Workflow: Annotated pdf, CrossRef and tracked changes

PROOF COVER SHEETAuthor(s): Sebastian H. Mernild

Article title: Glacier changes in the circumpolar Arctic and sub-Arctic, mid-1980s to late-2000s/2011

Article no: RDGS 1026917

Enclosures: 1) Query sheet

2) Article proofs

Dear Author,

1. Please check these proofs carefully. It is the responsibility of the corresponding author to check these and approveor amend them. A second proof is not normally provided. Taylor & Francis cannot be held responsible for uncorrectederrors, even if introduced during the production process. Once your corrections have been added to the article, it will beconsidered ready for publication.

Please limit changes at this stage to the correction of errors. You should not make trivial changes, improve prose style,add new material, or delete existing material at this stage. You may be charged if your corrections are excessive(we would not expect corrections to exceed 30 changes).

For detailed guidance on how to check your proofs, please paste this address into a new browser window:http://journalauthors.tandf.co.uk/production/checkingproofs.asp

Your PDF proof file has been enabled so that you can comment on the proof directly using Adobe Acrobat. If you wishto do this, please save the file to your hard disk first. For further information on marking corrections using Acrobat,please paste this address into a new browser window: http://journalauthors.tandf.co.uk/production/acrobat.asp

2. Please review the table of contributors below and confirm that the first and last names are structured correctlyand that the authors are listed in the correct order of contribution. This check is to ensure that your name willappear correctly online and when the article is indexed.

Sequence Prefix Given name(s) Surname Suffix

123456

Sebastian H.Jeppe K.Jacob C.Simon

Niels TvisRyan

MernildMalmrosYde

De VilliersKnudsenWilson

Queries are marked in the margins of the proofs, and you can also click the hyperlinks below.Content changes made during copy-editing are shown as tracked changes. Inserted text is in red font and revisions havea red indicatorå. Changes can also be viewed using the list comments function. To correct the proofs, you shouldinsert or delete text following the instructions below, but do not add comments to the existing tracked changes.

AUTHOR QUERIES

General points:

1. Permissions: You have warranted that you have secured the necessary written permission from the appropriatecopyright owner for the reproduction of any text, illustration, or other material in your article. Please seehttp://journalauthors.tandf.co.uk/permissions/usingThirdPartyMaterial.asp.

2. Third-party content: If there is third-party content in your article, please check that the rightsholder details forre-use are shown correctly.

3. Affiliation: The corresponding author is responsible for ensuring that address and email details are correct for allthe co-authors. Affiliations given in the article should be the affiliation at the time the research was conducted.Please see http://journalauthors.tandf.co.uk/preparation/writing.asp.

4. Funding: Was your research for this article funded by a funding agency? If so, please insert ‘This work wassupported by <insert the name of the funding agency in full>’, followed by the grant number in square brackets‘[grant number xxxx]’.

5. Supplemental data and underlying research materials: Do you wish to include the location of the underlyingresearch materials (e.g. data, samples or models) for your article? If so, please insert this sentence before thereference section: ‘The underlying research materials for this article can be accessed at <full link>/ description oflocation [author to complete]’. If your article includes supplemental data, the link will also be provided in thisparagraph. See <http://journalauthors.tandf.co.uk/preparation/multimedia.asp> for further explanation of supplementaldata and underlying research materials.

6. The CrossRef database (www.crossref.org/) has been used to validate the references. Changes resulting frommismatches are tracked in red font.

AQ1 Please provide an institutional e-mail address, if available, to be included in the article, as per jour-nal style.

AQ2 Would you consider changing the title from ‘Glacier changes in the circumpolar Arctic and sub-Arctic, mid-1980s to late-2000s/2011’ to ‘Glacier changes in the circumpolar Arctic and sub-Arctic,mid-1980s to late 2000s/2011?’ Kindly suggest.

AQ3 The reference ‘Hansen et al., 2010’ is cited in the text but is not listed in the references list. Pleaseeither delete in-text citation or provide full reference details following journal style[http://www.tandf.co.uk/journals/authors/style/reference/tf_APA.pdf].

AQ4 The reference ‘Bolch et al., 2010’ is cited in the text but is not listed in the references list. Pleaseeither delete in-text citation or provide full reference details following journal style[http://www.tandf.co.uk/journals/authors/style/reference/tf_APA.pdf].

AQ5 Please note that figures have been renumbered to maintain sequential order in the text. Please checkand confirm.

AQ6 The reference ‘Bahr, 2011’ is cited in the text but is not listed in the references list. Please eitherdelete in-text citation or provide full reference details following journal style[http://www.tandf.co.uk/journals/authors/style/reference/tf_APA.pdf].

AQ7 The reference ‘Carr et al., 2012’ is cited in the text but is not listed in the references list. Pleaseeither delete in-text citation or provide full reference details following journal style[http://www.tandf.co.uk/journals/authors/style/reference/tf_APA.pdf].

AQ8 Please provide the Notes on contributors (regarding current academic position, research interests,books/articles published).

AQ9 The reference ‘ACIA, 2005’ is listed in the references list but is not cited in the text. Please eithercite the reference or remove it from the references list.

AQ10 Please provide missing editor names for the “AMAP (2011)” references list entry.

AQ11 The reference ‘AMAP, 2011’ is listed in the references list but is not cited in the text. Please eithercite the reference or remove it from the references list.

AQ12 Please provide missing city for the “Andreassen et al., 2001” references list entry.

AQ13 Please provide missing city for the “Andreassen et al., 2012” references list entry.

AQ14 The reference ‘Andreassen et al., 2012’ is listed in the references list but is not cited in the text.Please either cite the reference or remove it from the references list.

AQ15 The reference ‘Arendt et al., 2002’ is listed in the references list but is not cited in the text. Pleaseeither cite the reference or remove it from the references list.

AQ16 The reference ‘Bahr, 1997’ is listed in the references list but is not cited in the text. Please eithercite the reference or remove it from the references list.

AQ17 The reference ‘Bahr, 2012’ is listed in the references list but is not cited in the text. Please eithercite the reference or remove it from the references list.

AQ18 The reference ‘Berthier et al., 2010’ is listed in the references list but is not cited in the text. Pleaseeither cite the reference or remove it from the references list.

AQ19 The reference ‘Bolch et al., 2012’ is listed in the references list but is not cited in the text. Pleaseeither cite the reference or remove it from the references list.

AQ20 Please provide missing city and page numbers for the “Cogley 2012” references list entry.

AQ21 Please provide missing city for the “Dyurgerov (2010)” references list entry.

AQ22 Please provide missing publisher for the “Glazovsky and Macheret, 2006” references list entry.

AQ23 The reference ‘Hansen et al., 2012’ is listed in the references list but is not cited in the text. Pleaseeither cite the reference or remove it from the references list.

AQ24 The reference ‘Huss and Farinotti, 2012’ is listed in the references list but is not cited in the text.Please either cite the reference or remove it from the references list.

AQ25 Please provide missing city and publisher for the “Kotlyakov et al., 2010” references list entry.

AQ26 Please provide missing page numbers for the “Moholdt et al. 2012” references list entry.

AQ27 Please provide missing page numbers for the “Paul et al. 2013” references list entry.

AQ28 Please provide missing volume number and page numbers for the “Radić et al. 2014” references listentry.

AQ29 The reference ‘Rayner, 2003’ is listed in the references list but is not cited in the text. Please eithercite the reference or remove it from the references list.

AQ30 Please provide missing city for the “Rouse et al., 1973” references list entry.

AQ31 The reference ‘Shahgedanova et al., 2010’ is listed in the references list but is not cited in the text.Please either cite the reference or remove it from the references list.

AQ32 Please provide missing editor names for the “Sharp et al., 2003” references list entry.

AQ33 Please provide missing city for the “World Glacier Monitoring Service, 2011” references list entry.

AQ34 Please provide missing page numbers for the “Vaughan et al., 2013” references list entry.

AQ35 The Year for ‘Barrand and Sharp (2010)’ has been changed to match the entry in the references list.Please confirm this is correct and provide revisions if needed.

AQ36 The Year for ‘Kargel et al. (2012)’ has been changed to match the entry in the references list.Please confirm this is correct and provide revisions if needed.

AQ37 The reference ‘Andreassen and Winsvold (2012)’ is cited in the text but is not listed in the refer-ences list. Please either delete in-text citation or provide full reference details following journalstyle[http://www.tandf.co.uk/journals/authors/style/reference/tf_APA.pdf].

AQ38 The disclosure statement has been inserted. Please correct if this is inaccurate.

How to make corrections to your proofs using Adobe Acrobat/Reader

Taylor & Francis offers you a choice of options to help you make corrections to your proofs. Your PDF proof file hasbeen enabled so that you can mark up the proof directly using Adobe Acrobat/Reader. This is the simplest and best wayfor you to ensure that your corrections will be incorporated. If you wish to do this, please follow these instructions:

1. Save the file to your hard disk.

2. Check which version of Adobe Acrobat/Reader you have on your computer. You can do this by clicking on the“Help” tab, and then “About”.

If Adobe Reader is not installed, you can get the latest version free from http://get.adobe.com/reader/.

3. If you have Adobe Acrobat/Reader 10 or a later version, click on the “Comment” link at the right-hand side to viewthe Comments pane.

4. You can then select any text and mark it up for deletion or replacement, or insert new text as needed. Please notethat these will clearly be displayed in the Comments pane and secondary annotation is not needed to draw attentionto your corrections. If you need to include new sections of text, it is also possible to add a comment to the proofs.To do this, use the Sticky Note tool in the task bar. Please also see our FAQs here: http://journalauthors.tandf.co.uk/production/index.asp.

