Glaciers of the Subantarctic Islands Global Land Ice Measurements from Space, Chapter 37B J. Graham Cogley1, E. Berthier2 and S. Donoghue3 1: Department of Geography, Trent University, Peterborough, Ontario, Canada K9J 7B8 2: OMP-LEGOS, 14 av E. Belin, 31400 Toulouse, France 3: Antarctic Climate and Ecosystem Cooperative Research Centre, Hobart, Australia; Institute of Antarctic and Southern Oceans Studies, University of Tasmania, Hobart, Australia

ABSTRACT Through case studies, we summarize progress for the Subantarctic in the drive to complete the World Glacier Inventory. Most of the 29 Heard Island glaciers have shrunk since the first observations in 1947, several having begun to retreat from former tidewater termini. Total glacier area has decreased from 288 km2 in 1947 to 257 km2 in 1988 and 231 km2 in 1988. On Kerguelen, shrinkage has been more dramatic. In 1963-64, the date of the earliest complete coverage, glacier extent was 703 km2. In 2001, it was 552 km2. We have compiled the first objective topographic map of Montagu Island, the largest of the South Sandwich Islands (glacierized area 94 km2), using the recently-released ASTER Global Digital Elevation Model. We assess the shortcomings of this new tool, and conclude that it may indeed prove valuable for first-time mapping. It will also be useful as an aid in selecting accurately-dated scenes for the estimation of multi-decadal changes. An inventory based on cartographic sources yields an estimate for total glacier area in the Subantarctic during the late 20th century of 7863 km2. Limited measurements on other islands suggest that Heard Island and Kerguelen are typical of glacier shrinkage, thinning and negative mass balance across the region.

1. Introduction In this chapter we summarize progress for the Subantarctic in the attempt to complete the World Glacier Inventory (WGI; http://nsidc.org/data/docs/noaa/g01130_glacier_inventory/), and illustrate through case studies how progress in the systematic monitoring of glaciological change is now accelerating. Many of the Subantarctic islands are poorly known even today, and until recently there was little information about glacier changes. Documenting change is not a realistic ideal while information about the first of the two required epochs remains uncompiled. Thus the aim of much glaciological work in the Subantarctic is that originally articulated for the WGI, namely to develop a snapshot, necessarily diachronous, of glacierization in the late 20th century. However information about changes is now accumulating more rapidly. Here we present case studies of glacier mapping and glacier change, relying on imagery from SPOT, ASTER, Landsat and other sensors, from Heard Island, Kerguelen, and Montagu Island. We also summarize the results of an almost complete map-based inventory of the Subantarctic as a whole.

2. The Regional Setting We define the Subantarctic as in Figure 1. The northern boundary, with excursions to exclude Patagonia and New Zealand, lies at –45° so as to include the summit ice on Marion Island. The boundary with Antarctica is drawn to the north of islands that are considered ―close‖ to the mainland. Scott Island, Peter I Island, the South Shetlands, the South Orkneys and the Balleny

1

Islands are all regarded as Subantarctic, partly because they are at risk of being neglected in studies focused on the much larger Antarctic Ice Sheet.

Figure 1. The Subantarctic (white background) as defined for the present purpose. Islands with no glaciers are in italics. “Is”: “Islands”; “Peter I” is “Peter the First Island”. The area of land in the Subantarctic is 30160 km2, of which 26% is occupied by glacier ice (section 4). Most of the islands lie on oceanic crust. Many are volcanic, and several are currently active or have been so recently (e.g. Bauer 1963; Lachlan-Cope et al. 2001; Smellie et al. 2002; Patrick et al. 2005; Stephenson et al. 2005). The Subantarctic islands, scattered through 360° of longitude and 25° of latitude, have distinctly different climates. Weather stations are few and widely dispersed (Jacka et al. 2004). The main features of the evolution of temperatures in the Subantarctic during the 20th century can, however, be seen in Table 1 and Figure 2.

2

Figure 2. Annual positive degree-day sums based on records of near-surface air temperature from stations near sea level: Transvaal Cove (Marion I), Port aux Français (Kerguelen), Grytviken (South Georgia), Orcadas (South Orkney Is) and Young I (Balleny Is). Young I record: Antarctic Meteorological Research Center, University of Wisconsin-Madison (http://amrc.ssec.wisc.edu/stations.html). Other records: Turner et al. 2004 (http://www.antarctica.ac.uk/met/READER/). Thick lines: 5-year averages. Monthly mean temperatures were corrected for within-month variability following Braithwaite (1984), a standard deviation of 4°C being assumed. Table 1. Temperature at Subantarctic weather stations. Number of years of observation in parentheses; Heard I (proxy): based on Atlas Cove measurements and linear regression against measurements at Port aux Français (Thost and Truffer 2008). Trends in bold type differ from zero by more than 2 standard errors. Station Transvaal Cove, Marion I Port aux Français, Kerguelen Atlas Cove, Heard I Heard I (proxy) Grytviken, S Georgia Orcadas, Laurie I (S Orkney Is) Young I, Balleny Is

Mean Tann (°C) 1951–1980 5.3 (30) 4.5 (29) 1.5 (7) 1.6 (30) 2.0 (28) –3.7 (30) —

Mean Tann (°C) 1981–2007 6.1 (26) 4.9 (27) 2.0 (12) 2.1 (22) — –3.2 (26) –7.1 ( 3)

Trend (°C a-1), Period +0.027, 1949–2006 (56) +0.010, 1951–2006 (56) +0.017, 1948–2006 (19) +0.014, 1951–2002 (52) +0.014, 1951–1980 (28) +0.033, 1981–2007 (26)

