from information hydro-physiographical provinces as shown in Figure 1, on the summer and winter flow status. Spring freshets, with permafrost underlying the northern basin. The generated by snowmelt and accentuated by ice breakup, summer flow regimes of rivers in these regions are have been well investigated (Prowse and Carter 2002; de influenced by basin attributes such as topography, land Rham et al. 2008). The present study focuses on cover, and glaciers and lakes. Northern mountains of the comparing the flow conditions during the summer and the Western Cordillera lie within the continuous permafrost winter seasons. One recent concern is the changing zone whereas permafrost occurs preferentially beneath tendency of river flow in Arctic basins. Possible factors, some lower slopes in the discontinuous permafrost belt including the role of permafrost, will be examined. and is present only at high altitudes further south. Several Occupying an area of 1.8 million km2, the sets of natural and artificial environments affect the Mackenzie Basin covers about one-fifth of the total land hydrographs, as demonstrated by the examples in Figure surface of Canada. Rivers that cross different parts of this 2. Athabasca River at Jasper (Fig. 2a) in the south vast basin acquire flow characteristics according to the receives runoff contributions from snowmelt, rainfall and diverse hydro-physiographic conditions of the regions they glacier melt, causing the hydrograph to fluctuate traverse. With the large latitudinal and altitudinal extents frequently in the summer. Liard River at Upper Crossing of the basin, the arrival of spring and autumn differs by (Fig. 2b) is >7¡ latitude further north and without runoff weeks or even months between its northern and southern contribution from glaciers. Its spring freshet arrives later extremities. To enable comparison of flow among rivers, a than for the Athabasca, and the recession to summer flow common time frame is adopted for the definition of is often interrupted by rain-driven hydrograph rises. The summer (taken as the months of June to September) and hydrograph of Peace River at Hudson Hope (Fig. 2c)

89

Figure 1. Hydro-physiographic regions of Mackenzie River Basin and southern limits of continuous and discontinuous Figure 2. Long term (1972-2012) hydrographs showing permafrost. the means, standard deviations, maxima, and minima for every calendar day between June and March. Examples reflects the effect of water release from Williston Lake for are provided for rivers from mountainous regions: (a) hydropower production, sometimes yielding more flows in Athabasca at Jasper, (b) Liard at Upper Crossing, (c) the winter than in summer. Peace at Hudson Hope; from Canadian Shield: (d) In the east lies the Canadian Shield with rolling Camsell River; and from Interior Plains: (e) Hay River. topography. The crystalline bedrock is largely impervious, but major fracture zones are exploited by valleys, infilled with soil and often occupied by lakes and wetlands. The 3 WINTER FLOW AND ICE SEASON considerable storage volume of large lakes or chains of lakes dampens flow fluctuations and their outflow tends to Extreme cold conditions prevail in the winter, when the be relatively even (e.g. Camsell River shown in Fig. 2d). land is blanketed with snow and shallow water bodies turn The Interior Plains generally have low gradients, with into ice, decoupling most hydrologic activities from continuous permafrost in the northern tundra, atmospheric impetus such as evaporation, storms or mid- discontinuous permafrost in the Subarctic, and winter warm spells. River condition varies: some rivers permafrost-free further south. Lakes and wetlands occupy cease to flow and their channels are infilled with drifted many areas with pitted topography, with annual snow or icing, while those rivers with continuous flow evaporation from open water surfaces ranging from >500 acquire a cover of ice. The ice covered duration is taken mm in the south to <200 mm in the tundra (Lins et al. as the period between freeze-up and breakup. As noted 1990). The Plains present little obstruction for north-south by Mackay (1960), the freeze-up date may be based on airflow that can alternately bring forth dry or wet the formation of a given amount of ice on the river but a conditions, giving rise to large intra- and inter-annual warm spell can destroy a fragile early ice cover. Breakup fluctuations in summer flow for Plains rivers such as the may be deemed as the initiation of ice movement or the Hay (Fig. 2e). clearance of ice at the observation site. Water and ice At the northern extremity of the Mackenzie Basin lies held in storage are discharged as part of the freshet. the Mackenzie Delta. It is an extension of the Interior Mackay (1960) noted that breakup at Fort Good Hope can Plains and consists of a watery maze of distributaries and be correlated with the air temperatue of Fort Simpson for myriad lakes. The Delta is flooded in the spring and dries April and early May. Warmer spring may therefore out gradually in the summer except when storm surges translate to an earlier breakup, through both processes of bring about inundation from the seaward side and thermal and mechanical breakup. The latter is initiated by backwater effect extending to the upper Delta (Marsh and hydrodynamic forces that lift and fracture the ice, and Schmidt 1993).

90 should be taken into consideration together with ice-melt association of several mechanisms with the identified in spring (Hiks and Beltaos 2008). trends in the Mackenzie Basin. An increasing flow trend The timing of freeze-up and breakup varies can be related to an increase in water supply and/or considerably within the Mackenzie Basin, depending enhanced efficiency in flow delivery. Precipitation input, notably on altitude and latitude. For most rivers, winter is cumulative storage release and permafrost degradation a low flow period with the flow sustained by groundwater can be contributing factors. discharge. Exceptions are most rivers in the Canadian Shield where outflow from large lakes continuously 6.1 Climate warming maintain moderate flows. A tendency for air temperature to rise in late summer can be detected in most of Mackenzie Basin (Fig. 4a), 4 FLOW IN MAIN STEM OF MACKENZIE especially in the central and southern areas. Despite the weak trend, warmer conditions can enhance evaporation The Mackenzie River collects runoff from its separate loss and degradation of permafrost. An increase in tributary basins. Integration of tributary inflow with evaporation may account for a reduction of summer flow discharge of the main stem can take some distances. This in non-permafrost areas while in permafrost terrain, is shown by Mackay (1970) who investigated the joining of limited CALM data (Circumpolar Active Layer Monitoring; Liard River with the Mackenzie in June 1968. Using water http://www.gwu.edu/~calm/data/north.html) indicate that temperature, turbidity and chemical (sodium and chloride) the active layer at Fort Simpson has deepened concentration, he found that the waters of these two rivers significantly between 1992 and 2014 and the active layer were not fully mixed at least 480 km downstream from at Willowlake River has also thickened between 1993 and their confluence at Fort Simpson. The Liard is a main 2012. Both ground ice melt and thawing of northern contributor of flow to Mackenzie River. Woo and Thorne wetlands, which opens pathways for flow connectivity, can (2003) calculated that the Peace and the Liard together lead to a positive tendency for summer flow in the provide about half of the of the Mackenzie annual total northern Plains, as revealed by the rivers around Fort flow before the Delta, other rivers of the northern Simpson. mountains about one-tenth, the Athabasca River about one-fifth, the Shield rivers yield another fifth, and the least 6.2 Fire amount comes from rivers of the Interior Plains. In view of the complex origins of runoff from different Forest fire directly impacts ground thermal regime and hydro-physiographic provinces, the flow of the main severe wild fire can increase the active layer thickness, as Mackenzie River carries mixed hydrological signatures documented by Mackay (1995). With forest fire, the loss that prevent the deciphering of regional factors that of vegetation and surface cover is accompanied by influence streamflow trends. For this reason, main stem changes in ground thermal properties that affect insulation Mackenzie River is excluded from the trend study below. and heat conduction, the radiation regime as well as surface water balance. Comparing a burnt and an unburnt site in southern Yukon, Burn (1998) found that the burned 5 RECENT STREAMFLOW TENDENCIES site had higher ground temperature at all depths, and attained annual maximum thaw depth in October and Studies of Arctic streamflow have identified monotonic November instead of mid-November. Permafrost linear trends in the mean and extreme discharges during degradation proceeded from both the top and the bottom, the recent decades (e.g. McClelland et al. 2004, and active layer thickness reached 3.8 m compared with Walvoord and Striegl 2007, Yang et al. 2014). Most of the 1.4 m beneath the forest, representing permafrost thaw of Mackenzie Basin other than the northern Shield and 2.5 m between 1958 and 1997. While permafrost northern Plains display negative summer flow trends degradation is evident after a forest fire, a considerable though their correlation coefficients (r-values) are not extent of devastation is needed to provide sufficient water statistically significant. The tendency is stronger in the from intensified ground ice melt to sustain increased flow. south (Fig. 3a). Rivers fed by lakes in the northern Shield Nevertheless, in discontinuous permafrost areas, and northern Plains maintain significant positive trends in degradation proceeds both vertically and along the edges both summer and winter, as do the smaller rivers on the of the permafrost bodies, as noted by Quinton et al. Plains around Fort Simpson. Most western rivers, from the (2011), while groundwater movement can convey heat to Peace and continuing northward, undergo a switch to further the thaw. Percolation would then be enhanced and statistically significant positive r-values in October, yet subterranean flow conduits can be expanded, as negative correlation intensifies in the Athabasca. Of note suggested by St Jacques and Sauchyn (2009). is that the southern limit of permafrost approximately delineates a separation between positive and negative 6.3 Precipitation tendencies in summer flow (Fig. 3). The precipitation trends of early and mid-summer do not 6 TREND-MAKING MECHANISMS form discernible spatial clusters but the trends of a number of stations shift in August to approximate those of Various hypotheses have been advanced to account for streamflow trends. Here, we examine possible

91

Figure 3. Streamflow trends indicated by correlation of flow with years (1972-2013); correlation coefficients are given in percent: (a) summer (June to September) and (b) winter (November to March). Also shown is the southern limit of dis- continuous permafrost.

Figure 4. Trends of (a) September air temperature and (b) September precipitation across Mackenzie Basin.

92 the following month. By September a general pattern 6.5 Uncertainties emerges, separating stations with declining tendency in the south from those with a positive trend elsewhere in the There are uncertainties regarding accuracy of streamflow basin (Fig 4b). A positive precipitation tendency in both data and reliability of results of analysis. August and September suggests that such a spatial (1) The agents of permafrost degradation cannot be configuration is not random (coincidentally, these zones unequivocally ascertained. Wild fire may be instrumental correspond roughly with the permafrost regions). The fall in permafrost thaw, but fires are not regionally extensive. is when a cold air mass returns to the Subarctic, bringing Climate warming is pronounced in winter, yet a frontal storms that deposit rain to raise autumnal temperature increase of several degrees during extreme streamflow and increase basin storage in support of cold may be ineffective in forcing the thaw. winter baseflow. (2) Permafrost degradation with its attendant effects on Combining late summer precipitation with warming flow augmentation may occur througout summer and of the basin, it is plausible that most of the non-permafrost extend into winter, but the magnitude of change is small areas in the south have diminished pre-winter moisture relative to the volume of summer runoff and the effect storage due to elevated evaporation and reduced would escape statistical detection. precipitation. In the north, however, increased late- (3) Measurement of discharge under ice remains a summer precipitation and enhanced permafrost challenge for stream gauging. Trend analysis requires degradation (through ice melt contribution and subsurface historical data, yet the reliability of winter flow flow connectivity) would enrich moisture storage that measurements depends on river size and winter severity supports baseflow. The amount of baseflow in relation to (Rosenberg and Pentland, 1966). the storage status prior to the onset of winter can be These consideratins should not be dismissed in estimated by an empirical relationship between winter flow the investigation of winter flow trends. and the discharge on October 15, which is indicative of the pre-winter storage status (Woo and Thorne, 2014). Thus, storage provides a plausible link between the trends 7 DISCUSSION AND CONCLUSIONS of autumn precipitation, late summer warming and enhanced winter flow. Warm season hydrological processes such as rainfall, snow and ice melt, surface runoff and evaporation cease 6.4 Lake storage to be active in the winter, when intense cold together with snow and ice cover restrict moisture exchanges between Large lakes considerably retard outflow response to water the land, the water surface and the atmosphere. Then, gains and losses, atmospheric inputs and the impetus of other than rivers with discharge dominated by lake inflow. However, the storage factor is complicated by the outflow, streamflow of the Mackenzie system switches cumulative amount of net atmospheric input (precipitation from a surface-flow controlled summer regime to a minus evaporation), inflows including spring freshet, groundwater-flow dictated winter regime. outflows including artificial release, storage capacity and Hydrometric data from the past four decades antecedent storage of the past seasons and even of indicate that summer flow has a weak negative trend, multiple years. except in the northern Shield and Plains with myriad Most lake-fed rivers are located in the Plains and lakes. In winter, the negative trend intensifies in the Shield regions. The Plains can be impacted by permafrost southern corner but the trend in the north reverses to degradation but the Shield is dominated by bedrock that is become positive. Given that the trends are genuine, relatively insensitive to permafrost change. Rivers in the possible mechanisms that lead to winter rise in river flow north, such as Great Bear, Camsell, and Lac LaMartre, include precipitation increase in late summer that show decadal increase in streamflow, yet no such enhances basin storage to sustain higher baseflow, and tendency is found in southern Shield rivers like Fond du permafrost degradation that augments contribution from Lac and Lockhart. Superimposed on these trends are ground ice melt and develops subterranean pathways to multi-year cyclical variations, but the presence of a long- facilitate groundwater discharge. These factors need not term trend, in both summer and winter, is attributable to be mutually exclusive and can reinforce each other. gradual buildup of lake storage. The chief contributor may Permafrost, for example, is one of the environmental be increased precipitation in the fall (Spence et al. 2011), conditions that respond to air temperature but itself at a time when northward migration of frontal storms influences runoff production and delivery. In the absence brings much precipitation to the Subarctic but when of definitive confirmation of one or more hypotheses, seasonal evaporation has declined. continued investigation is warranted. Under human influence, more water may be released in winter than in summer from Williston Lake, depending on hydroelectric power demand, thus reversing ACKNOWLEDGEMENT the normal flow regime of Peace River. Usually being the principal contributor of winter flow to the Mackenzie, the We wish to pay tribute to the late Professor Ross Mackay Peace will carry its winter flow tendency to the main for his pioneering research on permafrost and his early Mackenzie River. work on Mackenzie River, both of which provided inspiration for this paper.

93 lateral export of carbon and nitrogen. Geophysical Research Letters 34, L12402, REFERENCES doi:10.1029/2007GL030216. Woo, M.K. and Thorne, R. 2003. Streamflow in the Burn, C.R. 1998. The response (1958.1997) of permafrost Mackenzie Basin, Canada. Arctic 56: 328-340. and near-surface ground temperatures to forest fire, Woo, M.K. and Thorne, R. 2014. Winter flows in the Takhini River valley, southern Yukon Territory. Mackenzie drainage system. Arctic 67: 238-256. Canadian Journal of Earth Sciences 35: 184-199. Yang, D.Q., Shi, X.G. and Marsh, P. 2014. Variability and de Rham, L.P., Prowse, T.D. and Bonsal, B.R. 2008. extremes of Mackenzie River daily discharge during Temporal variations in river-ice break-up over the 1973-2011. Quaternary International doi: Mackenzie River Basin, Canada. Journal of Hydrology 10.1016/j.quaint.2014.09.023. 349: 441-454. Hicks, F.E. and Beltaos, S. 2008. River ice. In: Woo, M.K. (ed) Cold region atmospheric and hydrologic studies, the Mackenzie GEWEX experience, vol 2, Hydrologic processes. Springer, Berlin: 281Ð305. Lins, H.F., Hare, F.K. and Singh, K.P. 1990. Chapter 2, Influence of the atmosphere. In. Wolman, M.G. and Riggs, H.C. (eds), Surface Water Hydrology. The Geology of North America, vol O-1. Geological Society of America, Boulder, Colorado: 11-53. Mackay, J.R. 1960. Freeze-up and break-up of the Lower Makenzie River, Northwest Territories. In: Raasch, G.O. (ed.) Geology of the Arctic, vol. 2, Univ. of Toronto Press, Toronto: 1119-1134. Mackay, J.R. 1970. Lateral mixing of the Liard and Mackenzie rivers downstream from their confluence. Canadian Journal of Earth Sciences 7:111-124. Mackay, J.R. 1995. Active layer changes (1968 to 1993) following the forest-tundra fire near Inuvik, N.W.T. Canada. Arctic and Alpine Research 27: 323-336. Marsh, P. and Schmidt, T. 1993. Influence of a Beaufort Sea storm surge on channel levels in the Mackenzie Delta. Arctic 46: 35Ð41. McClelland, J.W., Holmes, R.M. and Peterson, B.J. 2004. Increasing river discharge in the Eurasian Arctic: consideration of dams, permafrost thaw, and fires as potential agents of change. Journal of Geophysical Research 109, doi: 10.1029/2004/2004JD004583. Quinton, W.L., Hayashi, M. and Chasmer, L.E. 2011. Permafrost-thaw-induced land-cover change in the Canadian subarctic: implications for water resources. Hydrological Processes 25:152Ð158. Prowse, T.D. and Carter, T. 2002. Significance of ice- induced storage to spring runoff: a case study of the Mackenzie River. Hydrological Processes 16:779Ð788. Rosenberg, H.B. and Pentland, R.L. 1966. Accuracy of winter streamflow records. Proceedings 23rd Eastern Snow Conference, 10-11 February 1966, Hartford, Connecticut: 51-72. Spence, C., Kokelj, S.V. and Ehsanzadeh, E. 2011. Precipitation trends contribute to streamflow regime shifts in northern Canada. IAHS Publication 346, Wallingford, UK: 3-8 St. Jacques, J-M., and Sauchyn, D.J. 2009. Increasing winter baseflow and mean annual streamflow from possible permafrost thawing in the Northwest Territories, Canada. Geophysical Research Letters 36, L01401, doi:10.1029/2008GL035822. Walvoord, M. A. and Striegl, R.G. 2007. Increased groundwater to stream discharge from permafrost thawing in the Yukon River basin: Potential impacts on

94

Part 4 Problems in geocryology

95

96 Subdivision of ice-wedge polygons, western Arctic coast

C.R. Burn and H.B. O’Neill, Department of Geography and Environmental Studies – Carleton University, Ottawa, ON, Canada

ABSTRACT Ice-wedge polygons are characteristic features of unconsolidated sediments in the continuous permafrost zone. They commonly have a well-defined surface expression in lowland basins, but are also ubiquitous on hill slopes. The polygons are outlined by a network of primary ice wedges, in places subdivided by secondary wedges, and even tertiary features. The evolution of ice-wedge networks was thought to be the result of winter climate variation. Three sets of observations suggest that, instead, the development of secondary and tertiary wedges may be due to growth of the primary ridges and troughs influencing snow depth. (1) J.R. Mackay showed that smaller, secondary wedges may crack more frequently than primary wedges. (2) Hill slope polygons are not characteristically subdivided by secondary ice wedges. (3) Thermal contraction cracks expand over winter, responding to cooling of the ground as the season progresses.

RÉSUMÉ Les polygones à coins de glace sont communs dans les dépôts meubles de la zone de pergélisol continu. On les distingue aisément dans les basses-terres, où leur contour est bien défini, mais ils sont aussi omniprésents en terrain pentu. Les polygones sont délimités par un réseau de coins de glace primaires, et sont par endroits subdivisés par des coins secondaires ou tertiaires. On a cru que ces subdivisions étaient dues aux variations du climat hivernal, mais trois observations indiquent que les coins secondaires et tertiaires se développent parce que l'accumulation de neige dans les fosses des coins primaires inhibe le craquement de ceux-ci. (1) J.R. Mackay a démontré que les coins secondaires, plus petits, peuvent craquer plus fréquemment que les coins primaires. (2) Les polygones sur les pentes ne sont généralement pas subdivisés par des coins secondaires. (3) Les craques de contraction thermique s'élargissent au cour de l'hiver en réponse au refroidissement progressif du sol.

1 INTRODUCTION

Ice-wedge polygons are one of the most well-known geomorphological features of Arctic environments (Figure 1). They were a key research interest of J.R. Mackay, who published the most comprehensive field assessments of the development of both the polygons and their bounding ice wedges (e.g., Mackay 1974, 1993, 2000). Ice wedges are amongst the few diagnostic features of permafrost terrain, and, as bodies of near-surface massive ice, may create considerable terrain sensitivity to changes in surface conditions or warming climate (e.g., Hayley 2015). Ice wedges themselves are useful as paleoenvironmental indicators, because truncated ice wedges may indicate the depth of a former active layer (e.g., Mackay 1975, 1978a; Burn et al. 1986; Burn 1997). Ice-wedge polygons are commonly recognized in lowland settings (Figure 2), the environment in which Leffingwell (1915) developed the theory of ice-wedge Figure 1. Ice-wedge polygons in sedge tussock tundra of origin by thermal contraction cracking. Leffingwell’s (1915, the Old Crow Flats, northern Yukon. Primary, 1919) work was the basis for Lachenbruch’s (1962) secondary, and tertiary ice-wedge troughs are visible. magisterial quantitative model of thermal contraction Photo by C.R. Burn, July 2013. leading to repeated cracking of the ground and development of the wedges. Ice wedges and their polygons that associated higher order wedges, which polygons were almost exclusively examined in lowlands crack less frequently, with more extreme winter until Mackay (1990, 1995) described anti-syngenetic ice conditions. These authors recognized the ground wedges on hill slopes. temperature gradient as a prime factor in developing In lowland basins, the polygons are bounded by a thermal stress. They stated: “(i)n relatively warm winters network of primary ice wedges, and in some cases may the temperature gradients are small, and only fissures of be subdivided by secondary or tertiary features. At the low orders are formed; in cold winters (with great relative first International Conference on Permafrost, Dostovalov temperature) minimum temperature gradients are high, and Popov (1966) provided a theory for subdivision of and cracking reaches higher orders of fissure generation.”