5. Make sure that you save the file when you close the document before uploading it to CATS using the “Upload File”button on the online correction form. If you have more than one file, please zip them together and then upload thezip file.

If you prefer, you can make your corrections using the CATS online correction form.

Troubleshooting

Acrobat help: http://helpx.adobe.com/acrobat.htmlReader help: http://helpx.adobe.com/reader.html

Please note that full user guides for earlier versions of these programs are available from the Adobe Help pages byclicking on the link “Previous versions” under the “Help and tutorials” heading from the relevant link above. Comment-ing functionality is available from Adobe Reader 8.0 onwards and from Adobe Acrobat 7.0 onwards.

Firefox users: Firefox’s inbuilt PDF Viewer is set to the default; please see the following for instructions on how touse this and download the PDF to your hard drive: http://support.mozilla.org/en-US/kb/view-pdf-files-firefox-without-downloading-them#w_using-a-pdf-reader-plugin

Glacier changes in the circumpolar Arctic and sub-Arctic, mid-1980s to late-2000s/2011

Sebastian H. Mernilda,b*, Jeppe K. Malmrosa,c, Jacob C. Yded, Simon De Villiersd, Niels Tvis Knudsene andRyan Wilsona

5 aGlaciology and Climate Change Laboratory, Center for Scientific Studies/Centro de Estudios Cientificos (CECs), Valdivia, Chile;bClimate, Ocean, and Sea Ice Modeling Group, Computational Physics and Methods, Los Alamos National Laboratory, Los Alamos,NM, USA; cDepartment of Geosciences and Natural Resource Management, University of Copenhagen, Copenhagen, Denmark;

dFaculty of Engineering and Science, Sognog Fjordane University College, Sogndal, Norway; eDepartment of Geoscience, AarhusUniversity, Aarhus, Denmark

10 (Received 13 March 2014; final version received 4 March 2015)

A new inventory record of satellite-derived area, length, elevation range and surface slope changes from the mid-1980sto late 2000s/2011 for 317 land-terminating glaciers and ice caps (GIC) is presented. The investigated GIC are locatedin 12 geographic regions throughout the circumpolar Arctic and sub-Arctic. This geographic subdivision allows us toexamine regional variations in recent glacier changes. The method is based on a semi-automated classification method

15 that extracts GIC extent from satellite scenes. Most of the observed GIC show a reduction in area, length, elevation ran-ge and slope. On regional scale, the observed GIC changed in area between −4 ± 3% (Nuuk, West Greenland; 1987–2003) and −40 ± 4% (Talkeetna, southern Alaska; 1987–2011), equal to shrinking rates between −0.2% yr−1 and −1.7%yr−1. The regional change in length was between −36 ± 13 m (southern British Columbia; 1985–2011) and −481 ± 85 m(southern Ellesmere Island; 1988–2009), equal to −1 ± 0.5 m yr−1 and −23 ± 4 m yr−1. Regional GIC changes can be

20 illustrated by power-law scaling relationships between GIC area and length, elevation range, and surface slope. Here, wefind regional variability in scaling parameters in both time and space, which should be considered when estimatingglobal assessments of GIC conditions and changes over time.

Keywords: area change; Arctic; glacier and ice caps; Landsat; length change; remote sensing

1. Introduction

25 Glaciers and ice caps (henceforth GIC) are influenced byclimate change (AMAP, 2011; Mernild, Liston, et al.,2014; Vaughan et al., 2013). Global warming in recentdecades has been more pronounced at high latitudes(Hansen et al., 2010), resulting in negative surface mass

30 balance, thinning and shrinking of GIC (e.g. AMAP,2011; Dowdeswell et al., 2007; Dyurgerov, 2010; Jacobet al., 2012; Kotlyakov et al., 2010; Leclercq et al.,2014; Mernild, Hanna et al., 2014; Mernild, Knudsen,et al., 2013; Mernild, Lipscomb, et al., 2013; Moholdt

35 et al., 2012; Sharp et al., 2003; World Glacier Monitor-ing Service, 2011). The area of GIC in the Arctic andsub-Arctic region (covering e.g. parts of Alaska, Canada,Greenland, north of Scandinavia and Siberia) is~440,000 km2 (Radić et al., 2014; Pfeffer et al., 2014),

40 which amounts to about ~60% of the total global GICarea of ~740,000 km2. This highlights the need toobserve and map GIC area change in the Arctic and sub-Arctic regions.

GIC area changes and frontal positions have become45 far more extensively observed in the satellite era (e.g.

Bjørk et al., 2012; Bolch et al., 2010; Glazovsky & Mac-heret, 2006; Leclercq et al., 2014; Sharp et al., 2003;

Wolken et al., 2008; Yde & Knudsen, 2007). However,only a small fraction of the thousands of individual GIC

50that are irregularly distributed across the Arctic and sub-Arctic have been directly field observed. Consequently,there is a lack of information about present GIC changesin mass balance, area, length, elevation, and slope andtheir relationship tolocal climate. Existing area analyses

55of Arctic and sub-Arctic GIC changes are based on his-torical accounts and aerial/satellite images spanning dif-ferent time intervals from the late ninetieth century tothe present day. These studies suggest that GIC areachange has ranged on average between −3% (1952–

602001) for the Russian Arctic islands (Glazovsky & Mac-heret, 2006) and −67% (1950–2003) for Koryak Upland(near the Kamchatka Peninsula) (Ananicheva & Kapu-stin, 2010). Mean length change rates range from−8 m yr−1 (1953–2005) to −12 m yr−1 (1900–1999) in

65West Greenland (Leclercq et al., 2012; Yde & Knudsen,2007), but change rates of up to −19 m yr−1 for a shorterand more recent period, 2005–2010, have been reportedfor southern and northern Norway (Andreassen et al.,2011). A global data-set of 471 GIC length fluctuation

70time series (excluding data from Canadian Arctic) indi-cated an average length change of −1560 ± 30 m over

AQ3

AQ2

AQ1AQ4

*Corresponding author. Email: smernild@gmail.com

© 2015 The Royal Danish Geographical Society

Geografisk Tidsskrift-Danish Journal of Geography, 2015http://dx.doi.org/10.1080/00167223.2015.1026917

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

the twentieth century (Leclercq et al., 2014). Whilechanges in length, area and volume receive increasingattention due to their relationship with climate change

5 (Vaughan et al., 2013), changes in other physicalcharacteristics such as slope and elevation remain to beexamined.

Here, we present a new inventory of GIC area,length, elevation and surface slope conditions (derived

10 from multispectral Landsat satellite imagery) from themid-1980s to late 2000s/2011, for 317 land-terminatingGIC (Figure 1). Although not giving a full description ofGIC changes in the entire Arctic and sub-Arctic, thisinvestigation includes GIC located in 12 geographic

15 regions distributed across the circumpolar Arctic andsub-Arctic (between 89.3°N and 52.4°N latitude; Fig-ure 1). This analysis allows us for the first time to studyboth temporal and spatial variability in GIC area, length,elevation and slope across large regions of the Arctic

20 and sub-Arctic for an almost common period. In thepast, earlier studies have been focused on specific timeperiods and regions. The new data-set can be used inconjunction with existing glacier inventories, which typi-cally present data from a single time period (e.g. the

25 Randolph Glacier Inventory (RGI) (Pfeffer et al., 2014)).In addition, it can be linked to the Global Land Ice Mea-surements from Space Glacier database (a projectdesigned to monitor the world’s glaciers primarily usingdata from optical satellite instruments; www.glims.org).

30 Using this unique GIC data-set, we further analysed thepower-law relationships between: (1) GIC area andlength; (2) area and elevation and (3) area and surfaceslope to emphasize the variability in scaling parametersin time and space since previous glacier-projecting stud-

35 ies have used fixed scaling parameters when estimatingglobal-scale GIC conditions. Also, because it is presentlynot viable to measure the area, length, elevation andslope of every remote GIC on Earth, power relations canbe applied to investigate temporal and spatial variability

40 on GIC conditions.

2. Methodology

2.1. Satellite data preparation and GIC classification

The GIC planimetric area data-set was obtained from 28image scenes captured by the Landsat 5 TM (Thematic

45 Mapper) and Landsat 7 ETM+(Enhanced Thematic Map-per Plus) satellite sensors, having a ground resolution of30 m (Table 1). The selected satellite images were: (1)obtained from a pair of Landsat scenes covering thesame area, recorded at least 10 years apart and with end

50 scenes obtained no earlier than 2003 (within the last dec-ade); (2) acquired around the end of the ablation season,typically from the end of July to the end of Septemberin order to avoid and minimize snow-covered

interference since small GIC can easily be misclassified55as snow patches; (3) obtained from scenes with less than

50% total cloud cover (cloud cover did not occur overthe sampled GIC margins); (4) visually inspected beforeuse in the supervised classification process to avoid mis-classification caused by snow-covered scenes; (5) when

60using Landsat 7 ETM+(2003 to present) subject to theScan Line Instrument malfunction, two (approximatelysame date) scenes were combined to replace data gaps;and (6) obtained where the image gaps did not interferewith the glacier area mapping.

65The digital elevation data used were from theAdvanced Spaceborne Thermal Emission and ReflectionRadiometer (ASTER) Global Digital Elevation Model(GDEM) version 2 (30 m resolution) (Tachikawa et al.,2011), obtained from the United States Geological Sur-

70vey Earth Explorer internet portal (http://earthexplorer.usgs.gov/).The surface elevation is to be considered as aparameter elevation since the same DEM is usedthroughout the time period.