Mean annual temperature decreases polewards. However the potential for ablation by melting, as measured by the positive degree-day sum near to sea level (Figure 2), remains substantial as far south as the South Orkneys in latitude –61°. Subantarctic climate is strongly maritime. Mean annual range of temperature is small, but increases southwards from 4.6°C at Marion Island to 12.6°C in the South Orkney Islands. There is evidence from some islands for the expected 3

contrast between windward and leeward coasts. On Heard Island, for example, föhn winds are documented on the eastward side, in the lee of the main peak, and on South Georgia the equilibrium-line altitude (ELA) is higher in the lee (northeast) of the axial mountain ranges (Clapperton et al. 1989). Temperatures throughout the Subantarctic have increased in recent decades. All of the trends in Table 1 are statistically significant, some strongly so, but interdecadal variability is also notable. For example much of the warming at Port aux Français happened between 1960 and the early 1980s, since when temperatures have changed little (Berthier et al. 2009). The pattern of interdecadal temperature variability shown by Berthier et al. is comparable to that seen also in the annual positive degree-day sum, as shown in Figure 2, but this is not true everywhere. The strong warming at Orcadas (Table 1) is not mirrored in Figure 2, suggesting, consistently with the findings of Zazulie et al. (2010), that much of the warming there is wintertime warming. Table 2. Precipitation at Subantarctic weather stations. Data from the Global Historical Climate Network (http://www.ncdc.noaa.gov/oa/climate/ghcn-monthly/index.php) except for Port aux Français (Berthier et al. 2009) and Arctowski (Marsz 2002). Station Transvaal Cove, Marion I Port aux Français, Kerguelen Grytviken, S Georgia Arctowski, King George I (S Shetland Is) Bellingshausen, King George I (S Shetland Is) Arturo Prat, Greenwich I (S Shetland Is)

Mean annual precipitation (mm) 2294 759 1473 507 707 616

Period, Number of years 1948-2009 (62) 1951-2005 (55) 1906-1981 (65) 1978-1996 (16) 1968-2009 (40) 1966-2003 (32)

Precipitation varies considerably from island to island (Table 2). However weather-station averages are not reliable as guides to accumulation on glaciers. All of the stations are close to sea level, and all are on the lee sides of islands. For example, for 1995-2001 the mean annual precipitation was 692 mm at Port aux Français, while 70 km to the west it was 3155 mm near to the eastern margin of Cook Ice Cap (Institut Paul Émile Victor, dataset 136; Berthier et al. 2009). On King George Island, for 1987-1991 the mean annual precipitation was 472 mm at Arctowski (Marsz 2002), while some 10 km to the north, at 700 m a.s.l., mean annual point mass balance was 2480 mm water equivalent for the same period (Wen et al. 1998). Table 3. Observations of the equilibrium-line altitude on Subantarctic islands. Island, Glacier Marion I Kerguelen, Cook Ice Cap Heard I Bouvet I S Georgia Signy I (S Orkney Is) King George I, Little Dome Nelson I Livingston I, Rotch Ice Dome Deception I, Glacier G1

Latitude –46.9° –49.4° –53.1° –54.4° –54.6° –60.7° –62.1° –62.3° –62.6° –62.9°

4

ELA (m a.s.l.) >1150 ~700 100–700 200–350 370–860 ~200 160 110 140–170 275–330

Source Sumner et al. 2004 Vallon 1977a Ruddell 2006 Orheim 1981 Clapperton et al. 1989 Cogley (unpubl.) Wen et al. 1998 Wen et al. 1998 Orheim and Govorukha 1982 Orheim and Govorukha 1982

Moving polewards, the ELA drops towards sea level, and can be assumed to reach it somewhere between latitudes –65° and –70° (Table 3). Observations for South Georgia are partly cartographic; the substantial range reflects the contrast between leeward and windward sides of the island. Tidewater glaciers – those whose terminuses stand in the sea – are known as far north as Kerguelen, where, of two calving glaciers on the west coast in 1963, Glacier Pasteur is still at sea level as of April 2009. Calving termini become the norm south of –55° to –60°.

3. Case Studies 3.1 Heard Island Heard Island is a volcanically active island located at 53.1º S, 73.5º E in the southern Indian Ocean (Plate 1). Heard Island’s position south of the Polar Front is unique among the islands of the southern Indian Ocean (e.g., Kerguelen, Crozet, and Marion). This high-elevation, roughly circular island (367 km2), culminating in Mawson Peak at 2745 m a.s.l. on the Big Ben Plateau, had an estimated glacier cover of 288 km2 in 1947 (Ruddell 2006) . The physiography and orographic effects of the island have resulted in the glaciers on the leeward and windward sides reacting differently to changes in the climate. Understanding of changes in the extent of the Heard Island glaciers has been derived from brief journal accounts, photographs and drawings made during the early sealing period and later, as scientists began to explore this remote island, from published reports, photographs, satellite images and eventually mass-balance studies. The fluctuations of these glaciers have been discussed by Budd and Stephenson (1970), Allison and Keage (1986), Budd (2000), Ruddell (2006) and Donoghue (2009). Two SPOT images from January 1988 and March 1991 are the first clear images of the majority of the island. Ruddell (2006) used the SPOT images and aerial photographic surveys to estimate glacier extent between 1947 and 1988 and to provide the first complete inventory of Heard Island’s glaciers. As of 1988, there were 29 glacierised basins (41 termini) and 3 glaciated basins (that may have hosted glaciers during the Last Glacial Maximum) on Heard Island with a total area of 256.9 km2 and an estimated ice volume, by area-volume scaling, of 14.2 km3 (Ruddell 2006). The Australian Antarctic Division Data Centre has analysed the 23 March 2008 Worldview-1 image (Plate 2). Glacier outlines were remapped, resulting in a total glacierized area of 231 km2 (Harris 2009; Lucieer et al. 2009). A new DEM based on this image is in preparation. It will update the 1997 (Ryan 2004) and 2002 (Brolsma and Smith 2008) RADARSAT DEMs. The glaciers on Heard Island can be divided into four regions, Laurens Peninsula, northern Big Ben, eastern Big Ben and southwestern Big Ben (Plate 1). The temperate glaciers of Heard Island were first systematically observed in the early 1900s. The majority of these glaciers were stable or slightly thinning until a marked retreat began in the early 1960s (Ruddell 2006). The retreat from the 1960s has continued through the last recorded observation in 2009 (Donoghue 2009). Recent (1947-2009) changes in the eastern Big Ben glaciers are known in more detail than for the other regions of Heard Island due to relatively cloud-free satellite images and glacier-based field surveys in 2000/01 and 2003/04 (Donoghue 2009). These eastern Big Ben glaciers include larger glaciers that extend from the summit of Big Ben to the coast (Compton, Stephenson, and Winston glaciers) and smaller glaciers that extend from around 1000 m a.s.l. to the coast (Brown, AU1121, and AU1141glaciers).