97 Dostovalov and Popov’s understanding of thermal contraction cracking was consistent with Lachenbruch’s (1962) model. The approach was the basis upon which Plug and Werner (2002) simulated the development of ice-wedge networks by using celluar automata. A fundamental assumption underlying the approach of Dostovolov and Popov (1966) and Plug and Werner (2002) is that changing surface conditions as ice-wedge polygons grow do not affect development of the thermal stress field. The purpose of this paper is to examine this assumption in light of the work of J.R. Mackay.

Figure 3. Location of Garry Island and Illisarvik, NWT.

Figure 2. The ice-wedge polygons of site C at Garry Island, N.W.T. (Also known as the “Drinking Lake” site.) Photo by J.R. Mackay.

2 ICE-WEDGE SURFACE EXPRESSION

Ice wedge development leads to a characteristic morphology with ridges on each side of the wedge separated by a trough above the wedge (Burn 2004). “Single-ridge” morphology has been seen, at first inspection, above syngenetic ice wedges of the outer Mackenzie delta area, but upon closer examination, the characteristic morphology was apparent but muted (Morse and Burn 2013). At Garry Island, NWT (Figure 3), where Figure 4. New ice wedge trough at Illisarvik. The ice Mackay conducted much of his research on ice wedges, wedge (site 3 of Mackay and Burn, 2002) has been the surface of troughs is characteristically ≤ 1 m below the rejuvenated by cutting vegetation to reduce snow depth. ridges, and the separation of ridges is about 2 m. These values are characteristic of much of the western Arctic coastlands. lee of topographic obstacles (Morse et al. 2012). As ice The development of ridges occurs due to wedges grow and displace soil upwards to form ridges, displacement of ground by growing ice wedges and snow depth in troughs increases, and there is also outward movement of soil from polygons towards the ice potential for snow depth within polygons to increase. As a wedges (Mackay 1992, 2000). This is due to the result, winter temperatures within a trough may increase, confinement of soil within the polygons during thermal and the frequency of winter cracking may decline. contraction, but accommodation of expansion in the Mackay (1978b) terminated ice-wedge cracking in a field troughs. Freshly growing ice wedges in recently drained experiment at Garry Island by erecting snow fences to lakes have little ridge development, although troughs may increase snow depth. Sites unaffected by the experiment be observed after a few years (Figure 4). Growth of the did not stop cracking. ridges therefore occurs over time as the ice wedges Mackay (1992) presented a remarkable set of data on develop. the cracking frequency of ice wedges at Garry Island. In tundra regions, snow depth is partly controlled by These data were collected over each of 21 years at 32 snow fall, but also by accumulation in vegetation or in the ice-wedge sections at the Drinking Lake site (Figures 2, 5). These ice-wedge sections included both primary and

98

Figure 5. The ice wedge network and crack frequency, 1967-87, at site C, Garry Island (after Mackay 1992, Fig. 6)

secondary features. The primary wedges ranged in crack widen after they open (Mackay 1989, p. 366). We might frequency from 20 out 21 years to 0 out of 21 years. The assume that cracks may also deepen after they have secondary wedges showed a similar range in crack been initiated. frequency. In other words, the secondary wedges cracked Crack widths and depths have been measured during as frequently as their primary counterparts. These data do field visits, but, to our knowledge, sequential not suggest that secondary cracking only occurs in measurements of crack depth following initiation have not relatively cold winters, because, if this were so, cracking been presented. Variation in crack width may be inferred of secondary ice wedges would be less frequent than of with data presented by Allard and Kasper (1998), who primary ice wedges. reported that electrical cables across ice wedges were strained, but not necessarily broken during two years of measurements at Salluit, QC. 3 OPENING OF CONTRACTION CRACKS At the Illisarvik experimentally drained lake, we have recently installed several TinyTag shock loggers at the At the western Arctic coast, Mackay (1993a) determined suggestion of H.H. Christiansen. These have been used that most thermal contraction cracks open between mid- with a thermistor cable attached to a data logger, to January and late March. The size of and depth of cracks determine the time of ground cracking. The miniature varies between ice wedges that crack in any given year shock loggers are cubic, and of side approximately 3 cm. and from year to year (Mackay 1974). The initial cracking They are placed firmly into the active layer in the side of event may be sudden, producing a report like that of a rifle an ice-wedge trough at the end of summer. The (Mackay 1993b), or, alternatively, cracking may be slow. accelerometer senses ground shocks and the logger Mackay (1993b) found that, in general, crack propagation stores them every 20 min. The thermistor cable has been at Garry Island was several orders of magnitude slower laid across the permafrost table of the same trough, within than the 200 m/s required for an audible report. 2 m of the shock loggers. It has been attached to a HOBO Nevertheless, the data indicate that once initiated, cracks HTEA miniature logger, recording five times each day may grow. Mackay (1993b) reported horizontal (every 4 hr 48 min). When the thermistor cable breaks, propagation, but he had noted earlier that cracks may the ground temperature reading defaults to -37 ¡C and

99 remains constant. The shock loggers and thermistor Richards islands, LeCompte (2004) digitized and cables have been installed at ice wedge #3 of Mackay measured over 3400 polygons near the northern point of and Burn (2002), which has been rejuvenated by annually Richards Island and on NW Garry Island. He found that cutting vegetation and hence reducing the snow cover the average size of upland polygons was 796 m2 (n = (Figure 4). The annual thermal contraction crack has been 1480) and of lowland polygons 309 m2 (n = 1998). A seen in March or April each year. All shock loggers have histogram of the distribution of upland and lowland measured one and only one ground movement in winter, polygons is presented in Figure 7. in late January or February. The thermistor cable has In part the difference in size is due to the reduced broken, but considerably later in winter, commonly in late thermal contraction coefficient of mineral materials in March. We interpret the delay between ground shock and comparison with ice. Anderson and Ladanyi (2004, p.53) rupture of the cable as due to expansion of the thermal indicate the coefficient is approximately 5 x 10-5 ¡C-1 for contraction crack, as the ground continues to cool during ice, and ≤ 2 x 10-5 ¡C-1 for mineral soil materials. winter. Therefore, if the ice content of near-surface permafrost is For the purposes of this paper, this interpretation less on well-drained hillslopes than in wet lowlands, we implies that increased thermal stress after initial crack might expect larger distances between ice wedges. In opening may be relieved in extant cracks by their growth, addition, upland polygons rarely appear to be subdivided, rather than necessarily by opening of new cracks. but rather the ice wedge networks are dominated by primary features. Ice-wedge troughs are discernable on hill slopes, but 4 HILLSLOPE ICE-WEDGE POLYGONS are difficult to see unless degradation of the ice wedge has occurred. In many cases, the ridges are eroded by Ice wedges are widespread on hill slopes near the slope movement, which may also fill in the troughs. As a western Arctic coast (Mackay 1990, 1995). The wedges result, the ridge-trough-ridge morphology of lowland ice form polygons, as in lowlands (Mackay 1995). However, wedges is replaced by a smoother surface on hill slopes. the polygons tend to be larger on hillslopes (Figure 6). In a comparative study of upland and lowland ice wedge polygons using aerial photographs from Garry and

Figure 6. Lowland and upland ice-wedge polygons at Garry Island. The images are at the same scale. The lowland is site C.

observations of thermal contraction cracking and polygon 5 ICE-WEDGE DEVELOPMENT AND POLYGON development at Garry Island and Illisarvik. The most EVOLUTION important observation is the similar contraction crack frequency distribution for primary and secondary wedges Dostovalov and Popov’s (1966) account of polygon at Garry Island site C (Mackay (1992). In greater detail, evolution in response to climate variation may be Mackay (1992, p. 246) also noted that “(s)econdary reconsidered in light of J.R. Mackay’s subsequent field wedges in topographically well-defined polygons tend to

100 crack more frequently than those in less well-defined polygons.” Given that ice wedges crack in response to 6 CONCLUSION thermal stress, we may infer from the last observation that primary ice wedges with well-developed topographic The purpose of this paper was to discuss a theory for expression experience less thermal stress than those with development of primary, secondary, and tertiary phases of lower ridges. As a result, the thermal stress within the ice wedge networks that accounted for these phases with reference to climate variation. The position advanced is that evolution of topography associated with ice wedges alters the thermal stress field within a polygon field, mainly by altering snow accumulation patterns. Therefore, secondary or tertiary ice wedges may develop without climate variation.

ACKNOWLEDGEMENTS

The research at Garry Island and Illisarvik has been supported by NSERC, most recently through the Frontiers ADAPT program; by PCSP; and by the Aurora Research Institute. From 1988-2010 visits were made in most years to Garry Island and every year to Illisarvik with J.R. Mackay. I would like to thank the many past and present graduate students and field assistants who helped to make these visits so enjoyable, especially Andrew Burn, Douglas Esagok, Graham Gilbert, and Steve Kokelj. We thank Pascale Roy-Léveillée for translating the abstract.

REFERENCES

Allard, M., and Kasper, J.N. 1998. Temperature conditions Figure 7. Relative frequency of ice-wedge polygons from for ice-wedge cracking: field measurements from a sample of 3478 in uplands and lowlands on the basis of Salluit, northern Québec. 7th International Conference polygon area. After LaCompte (2004, Fig.3.2). on Permafrost, National Research Council of Canada, Yellowknife, NWT. Centre d’études Nordiques, Université Laval, Collection Nordicana, polygons is relieved by cracking of secondary ice wedges. 57: 5- 12. A physical hypothesis to account for this observation is Andersland, O.B., and Ladanyi, B. 2004. Frozen ground that the development of ridge-trough-ridge topography at engineering. 2nd ed. John Wiley and Sons, Hoboken, the edges of polygons alters snow accumulation patterns NJ, USA.. sufficiently to affect the locus of maximum thermal stress. Burn, C.R. 1997. Cryostratigraphy, paleogeography, and If primary ice wedges do crack, then increasing thermal climate change during the early Holocene warm stress may be relieved by expansion of the cracks rather interval, western Arctic coast, Canada. Canadian than initiation of new cracks within a polygon. Journal of Earth Sciences, 34: 912-925. The hypothesis accounts for the relatively rare Burn C.R. 2004. A field perspective on modelling “single- subdivision of polygons on hill slopes, because the ridge” ice-wedge polygons. Permafrost and primary wedges are not shielded from thermal stress by Periglacial Processes, 15: 59-65. snow accumulation in troughs. Ridge-trough-ridge Burn, C.R., Michel, F.A., and Smith, M.W. 1986. morphology is muted on hill slopes, except where ice Stratigraphic, isotopic and mineralogical evidence for wedges are degrading and troughs are evident. an early Holocene thaw unconformity at Mayo, Yukon The hypothesis implies that while secondary ice Territory. Canadian Journal of Earth Sciences, 23: wedge initiation may be associated with a very cold 794-803. winter, it is also likely associated with development of Dostovalov, B.N. and Popov, A.I. 1966. Polygonal specific conditions restricting cracking of nearby primary systems of ice-wedges and conditions of their ice wedges, so that the additional stress cannot be development, Permafrost International Conference relieved by crack expansion. 1963, Lafayette, Indiana. National Academy of In several papers, Mackay indicated that ice-wedge Sciences – National Research Council, Washington, networks, although conceptually simple and accounted for DC. Publication 1287: 71-76 elegantly by Lachenbruch (1962), are in reality complex Lachenbruch, A.H. 1962. Mechanics of thermal systems because of changes in surface conditions contraction cracks and ice-wedge polygons in brought about by wedge growth (e.g., Mackay 1989, permafrost. Geological Society of America, New 1992). York, NY. Special Paper 70.

101 LeCompte, S. 2004. A comparison of properties of upland Delta, western Arctic coast, Canada. Journal of and lowland ice-wedge polygons on Richards Island Geophysical Research (Earth Surface), 118: 1320- and Garry Island, Northwest Territories. BSc thesis, 1332. Department of geography and Environmental Morse, P.D., Burn, C.R., Kokelj, S.V. 2012. Influence of Studies, Carleton University, Ottawa, ON. snow on near-surface ground temperatures in upland Leffingwell, E. deK. 1915. Ground-ice wedges, the and alluvial environments of the outer Mackenzie dominant form of ground ice on the north coast of Delta, N.W.T. Canadian Journal of Earth Sciences, 49: Alaska. Journal of Geology, 23: 635-654. 895-913. Leffingwell, E.deK. 1919. The Canning River region, Plug, L.J., and Werner, B.T. Nonlinear dynamics of ice- northern Alaska. United States Geological Survey, wedge networks and resulting sensitivity to severe Washington DC, USA. Professional Paper 109. cooling events. Nature, 417: 929-933. Mackay, J.R. 1974. Ice-wedge cracks, Garry Island, N.W.T. Canadian Journal of Earth Sciences, 11: 1366- 1383. Mackay, J.R. 1975. Relict ice wedges, Pelly Island, N.W.T. (107 C/12) In Report of Activities, Part A. Geological Survey of Canada, Ottawa, ON. Paper 75- 1 Part A: 469-470. Mackay, J.R. 1978a. Freshwater shelled invertebrate indicators of paleoclimate in north-western Canada during late glacial times: Discussion. Canadian Journal of Earth Sciences, 15: 461-462. Mackay, J.R. 1978b. The use of snow fences to reduce ice-wedge cracking, Garry Island, Northwest Territories. In Current Research, Part A. Geological Survey of Canada, Ottawa, ON. Paper 78-1A: 523- 524. Mackay, J.R. 1986. The first 7 years (1978-1985) of ice wedge growth, Illisarvik experimental drained lake site, western Arctic coast. Canadian Journal of Earth Sciences, 23: 1782-1795. Mackay, J.R. 1989. Ice-wedge cracks, western Arctic coast. The Canadian Geographer, 33: 365-368. Mackay, J.R. 1990. Some observations on the growth and deformation of epigenetic, syngenetic, and antisyngenetic ice wedges. Permafrost and Periglacial Processes, 1: 15-29. Mackay, J.R. 1992. The frequency of ice-wedge cracking (1967-1987) at Garry Island, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 29: 236-248. Mackay, J.R. 1993a. Air temperature, snow cover, creep of frozen ground, and the time of ice-wedge cracking, western Arctic coast. Canadian Journal of Earth Sciences, 30: 1720-1729. Mackay, J.R. 1993b. The sound and speed of ice-wedge cracking, Arctic Canada. Canadian Journal of Earth Sciences, 30: 509-518. Mackay, J.R. 1995. Ice wedges on hillslopes and landform evolution in the late Quaternary, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 32: 1093-1105. Mackay, J.R. 2000. Thermally induced movements in ice- wedge polygons, western Arctic coast: a long-term study. Géographie physique et Quaternaire, 54: 41-68. Mackay, J.R., and Burn, C.R. 2002. The first 20 years (1978/79 to 1998/99) of ice-wedge growth at the Illisarvik experimental drained lake site, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 39: 95-111. Morse, P.D., and Burn, C.R. 2013. Field observations of syngenetic ice-wedge polygons, outer Mackenzie

102 Retrogressive thaw slumps: From slope process to the landscape sensitivity of northwestern Canada.

S.V. Kokelj1, J. Tunnicliffe2, D. Lacelle3, T.C. Lantz4, R.H. Fraser5 1Northwest Territories Geological Survey, Government of the Northwest Territories, Yellowknife, Canada 2School of Environment, University of Auckland, Auckland, New Zealand 3Department of Geography University of Ottawa, Ottawa, ON, Canada 4School of Environmental Studies, University of Victoria, Victoria, BC, Canada 5Canada Centre for Remote Sensing, Natural Resources Canada, Ottawa, ON, Canada

ABSTRACT Retrogressive thaw slumping is an important driver of geomorphic change in ice-rich, glaciogenic landscapes. Here we summarize research on the processes of thaw slump development, with focus on studies from northwestern Canada. In the Peel Plateau, individual slumps commonly exceed 20 ha in area. These “mega slumps” displace downslope up to 106 m3 of previously frozen materials, reconfigure drainage networks, and significantly increase stream sediment and solute loads. The significant acceleration of slump activity has caused this process to become a dominant driver of geomorphic change in several ice-rich environments across the western Arctic. Landsat satellite imagery (1985 to 2011) and high frequency climatic and photographic time-series from the Peel Plateau indicate that an increase in rainfall has accelerated downslope sediment flux from slump scar zones, perpetuating slump activity and intensifying this disturbance regime. Increasing summer rainfall is likely the main factor driving recent geomorphic change. Remotely sensed mapping of slump-impacted terrain across a 1, 275, 000 km2 area of northwestern Canada indicates the close association with ice-rich hummocky moraine landscapes deposited at the margins of the former Laurentide Ice Sheet. This mapping provides a quantitative basis for evaluating the potential for climate-driven landscape change and assessing the spatial distribution of ice-cored permafrost across northwestern Canada.

RÉSUMÉ Les glissements régressifs de dégel représentent un processus thermokarstique dynamique et influencent la géomorphologie des environnements circumpolaires d’origine glaciaire. Les glissements présents sur le plateau de la rivière Peel, affectent fréquemment des superficies excédant 20 ha. Ces Çméga-glissementsÈ déstabilisent jusqu’à 106 m3 de sédiments, reconfigurent le drainage et augmentent la charge de transport des cours d’eau en aval. Une augmentation de leur activité est observée dans les environnements de pergélisol riche en glace de l’ouest de l’arctique. Les données satellitaires Landsat (de 1985 à 2011) ainsi que les imageries et reconstitution climatique à hautes fréquences indiquent qu’une augmentation des précipitations sous forme de pluie accélère le taux de transport sédimentaire en provenance de zones des parois d’instabilités. Ce mécanisme empêche la stabilisation naturelle des parois et intensifie le processus de glissement régressif. Une cartographie par télédétection des zones affectées par des glissements régressifs sur une superficie de 1, 275, 000 km2 du Nord-Ouest Canadian indique une forte association entre la présence de glissements et les moraines frontales riches en glace de l’inlandsis laurentien. Cette cartographie, considérée conjointement avec les processus de formation des glissements, permet une évaluation quantitative des changements géomorphologiques liés au climat, ainsi qu’une mesure de la distribution géographique des zones de pergélisol riche en glace du Nord-Ouest Canadian.