The acquired Landsat imagery covered 12 Arctic and75sub-Arctic glaciated regions (Figure 1) meeting the selec-

tion criteria. The regions were chosen to represent thephysiographic variability of glaciated regions within thecircumpolar Arctic and sub-Arctic, for example, fromrelatively high elevated, steep slope and humid condi-

80tions in British Columbia to relatively low elevated,gradual slope and dry conditions on Bolshevik Island(Figure 1). No data from Svalbard or Iceland were useddue to the lack of available scenes (the scenes not meet-ing the selection criteria): these two areas represent only

85~10% of the total Arctic GIC area (Radić et al., 2014).For each region, an observation period from the

mid-1980s to late 2000s/2011 was ideally sought. Forthe majority of the regions (8 out of 12), a minimumtime period of 21 years was covered, where Thule and

90Narvik cover a period of 19 years, Nuuk (16 years) andnorthern Kamchatka (12 years). Also, for 9 out of 12regions, satellite imagery were available from 2009 to2011, otherwise from 2003 to 2006. The non-identicalobservation periods used for some of the sampled

95regions might be a limiting factor for subsequent GICchange comparisons.

For all the Landsat scenes, the local UTM coordinatesystem was used with the geodetic datum WorldGeodetic System 1984 (WGS84). The scenes were

100atmospherically and radiometrically calibrated using theLandsat calibration tool in ENVI™ software package(http://www.ittvis.com/ProductServices/ENVI.aspx), con-verting the band values to surface reflectance. Theindividual bands (TM and ETM+ bands 1–5, and 7)

105were standardized using the ENVI™ Dark Subtract toolbefore ratios and indices were calculated.

The main GIC classification used was based on asupervised semi-automated classification process which

AQ5

2 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

Monofor

print

colour o

nline

Figure 1. Satellite-derived area changes of 317 GIC in the Arctic and sub-Arctic. Changes are shown as rates during the observationperiod from c. mid-1980s to late 2000s/2011 (the period varies between regions, and for northern Kamchatka the observation periodis 1999–2011). The data were divided into 12 local/minor regions: (1) Talkeetna area (15.4 × 103 km2, upper left); (2) an area insouthern British Columbia (5.6); (3) an area on southern Ellesmere Iceland (10.5); (4) Thule area (8.9); (5) Nuuk area (2.7); (6)Sermilik Fjord area (10.9); (7) Danmarkshavn area (9.1); (8) an area in eastern Sognefjord (6.6); (9) Narvik area (22.4); (10) an areaon central Novaya Zemlya (4.4); (11) an area on southern Bolshevik Island (6.6); and (12) northern Kamchatka area (6.5). Red circlesshow shrinking GIC and blue circles expanding GIC (percentage, %). Background satellite images are from Landsat 5 TM (the datesare illustrated at the bottom of each image).

Geografisk Tidsskrift-Danish Journal of Geography 3

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

delineated GIC area from each satellite scene utilizing5 multicriteria analysis involving the calculation of a series

of indices. The GIC classification workflow is illustratedin Figure 2 (Mernild et al., 2012). Indices used duringthis process include: (1) the Normalized Difference SnowIndex (NDSI) (Hall et al., 1995); (2) the Normalized Dif-

10 ference Vegetation Index (NDVI) (Rouse et al., 1973) tofilter out vegetation; and (3) the Normalized DifferenceWater Index (NDWI) (Gao, 1996) to filter out ice-mar-ginal lakes.

Derivatives of the aforementioned indices were com-15 bined into a model along with specified threshold values

(see further down in Section 2.1), in order to automateclassification outputs (Figure 2). The resulting GIC clas-sifications were converted to polygon files, representingthe GIC area, and were manually edited whereever nec-

20essary in ESRI™ ArcMap. After the satellite-derivedGIC classification, terminus positions were visuallyinspected and manually corrected for each observationperiod (beginning and end), and area shrinkage/advancerates were calculated. An example of GIC area estima-

25tion is shown in Figure 3, where ice margins were digi-tized for a specific glacier for both 1986 and 2011.Manual editing was usually necessary for debris-covered

Table 1. Satellite platform, sensors, band information, scenes used in the analysis and precision errors. A geographical distributionof the glaciated regions can be seen in Figure 1.

Glaciatedregions Platform Sensor and bands

Groundresolution,

mPrecisionerror, m Scenes

Survey year,month, and

date

Talkeetna Landsat-5 TM (bands 1–5, and 7) 30 × 30 ±15 LT50700161987233XXX02 19872108Landsat-5 TM (bands 1–5, and 7) LT50700162007240GLC00 20072808

SouthernBritishColumbia

Landsat-5 TM (bands 1–5, and 7) LT50510231985270PAC00 19852709Landsat-5 TM (bands 1–5, and 7) LE70510232011254EDC00 20111109

SouthernEllesmereIsland

Landsat-5 TM (bands 1–5, and 7) LT50450051988205PAC00 19882307Landsat-5 TM (bands 1–5, and 7) LT50450052009214GLC00 20090208

Thule Landsat-5 TM (bands 1–5, and 7) LT50370021987226XXX01 19871408Landsat-5 TM (bands 1–5, and 7) LT50380022006221KIS00 20060908

Nuuk Landsat-5 TM (bands 1–5, and 7) LT50050151987242XXX03 19873008Landsat-7 ETM+ (bands 1–5, and 7) LE70070152003260EDC02 20041709Landsat-7 ETM+ (bands 1–5, and 7) LE70070152004247EDC02 20030409

Sermilik Fjord Landsat-5 TM (bands 1–5, and 7) LT52310141986254XXX03 19860911Landsat-7 ETM+ (bands 1–5, and 7) LE72320132011226EDC01 20111408

LE72320142011226EDC00 20111408LE72310142007256EDC00 20070409

Danmarkshavn Landsat-5 TM (bands 1–5, and 7) LT52310061985219XXX03 19850708Landsat-5 TM (bands 1–5, and 7) LT52310062009221KIS00 20090908

EasternSognefjord

Landsat-5 TM (bands 1–5, and 7) LT51970121987227XXX01 19871508Landsat-5 TM (bands 1–5, and 7) LE71960122006232ASN00 20061908

m Landsat-5 TM (bands 1–5, and 7) LT52000171988219KIS00 19880608Landsat-5 TM (bands 1–5, and 7) LT51990172011259MOR00 20110609

CentralNovayaZemlya

Landsat-5 TM (bands 1–5, and 7) LT51770081987215XXX02 19870308Landsat-7 ETM+ (bands 1–5, and 7) LE71780082011232ASN00 20112008Landsat-7 ETM+ (bands 1–5, and 7) LE71780082011248ASN00 20110509

BolshevikIsland

Landsat-5 TM (bands 1–5, and 7) LT51570041985213XXX02 19850108Landsat-7 ETM+ (bands 1–5, and 7) LE71600032011234PFS00 20112208

NorthernKamchatka

Landsat-5 TM (bands 1–5, and 7) LE70990201999222EDC00 19991008Landsat-5 TM (bands 1–5, and 7) LT50990202011215MGR00 20110308

4 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

ice areas. Supraglacial debris effectively masks thespectral characteristics of ice resulting in classification

5 errors during automated mapping procedures.The index threshold input values used during the

GIC classification process varied depending on theregion and the date. Differences were mainly accountedfor by temporal changes in inherent surface reflectivity

10 and sun angle. Individual threshold values were selectedbased on classification performance, in regard to theaccurate identification of snow/ice, vegetation and water(NDSI: 0.4–0.5, NDVI: 0.25–0.35, NDWI: 0.2–0.5).Changes in the reflectivity of snow cover in particular,

15 due to compaction and changes in snow wetness, madeit difficult to apply an all-round classification algorithmthat worked everywhere at all times (Brest, 1987; Hallet al., 2001).

To aid in the visual correction of the automatic ice20area classifications, a simple ratio between visible light

and short wave infrared (SWIR) landsat bands was cal-culated using a similar but non-normalized approach tothat taken for the NDSI. The most commonly used ratiobetween bands 3 and 5 (red and SWIR) channels was

25introduced by Crane and Anderson (1984), and utilizedby Dozier and Marks (1987). An alternative ratiobetween bands 2 and 5 (Blue and SWIR) was found tooften produce better contrast in mountainous areas withsteep slopes (slope ≥ 55°) and shadow areas than the 3/5

30ratio, enabling ice and snow-covered areas to be moreaccurately identified. Similar results were found by Paul(2004) when applying ratios to Landsat TM and ETM+scenes covering the Swiss Alps. The 2/5 band ratiowas only used as an extra visual aid in the correction/

Figure 2. A schematic diagram of the workflow for the GIC classification of Landsat images. The trapeze shapes indicate stepswhere input data were added or output data generated, and the diamond shapes indicate steps where data processing occurred.

Monofor

print

colour o

nline

Figure 3. An example of the margin position for the glacier L29, in the Sermilik Fjord area (Southeast Greenland), for 1986 (blue)and 2011 (red), and the 1986 to 2011 area change (yellow-black) (the location of glacier L29 is 65.99°N; 35.99°W). Backgroundsatellite images are from Landsat 5 TM (11 September 1986).

Geografisk Tidsskrift-Danish Journal of Geography 5

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

5 digitizing process, not part of the auto-classificationmodel (Figure 2). Specific classification notes can befound in the Supplementary Material.

2.2. GIC auto-classification uncertainties

To estimate classification errors (misclassification), the10 raw GIC classifications produced by the automated

model were compared with the finalized visually cor-rected classifications. The overall misclassification rateproduced by the auto-classification model was found tobe on average 4.9% (3.0% due to the occurrence of

15 snow patches, and by 1.9% due to heavily debris cov-ered terrain and effects of shadows) (Table 2), rangingfrom 2.7% in the Thule region to 6.2% in the Talkeetnaregion. The estimated errors are higher than thosereported by Bolch et al. (2010) of ±2% for debris-free

20 GIC, but in the same range as the mapping errors fordebris-covered glacier tongues of ±2.5% reported byPaul et al. (2013).