5

The first flight over Heard Island in 1947 indicated that Winston Glacier had already begun to retreat. Between 1947 and 1963 Winston retreated 1.6 km (Budd and Stephenson 1970) and therefore appeared to have the most rapid retreat during that interval. However this was followed by a period of advance in the 1970s, which was later lost again in the 1980s (Ruddell 2006). By the 1960s several of the other glaciers along the east coast had begun to retreat (Budd and Stephenson 1970): AU1121 had retreated to a stagnant ice field; Brown and Stephenson glaciers had begun to retreat. By the 1980s Compton Glacier had the most substantial retreat of any of the glaciers on Heard Island. It was then 1.6 km inland from its 1947 tidewater glacier front (Allison and Keage 1986) and was 2.5 km inland by 1987 (Ruddell 2006). By the mid 1980s Brown and Stephenson were beginning to show signs of increased retreat. Both had developed proglacial lagoons. The Brown Glacier terminus retreated by a total of 1.7 km from the coast between 1947 and 2004 at an average rate of 30 m a-1 (Thost et al. 2004). GPS surveys from 2000 and 2003 indicate significant thinning of Brown Glacier by up to –11.7 m on the lower glacier and –8.5 m on the upper glacier (Thost and Truffer 2008). Between 1947 and 1987 Stephenson decreased in area by 18 % (Ruddell 2006) although this was not a steady retreat; instead there was a dramatic acceleration (to ~100 m a-1) in retreat of the northern margin from 1987 to 2000 (Kiernan and McConnell 2002). This increased retreat has continued to 2006 with the opening of a waterway between the two proglacial lagoons that have formed near the terminus of Stephenson Glacier. Quickbird images were used to measure the changes in eastern glaciers between January 2004 and January 2006 (Donoghue 2009) (Plate 2). Brown Glacier, and Compton Glacier along most of its lagoon-based terminus, each retreated by less than 0.1 km. The northern terminus of Stephenson Glacier, which has broken up into the lagoon, retreated by ~0.25 km, and the southern terminus by more than 0.75 km. Winston Glacier retreated by ~0.2 km. 3.2 Kerguelen The Kerguelen Islands (49° S, 69° E) are a group (7215 km2) of isolated islands in the southern Indian Ocean. They were discovered in 1772 by Yves de Kerguelen. Here, we summarize and update some recently published results of glacier change on the Kerguelen Islands (Berthier et al. 2009). We measured the extent of glaciers and ice caps on these islands by digitizing and comparing successive glacier outlines on a map and satellite images. Our oldest data set is the map published by Institut Géographique National (IGN) at a scale of 1:200 000. It was produced using aerial photographs taken in 1963 of Cook Ice Cap and in 1964 of other glaciated areas. SPOT (1991, 1994 and 2003), Landsat (November 2001) and ASTER (2005, 2006, 2009) images were used to generate a time series of ice cap extent. Our glacier inventory has been incorporated into the Global Land Ice Measurements from Space (GLIMS) Glacier Database (http://nsidc.org/glims/). The areal changes for the four main glacierized regions on the Kerguelen Islands are summarized in Table 4. Between 1963–64 and 2001, the total ice-covered area on the Kerguelen Islands shrank from 703±51 km2 to 552±11 km2.

6

Table 4. Changes in Kerguelen ice cover. Isolated glaciers (covering 18 km2 in 1963-1964) are not tabulated. Region Cook Ice Cap Rallier du Baty Peninsula Mont Ross Presqu’île de la Société de Géographie

Year 1963 2001 1964 2001 1964 2001 1964 1994

Area (km2) 500.9 410.0 102.3 79.2 59.3 35.1 14.7 4.7

Ice loss, km2 (%) 90.9 (18.2) 23.1 (22.5) 24.2 (40.8) 10 (68)

Between 1963 and 2001, Cook Ice Cap shrank by 18% (Figure 3a) due to the strong retreat of all outlet glacier fronts and the growth and appearance of nunataks (isolated rock outcrops within the ice cap). We estimated the overall rate of contraction for two time periods (Figure 3b). The ice loss was 1.9±1.3 km2/yr between 1963 and 1991 and increased to 3.8±0.7 km2/yr after 1991. Thus, in recent years, nearly 1% of the ice cap has disappeared annually. The temporal resolution of our data on glacier change is rather low and does not permit us to observe the inter-annual or decadal variability of the ice loss. However, we detect a recent acceleration of ice loss.