1 INTRODUCTION In this paper we consider the processes and Retrogressive thaw slumping is a dynamic form of feedbacks associated with thaw slump development, thermokarst slope disturbance that couples thermal and linkages with climate, and the factors that control the geomorphic processes to rapidly degrade ice-rich distribution of these disturbances. The broad goal of this permafrost (Fig. 1) (Burn and Lewkowicz, 1990). The research is to understand the climate change sensitivity of process can modify slope morphology and transport large permafrost environments. The focus on thaw slumps has volumes of thawed materials downslope to lakes, valley been stimulated in part by the concerns raised by land bottoms and coastal zones. In ice-rich permafrost users, and environment and infrastructure managers who environments, thaw slumping can be the dominant driver have reported that the magnitude of slumping and of landscape change (Lewkowicz, 1987a; Lacelle et al., landscape impacts has intensified in northwestern NWT. 2010; Lantuit et al., 2012). Mounting evidence suggests Our specific objectives are to: A) summarize the that this disturbance regime has intensified across a processes contributing to thaw slump development, and range of environments, increasing terrain and fluvial document their environmental impacts; B) evaluate the impacts (Kokelj and Jorgenson, 2013). drivers that intensify thaw slump activity across a range of western Arctic landscapes; and C) examine the broad-

103

Figure 1. Thaw slump photo montage. A) Schematic cross-section of a large thaw slump. B) Scar zone and headwall of a “mega slump” on the Peel Plateau. C) The same slump viewed to show the extent of the debris tongue deposit. The total disturbance area is approximately 39 ha and the headwall is up to 25 m high. The disturbance has been on the landscape since the 1950s. Intensified thermokarst activity in the past 2 decades has resulted in rapid headwall retreat and an increase in scar area driven by an acceleration of scar zone sediment flux and the deposition of a large debris tongue.

scale patterns of thaw slump distribution across sediment flux (Kokelj et al., 2015). The geochemistry of northwestern Canada. Based on these results we have permafrost, slope runoff and streams were evaluated by refined a conceptual model linking climate conditions with sampling across a range of slump impacted and slump intensification, which informs our understanding of unimpacted streams. Selected streams were also the sensitivity of permafrost landscapes. monitored with data loggers to evaluate patterns of variation in water levels, chemistry and sediment flux (Malone et al., 2013; Kokelj et al., 2013; Lacelle et al., 2 STUDY AREA AND METHODS 2015). Aerial photograph interpretation, and novel analysis of Landsat satellite images from 1985 to 2011 2.1 Field-based and local-scale mapping studies: Slump were undertaken to determine the distribution of slumps, dynamics and change with time landscape associations and their change over time (Brooker et al., 2014, Lacelle et al., 2015; Kokelj et al., A series of collaborative mapping and field-based studies 2015). were initiated on the Peel Plateau, northwestern NWT in Thaw slump distribution and rates of growth were 2010 to investigate the distribution and development of documented using historical air photographs in the Peel thaw slumps, and to determine the environmental impacts Plateau and three additional study areas across the of this disturbance regime (Fig. 2). The Peel Plateau is a western Arctic (Fig. 2). The goal of this analysis is to fluvially-incised, ice-rich morainal landscape that extends assess landscape and climate factors that contribute to along the eastern margins of the Richardson Mountains regional variation in slump characteristics and rates of and northeastern Mackenzie Mountains. Permafrost change. depths are greater than 100 m (Mackay, 1967) and the mean annual ground temperatures are relatively warm (< 2.2 Broad-scale mapping: The distribution of slump -2.5¡C) due to atmospheric temperature inversions in impacted terrain across northwestern Canada winter (O’Neill et al., 2015). Terrain stability on the Peel Plateau is of broad interest because it constitutes a major The objective of this study was to evaluate the broad- sediment source for numerous tributaries of the Peel scale distribution of terrain affected by retrogressive thaw River and Mackenzie Delta (Kokelj et al., 2013), it is slumping across the western and central regions of traversed by the Dempster Highway, which links western Subarctic and Arctic Canada, including portions of the Arctic Canada with the south (Gill et al., 2014), and it lies Yukon, Northwest Territories, and Nunavut (Fig. 2). This at the heart of Gwich’in traditional lands (Slobodin, 1981). project was made possible by the compilation of A range of field techniques were implemented to georeferenced SPOT 4 and 5 orthomosaics from 2005- investigate the processes and feedbacks of slump growth. 2010 (http://www.geomatics.gov.nt.ca/sdw.aspx). To map These included terrain surveys, deployment of remote the areas impacted by active retrogressive thaw slumps, a cameras to track slump sediment transport, collection of 15 x 15 km grid system was developed to cover automated climatic data and compilation of historical the1,274,625 km2 study area (Fig. 2). Trained technicians climate records to investigate drivers of scar zone assessed disturbance and landscape attributes for each

104 Figure 2. Broad-scale mapping study area in Northwestern Canada, including the western Arctic and subarctic Northwest Territories, northern Yukon and central eastern Nunavut. The star indicates the Peel Plateau, small white blocks indicate areas of air photograph analysis for slump change. The lighter coloured landscape delineated by a solid line indicates the 1,275,000 km2 broad-scale slump mapping project area. The red line indicates the Wisconsinan Glacial extent and ice front positions and the grey lines are major moraine features from Fulton (1995). The inset map at the bottom right shows the position of the study area in northern Canada. grid cell by viewing the georeferenced SPOT 4 and 5 due to fluvial activity (Kokelj et al., 2015), thermally driven orthomosaics. A classification scheme described in subsidence along lakeshores (Kokelj et al., 2009a), or Segal et al., (2015) provided a systematic basis for mass wasting triggered by extreme thaw or precipitation assessing each grid cell and populating all required (Lacelle et al., 2010). Active thaw slumps are comprised attribute fields which included: 1) the slump density; 2) of an ice-rich headwall, a low-angled scar zone consisting primary and secondary slump-landscape associations of thawed slurry and in some cases a periodically mobile (valley, lakeside, coastal); and 3) area of the largest tongue of debris that develops as the saturated materials slump. flow downslope (Figs. 1, 3). Surface energy fluxes, The SPOT imagery used a pixel size of 10 m, and was ground ice and headwall characteristics (size and viewed at a scale of 1:20 000 to 1:30 000 and mappers orientation) influence ablation of ground ice and rates of were able to confidently identify disturbances as small as thaw slump enlargement (Lewkowicz, 1987b; Lacelle et 0.75 ha. To ensure that slumps were mapped accurately, al., 2015). Gravity-driven flow and collapse move thawing each grid cell was examined by at least one trained materials to the base of the headwall. Meltwater and observer and one expert reviewer. Stable slump scars debris are transported from the slump headwall by fluvial were not considered in this analysis because of processes and shallow and deep-seated mass flows challenges associated with their consistent identification (Kokelj et al., 2013, 2015). The nature and rate of across a range of landscapes using the SPOT imagery. downslope sediment removal from the slump scar zone is As such, the map products represent conservative a function of slope, soil characteristics and the moisture estimates of slump distribution. content of the thawed materials (Kokelj et al 2015). A thaw slump can enlarge for decades until the exposed ground ice is covered by an accumulation of thawed 3 RESEARCH RESULTS AND DISCUSSION sediments (Fig. 3b) (Burn and Lewkowicz, 1990). Vegetation colonization on the nutrient-rich scar zone 3.1 Thaw slump processes, Peel Plateau occurs when material accumulation has stabilized the disturbance (Lantz et al., 2009). Retrogressive thaw slumping is an important driver of Our research on thaw slump processes and geomorphic change in landscapes underlain by ice-rich environmental impacts has been focused on large, valley- permafrost, and it is the most common form of mass side slumps in the Peel Plateau. The high relative relief of wasting on the Peel Plateau (Fig. 1) (Lacelle et al., 2015). fluvially-incised slopes and ice-rich permafrost provide In this environment, thaw slumps are initiated when favourable conditions for the development of extremely ground ice is exposed by mechanical and thermal erosion large thaw slumps now common in the Peel Plateau (Fig.

105 1) (Lacelle et al., 2015). These “mega slumps” grow A) beyond the break of slope, thaw thick layers of ice-rich permafrost and translocate large volumes of sediment Ground downslope, reconfiguring slopes and depositing debris in ice valley bottoms (Figs. 1, 3, 4). As thaw slumps enlarge they tend to increase in geomorphic complexity to include different modes and intensities of downslope sediment displacement, which operate across a range of temporal and spatial scales (Kokelj et al., 2015). These processes include: (1) exposure of, and ablation of large massive ice bodies, backwasting of the headwall by retrogressive Debris accumulation Increased rainfall failure and supply of sediments and meltwater to a low- and stabilization and sediment flux angled scar zone; (2) evacuation of debris from the slump scar zone facilitated by a combination of diurnal, B) C) meltwater driven fluvial transport, intermittent rainfall- induced mass wasting (gravitational collapse, slumps, torrents, plug-like debris flow), and quasi-continuous fluidized mass flows; (3) base-level erosion, or evacuation of outlet detritus; and (4) valley-confined downstream aggradation of debris derived from the scar zone, leading to cascading effects including development of debris dammed lakes and enhanced valley-side erosion and development of secondary slumps. These processes and feedbacks can prolong slump activity over periods of several decades to produce Figure 3. A) Schematic showing a valley-side slump, headwall, disturbances that are tens of hectares in area. scar area and vegetated debris tongue (green). B) Stabilization of the slump can occur when slump grows to the upper slope and when materials accumulate to insulate exposed ground ice. The 3.2 Intensification of slumping, feedbacks and cascading green shaded scar and debris tongue indicate the establishment effects of vegetation cover. C) The influence of accelerated sediment flux from the scar zone on headwall height and debris tongue Thaw slump activity has increased across a range of development. The dotted lines upslope of the headwall indicate landscapes in northwestern Canada (Lantz and Kokelj, the increased upslope growth potential of the disturbance. 2008; Lacelle et al., 2010; Kokelj et al., 2015). Warmer air temperatures and increasing rainfall can influence thaw slump initiation. Around lakes, rising permafrost and water temperatures can alter talik configuration and thaw ice- rich sediments subadjacent to lakeshores causing lake- bottom subsidence, shoreline collapse and thaw slump initiation (Kokelj et al., 2009a). Deeper seasonal thaw or higher rainfall can also cause active-layer detachment slides. The intensification of precipitation regimes and high summer streamflow events can accelerate thermoerosion along flow tracks and channels, exposing ground ice and leading to slump initiation. Several climate factors can intensify the processes of thaw slump development. On the Peel Plateau, a major increase in the number and size of active slump surfaces and debris tongue deposits since the mid-1980s (Table 1) has occurred in concert with a statistically significant increase in the magnitude and intensity of rainfall (Kokelj et al., 2015). Air temperature and precipitation interact to influence the moisture regime of slump soils and downslope sediment transport from the slump scar zone. Diurnal pulses of ground ice meltwater drive high frequency, low magnitude surficial-sediment movement, Figure 4. An active disturbance regime is evident in glacially- and intense rainfall events stimulate major mass flows conditioned, ice-rich terrain on the Peel Plateau. Fluvial incision and downslope debris tongue enlargement (Kokelj et al., and thermal erosion have exposed ice-rich sediments, which has led to the development of large retrogressive thaw slumps and 2015). These lower-frequency high magnitude rainfall- debris tongues within the network of incised valleys in the induced flows evacuate sediments from the slump scar Plateau. Kettle lakes suggest ice-cored terrain, which is zone (Fig. 3c). Where the floor of the slump scar is confirmed by the ice-rich permafrost at least 25 m thickness underlain by ice-rich permafrost or massive ground ice, exposed in the nearby slumps.

106 meltwater runoff and downslope removal of debris can Table 1. Summary showing differences in large active contribute to further thaw settlement of the slump floor or disturbance surface areas between 1985-1990, and 2011 as polycyclic behaviour (Lantuit et al., 2012). Together, these determined by using Landsat imagery to map 14 slump affected study areas each 100 km2 in area . This table was processes help maintain a headwall of exposed ground derived from data presented in Table 1 in Kokelj et al., 2015. ice and inhibit slump stabilization (Fig. 3c). The enlargement of thaw slumps can strengthen Disturbance and date N Mean A and A(ha)/100km2 feedbacks that amplify this disturbance regime. At Stdev (ha) decadal timescales, large slumps retreat more rapidly Total slump 1985 41 3.8 (3.7) 11.2 than small ones because they are less likely to Debris tongue 1985 14 3.2 (2.9) 3.2 accumulate sufficient debris to arrest headwall retreat (Lacelle et al., 2015). The exposure to incoming solar Total slump 2011 68 9.9 (11.2) 47.9 radiation and rate of ground ice ablation also increase Debris tongue 2011 31 6.6 (4.3) 14.6 with headwall height and length so that large slumps may grow upslope more rapidly than smaller disturbances. More rapid ablation of these large headwalls produce 3.3 Broad-scale mapping of thaw slumps: Relationships saturated scar zone soils preconditioned for the with the distribution of massive ground ice and occurrence of rainfall-induced mass flows (Kokelj et al., glacial legacy 2015). As thaw slumps enlarge so do their hydrological contributory areas, which can enhance the soil moisture A review of ground ice and surficial studies suggests an regime and further accelerate sediment removal from the association between thaw slumps and ice-rich glaciogenic scar zone. deposits (Lewkowicz, 1987a; Rampton, 1988; St-Onge These feedbacks accelerate scar zone sediment flux and McMartin, 1999; Dyke and Savelle, 2000; Lacelle et and debris tongue progradation, which can in turn al., 2015). Development of large retrogressive thaw intensify fluvial impacts and disturbance of adjacent slumps requires the presence of ice-rich permafrost slopes (Fig. 1, 3, 4). Debris tongue deposits can raise the indicating that these disturbances may be a valuable base level of the trunk stream causing the formation indicator of ice-cored or ice-rich terrain. An online mosaic of debris-dammed lakes and channel diversions. of SPOT imagery enabled the mapping of the distribution Lateral stream displacement commonly initiates of thaw slumps across northwestern Canada (Fig. 2). valley-side thermoerosion and slumping of adjacent Preliminary results indicate that retrogressive thaw slopes to produce localized areas of high slumps are widely distributed throughout northwestern disturbance intensity (Kokelj et al., 2015). Canada. Grid cells with terrain impacted by slumps totaled 2 An increase in the frequency and magnitude of thaw about 140 000 km or about 11% of the entire mapped slumps can elevate the supply of sediments and solutes area (Table 2). As an example, mapping results from to fluvial and lacustrine environments (Kokelj et al., Banks Island show that the distribution of thaw slump 2009b, 2013). Sediment and solute concentrations in affected terrain is clearly bounded by the extents of the slump-impacted streams are several orders of magnitude Sandhills and Jesse Moraines (Fig. 5). Other notable greater than in undisturbed streams (Kokelj et al., 2013; areas impacted by slumping include the Peel Plateau Malone et al., 2013). Hundreds of small to medium sized (Lacelle et al., 2015), the Yukon Coastal Plain to Herschel watersheds on the Peel Plateau are impacted by thaw Island (Rampton, 1982, Lantuit et al., 2012), the slumps (Brooker et al., 2014). A significant rising trend in Tuktoyaktuk Coastlands (Rampton, 1988), the Bluenose SO4 concentrations and the SO4/Cl ratio in the Peel River Moraine (St Onge and McMartin, 1999) and moraines on between 1960 and 2012 suggest that the acceleration of Victoria Island, including the Wollaston Peninsula (Dyke thaw slump activity has increased the weathering of and Savelle, 2000). The final map produced for this glaciogenic materials and altered the geochemical flux of project illustrates the broad-scale association between this 70, 000 km2 watershed (Kokelj et al., 2013). The rapid slump affected terrain and ice-rich moraine deposits downstream responses to thaw slump enlargement bounded by the maximum westward extent of late highlight the efficacy of this process in mobilizing Wisconsinan glaciation. previously frozen glaciogenic materials from slope storage This mapping project also showed that slumps are to the fluvial system (Fig. 1, 3c). This disturbance regime much more widespread in fluvial and lake environments and the geomorphic consequences described here can be than in coastal settings (Table 2), where they typically expected to increase as glacially conditioned, ice-rich have been studied (Lewkowicz, 1987a; Lantuit et al., permafrost landscapes adjust to a rapidly changing 2012). Coastal slumping was abundant along the Yukon climate. These observations highlight the importance of coastal plain, throughout the Tuktoyaktuk Coastlands and understanding spatial variability in the landscape along the Jesse Moraine on eastern Banks Island, susceptibility to this disturbance regime. although dominant landscapes impacted on Banks Island were inland (Fig. 5; see inset). Lakeside slumps were common in impacted areas across the study region, except in the fluvial landscapes of the Peel Plateau and eastern foothills of the Mackenzie Mountains. These regions, in addition to eastern Banks Island, were dominated by valley-side slumps that impacted streams and rivers.

107

Figure 5. Broad-scale slump mapping results for Banks Island and adjacent Victoria Island. The inset on the right shows an area of south-eastern Banks Island where a detailed assessment of thaw slump distribution was conducted using a combination of air photographs and SPOT imagery. The coloured grid cells indicate relative disturbance intensity based on the broad-scale assessment and black dots are individual thaw slumps determined by detailed mapping.

Table 2. Summary showing the grid cell area affected by thaw stabilization, perpetuating processes of slump growth and slumping for different geomorphic environments. Note that increasing the magnitude and duration of downstream proportional areas are calculated on the basis of a total terrestrial 2 impacts. area of 1,274,625 km . In some cells thaw slumps occur in Slumps are widespread across subarctic and Arctic association with more than one type of environment. regions of northwestern Canada. Their distribution is

Geomorphic Grid cell % of % of largely constrained to ice-rich moraine deposits bounded environment area (km2) slump total by the maximum westward extent of late Wisconsinan impacted by affected area glacial ice. Permafrost has preserved thick layers of relict slumping area mapped ground ice and glaciogenic sediments, making these Coastal 14,850 10.7 1.2 glacially-conditioned landscapes particularly susceptible Lake 76,725 55.4 6.0 to thaw slump disturbance, and inherently sensitive to Fluvial, valley-side 81,675 59.0 6.4 climate driven geomorphic change (Kokelj et al., 2015). Slump totals 138,375 100.0 10.9 Broad-scale mapping of slumps provides a spatial basis for evaluating the distribution of ice-cored permafrost, and the terrain sensitivity of northwestern Canada. 4 CONCLUSIONS

Thaw slumping is a dominant driver of geomorphic 5 ACKNOWLEDGEMENTS change in ice-rich glaciogenic landscapes. This process can rapidly degrade thick layers of ice-rich permafrost and This work was supported by the NWT Cumulative Impact release vast quantities of glaciogenic materials downslope Monitoring Program, the Aurora Research Institute and to fluvial, lacustrine and coastal environments. Since thaw the Northwest Territories Geological Survey, Government slumping involves the coupling of geomorphic and thermal of the Northwest Territories, Natural Sciences and processes this disturbance regime is particularly sensitive Engineering Research Council of Canada and the Polar to climatic variation. Recent research demonstrates that Continental Shelf Project. Institutional support from the an increase in rainfall has accelerated slump activity in the Gwich’in Tribal Council, Gwich’in Renewable Resources Peel Plateau, and may be an important factor contributing Board and the Tetlit Gwich’in Renewable Resources to increased slump activity in many parts of the Arctic and Council and field assistance from several residents of Fort subarctic of northwestern Canada. Enhanced moisture McPherson is gratefully acknowledged. The authors thank regimes can increase rates of downslope sediment Rebecca Segal for producing figures 1 and 5. removal from the slump scar zone inhibiting slump

108 6 REFERENCES northwestern Canada. Geomorphology, 235: 40Ð51. doi:10.1016/j.geomorph.2015.01.024 Lantuit, H., Pollard, W.H., Couture, N., Fritz, M., Brooker, A., Fraser, R.H., Olthof, I., Kokelj, S.V., Lacelle, Schirrmeister, L., Meyer, H., Hubberten, H.-W., 2012. D., 2014. Mapping the Activity and Evolution of Modern and Late Holocene Retrogressive Thaw Retrogressive Thaw Slumps by Tasselled Cap Trend Slump Activity on the Yukon Coastal Plain and Analysis of a Landsat Satellite Image Stack. Herschel Island, Yukon Territory, Canada. Permafrost Permafrost and Periglacial Processes, 25: 243Ð256. and Periglacial Processes, 23: 39Ð51. doi:10.1002/ppp.1819 doi:10.1002/ppp.1731 Burn, C. R., Lewkowicz, A. G., 1990. Canadian Landform Lantz, T.C., Kokelj, S.V., 2008. Increasing rates of Examples - 17 Retrogressive Thaw Slumps. Canadian retrogressive thaw slump activity in the Mackenzie Geographer, 34: 273Ð276. doi:10.1111/j.1541- Delta region, N.W.T. Geophysical Research Letters, 0064.1990.tb01092.x 35, L06502. Dyke, A.S., Savelle, J.M., 2000. Major end moraines of Lantz, T.C., Kokelj, S.V., Gergel, S.E., Henry, G.H.R., Younger Dryas age on Wollaston Peninsula, Victoria 2009. Relative impacts of disturbance and Island, Canadian Arctic: implications for paleoclimate temperature: persistent changes in microenvironment and for formation of hummocky moraine. Canadian and vegetation in retrogressive thaw slumps. Global Journal of Earth Sciences, 37: 601Ð619. Change Biology, 15: 1664Ð1675. doi:10.1111/j.1365- doi:10.1139/e99-118 2486.2009.01917.x Fulton, R.J., (Compiler) 1995. Surficial materials of Lewkowicz, A.G., 1987a. Nature and importance of Canada. Geological Survey of Canada, Map 1880A. thermokarst processes, Sand Hills moraine, Banks Gill, H.K., Lantz, T.C., O’Neill, B., Kokelj, S.V., 2014. Island, Canada. Geografiska Annaler. Series A, Cumulative impacts and feedbacks of a gravel road on Physical Geography, 69 (2): 321-327. shrub tundra ecosystems in the Peel Plateau, Lewkowicz, A.G., 1987b. Headwall retreat of ground-ice Northwest Territories, Canada. Arctic, Antarctic, and slumps, Banks Island, Northwest Territories. Canadian Alpine Research, 46: 947Ð961. doi:10.1657/1938- Journal of Earth Sciences, 24: 1077Ð1085. 4246-46.4.947 doi:10.1139/e87-105 Kokelj, S.V., Lantz, T.C., Kanigan, J., Smith, S.L., Coutts, Mackay, J.R., 1967. Permafrost Depths, Lower Mackenzie R., 2009a. Origin and polycyclic behaviour of tundra Valley, Northwest Territories. Arctic, 20: 21Ð26. thaw slumps, Mackenzie Delta region, Northwest Malone, L., Lacelle, D., Kokelj, S., Clark, I.D., 2013. Territories, Canada. Permafrost and Periglacial Impacts of hillslope thaw slumps on the geochemistry Processes, 20: 173Ð184. doi:10.1002/ppp.642 of permafrost catchments (Stony Creek watershed, Kokelj, S.V., Zajdlik, B., Thompson, M.S., 2009b. The NWT, Canada). Chemical Geology, 356: 38Ð49. impacts of thawing permafrost on the chemistry of doi:10.1016/j.chemgeo.2013.07.010 lakes across the subarctic boreal-tundra transition, O’Neill, H.B., Burn, C.R., Kokelj, S.V., Lantz, T.C., 2015. Mackenzie Delta region, Canada. Permafrost and “Warm” Tundra: Atmospheric and near-surface ground Periglacial Processes, 20: 185-199. temperature inversions across an alpine treeline in Kokelj, S.V., Jorgenson, M.T., 2013. Advances in continuous permafrost, western Arctic, Canada. Thermokarst Research. Permafrost and Periglacial Permafrost and Periglacial Processes. Processes, 24: 108Ð119. doi:10.1002/ppp.1779 doi:10.1002/ppp.1838 Kokelj, S.V., Lacelle, D., Lantz, T.C., Tunnicliffe, J., Rampton, V.N., 1988. Quaternary geology of the Malone, L., Clark, I.D., Chin, K.S., 2013. Thawing of Tuktoyaktuk Coastlands, Northwest Territories massive ground ice in mega slumps drives increases (Memoir No. 423). Geological Survey of Canada. in stream sediment and solute flux across a range of Segal, R.A., Kokelj, S.V., Lantz, T.C, Durkee, K., Gervais, watershed scales. Journal of Geophysical Research: S., Mahon, E., Snijders, M., Buysse, J., and Schwarz, Earth Surface, 118: 681Ð692. S. 2015. Broad-scale inventory of retrogressive thaw Kokelj, S.V., Tunnicliffe, J.F., Lacelle, D., Lantz, T.C., slumping in Northwestern Canada. Northwest Chin, K.S., Fraser, R., 2015. Increased precipitation Territories Geoscience Office, Yellowknife NWT. Open drives mega slump development and destabilization of File 2015-04 ice-rich permafrost terrain, northwestern Canada. Slobodin, R., 1981. Kutchin [Gwich’in], in: Subarctic, Global and Planetary Change, 129: 56-68. doi: Handbook of North American Indians. Smithsonian, 10.1016/j.gloplacha.2015.02.008. Washington, D.C., pp. 514Ð532. Lacelle, D., Bjornson, J., Lauriol, B., 2010. Climatic and St-Onge, D.A., McMartin, I., 1999. The Bluenose Lake geomorphic factors affecting contemporary (1950Ð Moraine, a moraine with a glacier core. Géographie 2004) activity of retrogressive thaw slumps on the physique et Quaternaire, 53: 287Ð296. Aklavik Plateau, Richardson Mountains, NWT, Canada. Permafrost and Periglacial Processes, 21: 1Ð 15. doi:10.1002/ppp.666 Lacelle, D., Brooker, A., Fraser, R.H., Kokelj, S.V., 2015. Distribution and growth of thaw slumps in the Richardson MountainsÐPeel Plateau region,

109

110 Geochemistry of the active layer and permafrost in northwestern Canada: from measurements to Quaternary stratigraphy

Denis Lacelle1, Marielle Fontaine1, Steve V. Kokelj2 1Department of Geography, University of Ottawa, Ottawa, ON, Canada 2Northwest Territories Geological Survey, Government of the Northwest Territories, Yellowknife, NT, Canada

ABSTRACT The amount and distribution of soluble ions and salts in active layer and permafrost provide a tool to assess present and past changes in active layer thickness, and chemical weathering regimes in a given landscape. This short paper summarizes various methodologies used for the extraction of soluble salts and ions in the active layer and permafrost, and then tests two methods from two sites in northwestern Canada. The results are discussed with respect to the extraction techniques, the timing of permafrost aggradation following sediment deposition (i.e., syngenetic vs epigenetic permafrost) and the history of permafrost (i.e., relict active layer vs residual thaw layer).