The standard error was expected to be half the pixelsize of the imagery used: ±15 m for both TM and ETM+

25 (Table 1) (Hall et al., 2003; Mernild et al., 2012). As atest, the Landsat 7 satellite-derived margin of MittivakkatGletscher (14 August 2011), located in the SermilikFjord region, Southeast Greenland, was validated againstdirect GPS ice margin observations (measured 9 August

30 2011), indicating an overall RMS error of 22 m betweenLandsat 7 satellite-derived and GPS measurements(Mernild et al., 2012).

2.3. GIC data-set

Landsat satellite-derived planimetric area changes are35 presented for 317 land-terminating GIC in the Arctic and

sub-Arctic between 52.4°N and 89.3°N latitude, from themid-1980s to late 2000s/2011. Observed GIC weredivided into 12 geographic regions, where the regions

varied in size from 2.7 to 22.4 × 103 km2 (Figure 1).40The number of GIC in each region is shown in Table 3.

The size of the sampled GIC data-set equates to <1%(3445 km2 in mid-1980s and 3019 km2 in late 2000s/2011) of the estimated total area of GIC in the Arcticand sub-Arctic (estimated to be ~440,000 km2 by Radić

45et al. (2014)). The GIC for each of the individual 12regions were selected: (1) to cover the GIC size rangewithin the same 12 regions and (2) so the number/size(frequency) distribution for the sampled GIC correlatedsignificantly with that of the RGI v. 2.0 data-set within

50the same specific 12 regions (Arendt et al., 2012; forGIC in the range from 0.028 to 339.7 km2), except forsouthern Ellesmere Island (Figure 4). Herein, the term“significantly” is only used for relationships that are sta-tistically significant at the 5% level or better, based on a

55linear regression t-test. These two requirements areneeded to make sure that our relatively ‘small’ GIC sam-ple is representative in frequency and in size as a sub-sample compared to the RGI GIC area sample coveringthe exactly same individual regions. For central Novaya

60Zemlya, for example, as the only GIC observations arein the eastern part of the island, the sampled GIC are notfully representative for the entire island, and may bebiased towards local trends due to climatic and oceanicdifferences between the eastern and western parts of the

65island.

2.4. Power-law scaling relations

Power-law scaling relationships (Equation (1)) have beenused when understanding and investigating relationshipsbetween different GIC metrics, such as planimetric area

70(A), length (L), elevation range (R) and surface slope (S)(Bahr 2011; Bahr et al., 1997; Grinsted, 2013; Mernild,Lipscomb, et al., 2013). Although GIC area has beenderived from remote sensing, length, elevation range andslope have only been measured on a very limited number

Table 2. Regional raw classification errors when compared with the final classification.

Glaciated regions Snow patch error, % Debris cover and shadow error, % Overall classification error, %

Talkeetna 3.0 3.2 6.2Southern British Columbia 2.9 3.1 6.0Southern Ellesmere Island 2.5 1.2 3.7Thule 2.7 1.9 2.7Nuuk 3.3 2.5 5.8Sermilik Fjord 3.4 1.8 5.2Danmarkshavn 2.7 1.9 4.6Eastern Sognefjord 3.7 1.6 5.3Narvik 3.1 1.5 4.6Central Novaya Zemlya 5.3 3.3 5.3Bolshevik Island 2.2 0.7 2.9Northern Kamchatka 3.9 2.1 6.0Average 3.0 1.9 4.9

AQ6

6 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

5 of GIC (e.g. Cogley, 2012; Leclercq et al., 2014). Here,the relationship is illustrated using the example of GICarea versus GIC length:

A ¼ cLc (1)

where c and γ are scaling parameters (c and γ, can be10 derived either empirically or theoretically given assump-

tions for idealized GIC (e.g. Bahr et al., 1997; Grinsted,2013)). Here, the scaling parameters are estimated inorder to investigate regional and temporal variabilitybetween physical characteristics of GIC length and area,

15 elevation range and area, and slope and area in order toelucidate connections to climate and topography.

3. Results and discussion

3.1. GIC initial area versus GIC area change

Analysis of the relationships between initial GIC areas20 from the mid-1980s and absolute area change over the

period are shown in Figure 5. These relationships showsignificant trends both for expanding (r2 = 0.49, where r2

is the explained variance, n = 317, n is the number of

observations, and p < 0.01, p is the level of significance)25and shrinking GIC (r2 = 0.75, p < 0.01). The largest GIC

areas exhibit the greatest absolute rates of area change,and the smallest GIC exhibit the lowest absolute rates.The reverse is true for rates of area change expressed asa percentage of the initial area of the GIC (Figure 6).

30This is confirmed by AMAP (2011) with smaller GICtending to respond more quickly to climate change.Shrinking GIC with initial areas of 0.1–1.0 km2 (follow-ing the logarithmic scale) lost 30 ± 2% of their area,while those with areas of 1.0–10.0 km2 lost 24 ± 2%. In

35the case of expanding GIC, those with an area of 0.1–1.0 km2 increased by 13 ± 4% and those with an area of1.0–10.0 km2 increased by 8 ± 2% (Figure 6).

Different theories claiming to explain why small GIChave lost the greatest percentage of their area have been

40put forward by Liston (1999), Granshaw and Fountain(2006), Demuth et al. (2008), and Tennant et al. (2012).Probable contributions to the explanation include: (1)large GIC are usually characterized by large variabilityin thickness, while small GIC are typically thinner and

45have more uniform thicknesses (Liston, 1999). Also, a

Table 3. The sampled Landsat GIC data-set were divided into the 12 glaciated regions.

Glaciatedregions

Numbersof GIC

Observationperiod

(number ofyears)

ObservedGIC area at

thebeginning ofthe period,

km2

Thepercentage

ofexpandingGIC, %

Mean GICarea changeand standarderror, % (%

yr−1)

Mean GIClength change

rate andstandard error,m (m yr−1)

Mean GICelevation

range changeand standarderror, m

Mean GICsurfaceslope

change andstandarderror

Talkeetna 26 1987–2011(24)

59.2 0 −40 ± 4(−1.7 ± 0.2)

−321 ± 77(−13 ± 3)

76 ± 13 −0.009± 0.009

SouthernBritishColumbia

31 1985–2011(26)

30.3 13 −12 ± 3(−0.4 ± 0.1)

−36 ± 13 (−1± 0.5)

29 ± 11 −0.013± 0.006

SouthernEllesmereIsland

30 1988–2009(21)

558.5 0 −35 ± 4(−1.7 ± 0.2)

−481 ± 85(−23 ± 4)

32 ± 5 0.004± 0.008

Thule 22 1987–2006(19)

666.9 0 −19 ± 3(−1.0 ± 0.2)

−224 ± 67(−12 ± 4)

10 ± 5 0.001± 0.003

Nuuk 21 1987–2003(16)

312.9 24 −4 ± 3(−0.2 ± 0.2)

−89 ± 22 (−6± 1)

19 ± 6 −0.001± 0.002

Sermilik Fjord a 33 1986–2011(25)

118.7 9 −28 ± 3(−1.1 ± 0.1)

−135 ± 23(−5 ± 1)

35 ± 10 0.008± 0.014

Danmarkshavn 27 1985–2009(24)

452.9 0 −21 ± 3(−0.9 ± 0.1)

−179 ± 36(−7 ± 2)

24 ± 6 −0.003± 0.003

EasternSognefjord

31 1988–2011(23)

135.4 0 −22 ± 2(−1.0 ± 0.1)

−121 ± 22(−5 ± 1)

38 ± 6 −0.011± 0.006

Narvik 29 1987–2006(19)

46.4 0 −19 ± 2(−1.0 ± 0.1)

−43 ± 7 (−2± 0.4)

14 ± 3 −0.004± 0.002

CentralNovayaZemlya

21 1987–2011(24)

706.1 0 −16 ± 2(−0.7 ± 0.1)

−374 ± 67(−16 ± 3)

32 ± 6 0.002± 0.002

BolshevikIsland

25 1985–2011(26)

305.8 44 −9 ± 5(−0.4 ± 0.2)

−275 ± 102(−11 ± 4)

14 ± 2 −0.002± 0.002

NorthernKamchatka

21 1999–2011(12)

52.0 10 −23 ± 3(−1.9 ± 0.3)

−104 ± 39(−9 ± 3)

8 ± 5 0.015± 0.006

aSoutheast Greenland GIC area data are updated from Mernild et al. (2012).

Geografisk Tidsskrift-Danish Journal of Geography 7

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

higher area-to-volume ratio indicates that for the sameablation rate, small GIC should shrink faster (Granshaw& Fountain, 2006); (2) a higher perimeter-to-area ratiowhich makes the small GIC more affected by long-wave

5radiation and convection of heat from the surroundingareas (Tennant et al. (2012) and Demuth et al. (2008));(3) small GIC were likely to have been formed in loca-tions that have been climatically marginal for continued

Monofor

print

colour o

nline

Figure 4. (a) An example of the location of satellite sampled GIC (large red diamonds) and GIC from the RGI (Arendt et al.,2012)) (small blue diamonds) for the Talkeetna region, Alaska (region 1) and the Talkeetna GIC area distribution for both the 2007sampled data-set (red curve) and RGI estimated data-set (blue curve) (the RGI GIC are limited by the highest and lowest sampledGIC latitudes and longitudes), and (b) sampled GIC (from the late 2000s/2011 sampled data-set) and RGI GIC area distributions fromthe remaining 11 regions, including the mean GIC area conditions for all 12 regions.