Figure 3. Retreat of Cook Ice Cap between 1963 and 2001. (a) 2001 Landsat image with the 1963 (white) and 2001 (black) glacier outlines. (b) Temporal changes of the ice-covered area. The 7

rates of glacier shrinkage (km2/yr) are indicated for 1963-1991 and 1991-2003. From Berthier et al. 2009, copyright 2009 American Geophysical Union; reprinted with permission. Ampère Glacier is the largest outlet glacier of Cook Ice Cap and, because of its relatively good accessibility, has received most attention in previous studies (Frenot et al. 1993; Vallon et al. 1977a,b). Between 1963 and 2006, the terminus retreated at a mean rate of 70±3.7 m/yr, the glacier length decreased from 15.3 km to 12.5 km and the glacier contracted by 18.3±5.1 km2 (Plate 3b). The area loss took place mostly close to the glacier front, where a proglacial lake (Ampère Lake) appeared in the 1960s and covered 1.6 km2 in 1991 and 3.5 km2 in 2006. A longer time series of terminus change for Ampère Glacier has been compiled by mapping seven frontal moraines (Frenot et al. 1993). These moraines were dated using the radial growth of Azorella selago Hooker, a cushion-forming plant of the Umbellifera group. The moraines were deposited between 1799 and 1962 following a maximum Little Ice Age advance in 1799, and indicate a total (but irregular) retreat of about 1 km in 160 year, at a slow mean rate of 6 m/yr. One can thus conclude that there has been a drastic (one order of magnitude) acceleration of the retreat of the Ampère Glacier front after 1963. An April 2009 ASTER image was used to update the outlines of Ampère Glacier (Plate 3a). The most striking feature is the disappearance of the part of the glacier tongue that previously existed in the lower reaches of the nunatak Lapparent. A lake is now found instead with a high density of icebergs, indicating that calving has become an important component of the mass balance. Intensification of calving and erosion of the glacier front by proglacial lake water are most likely responsible for the accelerating rate of retreat, that doubled from –0.35 km2/yr during 1963–1991 to –0.69 km2/yr during 2001–2009 (Plate 3b). Our analysis of area change has been complemented by estimates of the volume balance of Cook Ice Cap since 1963 using (i) area-volume scaling relationships, (ii) sequential DEM analysis (Berthier et al., 2004) and (iii) the integration of sparse but reliable measurements of ice elevation changes over the ice cap (Berthier et al. 2009). Although they are all highly uncertain, we obtained three independent and consistent measurements between 1963 and 2000, in the range of 25–30 km3. This is equivalent to an area-average thinning rate of 1.4–1.7 m/yr. The accelerated rates of ice loss observed since the 1990s suggest that ice masses on the Kerguelen Islands have not yet approached a steady state. New satellite acquisitions would be useful in assessing whether the apparent trend toward accelerated ice loss is maintained. A better understanding of the dynamics of glaciers and ice caps on the Kerguelen Islands is also needed in order to predict at what pace they will react to the 1–2°C warming predicted in this region by the end of the 21st century. 3.3 Montagu Island The ASTER global DEM, a product of METI (Ministry of Economy, Trade and Industry, Japan) and NASA released in June 2009 (Hayakawa et al. 2008; ASTER Global DEM Validation Team 2009), has attributes and shortcomings which will require considerable processing before it can be put to use in the estimation of glacier elevation changes. It can, however, be expected to be valuable in the topographic mapping of certain remote glacierized areas for the first time. In this section we assess the ASTER GDEM of one such area, Montagu Island in the South Sandwich Islands. Plate 4a shows Montagu Island according to the best available pre-ASTER map (Holdgate and Baker 1979). The highest elevation, 1370 m, is taken from Kemp and Nelson (1931), while that of Mount Oceanite in the southeast was estimated from a helicopter altimeter. The general topography is ―not accurately laid down‖. 8

Plate 4b shows the ASTER GDEM of Montagu Island, without editing. Plates 4a and 4b were georeferenced approximately to each other, relying on shaded-relief images of the GDEM. The shoreline and ice margin from Plate 4a are ~4 km NNE of their apparent positions in the GDEM. The ASTER position, with accuracy better than ±50 m (Fujisada et al. 2005), is clearly the more accurate. There are obvious artifacts in the ASTER topography. The irregular truncation of the southern limit of coverage is probably due to masking of the surrounding ocean based on the inaccurate Holdgate and Baker position. The numerous offshore ―islets‖ must be ice floes, misinterpreted by the GDEM processing algorithms. On the ―mainland‖, elevation errors are common. At the northeastern limit of coverage, spurious elongate ridges appear to rise to 2000 m. In the southeast, Mount Oceanite has been misinterpreted as sea. On the summit plateau, several pits and bumps can be seen. Close-ups with finer contour intervals (not shown), and comparison with Plate 4c, suggest strongly that these irregular details are artifacts. Pits and bumps were found in ―virtually every ASTER GDEM tile‖ examined during validation (ASTER Global DEM Validation Team 2009). GDEM elevations are obtained by ―stacking‖ stereo image pairs dating between 1999 and 2008. Averaging the results reduces root-mean-square errors significantly (ASTER Global DEM Validation Team 2009). For Montagu Island the number of image pairs ranges from 1 to 10, and is typically 3 or 4. Evidently this amount of averaging has not sufficed to reduce pit-and-bump noise below the threshold implied by the 100-m contour interval of Plate 4b. Indeed the stacking may be partly responsible for the noise: no cloud-free image has been acquired since the launch of the satellite in 1999. However, a problem peculiar to Montagu Island is that Mount Belinda has been erupting more or less continuously since 2001 (see Plate 4c). The MODIS Thermal Alert System (MODVOLC; Wright et al. 2004) recorded a strongly variable heat source at the location of Mount Belinda between 2001 and 2004 (Patrick et al. 2005), and an intensification of activity in September 2005 is documented by NASA Earth Observatory (2005) from ASTER imagery. Inspection of MODVOLC suggests that the eruptive activity has continued to at least 2007. Eruptive plumes are visible in all ASTER images examined for this study, and they have doubtless contaminated the topographic results. In fact, the plume of greyish steam off the north coast in Plate 4c is recognizable in Plate 4b, and the lava stream flowing northeastwards down the flank of Mount Belinda appears as a number of spurious high spots in Plate 4b. Plate 4c (NASA Earth Observatory 2005) shows the best single ASTER image of Montagu Island. Both the shoreline and the ice margin were transferred from Plate 4a to this image, but the approximate nature of the georeferencing meant that no reliable conclusions could be drawn about changes of glacier terminus positions. Instead, we accept Plate 4c as the most reliable source for the shoreline while, because Plate 4c is snow-covered, Plate 4a remains the best source for the ice margin. Minor adjustments were made by eye to match the Holdgate and Baker ice margin to the ASTER shoreline. The result of this merger of information is shown in Plate 4d, the first objective topographic map of Montagu Island. The truncation of the south coast could presumably be remedied rather easily, and flaws such as the misinterpretation of Mount Oceanite could be corrected by detailed analysis of suitable image pairs, although this would tend to nullify the convenience of the GDEM. A broader conclusion is that a general solution of the pit-and-bump problem will be essential if the ASTER GDEM is to fulfill its promise as an analytical tool. 9