RÉSUMÉ La quantité et la distribution des ions et des sels solubles dans la couche active et le pergélisol constituent un outil pour évaluer les changements passés et présents de l'épaisseur de la couche active et des régimes d'altération chimique d’un paysage. Cet article résume différentes méthodes utilisées pour l'extraction des ions et des sels solubles dans la couche active et le pergélisol, puis teste deux méthodes sur deux sites dans le nord-ouest du Canada. Les résultats sont discutés par rapport aux techniques d'extraction, à l’expansion du pergélisol suivant le dépôt des sédiments (pergélisol syngénétique vs épigénétique) et à l'histoire du pergélisol (couche active relique vs couche de dégel résiduelle).

1 INTRODUCTION Wisconsin, Illinoian), with a step-like change at the transition between interglacial-glacial age sediments Knowledge on the extent of permafrost degradation to (Figure 1). Variability in the distribution of soluble ions in past warm climate intervals (i.e., Medieval Warm Period, the Quaternary sediments was attributed to enhanced early Holocene thermal maximum) is of importance to freshwater leaching during the warmer-wetter interglacial assess the potential magnitude of changes permafrost periods. This finding was used to infer that geochemical environments may experience as a result of climate discontinuities in permafrost, combined with observations warming or disturbance. J.R. Mackay was one of the first on cryostructures, can become a powerful tool to infer to recognize the use of Quaternary stratigraphy and past conditions of permafrost (e.g., Mackay and identification of unconformities in frozen ground to infer Dallimore, 1992). past permafrost conditions, which evolved into the field of Recent studies of permafrost geochemistry focused on cryostratigraphy (e.g., Mackay et al. 1972; Mackay 1975; the active layer Ð permafrost boundary. For example, 1976). Paleo-thaw unconformity results from an increase studies on the geochemical composition of the active in active layer thickness followed by the aggradation of layer and near-surface permafrost from the Mackenzie permafrost and can be identified by distinct cryostructures Delta (Kokelj and Burn 2005; Lacelle et al. 2014), in the relict active layer or the former residual thaw layer Herschel Island (Kokelj et al. 2002) and the north slope of relative to the underlying permafrost (Murton and French Alaska (Ping et al. 1998; Bockheim et al. 1998; Keller et 1994; French 1998; French and Shur 2010). A well- al. 2007) have all shown that soluble ion content is lowest documented paleo-thaw unconformity is found in the in the active layer but can increase ca. 2 to 7 fold in the western Arctic where the active layer thickened during the top of permafrost. These studies have been conducted in early Holocene warm interval (Mackay 1974; Burn 1997). fine-grained material, which can contain significant In the absence of cryostratigraphic observations along amounts of unfrozen water at temperatures below 0¡C natural or anthropogenic exposures (i.e., coastal bluff, due to their higher specific surface area (Anderson et al. headwall of thaw slumps, mining section), alternative 1973; Williams and Smith 1989). As such, the increase in approaches must be used to assess the past permafrost soluble ions in the top of permafrost has been attributed to conditions. One of these approaches consists of leaching in the active layer and the downward transport of determining the concentration and distribution of soluble soluble ions during summer by water migrating along the ions in permafrost. This stems from the pioneering work negative soil thermal gradient. Therefore, the distribution on permafrost geochemistry done in the regions of of soluble ions in the top of permafrost may be used to Fairbanks and Barrow, Alaska, USA (e.g., O’Sullivan infer decadal to centennial stability of the active layer and 1966; Brown 1968; Péwé and Sellman 1973). These associated growth of the transient layer (e.g., Kokelj and studies demonstrated that soluble cations (Ca2+, Mg2+, Burn 2003; Lacelle et al. 2014). Na+, K+) and conductivity have a much lower The amount and distribution of soluble ions and salts concentration in sediments of interglacial ages (i.e., in active layer and permafrost provide a tool to assess Holocene, Sangamonian) relative to glacial age loess (i.e., present and past changes in active layer thickness, and

111 chemical weathering regimes in a given landscape (e.g., Soil water from saturated zones in the active layer can be Keller et al. 2007; Lacelle et al. 2014). However, results collected using a piezometer or shallow well. The and associated interpretations of the distribution and piezometer, consisting of a hollow steel or PVC pipe, is concentration of soluble salts and ions in the active layer installed in the soil with a screen mesh section placed at and permafrost are largely dependent on the the required sampling depth. The groundwater can be methodological approach used to extract the ions, and hand-pumped to the surface directly into a sample vial how the data are portrayed. Further, the extent to which through a Teflon tube (i.e., Lacelle et al. 2008). this approach can be used to identify past permafrost Unsaturated zone porewaters can be sampled using a conditions is unclear. This study first reviews various suction lysometer (i.e., Carey and Quinton 2005). methodologies used for the extraction of soluble salts and Alternatively, Rhizons samplers may be used in saturated ions in the active layer and permafrost, and then tests two and unsaturated soils to collect the small volume of pore different methods from two sites in northwestern Canada. water (i.e., Newman et al. 2015). The results are discussed with respect to the extraction Although not commonly used in continuous permafrost techniques, the distribution with depth of ions and salts of regions, direct sampling of soil water from the thawed different solubility, the timing of permafrost aggradation active layer has the advantage of characterizing the in situ following sediment deposition (i.e., syngenetic vs geochemical conditions of the water, calculating the epigenetic permafrost) and the history of permafrost (i.e., saturation of different minerals and assessing the relict active layer vs residual thaw layer). Although the evolution of the carbonate system, including PCO2 in the geochemistry of massive ice bodies has been used to soil water, a key component to the C-cycle in the active infer the source of water and the origin of the ice (e.g., layer (e.g., Lacelle et al. 2008). Mackay and Dallimore, 1992; Fritz et al. 2011), analyses of massive ice are not discussed here. 2.2 Extraction of pore water

The geochemistry of active layer and permafrost is commonly determined by indirect sampling of the water. Soil sampling pits are excavated in the active layer and underlying permafrost samples may be collected with a Cold Region Research and Engineering Laboratory (CRREL) core barrel and the soil samples are brought to the laboratory (e.g., Kokelj et al. 2002; Kokelj and Burn 2005). Extraction of porewaters is best done by decanting, squeezing, or centrifuging the sample. For the samples that are super-saturated, the soil water is readily collected by these methods. For under-saturated samples a known amount of deionized water is added to the sample to reach saturation (i.e. ca. 1:1 soil water ratio; Janzen, 1993), the sample is agitated for a few hours and the excess water is collected. Both the filtered super- saturated and leachate water are then analyzed for geochemical composition and ionic concentrations are corrected to field moisture content or to unit weight of dry soil.

2.3 Soil : water extractions

A similar approach to the extraction of pore water is the soil-water extraction technique. The active layer and Figure 1. Conductivity, sediment type and age at Eva permafrost samples are first brought to the laboratory and Creek Gold Placer mining cut, Fairbanks, Alaska are oven dried. To determine the most efficient and (modified from Péwé and Sellman 1973). complete extraction method (total soluble ions and salts), various soil-water ratios can be tested (1:1, 1:5, 1:10, 1:25, etc). The soil-water mixture is shaken for 1 hour, 2 REVIEW OF SAMPLING AND EXTRACTION centrifuged, and the leachate water decanted, filtered and TECHNIQUES analyzed for its ionic content. Figure 2 shows the results for a salt-rich permafrost sample collected near Navy Road (Inuvik, NT, Canada) using different soil-water A variety of methods are used to sample and analyze the ratios. Based on the results, soluble ions at this site are geochemistry of soil water and pore water in the active completely leached with a soil-water ratio > 1:5. Soils from layer and permafrost, including the soluble salts and ions. other regions may require different soil-water ratios due to Below, we review some of the most commonly used differing salt concentrations. It should be noted that the methods. pore water and soil:water extraction methods may 2.1 Direct sampling of soil water misrepresent cation concentrations due to cation-

112 exchange capacity, which is enhanced with the addition of 3 GEOCHEMISTRY OF ACTIVE LAYER AND low-salinity deionized water. PERMAFROST IN NORTHWESTERN CANADA

2.4 Acid extractions The Peel Plateau is an ice-rich hummocky moraine landscape that extends along the eastern margins of the For minerals that are insoluble at room temperature and Richardson Mountains (Catto 1996). The surface atmospheric PCO2 (i.e., carbonates), soil samples may be sediments consist of glacial sediments deposited during leached and digested with various acids. Some of the the Late Pleistocene advance and retreat of the most common procedures (e.g., Blum et al. 2002; Keller Laurentide Ice Sheet (Duk-Rodkin and Hughes 1992; et al. 2007; Conklin 2014) include initial leaching with 1M Lacelle et al. 2013). Climate data obtained from the Inuvik HNO3 at room temperature for 20 hours to dissolve easily and Fort McPherson meteorological stations indicates a soluble minerals (i.e., most carbonates, and minor amount mean annual air temperature of Ð7.3¡C, with the region of phosphates and sulfides). Subsequently, concentrated receiving <250 mm in annual precipitation. With mean HNO3 at 150¡C for 3 hours is used to partially digest less annual temperatures > Ð2.5¡C, permafrost is relatively soluble but still potentially weatherable minerals such as warm (O’Neill et al. 2015). biotite, plagioclase, potassium feldspar, sulfides, and In summer 2012 and 2013, active layer and near- some oxides and clays. However, considering that a surface permafrost samples were collected at four sites variety of acid extraction techniques are employed, the on the Peel Plateau and six sites in Mackenzie Delta reader should consult the literature on this topic to region. Here we present the results from two sites on the determine which one is best for their needs. Peel Plateau: i) CB slump: 67¡10’N, 135¡44’W, 595m; ii) Wilson slump: 67¡13’N, 135¡47’W, 610m. Active layer 2.5 Representing soluble salt and ion content in soils samples were collected at every 4 cm and placed in a sealed plastic bag. Permafrost samples were collected To calculate the soluble salt and ion content in soils, the with a hand-held Hilti power head with 3.5’’ diameter soil bulk density and ice content are required. Soluble salt diamond bit barrel by coring horizontally into the headwall contents (SSC) in the soil (expressed as equivalents mÐ2) of the slumps (Figure 3B). All samples were weighed at are calculated by summing over the required soil column the Aurora Research Institute laboratory (Inuvik, NT). the product of the salt concentration (C, meq kgÐ1), the When supernatant water was present in the active layer horizon thickness (T, m), the soil bulk density (BD, g cmÐ sample, the conductivity of the pore water was measured 3), and the weight fraction of the >2 mm soil fraction using a SympHony 3P90MS multimeter and the values and/or ice (CF): were then corrected to account for the amount of water in each sample. The melt water from ice-bearing permafrost [1] SSC = C x BD x T x (1-CF) was extracted, filtered, and conductivity was determined using the SympHony meter and δ18O and δD was This equation assumes that all of the soluble salts and determined using a LGR liquid isotope analyzer. ions in the soil are held in the <2 mm soil fraction. Using Gravimetric water content (GWC) was calculated for all this equation, the soluble salt and ion content in soils can samples after drying the wet sediment at 105oC for a be calculated for various soil depths. minimum of 24 hours. The total soluble ions in the active layer and permafrost were subsequently extracted using a 1:10 soil-water ratio. Samples were shaken for an hour, centrifuged, measured for conductivity and filtered prior to determination of cations and anions concentrations. The leachate conductivity and ionic values were corrected to account for the soil-water ratio.

3.1 CB slump

In late summer 2012, the depth of the active layer was 48 cm. Based on changes in cryostructures observed in the headwall of the slump (from reticulate to suspended), the depth of the paleo-thaw unconformity was inferred at 115 cm (Fig. 3). The GWC ranged from 35 to 170%. With the exception of the O horizon, GWC is lowest in the active layer and relict active layer (<60%) and highest at the top of permafrost and below the paleo-thaw unconformity. δ18O measurements were only made on the permafrost below the paleo-thaw unconformity where sufficient water Figure 2. Graph of soluble ion concentrations measured at was recoverable for analysis. The δ18O values of the different soil-water ratios. The permafrost sample (150 cm ground ice are highest immediately below the paleo-thaw depth) is from a hummock near Navy Road, Inuvik, NT, unconformity (ca. Ð26‰) and vary between Ð28.7 and Ð Canada. 27.3‰ below it. These values are similar to other sites in northwestern Canada where it has been suggested that

113 the ground ice formed during Late Pleistocene and has active layer (1.2-3.4 dS cm-1) and increases to 3.3 dS cm- not experienced thaw since (e.g., Mackay and Dallimore, 1 at the top of permafrost. Leachate conductivity values 1992; Lacelle et al. 2004; Murton et al. 2004; Fritz et al. progressively increase from the top of permafrost to the 2011). top of the paleo-thaw unconformity (115 cm depth). Below Like δ18O, pore water conductivity could only be this depth, leachate conductivity values range from 5.3 to measured in the permafrost below the paleo-thaw 7.2 dS cm-1. The dominant soluble ions in the active layer 2- 2+ 2+ unconformity. The pore water conductivity values follow a and permafrost are SO4 , Ca and Mg . Their similar trend to the GWC. Pore water conductivity is concentration is lowest in the active layer (<1000 meq 100 highest immediately below the paleo-thaw unconformity g-1 soils) and highest in the permafrost below the paleo- (2.0 dS cm-1) and then varies from 1.1 to 1.9 dS cm-1. The thaw unconformity (>1000 meq 100 g-1 soils). Cl- and - conductivity values determined from leachate (1:10 soil- NO3 , which are very soluble, have the lowest water ratio) is 3 to 6 times higher than the pore water concentration in the active layer and permafrost. values. Leachate conductivity values are lowest in the

Figure 3. A) Field photograph (July 2012) showing headwall of the CB slump on the Peel Plateau, NT, Canada. B-C) Photographs of two sections of the headwall showing a change in cryostructures, from reticulate to suspended. The contact between the two is interpreted as representing the early Holocene paleo-thaw unconformity. D) Gravimetric water content (GWC), δ18O, pore water and leachate (1:10 soil-water ratio) conductivity and leachate ion concentration in the active layer and shallow permafrost. The thick dashed line represents the active layer and the thin dashed line represents the paleo-active layer.

114 3.2 Wilson slump throughout the uppermost 3 m of permafrost, from 0.2 to 2.5 dS cm-1, the latter occurring as peaks at 80 and 192 In summer 2012, the active layer thickness in the cm depth, which corresponds to GWC maximum values. undisturbed terrain upslope of the Wilson slump was ca. Leachate conductivity values are 2 to 9 times higher then 45 cm. Based on the observed change from reticulate to those from the pore water. Leachate conductivity is lowest suspended cryostructures, the depth of the paleo-thaw in the active layer (<3.5 dS cm-1) and highest in the unconformity was inferred to be at 175 cm (Fig. 4). At this permafrost below the paleo-thaw unconformity (>5.1 dS site, the GWC ranged from 30 to 295% with peaks cm-1). Between the top of permafrost and the top of the occurring at 80 and 192 cm depths (295% and 200%, paleo-thaw unconformity, leachate conductivity values respectively). The GWC is lowest within the active layer have intermediate values, fluctuating between 3.8 and 6.5 (<50%). Unlike at the CB slump site, there is little dS cm-1, which contrast with the CB slump site. The difference in GWC above and below the paleo-thaw dominant soluble ions in the active layer and permafrost 18 2- 2+ 2+ unconformity. The δ O values of the ground ice show a are SO4 , Ca and Mg . Their concentration is lowest in progressive decrease from the bottom of the active layer the active layer (<2500 meq 100 g-1 soils) and highest (Ð22‰) to below the paleo-thaw unconformity where below the paleo-thaw unconformity (>5000 meq 100 g-1 values remain near Ð29‰. soils). Cl- and NO3- have the lowest concentrations and Pore water conductivity could only be measured in shows little variation with depth. the permafrost. Pore water conductivity fluctuates

Figure 4. A) Field photograph (July 2012) showing headwall of the Wilson slump on the Peel Plateau, NT, Canada. B-C) Two sections of the headwall showing a change in cryostructures, from reticulate to suspended. The contact between the two is interpreted as representing the early Holocene paleo-thaw unconformity. D) Gravimetric water content (GWC), δ18O, pore water and leachate (1:10 soil-water ratio) conductivity and leachate ion concentration in the active layer and shallow permafrost. The thick dashed line represents the active layer and the thin dashed line represents the paleo- active layer.