8 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

glaciation, while large ones were formed where climate5 conditions and topographic relief are more favourable;

therefore, small GIC might shrink more rapidly if climateconditions cause the GIC mass balance to become rela-tively more negative. However, GIC with a smaller area

than 0.1 km2 underwent less percentage area loss than10GIC with areas between 0.1 and 10.0 km2 (Figure 6).

This is probably because the smallest GIC tended to belocated in more sheltered locations with the possibilityof increased insulation (Debeer & Sharp, 2007; Demuth

Monofor

print

colour o

nline

Figure 5. Relationship for expanding GIC (0.11–82.9 km2) (blue squares) and shrinking GIC (0.03–339.7 km2) (red diamonds), andtheir initial area versus area change. The initial GIC area is based on data from mid-1980s, and the area rate is the annual area differ-ence between mid-1980s and late 2000s/2011. The inset histogram illustrates the GIC area distribution for the mid-1980s sampleddata-set, the late 2000s/2011 sampled data-sets, and for the RGI data-set between 52.4°N and 89.3°N latitude (Arendt et al., 2012).

Monofor

print

colour o

nline

Figure 6. Relationship between initial GIC area versus the percentage of area change for both expanding GIC (blue squares) andshrinking GIC (red diamonds). The initial GIC area is based on data from mid-1980s, and the area rate is the percentage annual areadifference between mid-1980s and late 2000s/2011. The bold lines (black, red and blue) illustrate the mean percentage of area changefor the logarithmic intervals: 0.01–0.1, 0.1–1.0, etc.

Geografisk Tidsskrift-Danish Journal of Geography 9

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

et al., 2008), since under some circumstances5 topographic shading becomes a major influence on mass

balance conditions.Despite these observations, a full understanding of

the link between a glacier’s initial area and area change,and why the shrinkage rates depends on initial areas has

10 not yet been established (Cogley, 2012), even though, asstated by AMAP (2011), changes in GIC area (inresponse to climate change) are influenced by the initialsize, shape, hypsometry and thermal conditions of theGIC. Links between area vs. length, area vs. elevation

15 range and area vs. surface slope have been estimated anddiscussed in Section 3.4.

3.2. Spatial distribution of area change

Throughout the ~25 years of satellite coverage, the Arc-tic and sub-Arctic GIC have undergone widespread, but

20 non-uniform rates of shrinkage. Only 8%, 24 out of the317 observed GIC, increased in area (Figure 7). On aregional scale, 7 out of the 12 regions studied showedubiquitous GIC reduction, whereas GIC expansion wasobserved for 6% of the Sermilik Fjord area (Mernild

25 et al., 2012), 10% of the northern Kamchatka area, 13%of the southern British Columbia area, 24% of the Nuukarea and 44% of the Bolshevik Island area (Table 3).

The GIC area changes on Bolshevik Island area arenot analysed sufficiently in the literature to identify the

30 potential presence of surge-type glaciers. It is thereforedifficult to conclude whether the GIC area expansionobserved on Bolshevik Island is a response to positivenet mass balances (from changes in climatic conditions,like increasing winter accumulation (snowfall) or

35 decreasing summer ablation) or due to surging activities

(climate-dynamic GIC response). If the GIC advance isexplained by surge-type behaviour or changes in the icedynamics, this should be clear from changes in theglacier surface characteristics, such as the distribution

40and style of crevassing or presence of looped medialmoraines or potholes (Yde & Knudsen, 2005), but noneof these surge diagnostic characteristics were observed.

Expanding GIC has also been recognized in otherArctic and sub-Arctic areas. However, these GIC are

45located as part of well-known surge clusters and in rela-tively high-precipitation regions, e.g. the western US andCanada (Post, 1969), West Greenland (Yde & Knudsen,2007) and East Greenland (Jiskoot et al., 2003). Theexpanding GIC in our study is predominantly north-fac-

50ing (85%) and is likely undergoing dynamic responsedue to increase in snow accumulation. In Novaya Ze-mlya and northern Kamchatka regions, where surgingbehaviour has been observed (Dolgoushin & Osipova,1975; Grant et al., 2009), it cannot be excluded that

55glacier advance is related to unobserved surging events.In Figure 7, the sampled data-set illustrates that the

highest frequency (number of observations) of GIC areachange rates between −5 and −20%, with an overallchange range from −99 to 47%. Seven of the observed

60GIC disappeared completely within the observation per-iod. Systematic counts of disappearing GIC have notbeen reported but disappearing GIC have been reportedin the Jostedalsbreen region in Norway (Paul et al.,2011), on Axel Heiberg Island in the Canadian High

65Arctic (Thomson et al., 2011) and in southeast Greenlandaround the Sermilik Fjord (Mernild et al., 2012), respec-tively.

On a regional scale, GIC in the Talkeetna regionhave changed by an average of -40 ± 4% and, in

Monofor

print

colour o

nline

Figure 7. Area changes for the 12 glaciated regions where each diamond represents an individual GIC. GIC are plotted againstlatitude. Red diamonds show shrinking GIC and blue squares expanding GIC. Black diamond’s illustrates the regional GIC areachange and the bars reflect one standard error. The histogram depicts the distribution of GIC area change rates.

10 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

5 southern Ellesmere Island, by −35 ± 4%. This indicatesshrinking rates of −1.7 ± 0.2% yr−1 for both regions.The two regions with the lowest rates of change werethe Nuuk area at −4 ± 3% (−0.2 ± 0.2% yr−1) andBolshevik Island area at −9 ± 5% (−0.4 ± 0.2% yr−1)

10 (Figure 7 and Table 3). The average shrinkage variationsbetween the most extreme areas in the Arctic and sub-Arctic (between the Talkeetna area with the highestshrinkage and the Nuuk area with the lowest shrinkage)were significantly different (97.5% quartile, based on the

15 null hypothesis). This indicates heterogeneous shrinkageconditions within the Arctic and sub-Arctic during a per-iod where warming has been more pronounced at highlatitudes than elsewhere. Differences in regional climate– variation in air temperature and precipitation conditions

20 – might be one of the reasons for the variation in aver-age GIC (mass-balance and) shrinkage between regions,together with the different GIC area size distributions.Given the warming in Greenland or Arctic during thefirst decade of the new millennium (e.g. Hanna et al.,

25 2012; Mernild, Hanna, et al., 2014), the different obser-vation periods used for each region may also have influ-enced GIC change results. However, this influence isexpected to be insignificant, for differences in areachange between regions due to the time lag between

30 changes in climate and changes in GIC area.On a regional scale, the mean GIC area distribution

(for the late 2000s/2011) correlates significantly with theshrinkage rate (r2 = 0.34; p = 0.01; based on linearregression) (not shown). Mean GIC area distribution

35 explains one-third of the variability in the regional GICshrinkage rate, supporting the hypothesis that the GICsize distribution is related to the shrinkage rate. Addi-tionally, AMAP (2011) mentions that GIC hypsometryplays an important controlling role in determining GIC

40 response to climate change and that in some cases low-lying GIC may decrease in area more quickly in contrastto the behaviour of GIC at higher elevations. The sam-pled data-set confirms this assumption based on the 12regions (not illustrated), as the regional area loss rate

45 was correlated (based on a multiple linear regression;r2 = 0.92) with: (1) the mean regional GIC elevation(p < 0.05; increasing mean GIC elevation equals decreas-ing loss rate); (2) the regional GIC length (p < 0.01;increasing mean GIC length equals increasing loss rate);

50 and (3) the regional GIC surface slope (p < 0.01;increasing mean GIC slope equals decreasing loss rate).

Earlier published GIC area studies covering similarregions are based on a variety of observation periods(Table 4). GIC area studies from British Columbia’s

55 northern coast and the St. Elias Mountains suggested anregional change of −8%, and −11% for British Columbia(for 1985–2005; Table 4) (Bolch et al. 2010), which iscomparable to the observed change of −12 ± 3%(Table 3) estimated from the sampled data-set for

60=southern British Columbia area. For the southernEllesmere Island area, for example, the sampled data-setindicated an area change of −35 ± 4% (−1.7% yr−1)(Table 3), which are not comparable to −55% for thenearest GIC observations located ~600–800 km away

65from northern Baffin Island (Anderson et al., 2008)(equivalent to a change rate of −1.2% yr−1 for the periodof 1958–2005; Table 4).

Even though differences are evident between therates of area change reported here and by other studies

70for the GIC of Ellesmere Island and northern BaffinIsland, both regions are characterized by relatively higharea change rates compared to other measured regions inthe circumpolar Arctic and sub-Arctic. Furthermore, forthe Bolshevik Island and central Novaya Zemlya regions,

75for example, the observed change equated to −9 ± 5%and −16 ± 2%, respectively. These observations are sup-ported by Carr et al. (2014), who emphasize that sub-stantial ice loss has occurred in the Russian High Arcticduring the past decade, predominantly on Novaya

80Zemlya.On the other hand, the sampled data-set from Novaya

Zemlya is higher than the values of up to ~−2% esti-mated by Glazovsky and Macheret (2006) for NovayaZemlya, from 1952 to 2001. This discrepancy in GIC

85area reduction for Novaya Zemlya could be caused by anumber of different issues, such as: (i) the difference inthe length of the observed time periods; (ii) regionalvariations in climate conditions (regional variability insummer temperature and winter snowfall) (Carr et al.,

902012; Moholdt et al., 2012); and (iii) differences in thesampled GIC area between the surveys. Across much ofthe Arctic, low winter precipitation means that most ofthe year-to-year variability in glacier mass balance arisesfrom changes in summer temperature. In more maritime

95regions such as southern Alaska, Iceland, western Scan-dinavia and Svalbard, variability in winter precipitationcan be an important influence on mass balance variabil-ity (AMAP, 2011).