Certain features of the topography are now clearer. The formlines of Plate 4a suggest a conical mountain, but in Plate 4d the summit region is recognizable for the caldera that Plate 4c shows it to be. Most notably, the ASTER GDEM shows that the peak of Mount Belinda is 2 km to the north of its formline position in Plate 4a, and that its maximum elevation, 1070 m, is 300 m less than previously estimated.

4. Cartographic Inventory of the Subantarctic Plate 4a is an example for Montagu Island of unpublished work which aims to inventory Subantarctic glaciers from cartographic sources. The work, summarized in Table 5 , is incomplete, but provides a regional perspective for section 3, further insight into the potential of the ASTER GDEM, and an estimate for total glacierized area. This total, 7863 km2, derives from maps published over several decades, mostly based on air photographs but in some cases on only ground surveys or even explorers’ sketches. The most extensively glacierized island, South Georgia, has not yet been inventoried and information presented here is from Smith (1960). The other islands not yet inventoried are King George Island and Deception Island, both in the South Shetlands. Information in Table 5 is from a complete inventory by the King George Island GIS Project (http://www.kgis.scar.org/mapviewer/kgis.phtml) and López-Martinez and Serrano (2002). Elsewhere, Table 5 summarizes on-screen digitizing of scanned paper maps, all transformed to the WGS84 datum and the UTM projection. The quality of these maps varies greatly. Maps of Kerguelen at 1:200 000 scale are good enough to yield acceptable estimates of elevation changes when compared with modern satellite images (section 3.2). The measurements of Kerguelen glacier area in 1963–64 given in Table 5, 698 km2, and in section 3.2, 703±51 km2, were obtained by different operators working on the same map. They are within 1% of each other, illustrating the agreement that is possible between duplicate cartographic measurements. Comparable agreement is seen between duplicate measurements of Heard Island glacierization, respectively 257 km2 (section 3.1) and 254 km2 (Table 5) for 1988. The map of Peter I Island in 1987 at 1:50 000 scale (Norsk Polarinstitutt 1988) is as good as the best maps available for any glacierized region anywhere. It was used as the basis for an inventory (Plate 5) in which the ice cover was subdivided into 26 glaciers with a total area of 151 km2. Unfortunately the ASTER GDEM of Peter I Island is unusable, being based on at most a single image pair which was evidently almost completely cloud-covered. At the opposite extreme, the best map of the Balleny Islands (Plate 6) is a crude sketch. The cartographic details are from the Soviet Atlas Antarktidy (ADD Consortium 2000; shorelines and ice margins) and Hatherton et al. (1965; conjectural formlines). Two of the largest islands have formlines, but on the third there is only a single spot height. Here the ASTER GDEM would represent a major advance in elementary knowledge, but it has no coverage at all of Sturge Island and Buckle Island. For Young Island, as for Montagu Island (section 3.3), the GDEM ocean mask is based on a wrong estimate of position and only a small part of the western side of the island appears in the GDEM. However, not all of the Subantarctic is represented disappointingly in the ASTER GDEM. Plate 7a shows Laurie Island in the South Orkney Islands from a 1:100 000 scale map based on January 1979 photography and older ground surveys (British Antarctic Survey 1988). The amount of detail is adequate for inventory purposes. A shaded-relief image of the ASTER GDEM shows pits 10

and bumps like those visible in Plate 4b, but they are few, and few or none have magnitudes detectable with the 50-m contour interval adopted for Plate 7b. The 1979 position of the island is about 2 km east of the ASTER position, and as for Montagu Island it was necessary to eliminate a large number of ice floes. In fact, Plates 4 and 7 show that a shoreline mask from an independent source is an essential tool in use of the GDEM.

5. Context and Conclusion Glaciers on Heard Island shrank at –0.18 % a-1 from 1948 to 1980, –0.68 % a-1 from 1980 to 1988 and –0.51 % a-1 from 1988 to 2008 (Ruddell 2006; section 3.1). On Kerguelen, Cook Ice Cap shrank at –0.38 % a-1 from 1963 to 1991 and at –0.85 % a-1 from 1991 to 2003 (section 3.2). Scattered measurements elsewhere tend to corroborate these results. For example, the summit glacier on Marion Island had an area of about 1 km2 in 1961 (Langenegger and Verwoerd 1971). Sumner et al. (2004) report that it ceased to exist during the 1990s, although they describe remnants of ice buried beneath scoria. In the South Shetland Islands, Calvet et al. (1999) documented shrinkage of the ice cover of Livingston Island between five dates spanning 1956 to 1996. The loss over the whole period was –31.6 km2 from a 1956 area of 734.12 km2, a rate of – 0.11 % a-1. The ice cover of Greenwich Island decreased from 145.4 km2 in 1957 to 136.7 km2 in 1991 (Ballester et al. 1993), a rate of –0.18 % a-1. Neither of these estimates include changes in the extent of nunataks. Braun and Gossmann (2002) measured the retreat of glacier terminuses in Admiralty Bay, King George Island, from 1956 to 1995. The 21 glaciers drained an area of 237.7 km2 in 1956 and 222.4 km2 in 1992, a decrease of –15.1 km2 at –0.18 % a-1. All of these rates are for tidewater glaciers without floating tongues, and are therefore roughly comparable with rates for Heard Island, where several glaciers remain in contact with the sea. The shrinkage rate on Kerguelen is less influenced by the tendency of calving termini that are not afloat to change little. The accelerated shrinkage of Cook Ice Cap is noteworthy, although on Heard Island retreat appears to have been fastest in the 1980s. Mass-balance measurements are few in the Subantarctic. Table 6 is a complete list. The recent geodetic measurements of Cook Ice Cap, Brown Glacier and two glaciers on Livingston Island are significant additions. All but one of the measurements is negative. Table 6. Mass-balance measurements on Subantarctic glaciers. Superscript G denotes a geodetic measurement, in each case assuming a density of 900 kg m-3 for converting volume balance to mass balance. Glacier