115 4 DISCUSSION 4.2 Source of soluble salts and ions

4.1 Effect of extraction technique Relations between the various ion species suggest that the source of SO4, Ca and Mg originates from the Soluble salts and ions in the active layer and shallow dissolution of gypsum or Ca+Mg sulfates (i.e., Malone et permafrost were determined using two different extraction al. 2013). Conversely, the Na/Cl ratio in the active layer at techniques: i) pore water analysis, a method derived from both sites approaches the sea-salt ratio (Rsea=0.86; studies on soil salinity developed in the context of Kleene et al. 1986), however the Na/Cl ratio is much agricultural soil studies (e.g., Conklin 2014); and ii) higher in the permafrost suggesting an alternate source of leachate analysis using a 1:10 soil-water ratio, the latter Na. Finally, NO3 shows little variations between the active reflecting total soluble ions (Figure 2). Using the pore layer and permafrost due to its high solubility. This topic water analysis technique, the soluble ion concentration is will be explored in greater details in a follow-up highly dependent on the amount of water present in the manuscript. sample, with more soluble ions being dissolved under higher water content (Figure 5). As such, the adaptation 4.3 Effect of history of permafrost on geochemical of the pore water analysis method to permafrost studies conditions may underestimate the total soluble ions associated with the soil. On the Peel Plateau, radiocarbon dating of horse Based on the active layer and permafrost samples mandible from a site nearby both slumps (Lacelle et al. used in this study, the pore water analysis technique 2013) and Bison priscus remains nearby Tsiigehtchic results in conductivity values being 2 to 9 times lower than (Zazula et al. 2009) constrains the presence of the the 1:10 soil-water extraction technique. As such, this Laurentide Ice Sheet between 18,500 and 13,500 14C cal method-dependent difference in soluble salt and ion BP, which deposited ca. 10-50 m of glacial sediments concentration may lead to different interpretations of the (Duk-Rodkin and Hughes, 1992). Permafrost in the newly results, especially on the abundance and distribution of deposited glacial sediments is likely syngenetic and soluble ions in active layer and permafrost profiles. For formed during the Aller¿d-B¿llering and Younger Dryas example, at the CB and Wilson slump sites, both soluble periods. During the early Holocene warm interval, near- ion extraction methods show increased conductivity in the surface permafrost experienced deep thaw, with active top of permafrost and just below the paleo-thaw layer thickness approaching 2 to 2.5 times that of today, unconformity; however the depth and magnitude of solute as evidenced from the change in cryostructures in the enrichments varies (Figures 3 and 4). At the Wilson headwall of slumps (e.g., Burn 1997). Following this warm slump, leachate conductivity shows solute enrichment at interval, permafrost subsequently aggraded to present 56 cm depth (6.5 dS cm-1), whereas it is found at 80 cm day conditions. Based on this history of permafrost and depth (2.5 dS cm-1) based on pore water conductivity. the ground ice content at both sites (Figures 3 and 4), Type F permafrost is found on the Peel Plateau (e.g., Jorgenson et al. 2010), which corresponds to ice-rich syngenetic permafrost overlain by a layer of thawed and refrozen soils containing lower ice content. At the CB and Wilson slump sites, total soluble ions concentration is highest below the paleo-thaw unconformity as a result of the syngenetic permafrost conditions that prevented leaching in the newly deposited glacial sediments and preserved soluble ions. We suggest that soluble salt and ions in type F permafrost may be 1-2 orders of magnitude higher below the maximum historical depth of thaw (i.e., below the paleo-thaw unconformity) relative to the active layer. In the case of epigenetic permafrost, one is not to expect similar soluble ion gradients across the paleo-thaw unconformity as substantial leaching could have occurred in the soils prior to the aggradation of permafrost. Future studies should investigate the distribution and concentration of soluble ions in Type A-E permafrost of the Jorgenson et al. (2010) classification. According to French (1998) and French and Shur (2010), permafrost above the historical maximum depth of thaw is classified either as: i) a relict active layer, if during Figure 5. Graph of conductivity measured in the pore its development the depth of seasonal freezing and water and leachate water (1:10 soil-water ratio). Pore thawing reaches the top of permafrost, or ii) a residual water conductivity is related positivity to field water thaw layer if the depth of seasonal freezing remained content. Variation in conductivity of the leachate water above the permafrost table. Assuming all other conditions provides a standardization that allows for comparison remain the same, the frequency and magnitude of between samples. freshwater leaching is likely different in a relict active layer

116 compared to in a residual thaw layer. For example, Lacelle et al. (2008) demonstrated that chemical weathering is highest at the base of the active layer, where water and PCO2 availability is higher, which leads to increased leaching and weathering; this could be enhanced within a residual thaw layer that remains perennially unfrozen for extended periods of time. Figure 6 compares the distribution and concentration of total soluble ions at the CB and Wilson slumps. Although similar abundance and vertical trends are observed within the active layer, differences are observable between the top of permafrost and the paleo- thaw unconformity and also below this unconformity. The low total soluble ions throughout the active layer is attributed to cumulative seasonal freshwater leaching since the time of deposition of the glacial sediments (i.e., > 15,000 years). A first peak in total soluble ions is Figure 6. Total soluble ions at the (A) CB slump, (B) observed just below the top of permafrost, corresponding Wilson slump. (C) Cumulative total soluble ions at both to a peak in ice content, and these likely developed by sites. downward migration of water and leached soluble ions during the summer along a negative thermal gradient over decadal to centennial time-scale (e.g., Cheng et al. 1983; ACKNOWLEDGEMENTS Mackay 1983; Kokelj et al. 2005). Between the top of permafrost and the paleo-thaw unconformity, the total This research was supported by NSERC Discovery Grant soluble ions at the CB slump site increases from the base and Northern Supplement, Polar Continental Shelf of the active layer to the top of the paleo-thaw Program, Northern Scientific Training Program and the W. unconformity; whereas at the Wilson slump, total soluble Garfield Weston Award for Northern Research. ions have a slightly lower abundance and vary little with depth within that zone. Below the paleo-thaw unconformity, the total soluble ions at both sites is greater REFERENCES than 400 mg 100 g-1 soils with a peak in abundance just below the thaw unconformity. However, the peak in Anderson, D.M., Tice, A.R., and McKim, H.L. 1973. The abundance is ca. 1.2 times higher at the Wilson site (650 unfrozen water and the apparent specific heat mg 100 g-1 soils). The distinct patterns at both sites are capacity of frozen soils. Proceedings of the Second tentatively interpreted as the CB slump site developing a International Conference on Permafrost, Yakutsk, relict active layer, whereas at the Wilson slump, where the USSR, National Academy of Sciences, Washington, unconformity occurs at much greater depth, resulted in DC. the development of a residual thaw layer; although the Blum, J.D., Klaue, A., Nezat, C.A., Driscoll, C.T., timing and duration of this layer is unknown. The Johnson, C.E., Siccama, T.G., Eagar, C., Fahey, T.J., demonstrate the effect of a residual thaw layer on soluble and Likens, G.E. 2002. Mycorrhizal weathering of salts and ions in permafrost, future studies should apatite as an important calcium source in base-poor investigate modern sites where the depth of seasonal forest ecosystems. Nature, 417, 729Ð731. freezing remained above the permafrost table. Given the Bockheim, J. G., Walker, D. A., Everett, L. R., Nelson, increase in winter air temperature and snow cover, the F.E., and Shiklomanov, N. I.. 1998. Soils and development of residual thaw layers in continuous cryoturbation in moist nonacidic and acidic tundra in permafrost may become more extensive. the Kuparuk River Basin. Arctic and Alpine Research, 30(2), 166-174. Brown, R. J. E., 1968: Occurrence of permafrost in 5 CONCLUSIONS Canadian peatlands. In: International Peat Congress. 3rd, Quebec, 18-23 August, Proceedings, 174-181. The following conclusions can be reached: Brown, J. 1969. Ionic concentration gradients in 1. Different extraction techniques of soluble salts and ions permafrost, Barrow, Alaska. Research Report 272. in active layer and permafrost can yield different patterns U.S. Army Cold Regions Research and Engineering of variation with depth. The use of a common extraction Laboratory, Hanover, New Hampshire, 25 p. technique of soluble salts and ions in the active layer and Burn, C.R. 1997. Cryostratigraphy, paleogeography, and permafrost is likely required to make results from different climate change during the early Holocene warm studies comparable. interval, western Arctic coast, Canada. Canadian 2. Combined with observations of cryostructures, Journal of Earth Sciences, 34, 912-925. analyses of pore water and leachate soluble ions can be Carey, S.K., Quinton, W.L. 2005. Evaluating runoff effective tools to infer past thaw depths in syngenetic generation during summer using hydrometric, stable permafrost. However, their application to epigenetic isotope and hydrochemical methods in a permafrost remains unclear. discontinuous permafrost alpine catchment. Hydrological Processes, 19(1), 95-114.

117 Catto, N.R., 1996. Richardson Mountains, Yukon- Lacelle, D., M. Fontaine, A.P. Forest, and S. Kokelj. 2014. Northwest Territories: The northern portal of the High-resolution stable water isotopes as tracers of postulated “Ice Free Corridor.” Quaternary thaw unconformities in permafrost: A case study from International, 32: 3Ð19. western Arctic Canada. Chemical Geology, 368, 85- Conklin, A. 2014. Introduction to soil chemistry: analysis 96. and instrumentation (2nd ed.). John Wiley & Sons, Mackay, J.R. 1974. Reticulate ice veins in permafrost, Inc.: Hoboken, New Jersey, USA. Northern Canada. Canadian Geotechnical Journal, 11, Duk-Rodkin, A., Hughes, O.L. 1992. Surficial geology, 230-237. Fort McPherson-Bell River. Yukon-Northwest Mackay, J.R. 1975. Relict ice wedges, Pelly Island, Territories. Geological Survey of Canada, Map 1745A, N.W.T. (107 C/12). In Report of Activities, Geological scale 1:250 000. Survey of Canada, Paper 75-1 A, pp. 469-470. Ecosystem Classification Group. 2010. Ecological regions Mackay, J.R. 1976. Pleistocene permafrost, Hooper of the Northwest Territories Ð Cordillera. Department Island, Northwest Territories. In Report of Activities, of Environment and Natural Resources, Government Part A, Geological Survey of Canada, paper 76-1 A, of the Northwest Territories, Yellowknife, NT, Canada. pp. 17-18. French, H.M. 1998. An appraisal of cryostratigraphy in Mackay, J.R., Rampton, V.N. and Fyles, J.G. 1972. Relic north-west Arctic Canada. Permafrost and Periglacial Pleistocene permafrost, western Arctic, Canada. Processes, 9(4), 297-312. Science, 176, 1321-1323. French, H., and Y. Shur. 2010. The principles of Mackay, J.R., Dallimore, S.R. 1992. Massive ice of the cryostratigraphy. Earth-Science Reviews, 101(3-4), Tuktoyaktuk area, western Arctic coast, Canada. 190-206. Canadian Journal of Earth Sciences, 29, 1235-1249. Fritz, M., Wetterich, S., Meyer, H., Schirrmeister, L., Murton, J.B., French, H.M. 1994. Cryostructures in Lantuit, H., Pollard, W.H. 2011. Origin and permafrost, Tuktoyaktuk coastlands, western arctic characteristics of massive ground ice on Herschel Canada. Canadian Journal of Earth Sciences, 31(4), Island (western Canadian Arctic) as revealed by 737-747. stable water isotopes and hydrochemical signatures. Murton, J.B., Waller, R.I., Hart, J.K., Whiteman, C.A., Permafrost and Periglacial Processes, 22(1), 26-38. Pollard, W.H, Clark, I.D. 2004.Stratigraphy and Janzen, H.H. 1993. Soluble salts. In Soil sampling and glaciotectonic structures of permafrost deformed methods of analysis. Edited by M.R. Carter. Lewis beneath the northwest margin of the Laurentide ice Publishers, Boca Raton, Fla., pp. 161Ð166. sheet, Tuktoyaktuk Coastlands, Canada. Journal of Keller K., J.D. Blum and G.W. Kling. 2007. Geochemistry Glaciology 50, 399-412. of soils and streams on surfaces of varying ages in Newman, B.D., and 11 co-authors. 2015. Arctic Alaska. Arctic, Antarctic and Alpine Research, Microtopographic and depth controls on active layer 39(1), 84-98. chemistry in Arctic polygonal ground. Geophysical Kokelj, S.V., C.A.S. Smith, and C.R. Burn. 2002. Physical Research Letters 42, 1808-1817. and chemical characteristics of the active layer and O’Neill, H.B., Burn, C.R., Kokelj, S.V., Lantz, T.C., 2015. permafrost, Herschel Island, western Arctic Coast, “Warm” Tundra: Atmospheric and near-surface ground Canada. Permafrost and Periglacial Processes, 13, temperature inversions across an alpine treeline in 171-185. continuous permafrost, western Arctic, Canada. Kokelj, S.V. and C.R. Burn. 2005. Near-surface ground Permafrost and Periglacial Processes. ice in sediments of the Mackenzie Delta, Northwest Péwé, T.L., Sellman, P.V. 1973. Geochemistry of Territories, Canada. Permafrost and Periglacial permafrost and Quaternary stratigraphy. Permafrost: Processes, 16, 291-303. North American Contribution to the Second Lacelle, D., Bjornson, J., Lauriol, B., Clark, I.D., Troutet, International Conference, 13-28 July, Yakutsk, Y., 2004. Segregated-intrusive ice of subglacial Siberia. 166-170. meltwater origin in retrogressive thaw flow headwalls, Ping, C. L., Bockheim, J. G., Kimble, J. M., Michaelson, Richardson Mountains, NWT, Canada. Quaternary G. J., and Walker, D. A., 1998: Characteristics of Science Reviews, 23: 681Ð696. cryogenic soils along a latitudinal transect in Arctic Lacelle, D., V. Juneau, A. Pellerin, B. Lauriol, and I.D. Alaska. Journal of Geophysical ResearchÐ Clark. 2008. Weathering regime and geochemical Atmospheres, 103: 28917Ð28928. conditions in a polar desert environment, Haughton Streletskii, D.A., Streletskaya, I.D., Rogov, V.V., Leibman, impact structure region, Devon Island, Canada. M.O. 2003. Redistribution of ions within the active Canadian Journal for Earth Sciences, 45, 1139-1157. layer and permafrost, Yamal, Russia. In: M. Philiips, Lacelle, D., Lauriol, B., Zazula, G., Ghaleb, B., Utting, N., S.M. Springman, L.U. Arenson (eds.) Permafrost: Clark, I.D., 2013. Timing of advance and basal Proceedings of the 8th International Conference on condition of the Laurentide Ice Sheet during the last Permafrost, Zurich, Switzerland, 20-25 July 2003. A.A. glacial maximum in the Richardson Mountains, NWT. Balkema Publishers, 2, 1117-1122. Quaternary Research, 80: 274Ð283. Williams, P. J., and M.W. Smith. 1989. The Frozen Earth. Fundamentals of Geocryology. Studies in Polar Research Series, 306 pp. Cambridge, New York, Port Chester, Melbourne, Sydney: Cambridge University Press.

118 Topoclimatic controls on active-layer thickness, Alaskan Coastal Plain

Frederick E. Nelson Department of Earth, Environmental, and Geographical Sciences Northern Michigan University, Marquette, MI, USA

Melanie A. Schimek Department of Geography, University of Delaware, Newark, DE, USA

ABSTRACT Although many studies of active layer development address spatial variations in soil thermal properties and moisture conditions, few have attempted to isolate topoclimatic influences on active-layer thickness. Observed patterns of thaw depth on various facets of an anthropogenic thermokarst landform near Prudhoe Bay, Alaska show systematic variation with slope orientation. This study demonstrates that aspect has a strong control on active-layer thickness, even at 70oN latitude, with south-facing slopes thawing to greater depth than north-facing slopes and northern exposure compensating even for the absence of an insulating layer of vegetation.

RÉSUMÉ Bien que de nombreuses études sur le développement de la couche active traitent de la variabilité spatiale dans les caractéristiques thermiques du sol et les conditions d'humidité, seulement quelques-unes ont tenté d'isoler les influences topoclimatiques sur l'épaisseur de la couche active. Les profondeurs de dégel mesurées sur une forme thermokarstique anthropique près de Prudhoe Bay, en Alaska démontrent une relation systématique entre l’épaisseur de la couche active et l'orientation de la pente. Cette étude prouve que l’orientation a un grand contrôle sur l'épaisseur de la couche active, même à 70¡N de latitude. Par exemple, la couche active sur les pentes exposées au sud dégèle à une plus grande profondeur que celles exposées au nord, de même que l’exposition au nord qui sert d’alternative à l'absence d'une couche de végétation isolante.

1 INTRODUCTION 2 STUDY AREA AND DEVELOPMENT HISTORY

Topography exerts substantial controls over local climate, Field investigations were undertaken during the summer creating pronounced differences in climatic parameters of 1981 on an abandoned road constructed during the over short lateral distances. These climatic contrasts, in early stages of oilfield development at Prudhoe Bay, turn, affect the rates at which geomorphic processes Alaska. The portion of the road investigated lies adjacent operate, and ultimately the form of the land surface. to the Deadhorse Airport, the main public air-transport These effects can be particularly pronounced in facility servicing the Prudhoe Bay oilfield (Figure 1). permafrost landscapes, which are highly susceptible to Nelson and Outcalt (1982) provided a description of the climatic influences where underlain by ice-rich sediments. road and a preliminary analysis of data collected from it. The term “topoclimatology,” introduced by C.W. The Prudhoe Bay oilfield is located on the northern edge Thornthwaite (1954), is a shorthand reference to the study of Alaska’s Arctic Coastal Plain physiographic region of the influence of Earth’s surface terrain on micro- and (Wahrhaftig 1965), an extensive area of low relief with meso-scale climate. Topoclimatic studies encompass tundra vegetation, shallow thaw lakes, drained lake diverse scientific disciplines, including plant ecology, basins, and anastomosing rivers. Much of the landscape habitat suitability modeling, energy balances, hydrology, is saturated with water during the short summer season. soil and ground water processes, geomorphology, The marine silts underlying the region contain relatively agriculture, and urban climate. Topoclimatic relations are large volumes of ice, mostly in the upper 5 to 10 m. Areas a key element in simulation studies of permafrost of low-centered ice-wedge polygons are typical of the distribution in mountainous areas. Prudhoe Bay region. A comprehensive description of the Surprisingly little empirical research has been region was provided by Walker et al. (1980), and the completed in mapping geographic variations in active- cumulative impacts of oil field activities have been layer thickness (ALT) on hillslopes. Robust statistical data described and analyzed by Walker et al. (1986) and are necessary to validate modeling efforts. Nelson et al. Raynolds et al. (2014). (1997) mapped active-layer thickness in the Kuparuk The discovery of large oil reserves at Prudhoe Bay in region of Alaska’s North Slope using a computational 1968 necessitated development of a means to transport algorithm that included a topoclimatic index to control for matériel to the North Slope. This major oil and gas aspect-induced differences. The data set used to calibrate discovery initiated strong pressure to develop overland topoclimatic parameterizations in that study is described transport facilities. During the winter of 1968-69, the State in detail in this paper. of Alaska constructed a 735 km road to transport materials to the North Slope. From Livengood to Bettles

119 the road followed an existing winter trail used by tractor into deep ruts, necessitating creation of new roadbeds trains when the airfield at Bettles was built in the early around eroded segments (Haynes, 1972). 1940s. The road from Bettles, south of the Brooks Range, A decade after construction, the road’s morphology to Prudhoe Bay on the coast of the Beaufort Sea, was consisted of two parallel trenches separated by a new (Figure 1). The oil companies built and operated a relatively stable berm (Figure 2). Areas where the winter road from Sagwon to the Prudhoe Bay oilfield. This insulating tundra vegetation was removed were eroded section of road was built between December 1968 and deeply within two years of the disturbance (Haynes, March 1969, at a cost of US$766,000 (Anonymous, 1972). Only ten years after the initial disturbance, two 1970). meters of subsidence had occurred in some areas To create a relatively level surface in hummocky and adjacent to the road’s flanks (Nelson and Outcalt, 1982). polygonal terrain, a rectangular 7 x 4 m steel frame with a Thermal erosion generated by running water in the bladed two-ton (1818 kg) crossbar was dragged across the sections removed sediment, annually exposing new layers surface (Anonymous, 1970). In wet areas requiring more of ice-rich permafrost and resulting in further settlement. than surface leveling, two parallel strips of tundra were The , underlain by three layers of organics, bulldozed and the mats stacked atop each other (Nelson is a dry and compacted microenvironment on which less and Outcalt, 1982). The result was an elevated berm on vegetation grows than the surrounding tundra. Permafrost which vehicles could operate. The mode of construction aggradation has occurred within the road berm, owing to resulted in the berm containing a triple layer of organic the insulating effects of the organic layers, and only material. Excavation of the road surface indicated that 10- shallow thawing occurs on the road surface. The road has 20 cm of the bladed organic layer was removed from each been modified primarily by erosion along its flanks through section now occupied by thermokarst troughs. Drainage slumping of organic-rich blocks and by water flowing in patterns have been severely disrupted by subsidence in the adjacent troughs. Examination of sequential these parallel troughs, creating a local base level for photographs demonstrates that slope angles on the running water in the surrounding area and eroding the thermokarst road have relaxed over the several decades tops of adjacent ice wedges. since our initial observations. Troughs are gradually filling At the time of construction, the road was the longest of in with material sloughed off the road’s flanks. In winter, its type ever built in Alaska (Anonymous, 1969). Spring snow drifts into the troughs, filling them completely. The thaw forced closure of the road only one month after its effect of deep snow cover protects the troughs from the completion in the year of construction (Johns, 1969). The full brunt of low winter temperatures. road caused extensive damage to the tundra insulation by removing the protective mat of vegetation and snow. Thermokarst processes transformed the damaged areas

Figure 1. Map of winter haul road in northern Alaska, Adapted from Anonymous (1970).

120 In spring, the troughs collect warm meltwater from the west-facing slopes. Measurements were recorded on five tundra, quickly melting and suspending any snow occasions during the course of the 1981 summer: 2 July, remaining in them. The transition of troughs from snow 14 July, 2 August, 19 August, and 1 September. Each catchment areas to running meltwater can occur in less sample was comprised of 50 replications of thaw depth than two days (Nelson and Outcalt, 1982). over the same marked section of the road used to make slope-angle measurements. Validation of the mechanical probing methodology was carried out on several dates during the thaw season, using a YSI 419 thermistor probe and a Wheatstone bridge measuring resistance to locate the level at which temperature in the substrate was 0.0oC (cf. Mackay 1977). Calibration was achieved using an ice bath prior to field work each day. The thermal probe was pushed into the substrate in 5 cm increments, allowing sufficient time for temperature to reach equilibrium. A very high degree of agreement was found between the mechanical and thermal probed values, except on the final day of sampling (September 2), by which time freezeback from the top of permafrost had been sufficient to cause large differences between the two procedures (Figure 3). Ice bonding of refreezing sediment at the base of the active Figure 2. Well-preserved section of the thermokarst road layer had progressed far enough by this date that south of Deadhorse airport during the summer of 1980. upfreezing (Mackay 1979, 1984) was detected, even in View is to west. the mechanically probed samples. The trough bottom was the sole exception, owing to the presence of ponded 3 METHODOLOGY water at some sampling points along this transect.

3.1 Sampling procedures The study area contains an east-west and a north-south section of the 1969 road between the Sagavanirktok River and Beechey Point (70˚11.4’ N 148˚28.9’ W), providing a rectilinear surface facing each of the cardinal directions. The east- and west-facing segments of road where data were collected lie northwest of the Deadhorse airfield, while the north- and south-facing segments are south of the airfield. To obtain an accurate characterization of slope steepness, 50 measurements of slope inclination were made with a Brunton compass at equal intervals of distance along each inclined facet of the road (Table 1).