Overall, the analysed GIC regions display a shrink-100age trend. Although there is variability within the

regions, the results shown here are in agreement with thefew observed GIC mass-balance time series in Arcticand sub-Arctic from mid-1980s to late 2000s/2011(Dyurgerov, 2010; Mernild et al., 2011), with the excep-

105tion of GIC observations from Scandinavia, where GICin the 1980s and 1990s showed positive mass-balance(AMAP, 2011; Vaughan et al., 2013), and with the globalGIC mass balance trend towards negative balances(AMAP, 2011; Cogley, 2012; Kaser et al., 2006). The

110global mean mass balance from the most recent pentad(2006–2010) has been less strongly negative, although itstill displays significant losses (Cogley, 2009, 2012).However, due to the delayed response of GIC areachanges to climate change(AMAP, 2011), we might

AQ7

Geografisk Tidsskrift-Danish Journal of Geography 11

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

Table 4. Selected GIC area and length change studies for the Arctic and sub-Arctic area.

Glaciated area PeriodGIC mean areachange, %

Recession rate,m yr−1 Source

Western Canada/United StatesYukon Territory 1958/1960 to

2006/2008−22 Barrand and Sharp

(2012)British Columbia at the northern coast and in the

St. Elias Mountains1985 to 2005 −8 Bolch et al. (2010)

Alberta −25British Columbia −11

Arctic CanadaBylot and Queen Elizabeth Island 1960 to 2000/

2001−2 to −5 Sharp et al. (2003)

Dowdeswell et al.(2007)

Interior northern Baffin Island 1958 to 2005 −55 Anderson et al. (2008)Southeast Baffin Island 1920 to 2000 −13 Paul and Svoboda

(2009)

GreenlandCentral East Greenland 2002 to 2009 −10 Kargel et al. (2002)Around the Sermilik Fjord, Southeast Greenland 1986 to 2011 −27 −10 Mernild et al. (2012)South east Greenland 1933 to 1943 −24 Bjørk et al. (2012)

1943 to 1965 −81965 to 1972 −51972 to 1981/

1985−12

1981/1985 to2000

−11

2000 to 2010 −171933 to 2010 −12

West Greenland 1900 to 1999 −12 Leclercq et al. (2012)Disko Island (Qeqertarsuaq) 1953 to 2005 −8 Yde and Knudsen

(2007)

NorwayJotunheimen and Breheimen Regions, Southern

Norway1930 to 2003 −23 Andreassen et al.

(2008)1965 to 2003 −12The 10 largest glaciers in Norway 1900s to 1999/

2006−13 Andreassen and

Winsvold (2012)Southern and northern Norway 2000 to 2010 −17 Andreassen et al.

(2011)2005 to 2010 −19

Russian ArcticFranz Josef land, Severnaya, and Novaya

Zemlya1952 to 2001 −3 Glazovsky and

Macheret (2006)

North AsiaNorthern Polar Urals 1953/1960 to

2000−22 Shahgedanova et al.

(2012)SuntarKhayata Region 1945 to 2002/

2003−19 Ananicheva et al.

(2006)Chersky Mountain Range 1970 to 2002/

2003−28 Ananicheva et al.

(2006)Koryak Upland (near Kamchatka Peninsula) 1950s to 2003 −67 Ananicheva and

Kapustin (2010)

AQ35

AQ36

AQ37

12 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

Figure 8. Distribution of GIC within the 12 regions in late 2000s/2011: (a) aspect; and (b) mean elevation (where the blackdiamond’s illustrates the regional mean and the bars reflect mean maximum and mean minimum elevation. The distance between themean maximum and mean minimum elevations are expressed as the elevation range), length and surface slope (where the bars reflectone standard error).

Geografisk Tidsskrift-Danish Journal of Geography 13

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

5 expect continuous GIC area and mass loss over the com-ing decades (Marzeion et al., 2012; Mernild, Lipscomb,et al., 2013; Radić et al., 2014).

3.3. Spatial distribution of length, elevation, and slopechanges

10 In Figure 8(a) and (b), the GIC mid-range elevation(maximum plus minimum elevation divided by two),horizontal length (the mean distance from the highestelevation to the margin for each glacier/ice capbranch), surface slope (the mean glacier branch eleva-

15 tion difference divided by the horizontal length) andaspect are illustrated for the sampled GIC for all 12regions for late 2000s/2011, illustrating the variabilitybetween regions.

Overall, the sampled GIC are predominantly facing20 north (29%) and east (19%). On a regional scale, the

mean elevation varied between 340 ± 80 m a.s.l. (Bol-shevik Island) and 1770 ± 180 m a.s.l. (eastern Sognefj-ord). The average lengths of GIC in the different regionsvaried from 1035 ± 120 m (southern British Columbia)

25 to 5220 ± 570 m (central part of central Novaya Ze-mlya), and the mean surface slope varied from 0.07 (thesouthern part of Bolshevik Island) to 0.34 (southern Brit-ish Columbia) (Figure 8(b)).

Regarding the regional GIC horizontal length reces-30 sion rates and the elevation range changes over time, the

sampled data-set (Table 3) shows variability in the lengthrange of −1 ± 0.5 m yr−1 (southern British Columbia) to−23 ± 4 m yr−1 (southern Ellesmere Island) and in theelevation range of 8 ± 5 m (northern Kamchatka) to 76

35 ± 13 m (Talkeetna) (Table 3) across the circumpolar Arc-tic and sub-Arctic. The mean length change rate of −1± 0.5 m yr−1 for southern British Columbia is expectedto be given so that the sampled GIC in this area exhibitthe steepest mean slope of around 0.35 ± 0.02 (Fig-

40 ure 8(b)) in contrast to the southern part of BolshevikIsland which had the lowest mean slope of 0.07 ± 0.01.For regions with relatively low mean slopes of less than0.10, e.g. southern Ellesmere Island, southern BolshevikIsland, and central Novaya Zemlya (Figure 8(b)), the

45 mean length change rates are relatively high (in therange of −11 ± 4 to −23 ± 4 m yr−1 (Table 3)). At thesame time, the mean GIC length in the steep southernBritish Columbia are relatively short at ~1 km comparedto the mean length of GIC in southern Ellesmere Island,

50 southern Bolshevik Island and central Novaya Zemlya ofmore than 3.2 km (Figure 8(b)), indicating that the regio-nal variability in topographic conditions might influencethe retreat rates. A comparison between the sampleddata-set and other available studies for the same period,

55 for example, from Greenland (Table 4), reveals that themean-sampled GIC length recession rates of 5–12 m yr−1

(Table 3), are in approximately the same range as, but

Monofor

print

colour o

nline

Figure 9. (a) GIC area vs. length; (b) area vs. elevation range;and (c) area vs. surface slope for the mid-1980s and the late2000s/2011.

14 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

generally slightly less than, the earlier published valuesof 11–17 m yr−1 (Bjørk et al., 2012; Table 4).

5 Regarding the regional GIC surface slope changesover time (Table 3), changes in slope were relativelysmall and random. Overall, for the sampled data-set, fourout of five GIC regions that had increasing mean slopeshad average elevations of beneath 670 m a.s.l. (Fig-

10 ure 8(b)).

3.4. Observed scaling relationships

For each of the three relations: area (A) vs. length (L),area vs. elevation range (R), and area vs. surface slope(S) power-law scaling relationships are estimated for

15 both the mid-1980s and late 2000/2011 for all sampledGIC. From this, it is possible to quantify the power-lawscaling parameters c and exponent γ for each of theregions to highlight the difference through time(Figure 9(a)–(c) and Table 5). Despite the occurrence of

20 local deviations, fixed power-law scaling relationshipsare often applied to investigate global or regional GICconditions to highlight the spatiotemporal GIC variabil-ity. It must be emphasized that the scaling parametersobtained serve to illustrate that there are large mean

25 regional deviations. The global values should thereforenot be used for extrapolation in space and time on aregional scale.

Since changes in GIC area are dependent on theinitial GIC area (Figures 5 and 6) in addition to the

30 shape and hypsometry (AMAP, 2011), the relationshipsbetween A/L, A/R and A/S change with time (Table 5).This temporal variability is shown here for all threescaling relationships, through the scaling parameters cand γ (Equation (1)), and differs between each region

35 (Table S1, Supplementary Material).On average, the A/L power-law scaling parameter c decreased in timefrom 0.62 to 0.54, whereas γ increased from 1.86 to1.92 (Table 5). For ice caps, the same trends occurredin time and between the regions, with c decreasing

40 from 2.30 to 1.59 and γ increasing from 1.40 to 1.58.Also, for the A/R and the A/S relationships for GIC,the overall mean temporal and spatial variability in cand γ were identical to the overall trends in c and γfor the A/L relationships. However, on the regional

45 scale, Table S1 shows a non-universal variability in c

and γ over time. Such variability in observed GIC met-ric changes and subsequently in the scaling parameters(influenced by local climatic and glaciological condi-tions, and topography) of the 12 regions around the

50circumpolar Arctic and sub-Arctic derivates from globalsteady-state parameters and is worth considering whenusing idealized GIC power-law scaling relationships forestimating GIC conditions in time and space. Such ide-alized relationships are often used in global assessments

55of e.g. GIC area and subsequent GIC volume condi-tions and changes, and GIC contributions to sea-levelrise (e.g. Grinsted, 2013). Importantly, these globalassessments could thus be subject to uncertainties,given the spatial and temporal variability in scaling

60parameters that has been observed.