Island

Hodges Gl Hamberg Gl Little Dome Flagstaff Gl Gl G1

S Georgia S Georgia King George I King George I Deception I

Hurd GlG Johnsons GlG Brown GlG

Livingston I Livingston I Heard I

Cook ICG

Kerguelen

Area (km2) 0.27 11.40 14.00 0.09 0.42

Period (No. of years) 1958 (1) 1958 (1) 1992 (1) 1958 (1) 1969–1974 (6)

Balance (kg m-2 a-1) –150 –254 +163 –537 –412

5.24 5.62 6.18 ~4.40 500.90

1956–2000 (44) 1956–1998 (42) 1947–2004 (57) 2000–2003 (3) 1963–2000 (37)

–74 –161 –450 –1644 ~ –1400

11

Source Smith 1960 Smith 1960 Wen et al. 1998 Noble 1965 Orheim and Govorukha 1982 Molina et al. 2007 Molina et al. 2007 Thost and Truffer 2008 Section 3.2

There is thus no evidence that the remote and little-known glaciers of the Subantarctic are exceptions to the rule, widely observed elsewhere, that glaciers are shrinking and losing mass. Where they have been observed, Subantarctic rates of glacier retreat, shrinkage and mass balance are comparable to those of better-known regions. Future monitoring will inevitably have to rely heavily on remote sensing. Measurements of shrinkage (reduction of area) are valuable in themselves, but future work should focus more aggressively on the measurement of elevation changes by subtraction of sequential DEMs obtained from radar and optical stereo imagery. In this regard the measurements reported here are notable early contributions.

12

Glaciers of the Subantarctic Islands: References Allison, I.F., and P.L. Keage, 1986, Recent changes in the glaciers of Heard Island, Polar Record, 23(144), 255-271. ADD Consortium, 2000, Antarctic Digital Database, Version 3.0, Database, Manual and Bibliography. Scientific Committee on Antarctic Research, Cambridge. 93p. and digital data (version 4.1; http://www.add.scar.org/add_main.html). ASTER Global DEM Validation Team, 2009, ASTER Global DEM Validation Summary Report, Jet Propulsion, California Institute of Technology, Pasadena, CA. 28p. [http://asterweb.jpl.nasa.gov/gdem.asp.] Australian Antarctic Division, 2009, Glacial retreat on Heard Island, SCAR Map Catalogue map no. 13691, accessed 19 November 2009. Ballester, N., F. Granada, J. Corbera and J. Calvet, 1993, Fluctuaciones del casquete glaciar de la isla Greenwich (Shetland del Sur) en el período 1956-1991, Quinto Simposio de Estudios Antárticos, Barcelona, 259-264 Bauer, A., 1963, Les glaciers de l’Île de Kerguelen, Comité National Français pour les Recherches Antarctiques, 2, 1-75, maps. Berthier, E., Y. Arnaud, D. Baratoux, C. Vincent and F. Rémy (2004). Recent rapid thinning of the ―Mer de Glace‖ glacier derived from satellite optical images. Geophysical Research Letters, 31, L17401, doi: 10.1029/2004GL020706. Berthier, E., R. Le Bris, L. Mabileau, L. Testut and F. Rémy, 2009, Ice wastage on the Kerguelen Islands (49°S, 69°E) between 1963 and 2006, Journal of Geophysical Research, 114, F03005, doi:10.1029/2008JF001192. Braithwaite, R.J., 1984, Calculation of degree-days for glacier-climate research, Zeitschrift für Gletscherkunde und Glazialgeologie, 20, 1-8. Braun, M., and H. Gossmann, 2002, Glacial changes in the areas of Admiralty Bay and Potter Cove, King George Island, maritime Antarctica, in Beyer, L., and M. Bölter, eds., Geoecology of Antarctic Ice-free Coastal Landscapes, 75-89. Springer, Berlin. British Antarctic Survey, 1988, South Orkney Islands, East Sheet. BAS 100 Series, Sheet 1, map at 1:100,000 scale. Brolsma, H., and D. Smith, 2008, Heard Island RADARSAT (2002) DEM, CAASM Metadata, Australian Antarctic Data Centre. Budd, G.M., 2000, Changes in Heard Island glaciers, king penguins and fur seals since 1947, Papers and Proceedings of the Royal Society of Tasmania, 133(2), 47-60. Budd, G.M., and P.J. Stephenson, 1970, Recent glacier retreat on Heard Island, International Association of Scientific Hydrology Publications, 86, 449-458. Calvet, J., D. García Sellés and J. Corbera, 1999, Fluctuaciones de la extensión del casquete glacial de la isla Livingston (Shetland del Sur) desde 1956 hasta 1996, Acta Geológica Hispánica, 34(4), 365-374. Clapperton, C.M., D.E. Sugden and M.S. Pelto, 1989, Relationship of land terminating and fjord glaciers to Holocene climatic change, South Georgia, Antarctica, in Oerlemans, J., ed., Glacier Fluctuations and Climatic Change, 57-75. Kluwer, Dordrecht. Cogley, J.G., 2009, A more complete version of the World Glacier Inventory, Annals of Glaciology, 50(53), 32-38. Donoghue, S., 2009, Changes in the morphology, mass balance, and dynamics of Brown Glacier, Heard Island, with comparisons to the surrounding sub-Antarctic islands, Ph.D. thesis, University of Tasmania. Frenot, Y., J.-C. Gloaguen, G. Picot, J. Bougère and D. Benjamin, 1993, Azorella selago Hook. used to estimate glacier fluctuations and climatic history in the Kerguelen Islands over the last two centuries, Oecologia, 95, 140-144.