Table 1. Slope inclination and orientation on thermokarst road. All values are expressed in degrees. Figure 3. Comparative results from mechanical and Facet min max mean std dev orientation thermal probing.

North 17.0 57.0 29.9 7.6 000 Daily air temperatures for the 1981 summer season South 11.0 51.0 24.7 8.4 180 were obtained from weather observations made atop the East 11.5 41.0 22.4 6.6 080 air traffic control tower at the Deadhorse airport. Because West 12.0 38.5 21.6 6.0 260 the airport and the site are in close proximity, air temperature measurements were assumed to be Measurements of thaw progression on various facets of representative of the site’s location. Hourly measurements the road were recorded throughout the summer of 1981 were only recorded at the air traffic control tower during by mechanical probing of the active layer. Thaw depth the time period 1400 to 0200 (local time) for the duration was determined using a standard 1 cm diameter Cold of the summer season. Mean daily temperature was Regions Research and Engineering Laboratory (CRREL) calculated by averaging the temperatures recorded at calibrated steel probe. The data set consists of thaw- 0200 and 1400 so as not to bias the results toward the depth measurements taken throughout the summer on the time of day when temperatures were recorded. The road surface, a trough bottom paralleling the road, north- cumulative degree days of thaw (DDT) was computed by and south-facing slopes, undisturbed tundra paralleling summing mean daily temperatures from June 1 through the road, and (on alternate sampling dates) east- and September 2.

121

Figure 4. Thaw progression versus the square root of thawing degree-day sums (DDT) on various topographic elements of the thermokarst road throughout the summer of 1981.

3.2 Statistical analysis 4 RESULTS

Descriptive statistics were computed for all data sets. The Figure 4 shows the “Stefan relation” for the 1981 data set, Lilliefors (1967) test was used to examine the conformity i.e., thaw progression versus the square root of thawing of sample frequency distributions with the normal degree-day sums on each facet of the road during the (Gaussian) probability density function. Four of the summer of 1981. At locations with relatively simple ground samples from July, when the active layer was in an early cover and homogeneous soil columns, the Stefan relation stage of development, showed minor departures from shows thaw penetration progressing as a linear function of normality at the 0.1 level. the square root of accumulated degree-days. Figure 4 Sets of samples for each measurement date were reflects the diverse orientations and material properties of analyzed using one-way analysis of variance (ANOVA). the thaw-depth samples collected from and adjacent to Because ANOVA is robust with respect to slight the thermokarst road. Thaw progressed at divergent rates departures from normality and heterogeneity of variance in different parts of the road, and several of the data when equal sample sizes are employed (Kirk, 1982, p.78), traces cross during the summer. A pronounced decrease the samples were deemed appropriate for ANOVA in the depth of thaw is also apparent between the last two procedures. Computed F values are significant at the 0.01 sampling dates, indicating that freezeback from the top of level for all sampling dates. the permafrost table (Mackay, 1979, 1984) affected most To determine which pairs of samples differ from one sampling locations. another, each sample was compared with all others from Comparative analysis of mean values and assessment the same measurement date using Tukey’s “honestly of their statistical significance can be achieved efficiently significant difference” (HSD; Tukey 1991; Hsu 1996) using box plots (Andrews et al., 1980). Figures 5 and 6 multiple comparison procedure (MCP). Comparisons were illustrate which facets of the thermokarst road also made using Dunnett’s MCP (Kirk, 1982, 112-118) to experienced significantly different values of thaw depth at compare sample means against a control, in this case approximately two-week intervals over the summer of samples obtained from the undisturbed tundra. 1981. Tukey’s HSD (Figure 5) procedure compares each sample with all others to determine if differences exist between topographic facets. Pronounced differences in thaw depth are apparent on slopes facing various

122

Figure 5. Summer 1981 box plots from Tukey’s multiple comparison procedure, 0.05 significance level. Sample means are compared to all other sample means. Confidence intervals for each sample are displayed as the upper and lower bounds of each box, and the line through the middle of the box represents the sample mean. When any portion of the box falls into the same area on the vertical scale as another box the two are not significantly different. If the upper and lower limits of two boxes do not overlap, the corresponding samples are considered to be significantly different.

UND = undisturbed tundra BO = bottom of “outer” (south) trough SS = south-facing slope RS = road surface NS = north-facing slope WS= west-facing slope ES = east-facing slope

123

Figure 6. Box plots showing thermokarst road directional facets relative to undisturbed tundra. These figures are graphical representations of results from Dunnett’s multiple comparison test, 0.05 level of significance. This procedure compares sample means to a control only; other samples cannot be compared to each other. Undisturbed tundra serves as the control and dotted lines are the bounds of the confidence interval.

UND = undisturbed tundra BO = bottom of “outer” (south) trough SS = south-facing slope RS = road surface NS = north-facing slope WS= west-facing slope

ES = east-facing slope

124 directions, and rankings change over the course of the they provide interesting opportunities for thaw season. geomorphological and topoclimatic field experiments in Dunnett’s test (Figure 6) uses the undisturbed tundra areas of otherwise homogeneous terrain. as a “control,” i.e., as the basis for comparison with thaw depth on other slope facets. Dashed lines show the calculated confidence interval for the undisturbed tundra. ACKNOWLEDGEMENTS Any sample that does not intersect the confidence interval of the undisturbed tundra can be considered significantly FEN extends thanks to Cecil Goodwin, Jeanne Makihara, different than undisturbed tundra. Jim Akerman, and Therese Spies for collaborative Results from MCPs for the thaw-depth data show that assistance along the very long road leading to this report. on 2 July the undisturbed tundra had experienced significantly shallower thaw than all other samples. By 2 August, the undisturbed tundra and the road surface were REFERENCES not significantly different, having a similarly thin active layer, while the north-facing slope had significantly greater Andrews, H. P., Snee, R. D., and Sarner, M. H. 1980. thaw than the undisturbed tundra. Maximum thaw values Graphical display of means. The American were recorded on 19 August. The south-facing slope had Statistician , 34(4): 195-199. the greatest average thaw on this date. Tukey’s HSD Anonymous. 1969. Alaskans built 400-mile road to North shows the depth of thaw on the south-facing exposure to Slope in one season. Contractors & Engineers be significantly greater than on all other slopes. The 19 Magazine, 66(4): 42-43. August and 1 September graphs for Dunnett’s procedure Anonymous. 1970. Winter road in northern Alaska, 1969- show that thaw depth on the road surface was 70. The Polar Record, 15(96): 352-355. significantly less than that of the undisturbed tundra and Haynes, J.B. 1972. North Slope oil: physical and political north-facing slope, which are not significantly different. problems. The Professional Geographer, 24(1): The triple organic layer on the road surface acts to 17-22. insulate the underlying permafrost, even though this Hsu, J.C. 1996. Multiple Comparisons: Theory and surface begins to thaw before the others in late spring, Methods. Chapman & Hall, New York, NY, USA, owing to its greater exposure to deflation of snow by wind. 277 pp. The ALT data demonstrate that, for road facets with Johns, G.E. 1969. A winter road. The Northern Engineer. similar material properties, north-facing slopes experience 1(3): 12-13. minimal end-of-season thaw and south-facing slopes Kirk, R.E. 1982. Experimental Design: Procedures for the experience the greatest thaw. The similarity of the Behavioral Sciences. Wadsworth, Inc., Belmont, samples from undisturbed tundra and north-facing slope, CA, USA: 911 pp. in particular, is remarkable. Even at 70oN, a north-facing Lilliefors, H. W. 1967. On the Kolmogorov-Smirnov test for exposure compensates for the absence of insulation normality with mean and variance unknown. provided by a vegetation cover and organic layer. Journal of the American Statistical Association, 62(318): 399-402. Mackay, J.R. 1977. Probing for the bottom of the active 5 CONCLUSIONS layer. Geological Survey of Canada Paper 77- 1A, 327-328. Active-layer thickness is modified by topoclimatic Mackay, J.R. 1979. Uplift of objects by an upfreezing processes and parameters at the surface, including air surface. Canadian Geotechnical Journal, 16: and soil temperature on each exposure, solar radiation, 609-613. surface cover, and slope angle. Mackay, J.R. 1984. The frost heave of stones in the active This study demonstrates that slope aspect exerts a layer above permafrost with downward and strong control over active-layer thickness, even at high upward freezing. Arctic and Alpine Research, 16: latitudes, with south-facing slopes thawing to greater 439-446. depth than north-facing slopes and northern exposure Nelson, F. E., and Outcalt, S. I. 1982. Anthropogenic compensating for the lack of an insulating layer of geomorphology in northern Alaska. Physical vegetation. Geography , 3 (1): 17-48. When 1981 averages are compared to those from Nelson, F.E., Shiklomanov, N.I., Mueller, G.R., Hinkel, 1995, 2006, and 2007 (Nelson and Schimek, unpublished K.M., Walker, D.A., and Bockheim, J.G. (1997). data), ALT is 10-15 cm greater in the latter years. Estimating active-layer thickness over a large The triple organic layer beneath the road surface acts region: Kuparuk River basin, Alaska, U.S.A. as an insulator, resulting in decreased active-layer Arctic and Alpine Research, 29(4): 367-378. thickness values even though this surface begins thawing Raynolds, M.K., Walker, D.A., Ambrosius, K.J., Brown, J., sooner, owing to its earlier emergence from snow cover. Everett, K.R., Kanevskiy, M., Kofinas, G.P., The rate of thaw on the road surface decreases Romanovsky, V.E., Shur, Y., and Webber, P.J. significantly when the thawing front contacts the buried 2014. Cumulative geoecological effects of 62 organic layers. years of infrastructure and climate change in ice- Although features such as the road investigated in this rich permafrost landscapes, Prudhoe Bay study impart great damage in permafrost environments,

125 Oilfield, Alaska. Global Change Biology, 20(4): 1211-1224. Thornthwaite, C. W. 1954. Topoclimatology. In T. W. Wormell et al. (Ed.), Proceedings of the Toronto Meteorological Conference 1953. Royal Meteorological Society , London, UK, 227-232. Tukey, J. W. 1991. The philosophy of multiple comparisons. Statistical Science, 6(1): 100-116. Wahrhaftig, C. 1965. Physiographic Divisions of Alaska. U.S. Geological Survey Professional Paper 482, 52 pp. + maps, appendices. Walker, D.A., Everett, K.R., Webber, P.J., and Brown, J. 1980. Geobotanical Atlas of the Prudhoe Bay Region, Alaska. U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, NH, USA, CRREL Report 80-14, 69 pp. Walker, D.A., Webber, P.J., Binnian, E.F., Everett, K.R., Lederer, N.D., Nordstrand, E.A., and Walker, M.D. 1987. Cumulative impacts of oil fields on northern Alaskan landscapes. Science, 238(4828): 757-761.

126 Permafrost degradation adjacent to snow fences along the Dempster Highway, Peel Plateau, NWT

H.B. O’Neill and C.R. Burn Department of Geography and Environmental Studies – Carleton University, Ottawa, ON, Canada

ABSTRACT The long-term ground thermal effects of a snow fence in continuous permafrost (Fence 1) were examined on Peel Plateau, west of Fort McPherson, NWT. Active-layer thicknesses and vegetation changes were also described at three additional fences (Fences 2-4). The fences were erected in the early 1980s, so present environmental conditions represent the effects of over 30 years of modification to snow conditions. We observed increased snow cover, active- layer thickness (ALT), and moisture content at Fence 1, which have combined to prevent winter freezeback, so that a talik now exists at the site. ALTs were statistically related to distance from the fence at flat sites, but not at sites of slope ≥5¡ with good drainage.

RÉSUMÉ Les effets thermaux à long terme d’une clôture de neige située dans le pergélisol continu (Fence 1) ont été examinés sur Plateau Peel, à l’ouest du Fort McPherson, TNW. Des changements d’épaisseur du mollisol et de la végétation près de trois clôtures additionnelles ont aussi été décrits (Fences 2-4). Les clôtures ont été érigées au début des années 1980, par conséquent les conditions environnementales actuelles représentent les effets des modifications des conditions de la neige sur plus de 30 ans. Nous avons observé une augmentation de la couverture de neige, de l’épaisseur du mollisol, et de la teneur d’humidité à Fence 1, dont les effets combinés ont empêché le regel de l’hiver, ce qui a eu pour conséquence de créer un talik à ce site. L’épaisseur du mollisol a été statistiquement reliée à la distance à partir de la clôture sur les sites plats, mais pas sur les sites où la pente est 5 avec du drainage.

1 INTRODUCTION subsidence around the fence has created the potential for a stream to form along the structure, which might drain an Snow fences are common engineering structures adjacent lake – the town’s source of drinking water (Nolan throughout cold regions. They are typically constructed to 2010). reduce snow accumulation and increase visibility at Past research has identified the ground thermal infrastructure by reducing wind velocity and promoting effects of enhanced snow accumuation at fences (Mackay snow deposition near the fence (Freitag and McFadden 1978, 1993; Hinkel et al. 2003; Karunaratne and Burn 1997). Snow fences are economic because their 2003; Hinkel and Hurd 2006; Burn et al. 2009; Lafrenière construction cost is easily recuperated by reducing et al. 2013), and reported on ground subsidence and expenditures on snow removal (Tabler 1991). In flat vegetation changes near the structures over several years ground, snow fences with a porosity of 50 % placed following their erection (e.g. Hinkel and Hurd 2006; Burn perpendicular to the prevailing wind may trap snow within et al. 2009; Johansson et al. 2013). However, to our a distance of 15H in the upwind direction and 30H in the knowledge, none have examined longer-term (multi- downwind direction, where H is the height of the snow decadal) effects of snow fences in different topographic fence (Tabler 1991, Fig. 9). settings. In permafrost regions, snow fences alter the ground In an applied context, these topics should be explored thermal regime because they trap blowing snow, limiting near highways because snow fences may initiate ground heat loss and prolonging active-layer freeze back. thermokarst processes in close proximity to the road As a result, near-surface ground temperatures in winter embankment. Infrastructure managers may benefit from are considerably higher near snow fences (Mackay 1978, an understanding of the long-term response of permafrost 1993; Hinkel et al. 2003; Kurunaratne and Burn 2003; to these installations, to help plan future projects. Hinkel and Hurd 2006; Burn et al. 2009). Over time, the Therefore, the objectives of this study were to (1) examine warming effect on the ground may cause subsidence if ground thermal effects of a long-standing snow fence ice-rich permafrost degrades (e.g. Hinkel and Hurd 2006). along the Dempster Highway on Peel Plateau, west of While snow fences may solve issues of local snow Fort McPherson, NWT, and (2) describe resulting active- accumulation and visibility, they may become problematic layer thicknesses and vegetation changes near this fence, if the thermal regime in ice-rich terrain is disturbed. For and at three additional snow fences in the region. example, in Kaktovik, Alaska, the construction of a snow fence to reduce snow accumulation in the village has caused a network of ice-wedge polygons to degrade. The

127 2 BACKGROUND

Snow cover is the most important local control on ground temperatures in the continuous permafrost zone (Mackay and MacKay 1974; Zhang 2005). The latent heat content of the active layer also influences the freezeback period and resulting permafrost temperatures. Complex relations exist between snow cover, ground thermal conditions, active layer soil moisture, and vegetation in tundra environments (e.g. Sturm et al. 2001; Kokelj et al. 2010; Morse et al. 2012). For example, winter ground temperatures commonly rise as tall shrubs establish and trap snow, causing an increase in active-layer thickness (ALT) (Sturm et al. 2001). However, ALT may thin over time as the shrub canopy develops (e.g. Mackay and Burn 2002; Blok et al. 2010), despite relatively high winter ground temperatures (Gill et al. 2014).

2.1 Study area Figure 1. Location of study sites. Fences 2 and 4 are at The study was conducted along the Dempster Highway approximately 67.179¡N 135.776¡W and 67.104N on Peel Plateau and in Richardson Mountains, between 136.096W, respectively, in Richardson Mountains. the Yukon/NWT border and Peel River (Figure 1). The region is in the continuous permafrost zone. The Dempster Highway is an all-season gravel road that 3 METHODS connects the communities of Fort McPherson, Tsiigehtchic, and Inuvik, NWT to southern Canada. The 3.1 Study design and sites highway is important to the communities because it facilitates the flow of goods and services and supports In 2012-2014, snow depth, ALT, shallow ground regional tourism. The Dempster Highway transportation temperatures, ground surface topography, and vegetation corridor is currently being extended to connect with the were examined at a snow fence (Fence 1; Figure 2) Beaufort Sea coast at Tuktoyaktuk. located on Peel Plateau, adjacent to the Midway The Peel Plateau region has a continental climate with emergency airstrip. ALT and vegetation heights were long, cold winters and short, cool summers. The mean measured at three other snow fences (Fences 2-4) in the annual air temperature (1987-2006) is -7.0 ¡C at the Fort region, to capture these conditions in a range of McPherson airport, the nearest long-term meteorological topographic settings. Fence 1 was in a poorly-drained station, which was in operation until 2007. Total annual peatland (Figure 2). The ground near the fence was precipitation averages 295 mm, with 148 mm falling as saturated, and wetland vegetation dominated near the rain (Environment Canada 2012). Precipitation falls mostly fence. Fence 2 was on a 20¡, well-drained tundra slope in in late summer and early fall. The dominant winter winds Richardson Mountains. Fence 3 was in relatively flat, blow from the north. shrubby ground, but was well drained in comparison with The substrate on Peel Plateau consists of fine- Fence 1. Fence 4 was on a well-drained tundra slope (5¡) grained, glacially-derived sediments underlain by marine in Richardson Mountains. The fences were erected in the shale and siltstone bedrock (Norris 1984). In Richardson early 1980s, and are approximately 1 m high. Mountains, near-surface sediments mostly consist of discontinuous till and colluvial deposits (Fulton 1995). 3.2 Field measurements and analyses Permafrost temperatures in the tundra environment on Peel Plateau are relatively high (~-2 ¡C) partly due to Snow depths were measured at Fence 1 by probing in strong winter air temperature inversions (O’Neill et al. March 2013 and 2014 every 2 m along a 40 m transect 2015). In addition, permafrost is characteristically ice-rich normal to Fence 1, to characterize snow accumulation in the near surface, and may contain massive ground ice near the fence. ALTs were estimated by probing along the at depth (Kokelj et al., 2013; Lacelle et al. 2015). As a transect in August 2012, 2013, and 2014, with probes of result, permafrost is sensitive to disturbances that affect either 150 or 175 cm length, to determine whether the the surface energy balance. snow fence had caused near-surface permafrost degradation. ALTs were also estimated by probing at Fences 2-4, typically every 2 m along transects that ran 50 m from either side of the structures. Some transects were shorter than 50 m because of the proximity of the fences to the road embankment. Two transects were established at Fence 4 because of the structure’s considerable length.

128 and distance from the snow fences. All statistical tests were conducted at the 0.05 significance level.

4 RESULTS

4.1 Fence 1

4.1.1 Snow depths

Snow was much deeper in 2013 than in 2014 on Peel Plateau (O’Neill et al. 2015). Consequently, snow depths near Fence 1 in 2013 were up to 150 cm, while further from the structure they were <1 m. In 2014, snow depths near the fence were about 1 m, and at either end of the transect ranged from about 30 – 80 cm (Table 1; Figure 3). Though the depths were varied between the two years of measurement, snow depths near the fence were consistently greater (Table 1; Figure 3).

4.1.2 Ground temperatures

Annual mean ground temperatures (AMGT) were higher near Fence 1 than at the control site. On the south side of Fence 1, AMGT at 100 cm depth was 0.6 ¡C in 2012-2013 (Table 2), but the instrument was inundated with melt water in 2014, so a value for the second year could not be determined.

Table 1. Active-layer thickness (ALT) and snow depths along the transect at Fence 1, 2012-2014. The date format is MM/DD/YYYY.

Dist. ALT (cm) Snow depth (cm) Figure 2. Field sites in summer 2014. Locations of the (m) 08/16/ 08/14/ 08/08/ 04/19/ 04/25/ transect (line) and shallow thermistor arrays (dots) are 2012 2013 2014 2013 2014 shown for Fence 1. 0 61 44 49 74 72 2 50 47 52 92 70 4 60 43 45 102 71 Ground temperatures were monitored with shallow 6 67 61 37 105 72 thermistor arrays at three locations at Fence 1 (Figure 2), 8 63 77 40 116 89 to compare conditions near the fence with those in 10 78 90 50 137 89 12 88 100 60 145 107 undisturbed terrain. The South and North instruments 14 97 120 92 150 96 were beside the fence. The Control instrument was placed 16 114 130 105 142 97 in nearby undisturbed terrain, 25 m upwind of the fence in 18 110 155 150+ 143 101 the direction of prevailing winter winds (N). The arrays 20 150+ 175+ 150+ 143 97 consisted of thermistors (HOBO TMC6-HA/D) placed at 5 22 100 175+ 150+ 140 100 and 100 cm depth, and attached to data loggers (HOBO 24 150+ 175+ 132 139 106 H08-006-04), which recorded temperatures every 2 hours. ------Fence------The thermistors have a reported accuracy of ±0.25 ¡C 26 150+ 175+ 150+ 141 101 28 130 125 150+ 139 100 over a measurement range of -40 to 50 ¡C, and about 30 140 140 150+ 138 85 ±0.45 ¡C precision with the logger used. 32 110 128 133 135 48 The ground surface profile along the transect at Fence 34 80 129 60 95 38 1 was surveyed in summer 2012 with an optical level. 36 62 41 57 90 33 Dominant vegetation species were recorded along the 38 70 66 50 88 33 transect. Vegetation height was measured along the 40 55 70 48 87 38 transects at Fences 2-4 to examine the effect of the fences on shrub vegetation. Vegetation heights were not measured at Fence 1 because low sedges, not shrubs, dominated there. The Spearman rank correlation test was used to examine associations between ALT and vegetation height

129

Figure 3. Cross-section of Fence 1 showing the ground surface, snow depths, and active-layer thicknesses along the transect. The gray shading represents sedge- dominated vegetation. Thermistor arrays are shown as black rectangles.