4. Conclusions

Here, satellite-derived GIC reduction and advancementmeasurements are presented for 317 GIC, divided into12 geographic regions spread over Arctic and sub-Arctic,

65which simultaneously provide insight into temporal andspatial variability in area, length, elevation range andsurface slope for the mid-1980s to late 2000s/2011.Landsat TM/ETM+ scenes were used as input in a semi-automated model, producing an average classification

70error of 4.9%. Over the observation period, the majority(92%) of the GIC lost area, with area changes varyingbetween −4 ± 3% (−0.2 ± 0.2% yr−1) for Nuuk and −40± 4% (−1.7 ± 0.2% yr−1) for Talkeetna since themid-1980s. Regarding observed power-law scaling

75relationships between GIC area and length, area and ele-vation range, and area and surface slope, a non-universalvariability for the observation period and between the 12regions was observed in the scaling parameters. Thissuggests that regional scaling relationships may be used

80to improve global scaling relationships and assess theirvariability.

Supplementary material

The supplementary material contain: (i) regional GICrelationships between area, length, elevation range and

85surface slope (Table S1) for both mid-1980s and late2000s/2011 and (ii) specific classification notes.

Table 5. Average GIC scaling parameters for the entire data-set for mid-1980s (illustrated at the top for each grid box) and late2000s/2011 (bottom): area (A, km2) versus length (L, km), area versus elevation range (R, km) and area versus surface slope (S).

Glaciated area Glacier (G) and ice cap (IC) Area vs. length Area vs. elevation range Area vs. surface slope

Entire data-set G A = 0.62L1.86, r2 = 0.80 A = 8.63R1.36, r2 = 0.34 A = 0.24S−1.17, r2 = 0.22A = 0.54L1.92, r2 = 0.78 A = 7.21R1.39, r2 = 0.34 A = 0.21S−1.05, r2 = 0.17

IC A = 2.30L1.40, r2 = 0.58 A = 80.44R0.88, r2 = 0.27 A = 4.21S−0.38, r2 = 0.04A = 1.59L1.58, r2 = 0.60 A = 72.77R0.91, r2 = 0.30 A = 5.06S−0.21, r2 = 0.01

Note: Significant trends (p < 0.05) are highlighted in bold.

AQ8

Geografisk Tidsskrift-Danish Journal of Geography 15

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

AcknowledgementsWe extend a very special thanks to the anonymous reviewersfor their insightful critique of this article. This work was

5 supported partly by the Earth System Modelling program withinthe U.S. Department of Energy’s Office of Science, by Los Ala-mos National Laboratory (LANL), and by the European Com-munity’s Seventh Framework Programme under grantagreement No. 262693. LANL is operated under the auspices of

10 the National Nuclear Security Administration of the U.S.Department of Energy under Contract No. DE-AC52-06NA25396). All satellite data were acquired through the USGSEarth Explorer internet portal (http://earthexplorer.usgs.gov/).

Disclosure statement15 No potential conflict of interest was reported by the authors.

Supplemental dataSupplemental data for this article can be accessed here [http://dx.doi.org.10.1080/00167223.2015.1026917].

References20 ACIA. (2005). Arctic climate impact assessment (p. 1024).

Cambridge University Press.AMAP. (2011). Snow, water, ice and permafrost in the Arctic

(SWIPA): Climate change and the cryosphere. In Mountainglaciers and ice caps (chapter 7, p. 61). Oslo: Arctic Moni-

25 toring and Assessment Programme (AMAP), xii + 538 pp.Ananicheva, M. D., & Kapustin, G. (2010). Change of glacier

state in mountain regions of the Russian sub-arctic: Assess-ment by satellite imagery and aerial photography in theUSSR glacier inventory [in Russian; English summary on

30 last page]. Russian contribution into IPY, 2007–2008. Cry-osphere, 3, 19–26.

Ananicheva, M. D., Kapustin, G. A., & Koreysha, M. M.(2006). Glacier changes in the Suntar-Khayate Mountainsand Chersky Range from the glacier inventory of the USSR

35 and satellite images 2001–2003. Data of GlaciologicalStudies, 101, 163–169.

Anderson, R. K., Miller, G. H., Briner, J. P., Lifton, N. A., &DeVogel, S. B. (2008). A millennial perspective on Arcticwarming from C-14 in quartz and plants emerging from

40 beneath ice caps. Geophysical Research Letters, 35,L01502.

Andreassen, L. M., Elvehøy, H., Jackson, M., Kjøllmoen, B.,& Giesen, R. H. (2011). Glaciological investigations inNorway in 2010 (p. 106). (B. Kjøllmoen, Ed.). The Norwe-

45 gian Water Resources and Energy Directorate.Andreassen, L. M., Paul, F., Kääb, A., & Hausberg, J. E.

(2008). Landsat-derived glacier inventory for Jotunheimen,Norway, and deduced glacier changes since the 1930s. TheCryosphere, 2, 131–145.

50 Andreassen, L. M., Winsvold, S. H., Paul, F., & Hausberg, J.E. (2012). Inventory of Norwegian glaciers (pp. 1–242).(L. M. Andreassen & S. H. Winsvold, Eds.). NorwegianWater Resources and Energy Directorate.

Arendt, A. A., Bolch, T., Cogley, J. G., Gardner, A., Hagen, J.-55 O., Hock, R., … Zheltyhina, N. (2012). Randolph glacier

inventory [v2.0]: A dataset of global glacier outlines. Glo-bal Land Ice Measurements from Space. Boulder, CO: Dig-ital Media. Retrieved from http://www.glims.org/RGI/randolph.html

60Arendt, A. A., Echelmeyer, K. A., Harrison, W. D., Lingle, C.S., & Valentine, V. B. (2002). Rapid wastage of Alaskaglaciers and their contribution to rising sea level. Science,297, 382–386.

Bahr, D. B. (1997). Global distributions of glacier properties:65A stochastic scaling paradigm. Water Resources Research,

33, 1669–1679.Bahr, D. B., Meier, M. F., & Peckham, S. (1997). The physical

basis of glacier volume-area scaling. Journal of Geophysi-cal Research, 102, 20355–20362.

70Bahr, D. B., & Radić, V. (2012). Significant contribution tototal mass from very small glaciers. The Cryosphere, 6,763–770.

Barrand, N. E., & Sharp, M. J. (2012). Sustained rapid shrink-age of Yukon glaciers since the 1957–1958 International

75Geophysical Year. Geophysical Research Letters, 37,L07501.

Berthier, E., Schiefer, E., Clarke, G. K. C., Menounos, B., &Rémy, F. (2010). Contribution of Alaskan glaciers to sea-level rise derived from satellite imagery. Nature Geosci-

80ence, 3, 92–95.Bjørk, A. A., Kjær, K. H., Korsgaard, N. J., Khan, S. A.,

Kjeldsen, K. K., Andresen, C. S., … Funder, S. (2012). Anareal view of 80 years of climate-related glacier fluctuationsin Southeast Greenland. Nature Geoscience, 5, 427–432.

85doi:10.1038/NGEO1481Bolch, T., Menounos, B., & Wheate, R. (2012). Landsat-based

inventory of glaciers in western Canada, 1985–2005.Remote Sensing of Environment, 114, 127–137.

Brest, C. (1987). Seasonal Albedo of an urban/rural landscape90from satellite observations. Journal of Climate and Applied

Meteorology, 26, 1169–1187.Carr, J. R., Stokes, C., & Vieli, A. (2014). Recent retreat of

major outlet glaciers on Novaya Zemlya, Russian Arctic,influenced by fjord geometry and sea-ice conditions. Jour-

95nal of Glaciology, 60, 155–170.Cogley, J. G. (2009). Geodetic and direct mass-balance mea-

surements: Comparison and joint analysis. Annals of Glaci-ology, 50, 96–100.

Cogley, J. G. (2012). The future of the world’s glaciers. In: A.100Henderson-Sellers & K. McGuffie (Eds.), The future of the

world’s climate. Elsevier.Crane, R. G., & Anderson, M. R. (1984). Satellite discrimina-

tion of snow/cloud surfaces. International Journal ofRemote Sensing, 5, 213–223.

105Debeer, C. M., & Sharp, M. J. (2007). Recent changes in gla-cier area and volume within the southern Canadian Cordil-lera. Annals of Glaciology, 46, 215–221.

Demuth, M., Pinard, V., Pietroniro, A., Luckman, B., Hopkin-son, C., Dornes, P., & Comeau, L. (2008). Recent and

110past-century variations in the glacier resources of theCanadian Rocky Mountains: Nelson River system, TerraGlacialis. Special Issue: Mountain Glaciers and ClimateChange of the Last Century (pp. 27–52).

Dolgoushin, L. D., & Osipova, G. B. (1975). Glacier surges115and the problem of their forecasting. IAHS Publications,

104, 292–304.Dowdeswell, E. K., Dowdeswell, J. A., & Cawkwell, F.

(2007). On the glaciers of Bylot Island, Nunavut, ArcticCanada. Arctic, Antarctic, and Alpine Research, 39, 402–

120411.Dozier, J., & Marks, D. (1987). Snow mapping and classifica-

tion from Landsat Thematic Mapper data. Annals of Glaci-ology, 9, 97–103.

AQ38

AQ9

AQ10

AQ11

AQ12

AQ13

AQ14

AQ15

AQ16

AQ17

AQ18

AQ19

AQ20

16 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

Dyurgerov, M. B. (2010). Data of glaciological studies –5 Reanalysis of glacier changes: From the IGY to the IPY,

1960–2008. Publication No. 108. Institute of Arctic andAlpine Research (University of Colorado).

Gao, B.-C. (1996). NDWI – A normalized difference waterindex for remote sensing of vegetation liquid water from

10 space. Remote Sensing of Environment, 58, 257–266.Glazovsky, A., & Macheret, Y. (2006). Eurasian Arctic. In: V.