13

Fujisada, H., G.B. Bailey, G.G. Kelly, S. Hara and M.J. Abrams, 2005, ASTER DEM performance, IEEE Transactions on Geoscience and Remote Sensing, 43(12), 2707-2714. Haeberli, W., H. Bösch, K. Scherler, G. Østrem and C.C. Wallén, eds., 1989, World Glacier Inventory: Status 1988, Wallingford, Oxon, IAHS Press; Nairobi, GEMS-UNEP; Paris, UNESCO. Harris, U., 2009, Heard Island digitizing 2009, CAASM Metadata, Australian Antarctic Division Data Centre (http://data.aad.gov.au/aadc/metadata/). Hatherton, T., E.W. Dawson and F.C. Kinsky, 1965, Balleny Islands reconnaissance expedition, 1964, New Zealand Journal of Geology and Geophysics, 8(2), 164-179. Hayakawa, Y.S., T. Oguchi and Z. Lin, 2008, Comparison of new and existing global digital elevation models: ASTER G-DEM and SRTM-3, Geophysical Research Letters, 35(17), L17404. doi:10.1029/2008GL035036. Holdgate, M.W., and P.E. Baker, 1979, The South Sandwich Islands: I. General Description. Scientific Report 91, British Antarctic Survey, Cambridge. 76p., 11 plates. Jacka, T.H., W.F. Budd and A. Holder, 2004, A further assessment of surface temperature changes at stations in the Antarctic and Southern Ocean, 1949-2002, Annals of Glaciology, 39, 331-338. Kemp, S., and A.L. Nelson, 1931, The South Sandwich Islands, Discovery Reports, Volume III, 133-198. Cambridge University Press, London. Kiernan, K., and A. McConnell, 2002, Glacier retreat and melt-lake expansion at Stephenson Glacier, Heard Island World Heritage Area, Polar Record, 38(207), 297-308. Lachlan-Cope, T., J.L. Smellie and R. Ladkin, 2001, Discovery of a recurrent lava lake on Saunders Island (South Sandwich Islands) using AVHRR imagery, Journal of Volcanology and Geothermal Research, 112, 105-116. Langenegger, O., and W.J. Verwoerd, 1971, Topographic survey, in van Zinderen Bakker, E.M., Sr., J.M. Winterbottom and R.A. Dyer, eds., Marion and Prince Edward Islands, 32-39. A.A. Balkema, Cape Town. 427p. López-Martinez, J., and E. Serrano, 2002, Geomorphology, in Smellie, J.L. et al., Geology and Geomorphology of Deception Island, BAS Geomap Series Sheets 6-A and 6-B, 31-39. British Antarctic Survey, Cambridge. Lucieer, A., A. Bender and U. Harris, 2009, Remote Sensing HIMI Project 2008/2009: Final Report, Australian Antarctic Division Data Centre (in preparation). Marsz, A.A., 2002, Ujemny trend rocznych sum opadowych na stacji im. H. Arctowskiego (Wyspa Króla Jerzego, Szetlandy Pd., Antarktyka Zach.), Problemy Klimatologii Polarnej, 8, 63-77. [The negative trend in total annual precipitation at Arctowski station (King George Island, South Shetland Islands, West Antarctica).] Molina, C., F.J. Navarro, J. Calvet, D. García-Sellés and J.J. Lapazaran, 2007, Hurd Peninsula glaciers, Livingston Island, Antarctica, as indicators of regional warming: ice-volume changes during the period 1956-2000, Annals of Glaciology, 46, 43-49. NASA Earth Observatory, 2005, Mount Belinda erupts, http://earthobservatory.nasa.gov/IOTD/view.php?id=5949. Posted 21 October 2005; accessed 28 October 2009. Noble, H.M., 1965, Glaciological observations at Admiralty Bay, King George Island, in 195758, British Antarctic Survey Bulletin, 5, 1-11. Norsk Polarinstitutt, 1988, Peter I Øy, map at 1:50 000 scale, contour interval 20 m, from 1987 air photos. Norsk Polarinstitutt, Tromsø. Orheim, O., 1981, The glaciers of Bouvetøya, Norsk Polarinstitutt Skrifter 175, 79-84. Norsk Polarinstitutt, Tromsø. Orheim, O., and L.S. Govorukha, 1982, Present-day glaciation in the South Shetland Islands, Annals of Glaciology, 3, 233-238.