On the north side, AMGT at 100 cm depth was 0.1 and 0.0 ¡C in 2012-2013 and 2013-2014, respectively. At the control site, AMGT at 100 cm depth was lower (-0.8 and - 1.1 ¡C) in each of the two years of record. Temperatures near the ground surface (5 cm depth) remained high (> -2 ¡C) during both winters beside Fence 1, but reached considerably lower values (< -8 ¡C) at the control site

(Figure 4). Figure 4. Ground temperatures at (a) 5 cm and (b) 100 cm In summer, ground temperatures at 5 cm at the control depth near the snow fence and at the control site. site are consistently lower than at the sites beside the fence (Figure 4). This is likely due to the contrasting moisture and surface vegetation conditions, discussed The ground around Fence 1 has subsided noticeably below. as a result of the disturbed thermal conditions and active-

layer deepening. We estimate that the subsidence is

about 50 cm (Figure 3). As a result, the ground is much Table 2. Annual mean ground temperatures (AMGT) at 5 wetter around the fence from water accumulation in cm and 100 cm (bold) at the snow fence and control site subsided areas (Figure 2). in 2012-2014. Minimum temperatures are shown in brackets beside the annual means. 4.2 Active-layer thickness and topography

Instrument AMGT (¡C) AMGT (¡C) 5 100 There was a significant negative correlation -0.20, p < 2012-13 2013-14 2012-13 2013-14 (ρ = South 2.2 (-1.4) n/a (-2.4) 0.6 (-0.2) n/a (-0.6) 0.01, n = 164) between ALT and the distance from North 1.8 (-1.5) 1.3 (-2.4) 0.1 (-0.6) 0.0 (-0.6) Fences 1-4 (Figure 5). However, the effect of the snow Control -0.2 (-4.8) -1.3 (-8.9) -0.8 (-1.5) -1.1 (-3.4) fences on the active layer was not consistent among the fences, but varied with topography. ALTs at the flattest sites (Fences 1 and 3) were statistically associated with 4.1.3 Active-layer thickness and ground subsidence distance from the fences (ρ = -0.59, p < 0.01, n = 47), but not at sites on sloping ground (Fences 2 and 4, Figure 5, ALTs were considerably greater near Fence 1 than near ρ = -0.13, p = 0.15, n = 117). the ends of the transect (Table 1, Figure 3). In each year, the thaw depth near the fence extended below the 4.3 Vegetation maximum length of the measurement probes, so the greatest ALTs could not be determined. The mean (2012- Two distinct vegetation responses were observed near 2014) ALTs at points further than 15H (~15 m) from the the snow fences. First, the vegetation community has fence was 65 cm in the upwind (N) direction, and within changed at Fence 1 from shrub-tundra species to wetland 15H, the mean was 141 cm. The transect only extended vegetation (Figure 2). At the control site in undisturbed 15 m beyond the fence in the downwind direction, due to tundra, cloudberry (Rubus chamaemorus) and mosses the proximity of the highway embankment. However, dominated vegetation cover, with some low-lying shrubs ALTs on the south side of the fence were also greater such as dwarf-birch (Betula glandulosa), cranberry than in the control areas (Figure 3). (Vaccinium vitis-idaea), and Labrador tea (Rhododendron During active-layer probing in August 2014, a thin (5- 10 cm) frozen layer was encountered near the fence at a depth of about 50 cm. This frozen layer was penetrated with the probe, and underlain by unfrozen ground, indicating that a talik now exists near the fence.

130 freezing season (Figure 4). In summer, ground temperatures at 5 cm depth were also typically higher near the fence than at the control site. This may occur since the saturated ground near the fence has a higher thermal conductivity than the dryer organic material at the control site. Johansson et al. (2013) also observed this effect in northern Sweden at snow fences in a peatland. The active layer is considerably thicker near the fence than at the control site, and appears to have increased by at least 1.25 m. This increase in ALT has resulted in an estimated 50 cm of ground surface subsidence, presumably from the thaw of near-surface ground ice. Subsidence around the fence has caused moisture to Figure 5. Active-layer thicknesses along transects at all accumulate, and there has been a shift from shrub-tundra fences in 2014. The blue data points are from sites on flat vegetation to moisture-tolerant species. These long-term ground and the red points are from sites on slopes. effects of the snow fence have caused a talik beneath Fence 1, indicated by probing in summer 2014. The ground beneath the snow fence now likely acts as a heat subarcticum) also present. However, vegetation beside source to degrade adjacent permafrost year-round. the fence consisted mostly of moisture-tolerant sedges In summary, the thermal disturbance from increased (Carex spp.), with limited cover of cloudberry, dwarf-birch, snow cover at Fence 1 has caused substantial permafrost mosses, and willows (Salix spp.). The second vegetation degradation and ground subsidence. These effects have response observed was a change in shrub structure at been amplified by poor drainage and saturated ground at Fences 2-4. Pooled data of vegetation height was the site. significantly and negatively correlated (ρ = -0.22, p < 0.01, n = 142) with the distance from the fences (Figure 6), 5.2 Active-layer thickness, topography, and vegetation indicating that tundra shrubs have increased in height response near Fences 2-4. At Fences 1 and 3, in relatively flat ground, ALTs were thicker near the fences than at other sites surveyed. No similar systematic effects were observed in sloping terrain at sites 2 and 4 (Figure 5). This may be because water, either from thawed ground ice or precipitation, is able to drain away and does not increase the latent heat content of the active layer that must dissipate during freeze back. However, this should be investigated further, as the number of fences examined was limited, and ground ice conditions were unknown at Fences 2-4. There appeared to be some ground subsidence at Fences 2-4, though not nearly to the extent observed in the wet environment at

Fence 1. Figure 6. Vegetation heights along transects at Fences 2- Higher shrubs were observed in proximity to the 4. The blue data is from the site in flat ground and the red structures at Fences 2-4 (Figure 2, Figure 5). This data are from sites on slopes. suggests that the snow fences enhance shrub growth, likely due primarily to protection from wind abrasion in winter within the snow pack, and to changed ground 5 DISCUSSION thermal, active layer, moisture, and nutrient conditions (e.g. Sturm 2001). However, at Fence 1, the significant 5.1 Long-term permafrost degradation and thermokarst depression around the fence and associated saturated initiation at Fence 1 conditions has instead caused establishment of sedge vegetation. Therefore, it appears that snow fences in well- Considerable snow cover was present in late winter at drained terrain are generally an anthropogenically- Fence 1. O’Neill et al. (2015) placed an iButton array at induced source of shrub enhancement for tundra, but at Site 1 and observed that deep (> 90 cm) snow cover poorly-drained sites, they may cause a shift in vegetation accumulates around the structure by early November, community. The observation of shrub enhancement near limiting heat loss early in the freezing season. infrastructure in the region is similar that observed by Gill Near-surface ground temperatures were higher near et al. (2014) beside the Dempster Highway embankment. Fence 1 than at the nearby control site, and above 0 ¡C on an annual basis. Seasonally, winter ground surface 6 SUMMARY AND CONCLUSIONS temperatures (5 cm depth) were considerably higher near the fence due to the insulating effect of deep snow cover, (1) A snow fence in operation for ~30 years alongside the but also due to release of latent heat throughout the Dempster Highway in the continuous permafrost zone has

131 caused deep snow accumulation, which has resulted in Environment Canada. 2012. Climate Data Online. permafrost degradation. Available from [31 January 2012]. (2) Active-layer thicknesses are significantly greater near Freitag, D.R., and McFadden, T. 1997. Introduction to fences in flat ground, but are similar to undisturbed tundra Cold Regions Engineering. ASCE Press, New York, at fences on slopes, presumably because water is able to N.Y., U.S.A. drain away in sloped areas. Fulton, R.J. 1995. Surficial Materials of Canada, Geological Survey of Canada, Map 1880A, Scale (3) Vegetation has responded to disturbed conditions near 1:5,000,000. snow fences. Increased shrub growth has occurred at Gill, H.K., Lantz, T.C., O’Neill, H.B., and Kokelj, S.V. well-drained sites, while sedges now dominate the 2014. Cumulative impacts and feedbacks of a gravel vegetation cover at a poorly-drained site. road on shrub tundra ecosystems in the Peel Plateau, Northwest Territories, Canada, Arctic, Antarctic, and (4) Shallow ground temperatures (5 cm) remain higher Alpine Research, 46: 947–961. near the fence than at the control site for most of the Hinkel, K.M., and Hurd, J.K. 2006. Permafrost summer, likely due to higher thermal conductivity in the destabilization and thermokarst following snow fence saturated ground near the structure. installation, Barrow, Alaska, U.S.A., Arctic, Antarctic, and Alpine Research, 38: 530-539. The results of this investigation highlight the effect of Hinkel, K.M., Bockheim, J.G., Petersen, K.M., and Norton, anthropogenic disturbance on thermokarst initiation in D.W. 2003. Impact of snow fence construction on tundra of the continuous permafrost zone, and potential tundra soil temperatures at Barrow, Alaska, 8th unforeseen effects of snow fences near linear International Conference on Permafrost, Balkema, transportation infrastructure, particularly in poorly drained Lisse, Zurich Switzerland, 1: 401-405. areas. Snow fences should, if possible, be placed in well- Johansson, M., Callaghan, T.V., Julia, B., Akerman, H.J., drained terrain to minimize permafrost degradation. Jackowicz-Korczynski, M., and Christensen, T. R. (2013), Rapid responses of permafrost and vegetation to experimentally increased snow cover in sub-artic ACKNOWLEDGEMENTS Sweden, Environmental Research Letters, 8 035025, doi:10.1088/1748-9326/8/3/035025. This research was supported by the Natural Sciences and Kokelj, S.V., Riseborough, D., Coutts, R., and Kanigan, Engineering Research Council of Canada, the Northwest J.C.N. 2010. Permafrost and terrain conditions at Territories Cumulative Impacts Monitoring Program, the northern drilling-mud sumps: Impacts of vegetation Northwest Territories Geological Survey, the Tetlit and climate change and the management implications, Gwich’in Renewable Resources Council, the Northern Cold Regions Science and Technology, 64: 46-56. Scientific Training Program of Aboriginal Affairs and Kokelj, S.V., Lacelle, D., Lantz, T.C., Tunnicliffe, J., Northern Development Canada, the W. Garfield Weston Malone, L., Clark, I.D., and Chin, K. 2013. Thawing of Foundation, and the Aurora Research Institute. The massive ground-ice in mega slumps drives increases research is a contribution from the NSERC Frontiers in stream sediment and solute flux across a range of ADAPT program and the Network of Expertise in Northern watershed scales, Journal of Geophysical Research Transportation Infrastructure Research. We are grateful 118: 681-692. for field assistance from Abe Snowshoe, Steven Tetlichi, Lacelle, D., Brooker, A., Fraser, R.H., and Kokelj, S.V. Christine Firth, and from Jeff Moore, Emily Cameron, 2015. Distribution and growth of thaw slumps in the Krista Chin, and Blair Kennedy. The research was Richardson Mountains–Peel Plateau region, conducted with the kind permission of G. Jagpal, Regional northwestern Canada, Geomorphology, 235: 40-51. Superintendant, Department of Transportation. S.V. Lafreniere, M.J., Laurin, E., and Lamoureux, S.F. 2013. Kokelj and G. Jagpal provided helpful comments that The impact of snow accumulation on the active layer improved the manuscript. We also thank Andrea Perna for thermal regime in high Arctic soils, Vadose Zone J. translating the abstract. 12, DOI:10.2136/Vzj2012.0058 Karunaratne, K.C., and Burn, C.R. 2003. Freezing n- factors in discontinuous permafrost terrain, Takhini REFERENCES River valley, Yukon Territory, Canada, 8th International Conference on Permafrost, Balkema, Blok, D., Heijmans, M.M.P.D., Schaepman-Strub, G., Lisse, Zurich Switzerland, 1: 519-524. Kononov, A.V., Maximov, T.C., and Berendse, F. Mackay, J.R. 1978. The use of snow fences to reduce ice- 2010. Shrub expansion may reduce summer wedge cracking, Garry Island, Northwest Territories. In permafrost thaw in Siberian tundra. Global Change Current Research, part A. Geological Survey of Biology, 16: 1296-1305. Canada Paper 78-1A, 523–24. Burn, C.R., Mackay, J.R., and Kokelj, S.V. 2009. The Mackay, J.R. 1993. Air temperature, snow cover, creep of thermal regime of permafrost and its susceptibility to frozen ground, and the time of ice-wedge cracking, degradation in upland terrain near Inuvik, N.W.T., western Arctic coast, Canadian Journal of Earth Permafrost and Periglacial Processes, 20: 221-227. Sciences, 30: 1720–1729.

132 Mackay, J.R., and Burn, C.R. 2002. The first 20 years (1978-1979 to 1998-1999) of active-layer development, Illisarvik experimental drained lake site, western Arctic coast, Canada, Canadian Journal of Earth Sciences, 39: 1657-1674. Mackay, J.R. and MacKay, D.K. 1974. Snow cover and ground temperatures, Garry Island, N.W.T., Arctic, 27: 287-296. Morse, P.D., Burn, C.R., and Kokelj, S.V. 2012. Influence of snow on near-surface ground temperatures in upland and alluvial environments of the outer Mackenzie Delta, Northwest Territories, Canadian Journal of Earth Sciences, 49: 895-913. Nolan, M. 2010. The State of the Arctic from above, on, and below the Surface: permafrost. Available from [5 March 2015]. Norris, D.K. 1984. Geology of the northern Yukon and northwestern District of Mackenzie. Geological Survey of Canada, Map 1581A, scale 1:500,000. O'Neill, H.B., Burn, C.R., Kokelj, S.V. and Lantz, T.C. 2015. ‘Warm’ tundra: atmospheric and near-surface ground temperature inversions across an alpine treeline in continuous permafrost, western Arctic, Canada, Permafrost and Periglacial Processes, 26: doi: 10.1002/ppp.1838. Sturm, M., McFadden, J.P., Liston, G.E., Chapin III, F.S., Racine, C.H., and Holmgren, J. 2001. Snow-shrub interactions in Arctic tundra: a hypothesis with climatic implications, Journal of Climate, 14: 336-344. Tabler R.D. 1991. Snow fence guide. Strategic Highway Research Program, National Research Council, Washington, D.C. Zhang, T. 2005. Influence of the seasonal snow cover on the ground thermal regime: an overview, Reviews of Geophysics, 43: 1-23.

133

134 The thermo-mechanical behavior of frost- cracks over ice wedges: new data from extensometer measurements.

Denis Sarrazin & Michel Allard Centre d’études nordiques, Université Laval, Québec city, Canada

ABSTRACT Specially adapted extensometers were deployed across eight frost cracks over ice wedges on Bylot Island in order to measure the timing of cracking and width variations of the open cracks over the winter. The cracking-contraction- expansion data were correlated with atmospheric and ground thermal temperature data acquired by automated meteorological stations and thermistor cables. Analysis of the data shows that narrow, sub-millimeter-size cracks first open abruptly early in winter when the active layer is frozen back. The cracks abruptly expand later when permafrost temperature falls below -10 ¡C. They keep enlarging over the winter, reaching widths between 6.8 and 18.2 mm by the end of March when permafrost temperatures oscillate around -18 to -20 ¡C. Short and small width variations in winter are associated with warmer spells of a few days duration. The cracks narrow by the end of the winter with warming ground and air temperatures. But they stay open at about half of their maximum width for several weeks in May-June, at the time of snowmelt.

RÉSUMÉ Des extensomètres spécialement adaptés furent déployés en travers de huit fentes de gel localisées au-dessus de coins de glace à l’Ile Bylot afin de déterminer le moment de la fissuration et de mesurer durant l’hiver les variations en largeur des fissures ouvertes. Les données de fissuration-contraction-expansion ont été corrélées avec des données de température de l’air et du sol acquises à l’aide de stations météorologiques et de câbles à thermistances automatisés. L’analyse des données révèle que d’étroites fissures de l’ordre du millimètre et moins s’ouvrent d’abord au début de l’hiver au moment où la couche active finit de regeler. Les fissures s’élargissent abruptement ensuite lorsque le pergélisol passe à des températures inférieures à -10 ¡C. Au cours de l’hiver, elles s’élargissent encore davantage, atteignant vers la fin de mars des largeurs entre 6.8 et 18.2 mm pendant que la température du pergélisol oscille autour de -18 à -20 ¡C. De petites et brèves variations en largeur des fissures en hiver sont associées à des redoux de quelques jours. Les fissures diminuent en largeur vers la fin de l’hiver quand les températures de l’air et du sol se réchauffent. Mais elles restent encore ouvertes d’environ la moitié de leur largeur maximum pendant quelques semaines en mai et juin, au temps de la fonte nivale.

1 INTRODUCTION Grechischev (1973). But we owe to John Ross Mackay the results of innovative field measurements that tested Tundra polygon networks which result from repeated the existing theories and brought a unique understanding thermal contraction cracking and ice-wedge growth are of the climate and terrain conditions that regulate frost among the most prominent features in terrain underlain by cracking of the ground (Mackay, 1974, 1975, 1978, 1984, permafrost. It has been thoroughly documented that 1992, 1993), soil movements in polygonal terrain (2000) frozen soils are affected by thermal contraction when the and, even, the growth of new cracks in newly exposed ground surface is submitted to deep winter cold terrain on drained lake bottoms (Mackay and Burn, 2002). temperatures, giving way to the opening of networks of Several field studies to measure the atmospheric cracks. Over the years, the repeated filling of the cracks and ground temperature conditions that induce frost in the permafrost with percolating surface water, mainly cracking made use of electrical cables laid across runnels from snowmelt in early summer, builds up the ice-wedges or cracks along the sides of tundra polygons. Mackay was that delineate tundra polygons. Thermal contraction truly a pioneer in that matter, starting in 1967 (Mackay cracking is also a widespread process that affects 1992). The principle of the breaking electric cables is that infrastructures such as road and airport runways, often a live wire that is laid across a potential crack is severed leading to major maintenance and repairs works. when the crack opens to release tensile stress generated Leffingwell’s pioneering field descriptions and by intense ground cooling at low temperatures. The interpretations in 1919 provided an understanding of the electric circuit is then open and the time of breaking is frost cracking and ice-wedge formation processes that recorded with a proper data logging device. Breaking has proven exact up to the present. The physics behind times (date and hour) of a set of cables can help interpret the thermo-mechanical behavior of frost cracks, ice temperature conditions under which cracking events took wedges and development of tundra polygons was place when recorded air and ground temperature data for theoretically formulated by Lachenbruch (1962, 1966) the study site are available, for example, from thermistors, whereas similar thinking based on observations was also dataloggers and automated meteorological stations presented and discussed by Russian authors such as (Allard and Kasper 1998; Fortier and Allard, 2005).

135 The breaking cable method has yielded significant with cold spells. The results show the potential of using and fundamentally important results. Generally, the that technology to monitor frost cracks dynamics. observed temperature values and the measured air and This study presents results obtained with newly ground cooling rates that induce frost cracking more or adapted extensometers laid across known frost cracks less match the theoretical calculations of Lachenbruch and ice wedges near Centre d’études nordiques’s (1962,1966). Comparable results from different regions research station on Bylot Island. with similar soil conditions (fine grained organic-rich soils of the Mackenzie delta region, from Salluit in Nunavik and 2 THE BYLOT ISLAND STUDY SITES from Bylot Island in Nunavut) but with different temperature and climate regime confirm the universality of The study sites are located in a glacial outwash the thermo-mechanical principles responsible for the frost valley on the southwestern side of Bylot Island cracking process, the required ground temperature (73¡09’38’’N, 80¡00’43’’W) (Figure 1). The valley plain is conditions and the robustness of the physics that describe bordered on both sides by an aggradation terrace 3 to 4 m it (Fortier and Allard, 2005; table 6). However, the above the glacio-fluvial outwash (sandur) surface. The measurement method is not perfect and the validity of the sediment cover is made of windblown silt eroded from the results remains subject to caution. Indeed, when a cable outwash plain and deposited over an organic-rich wetland. is buried in the active layer it becomes an integral part of Since this accumulation of fine-grained organic-rich the frozen soil after freeze back and normally would be permafrost has been taking place for over 3500 years, the expected to break when a crack opens. Some looseness soil stratigraphy consists of a series of layers of ice-rich may however be present around the buried cable allowing silty peat. This aggradation terrace is crisscrossed by a for a slight adjustment when it is put under tension; also complex network of large syngenetic ice wedges metal wires, mesh and rubber or plastic sheeting of wires polygons, ponds, tundra lakes, kettle lakes and pingos can be stretched to some extent and not break right away. (Allard, 1996; Fortier and Allard, 2004; Fortier and Allard, For instance, Mackay (1974) noted that despite observed 2005). crack openings, 5 to 10 % of his cables did not break. Allard and Kasper (1998) and Fortier and Allard (2005) also noticed that the plastic sheeting of some bigger gauge cables would be stretched at the crack location; the metallic wire of some cables would break, therefore detecting a crack expansion event, but without severing the sheeting. Partial closing of the cracks under soil expansion would close back the circuit. This fortuitous discovery demonstrated that open cracks widen and narrow and that the permafrost alternately contracts and expands in winter with colder and warmer temperature spells. Other methods to detect the timing of frost cracking can be terrain observations in winter (Christiansen, 2005) and the use of accelerators (shock loggers) buried in the ground (Matsuoka, 1999). Terrain observations are subjected to many misses because of one’s lack of capability to be constantly present on site and because of snow cover variations in thickness, density, etc. The timing of vibrations created by sudden movements, such as opening of cracks, is recorded by accelerators and interpreted with temperature data and sometimes concurrently with breaking cables (Matsuoka and Christiansen, 2008). The drawback with this method is that there is no absolute identification of the vibration source such as a given ice wedge or crack. Vibrations in the soil can be created by various minor cracks and sudden deformations at some other locations in the surrounding terrain. Nevertheless Matsuoka and Christiansen’s results show that many vibration events may be recorded over one winter near an ice wedge, likely due to dilation and cracking at times different than the breaking cables which, again, suggests that breaking cables probably leaves many cracking events undetected. Figure 1. Top: location of Bylot Island. Bottom: location of Extensometers were also used across polygon troughs by the study sites. Sites CMP and PP are at the north end of Matsuoka (1999) and by Matsuoka and Christiansen the upper lake. Sites PD are near the large lake in (2008) who measured winter dilation phases in relation forefront.

136 As calculated from the 1999 to 2010 period, the mean annual air temperature (MAAT) at the study site is - 3 METHODS 14.4¡C. The active layer is about 40 cm deep. The freezing index varies from 4700 to 6000 degree-days. The The extensometers used consist of rotating mean volumetric water content is 72% after thaw. At every potentiometers originally designed in a joint venture field season, abundant open frost cracks are visible along between the Finnish Forest Research Institute and the polygon sides and in troughs throughout the landscape. University of Oulu to measure tree stem diameter Many “virtual” cracks can also be found cutting through variations for dendrology work (Pesonen et al., 2004). the moss cover along polygons sides until late in the Instead of having the instrument tied as a girth around a summer. Coring through many ice veins in the active layer tree, the free end of the instrument’s stainless-steel band typically shows annual veins penetrating into ice wedges was screwed to a steel corner iron driven to the top of the below the permafrost table (figure 2a). Indeed the permafrost (maximum thaw depth in late summer). The permafrost and the climate conditions are conductive to instrument itself was firmly attached with tie-wraps to widespread and repeated frost-cracking as air another corner iron anchored into the ground on the other temperatures fall below -10 ¡C in October and stay in the - side of a crack (Figure 3). The electrical potentiometer in 20 and -30 ¡C range until the end of April. Temperature the instrument converts the length of extended stainless- variations and cold spells occur frequently. Published steel band into electrical resistance values that can be breaking cable data indicate average temperatures of - recorded with a Campbell Scientific datalogger. Tests for 34.3¡C in the air and of -18.6 ¡C at the permafrost table at forestry applications show that the instrument is thermally cable breaks (interpreted as openings of cracks). The insensitive with an accuracy of 0.03 mm. The steel average air cooling rate before crack opening was anchors were set about 25 cm apart, i.e. about 12 cm on calculated at -0.5 ¡C/h over a mean period of about 17 each side of a crack. The height of the steel band and the hours (see Fortier and Allard, 2006 for more details). top of the steel corner anchors was only about 3-5 cm above the soil surface. The extensometers were deployed across eight cracks in two distinct networks separated by a distance of 1700 m. The selected cracks were on flat terrain; that is to say over permafrost wedges not having ramparts and runnels (example in figure 2b). This set up makes it sure that only the crack opening, widening, narrowing and closing are measured and that no dilation-contraction movements either between ridges or between troughs and ridges interfere with the measurements of a single crack width. Being set deep enough and protruding only 3-5 cm above ground also ensure that tilting of the anchors with surface freezing and thawing is not affecting the measurements either. A plastic box was placed upside down over the extensometer to prevent malfunction from snow cover, freezing rain and damage by animals. All sites are equipped with thermistors cables, some of which are designed to monitor the ground thermal regime at the surface and at depths of 40 cm (about the permafrost table) and 80 cm (in the permafrost). A CEN’s SILA automated meteorological station measures atmospheric temperatures. The two field sites were also equipped with snow depth recorders (Campbell Scientific SR-50 snow depth recorders). The cracking-contraction-expansion data were graphed over time together with atmospheric and ground thermal temperature data for interpretation.

Figure 2. A. Ice veins penetrating into wedge ice at the permafrost table. B. A 2 m wide ice-wedge under flat terrain.

137 by a little less than a millimeter on 10 and 13 November when air temperatures fell in the -20 to -25 ¡C range. When extensometer PD_93521 started to open, the temperature was -11 ¡C near the soil surface, -4.5 ¡C 40 cm deep, about at the permafrost table, and -2 ¡C at the depth of 80 cm, in the permafrost. Then it expanded more rapidly to 3 mm at the same time as PD_93577 suddenly opened and expanded to more than 2 mm as temperatures fell to -18 ¡C near the surface, -10 ¡C at the permafrost table and -8 ¡C at the depth of 80 cm. PD_93522 had a somewhat different behavior: it opened gradually to 1 mm from late October when near surface soil temperatures fell below 0 ¡C until 29 November, on which day it suddenly expanded to about 4 mm. That day of abrupt opening, air temperatures were -24 ¡C and surface soil temperature was about -24 ¡C; it was -15 ¡C Figure 3 An extensometer installed across a crack. at the depth of 40 cm and -13 ¡C at the depth of 80 cm.

4 RESULTS

4.1 Thermo-mechanical behavior of frost-cracks

Two of the installed systems failed. Over the six years with data, the behavior of all the working extensometers was similar and repetitive (Figures 4, 5 and 6). We therefore present detailed results obtained from six extensometers tried at the two sites (PD and CMP; Figure 1) over the winters 2009-2010 and 2010- 2011 (Table 1). The first two ones (PD_95577 and Cmp_93698) were first installed in summer 2005. The four others (PD_93521, Cmp_93585, Cmp_93708, PP_93522) were installed in summer 2007-2008. Battery failures, water infiltrations and datalogger malfunctions are responsible for the absence of data in some years in the table. It has been noted that some extensometers on one Figure 4. Extensometer expansion across three active or two occasions did not measure any cracking activity frost cracks in the winter 2009-2010. Also shown: time- during a given winter even if the instrumentation was in temperature curves for air, 40 cm and 80 cm depths. working order. Fresh rootlets crossing the cracking planes indicated that no cracking had occurred the previous During December, the three extensometers recorded winter in those cases. expansion up to widths of 6 to 9 mm, with some short

pauses on warmer days (or spells of 2-3 days). Increased expansion took place from the end of December and in January as air temperatures rapidly dropped to reach a minimum near -40 ¡C and as the permafrost temperature dropped to -16 ¡C. Small oscillations in width by 0.5 to 1 mm followed air temperature variations during the cold winter months. Maximum opening occurred in March even as air temperatures (still cold) started to generally warm up but as the temperature in permafrost continued to cool down to about -20 ¡C (80 cm deep). The maximum expansion reached values between 11 and 14 mm.

Table 1. Data summary of the six extensometers. DGCI, With warming air and soil temperatures, starting Date of ground contraction initiation. DMW, Date of either in mid-March (PP_93522) or in April, depending on maximum crack width. MW, Maximum recorded crack sites, the gaps narrowed progressively, with small pauses width (rounded to nearest tenth of mm). during colder spells of one to three days. Then they abruptly narrowed again just as air temperatures rose Winter 2009-2010 over 0 ¡C. However the gaps never close back totally, Figure 4 depicts the behavior of three extensometers which indicates that the frost cracks stayed partially open from September 2009, i.e. before the start of the freezing after the winter was over. The residual gap values season, to late June 2010, i.e. late after air temperatures measured by the three extensometers vary in width from got back above freezing values, thus providing a complete 1.8 to 7.8 mm. Given that maximum expansion varied winter cycle. Two of the extensometers started to expand

138 from 11 to 14 mm, gap closing at the beginning of summer was from 46 to 93 %. 4.2 Snow cover and ground contraction and expansion Winter 2010-2011 Figure 6 shows the behavior of four extensometers The whole pattern was identical the following winter under snow cover at sites equipped with snow depth (Figure 5). Extensometer PD_93577 opened by about 1 recorders, from 2005 to 2009. The maximum snow bank mm in mid-October as the active layer started to freeze- thickness varied from 23 cm to a 37 cm peak of short back when the air temperature fell to -14¡C, the soil duration in 2006. Although snow cover thickness did not surface temperature (2 cm) fell to -3.4 ¡C and the ground vary by much from one year to the next, the gaps opened temperature 40 cm deep just fell below 0 ¡C. Then a spell to larger values in winters 2007-2008 (up to 20 mm for of 7-8 warmer days with surface soil temperature rising PD_93577) and 2008-2009 as snow accumulation was back to 0 ¡C corresponded with a closing of the incipient less abundant in early winter. The residual gaps at gap. Starting in mid-November, extensometer PD_93521 snowmelt time were wider in those years also. Most of the expanded to a width of 1 mm in a two-step phase gaps begin to close under the snow cover several weeks corresponding to air temperature fluctuations; the soil before it begins to melt. The time series of air and ground surface temperature had just fallen to -10 ¡C and the temperatures vs snow depth variations and crack dilation- ground temperature 40 cm deep had just fallen to -6 ¡C. It contraction rates however still need to be analyzed in still enlarged to 1.7 mm as PD_93577 expanded to 1 mm more details. over two weeks in mid-December over a long cold spell during which air temperatures oscillated between -19 ¡C and -37¡C. Then both of them enlarged rapidly and other extensometers (Cmp_93585 and Cmp_83698) rapidly opened as temperature in the permafrost (80 cm deep) fell below -10 ¡C. By the end of December, the four extensometers had opened with gaps varying between 2.2 and 5.5 mm in width. There was no widening during the first two weeks of January as air temperatures were warmer (up to -12 ¡C) and the permafrost temperature remained stable slightly below -10 ¡C. Expansion of all the extensometers resumed in late January and occurred until late April as permafrost temperatures were about -18 ¡C. The maximum expansion of the four extensometers varied from 8 to 12.6 mm. The expansion rate was at its fastest from mid-January to mid- March with short pauses when the air temperatures peaked and more expansion when it Figure 6. Snow height vs extensometer expansion (crack dropped. As warming begun, all gaps were reduced. The opening/enlarging and ground contraction) for a series of narrowing stopped as the permafrost temperature rose four successive winters. above -10 ¡C and the gaps stayed open by 4 to 7 mm. Gap closing from maximum aperture from late April to the 4.3 Comparison with breaking cables advent of above 0 ¡C air temperatures at the beginning of June was from 45 to 51%. A comparative test was run to check extensometers against breaking cables (Figure 7). Seven cables (same gauge as in Fortier and Allard, 2005) with a thermistor at their end were buried near the soil surface, running across two cracks. The results show that the cables broke at various dates during the coldest winter months when the cracks in the permafrost were already open. The cracking events occurred after steps in the expansion curves of the extensometer, i.e. when cold spells increased the rate of spreading. The latest event in March even took place during a slight phase of extra widening by a fraction of a millimeter. Breaking occurred at temperatures between - 14 and -22 ¡C.

Figure 5. Extensometer expansion-contraction across four active frost cracks in the winter 2010-2011. Also shown: time-temperature curves for air, 40 cm and 80 cm depths.

139 The cracks enlarge and double in width over the winter as the cold wave penetrates deeper and permafrost reaches temperatures of -18 to -20 ¡C. Over the winter they enlarge during drops of temperature and either stop widening or slightly reduce in width during warmer spells. Slight variations of a few tenths of a degree 80 cm deep in the permafrost are reflected by sub-millimeter variations in crack width at the surface, therefore indicating that the permafrost expands and contracts with temperature oscillations. Thinning of the cracks at the surface begins as the permafrost temperature starts to rise above -18 ¡C in May, but it becomes much faster when the air and

surface temperature rise above 0 ¡C. This indicates that Figure 7. Comparison between two extensometers and soil expansion begins near the surface and is delayed or five breaking thermistor cables installed near the surface slowly propagates at depth. The surface gaps stop closing across cracks in winter 2006-2007. Vertical lines are when as permafrost temperature reaches values between -14 cables broke. and -10 ¡C, in fact -10 ¡C in most cases. It is likely that a crack deep in the permafrost would 5 DISCUSSION narrow and eventually close later in the summer as the annual heat wave penetrates at deeper levels and the The excellent agreement between air, surface, permafrost re-expands. However the cracks are still open active layer and permafrost temperatures fluctuations on up to the surface at snowmelt and water can freely flow the one side and extensometer variations on the other into them, freeze as a new vein and contribute to ice- side brings confidence that thermal contraction and wedge growth (Figures 2A, 6 and 8). The new vein is an expansion of the soil and frost crack behavior are truly addition to wedge size and to the volume of ice in the recorded by the instruments. The high sensitivity of the permafrost. instruments when installed across known frost cracks over Comparison of two extensometers with a few active ice wedges brings insights on processes that were breaking cables show that most cables are severed when not possible previously with breaking cables and shock the cracks are already open by 1.5 to 3 mm at the surface loggers. There is very little doubt that the extensometers and as ground temperatures are below about -15 ¡C. measure the opening and variations in width of the frost Others break when temperatures fall in the -20 ¡C values cracks on the ground surface over the ice wedges. (Figure 7). The -13 to -24 ¡C temperatures at the top of Field observations show that existing cracks do not permafrost obtained from breaking wires and thought to totally seal in the active layer in summer. In fact, it is often be mean temperature for permafrost cracking in previous possible to split open coherent soils along the virtual studies (Mackay, 1984,1993; Kasper and Allard, 1998; cracks with bare hands or with the spade of a shovel. Fortier and Allard, 2005) seem therefore to correspond to Observers further report that vegetation roots do not run the temperature range at which the cracks are enlarging across active cracks. In fact, root spreading across cracks to their maximum width rather than when they first open. is an indication of cessation of activity (Kokelj et al., This would be the time when the cables are subject to the 2007). The extensometer data answer some questions strongest tensile stress. raised by Mackay in several papers and discussed by Lachenbruch (1962, 1969). For instance, do cracks extend downward from the surface or upward from the top of the ice wedges at the permafrost table (Mackay, 1984)? Our data indicate that, over already active ice wedges, many cracks open with a small gap of about 1 mm or less as soon as the active layer is frozen back but as the permafrost is still not cold enough to crack. This small gap opening is probably related to the low contraction coefficient of the active layer which contains relatively little ice. At the cracking-prone sites of Bylot Island, the surface cracks widen abruptly after the permafrost (80 cm deep) has reached a temperature below about -10 ¡C, thus suggesting that the cracks then open in the permafrost and in the ice wedges. Therefore some cracks also open first at the top of the ice wedges, as shown by some gaps larger than 2 mm that first open at permafrost temperatures below -10 ¡C. The larger gap generated when permafrost cracks is likely due to the high Figure 8. An open frost-crack recently contraction coefficient of the ice-rich ground and/or wedge filled with ice in Mid-June. ice.

140 6 CONCLUSION Archipelago. Permafrost and Periglacial Processes. 16: 145-161. The monitoring of cracks width at Bylot Island with Grechischev, S. Y. E. 1973. Basic laws of thermorheology new sensitive extensometers allowed a better and temperature cracking of frozen ground. In USSR understanding of the thermo-mechanical behavior of Contribution, Permafrost, Second International permafrost and ice wedge-polygons. Over some active ice Conference, national Academy of Science, wedges, cracking begins in the active layer when it has Washington, D.C., pp. 228-234. just frozen back at the beginning of winter. The permafrost Kokelj, S.V., Pisaric, M.F.J., and Burn, C.R. 2007 itself cracks a few days or weeks later when its Cessation of ice-wedge development during the 20th temperature falls below -10 ¡C. Some cracks start in the century in spruce forests of eastern Mackenzie Delta, active layer, other start at the top of ice-wedges. The Northwest Territories, Canada. Canadian Journal of small gaps initiated in the active at the beginning of winter Earth Sciences, 44:1503-1515. get suddenly larger when the permafrost underneath Lachenbruch, A.H., 1962. Mechanics of thermal cracks a few weeks later. The cracks enlarge under colder Contraction Cracks and Ice-wedge Polygons in temperatures in the cold winter months and begin to Permafrost. Geological Society of America, Special narrow with warming temperatures in April. Expansion Paper 70: New York, 69 p. and contraction of permafrost is a very sensitive Lachenbruch, A,H. 1966. Contraction theory of ice-wedge temperature driven process. Thermal contraction cracks polygons: a qualitative discussion. In Proceedings, are still open at about 50% of their maximum width when First International Permafrost Conference, National the snow cover melts. Water can percolate in them to Academy of Sciences-National Research Council, form new veins in the ice wedges. Washington, D.C., publication no 1287, pp. 63-70. The adapted extensometers originally designed for Leffingwell, E. de K. 1919. The Canning River Region, dendrology work yielded results of a precision not attained Northern Alaska. United States Geological Survey so far in frost-cracking studies. Professional Paper 109, 251 p. Mackay, J.R. 1974. Ice-wedge cracks, Garry Island, 7 ACKNOWLEDGEMENTS Northwest Territories. Canadian Journal of Earth Sciences, 11: 1366-1383. The authors express their thanks to Dr. Gilles Mackay, J.R. 1975. The closing of ice-wedge cracks in Gauthier (Centre d’études nordiques) for coordinating the permafrost, Garry Island, Northwest Territories. access to the Bylot Island research station in Sirmilik Canadian Journal of Earth Sciences, 12: 1668-1674. National Park over the years. Logistical support was Mackay, J.R. 1978. The use of snow fences to reduce ice- provided by Polar Continental Shelf Project and Centre wedge cracking, Garry Island, Northwest Territories. In d’études nordiques. The data is a contribution from the Current Research, part A, Geological Survey of SILA network of monitoring stations and thermistor cables Canada, Paper 78-1A, pp. 523-524. operated by CEN. Instrumentation and field work were Mackay, J.R. 1984. The direction of ice-wedge cracking in financially supported by grants from the National permafrost: upward or downward? Canadian Journal Research and Engineering Council of Canada and the of Earth Sciences, 21: 516-524. ArcticNet network of centers of excellence. The comments Mackay, J.R. 1992. The frequency of ice-wedge cracking of the reviewer were much appreciated. (1967-1987) at Garry Island, western Arctic coast, Canada. Canadian Journal of Earth Sciences, 29: 8 REFERENCES 236-248. Mackay, J.R. 1993. Air temperature, snow cover, creep of Allard, M. 1996. Geomorphological changes and frozen ground, and the time of ice-wedge cracking, permafrost dynamics: key factors in changing Arctic western Arctic coast. Canadian Journal of Earth ecosystems. An exemple from Bylot Island, Nunavut, Sciences, 30: 1720-1729. Canada. Geoscience Canada, 23: 205-212. Mackay, J.R. 2000. Thermally induced movements in ice- Allard, M. and Kasper, J.N. (1998) Temperature wedge polygons, western arctic coast: a long-term conditions for ice wedge cracking: field measurements study. Geographie physique et Quaternaire, vol 54: from Salluit, northern Québec. In Permafrost, 41-68. Proceedings, 7th International Conference, Mackay, J.R. and Burn, C.R. 2002. The first 20 years Yellowknife, Canada, Lewkowicz A.G., Allard, M. (1978-1979 to 1998-1999) of ice-wedge growth at the (eds). Centre d’études nordiques, Université Laval, Illisarvik experimental drained lake site, western Arctic Collection Nordicana 57, Québec; 5-11. coast, Canada. Canadian Journal of Earth Sciences, Christiansen, H.H. (2005) Thermal Regime of Ice-wedge 39: 95-111. Cracking in Adventdalen, Svalbard. Permafrost and Matzuoka, N. and Christiansen, H.H. 2008. Ice wedge Periglacial Processes. 16: 87-98. polygon dynamics in Svalbard: High-Resolution Fortier, D. and Allard, M. (2004) Late Holocene monitoring by multiple techniques. In: Kane, D.L. and syngenetic ice wedge polygon development, Bylot Hinkel, K.M. (eds.), Proceedings of Ninth International Island, Canadian Arctic archipelago. Canadian Journal Conference on Permafrost, 2, University of Alaska of Earth Sciences , 41(8): 997-1012. Fairbanks, p. 1149-1154. Fortier, D. and Allard, M. (2005) Frost-cracking Conditions, Bylot Island, Eastern Canadian Arctic

141 Matsuoka, N. 1999. Monitoring of thermal contraction cracking at an ice-wedge site, central Spitsbergen. Polar Geoscience 12: 258-271. Pesonen, E., Mielikäinen, K. & Mäkinen, H. 2004. A new girth band for measuring stem diameter changes. Forestry 77: 431-439.

142

143