14

Patrick, M.R., and 7 others, 2005, First recorded eruption of Mount Belinda volcano (Montagu Island), South Sandwich Islands, Bulletin of Volcanology, 67, 415-422. Ruddell, A., 2006, An inventory of present glaciers on Heard Island and their historical variation, in Green, K., and E. Woehler, eds., Heard Island: Southern Ocean Sentinel, 28-51. Surrey Beatty and Sons, Chipping Norton, NSW. Ryan, U., 2004, Heard Island RADARSAT (1997) digital elevation model. Australian Antarctic Data Centre: SnoWhite Metadata, http://www.aad.gov.au/default.asp?casid=3802. Smellie, J.L., and 9 others, 2002, Geology and Geomorphology of Deception Island. BAS Geomap Series Sheets 6-A and 6-B, maps at 1:25 000 scale and text (78p.). British Antarctic Survey, Cambridge. Smith, J., 1960, Glacier problems in South Georgia, Journal of Glaciology, 3, 705-714. Stephenson, J., G. Budd, J. Manning and P. Hansbro, 2005, Major eruption-induced changes to the McDonald Islands, southern Indian Ocean, Antarctic Science, 17(2), 259–266. Sumner, P.D., K.I. Meiklejohn, J.C. Boelhouwers and D.W. Hedding, 2004, Climate change melts Marion Island’s snow and ice, South African Journal of Science, 100, 395–398. Thost, D.E., and M. Truffer, 2008, Glacier recession on Heard Island, southern Indian Ocean. Arctic, Antarctic and Alpine Research, 40(1), 199-214. Thost, D.E., M. Truffer and S. Donoghue, 2004, The Heard Island glaciology program 2003-04: studies on the morphology, dynamics, mass balance and climatic setting of Brown Glacier. Unpublished report, Australian Antarctic Division, Hobart. 32p. Accessed 20 February 2006 from http://www.gi.alaska.edu/~truffer/research_main.html. Turner, J., S.R. Colwell, G.J. Marshall, T.A. Lachlan-Cope, A.M. Carleton, P.D. Jones, V. Lagun, P.A. Reid and S. Iagovkina, 2004, The SCAR READER project: toward a high-quality database of mean Antarctic meteorological observations, Journal of Climate, 17(14), 2890-2898. Vallon, M., 1977a, Bilans de masse et fluctuations récentes du glacier Ampère (Iles Kerguelen, T.A.A.F.), Zeitschrift für Gletscherkunde und Glazialgeologie, 13(1-2), 57-85. Vallon, M., 1977b, Topographie sous-glaciaire du glacier Ampère (Iles Kerguelen, T.A.A.F.), Zeitschrift für Gletscherkunde und Glazialgeologie, 13(1-2), 37-55. Wen, J.H., J.C. Kang, J.K. Han, Z.C. Xie, L.B. Liu and D.L. Wang, 1998, Glaciological studies on the King George Island ice cap, South Shetland Islands, Antarctica, Annals of Glaciology, 27, 105-109. Wright, R., L.P. Flynn, H. Garbeil, A.J.L. Harris and E. Pilger, 2004, MODVOLC: near-real-time thermal monitoring of global volcanism, Journal of Volcanology and Geothermal Research, 135, 29-49. Zazulie, N., M. Rusticucci and S. Solomon, 2010, Changes in climate at high southern latitudes: a unique daily record at Orcadas spanning 1903-2008, Journal of Climate, 23(1), 189-196.

15

Table 5. Cartographic glacier inventory of the Subantarctic (WGI region 6C) Region

Inventory Code

Scott Peter I S Shetland Is Clarence Elephant Gibbs Is King George Nelson Robert Greenwich Livingston Deception Snow Low Smith S Orkney Is Laurie Saddle Powell Robertson Is Coronation Signy S Sandwich Is Zavodovski Leskov Visokoi Candlemas Vindication Saunders Montagu Bristol Bellingshausen

6C10100 6C10200 6C201 6C20100 6C20101 6C20102 6C20103 6C20104 6C20105 6C20106 6C20107 6C20108 6C20109 6C20110 6C20111 6C202 6C20200 6C20201 6C20202 6C20203 6C20204 6C20205 6C203 6C20300 6C20301 6C20302 6C203030 6C203031 6C20304 6C20305 6C20306 6C203070

Longitude (deg)

Latitude (deg)

Glacierized area (km2)

-179.93 -90.80 -57.40 -54.10 -55.25 -55.75 -58.30 -59.06 -59.45 -59.85 -60.10 -60.55 -61.40 -62.15 -62.48 -45.00 -44.75 -44.80 -45.03 -45.12 -45.50 -45.63 -26.45 -27.58 -28.13 -27.17 -26.67 -26.78 -26.45 -26.33 -26.58 -27.10

-67.40 -68.85 -61.70 -61.25 -61.48 -61.48 -62.00 -62.32 -62.40 -62.40 -62.60 -62.95 -62.80 -63.30 -63.00 -60.70 -60.77 -60.62 -60.67 -60.74 -60.58 -60.72 -57.76 -56.30 -56.67 -56.70 -57.08 -57.10 -57.76 -58.42 -59.03 -59.43

0.05 151.0 3314 116.1 458.6 16.2 1085 156.6 148.1 145.4 756.6 62.3 117.6 130.4 121.6 486 55 0.23 18 2.3 404 6.5 277.6 (0.1) 0.0 24.1 4.2 0.3 25.6 93.5 94.5 0.1

16

Date of glacierized area ~2002 1987 1957 1957 1957 2000 1957 1957 1957 1957 1979 1956 1957 1957 1979 1979 1979 1979 1979 1968 1962 1962 1962 1964 1964 1964 1964 1964 1964

Area (km2)

Glacier cover (%)

0.10 156.0 3603 121.8 468.0 25.2 1150 164.8 154.2 155.3 849.7 109.3 124.9 135.0 145.2 575 74 3 25 5.9 446 21 335.1 15.5 0.3 30.7 10.2 3.1 34.7 101.4 97.7 1.4

50 96.8 92.0 95.3 98.0 64.3 94.4 95.0 96.0 93.6 89.0 57.0 94.2 96.6 83.8 84.5 74.3 7.7 72.0 39.0 90.6 31.0 82.8 0.6 0.0 78.5 41.2 9.7 73.8 92.2 96.7 7.1

Maximum elevation (m a.s.l.) 60 1640 ~2100 1924 973 734 ~700 325 385 ~600 1700 539 320