M. Kotlyakov (Ed.), Glaciation in north and central Eur-asia in present time (pp. 97–114, 438–445). Nauka. [InRussian with English summary]

15 Granshaw, F. D., & Fountain, A. G. (2006). Glacier change(1958–1998) in the North Cascades national park complex,Washington, USA. Journal of Glaciology, 52, 251–256.

Grant, K. L., Stokes, C. R., & Evans, I. S. (2009). Identificationand characteristics of surge-type glaciers on Novaya Zemlya,

20 Russian Arctic. Journal of Glaciology, 55, 960–972.Grinsted, A. (2013). An estimate of global glacier volume. The

Cryosphere, 7, 141–151.Hall, D. K., Bayr, K. J., Schöner, W., Bindschadler, R. A., &

Chien, J. Y. L. (2003). Consideration of the errors inherent25 in mapping historical glacier positions in Austria from the

ground and space (1893–2001). Remote Sensing of Envi-ronment, 86, 566–577.

Hall, D. K., Foster, J. L., Chien, J. Y. L., & Riggs, G. A.(1995). Determination of actual snow-covered area using

30 Landsat TM and digital elevation model data in GlacierNational Park, Montana. Polar Record, 31, 191–198.

Hall, D. K., Riggs, G. A., Salomonson, V. V., DiGirolamo, N.E., & Bayr, K. J. (2001). MODIS snow-cover products.Remote Sensing of Environment, 83, 181–194.

35 Hanna, E., Mernild, S. H., Cappelen, J., & Steffen, K. (2012).Recent warming in Greenland in a long-term instrumental(1881–2012) climatic context: I. Evaluation of surface airtemperature records. Environmental Research Letters, 7,045404. doi:10.1088/1748-9326/7/4/045404

40 Hansen, J., Ruedy, R., Sato, M., & Lo, K. (2012). Global sur-face temperature change. Reviews of Geophysics, 48,RG4004. doi:10.1029/2010RG000345

Huss, M., & Farinotti, D. (2012). Distributed ice thickness andvolume of all glaciers around the globe. Journal of Geo-

45 physical Research, 117, F04010. doi:10.1029/2012JF002523

Jacob, T., Wahr, J., Pfeffer, W. T., & Swenson, S. (2012).Recent contributions of glaciers and ice caps to sea levelrise. Nature, 482, 514–518.

50 Jiskoot, H., Murray, T., & Luckman, A. (2003). Surge potentialand drainage-basin characteristics in East Greenland.Annals of Glaciology, 36, 142–148.

Kargel, J. S., Ahlstrøm, A. P., Alley, R. B., Bamber, J. L., Ben-ham, T. J., Box, J. E., … Willis, I. (2002). Brief communi-

55 cation Greenland’s shrinking ice cover: “Fast times” butnot that fast. The Cryosphere, 6, 533–537.

Kaser, G., Cogley, J. C., Dyurgerov, M. B., Meier, M. F., &Ohmura, A. (2006). Mass balance of glaciers and ice caps:Consensus estimates for 1961–2004. Geophysical Research

60 Letters, 33, 1–5.Kotlyakov, V. M., Dyakova, A. M., Koryakin, V. S., Kravtsova,

V. I., Osipova, G. B., Varnakova, G. M., … Zverkova, N.M. (2010). Glaciers of the former Soviet Union. In: R. S.Williams Jr. & J. G. Ferrigno (Eds.), Glaciers of Asia (pp.

65 F1–F58). U.S. Geol. Surv. Prof. Pap., 1386-F.Leclercq, P. W., Oerlemans, J., Basagic, H. J., Bushueva, I.,

Cook, A. J., & Le Bris, R. (2014). A data set of worldwideglacier length fluctuations. The Cryosphere, 8, 659–672.

Leclercq, P. W., Weidick, A., Paul, F., Bolch, T., Citterio, M.,70& Oerlemans, J. (2012). Brief communication “Historical

glacier length changes in West Greenland”. The Cryo-sphere, 6, 1339–1343.

Liston, G. E. (1999). Interrelationships among snow distribu-tion, snowmelt, and snow cover depletion: Implications for

75atmospheric, hydrologic, and ecologic modeling. Journal ofApplied Meteorology, 38, 1474–1487.

Marzeion, B., Jarosch, A. H., & Hofer, M. (2012). Past andfuture sea-level change from the surface mass balance ofglaciers. The Cryosphere, 6, 1295–1322.

80Mernild, S. H., Hanna, E., Yde, J. C., Cappelen, J., &Malmros, J. K. (2014). Coastal Greenland air temperatureextremes and trends 1890–2010: Annual and monthly anal-ysis. International Journal of Climatology, 34, 1472–1487.doi:10.1002/joc.3777

85Mernild, S. H., Lipscomb, W. H., Bahr, D. B., Radić, V., &Zemp, M. (2013a). Global glacier changes: A revisedassessment of committed mass losses and sampling uncer-tainties. The Cryosphere, 7, 1565–1577.

Mernild, S. H., Liston, G. E., & Hiemstra, C. A. (2014). North-90ern hemisphere glacier and ice cap surface mass balance

and contribution to sea level rise. Journal of Climate, 27,6051–6073. doi:10.1175/JCLI-D-13-00669.1

Mernild, S. H., Knudsen, N. T., Hoffman, M. J., Yde, J. C.,Hanna, E., Lipscomb, W. H., … Fausto, R. S. (2013b).

95Volume and velocity changes at Mittivakkat Gletscher,southeast Greenland. Journal of Glaciology, 59, 660–670.

Mernild, S. H., Knudsen, N. T., Lipscomb, W. H., Yde, J. C.,Malmros, J. K., Hasholt, B. H., & Jakobsen, B. (2011).Increasing mass loss from Greenland’s Mittivakkat Glet-

100scher. The Cryosphere, 5, 341–348.Mernild, S. H., Malmros, J. K., Yde, J. C., & Knudsen, N. T.

(2012). Multi-decadal marine- and land-terminating glacierrecession in the Ammassalik region, southeast Greenland.The Cryosphere, 6, 625–639.

105Moholdt, G., Wouters, B., & Gardner, A. S. (2012). Recentmass changes of glaciers in the Russian High Arctic. Geo-physical Research Letters, 39, doi:10.1029/2012GL051466

Paul, F. (2004). The new Swiss glacier inventory 2000 – Appli-cation of remote sensing and GIS (p. 198) (Ph.D. thesis).

110Department of Geography, University of Zurich.Paul, F., Andreassen, L. M., & Winsvold, S. H. (2011). A new

glacier inventory for the Jostedalsbreen region, Norway,from Landsat TM scenes of 2006 and changes since 1966.Annals of Glaciology, 52, 153–162.

115Paul, F., Barrand, N. E., Baumann, S., Berthier, E., Bolch, T.,Casey, K., … Winsvold, S. (2013). On the accuracy of gla-cier outlines derived from remote-sensing data. Annals ofGlaciology, 54(63).

Paul, F., & Svoboda, F. (2009). A new glacier inventory on120southern Baffin Island, Canada, from ASTER data: II. Data

analysis, glacier change, and application. Annals of Glaciol-ogy, 50, 22–31.

Pfeffer, W. T., Arendt, A. A., Bliss, A., Bolch, T., Cogley, J.G., Gardner, A. S., … The Randolph Consortium (2014).

125The Randolph Glacier Inventory: A globally completeinventory of glaciers. Journal of Glaciology, 60, 537–552.

Post, A. (1969). Distribution of surging glaciers in westernNorth America. Journal of Glaciology, 8, 229–240.

Radić, V., Bliss, A., Breedlow, A. C., Hock, R., Miles, E., &130Cogley, J. G. (2014). Regional and global projections of

twenty-first century glacier mass changes in response to cli-mate scenarios from global climate models. ClimateDynamics,. doi:10.1007/s00382-013-1719-7

AQ21

AQ22

AQ23

AQ24

AQ25

AQ26

AQ27

AQ28

Geografisk Tidsskrift-Danish Journal of Geography 17

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial

Rayner, N. A., Parker, D. E., Horton, E. B., Folland, C. K.,5 Alexander, L. V., Rowell, D. P., … Kaplan, A. (2003).

Global analyses of sea surface temperature, sea ice, andnight marine air temperature since the late nineteenthcentury. Journal of Geophysical Research, 108, 4407.

Rouse, J. W., Haas, R. H., Schell, J. A., & Deering, D. W.10 (1973). Monitoring vegetation systems in the Great Plains

with ERTS. Third ERTS Symposium, NASA SP-351 (Vol.I, pp. 309–317.

Shahgedanova, M., Nosenko, G., Bushueva, I., & Ivanov, M.(2012). Changes in area and geodetic mass balance of

15 small glaciers, Polar Urals, Russia, 1950–2008. Journal ofGlaciology, 58, 953–964.

Shahgedanova, M., Nosenko, G., Khromova, T., & Muraveyev,A. (2010). Glacier shrinkage and climatic change in theRussian Altai from the mid-20th century: An assessment

20 using remote sensing and PRECIS regional climate model.Journal of Geophysical Research, 115, D16107.doi:10.1029/2009JD012976

Sharp, M. J., Copland, L., Filbert, K., Burgess, D., & William-son, D. (2003). Recent changes in the extent and volume

25 of Canadian Arctic glaciers. In Snow watch 2002 workshopand assessing global glacier recession. Glaciological Datareport (pp. 73–75). Boulder, CO: NSIDE/WDC forGlaciology.

Tachikawa, T., Kaku, M., Iwasaki, A., Gesch, D., Oimoen, M.,30 Zhang, Z., … Carabajal, C. (2011). ASTER global digital

elevation model version 2 – Summary of validation results,NASA land processes. Distributed Active Archive Centerand the Joint Japan–US ASTER Science Team. Retrievedfrom https://igskmncnwb001.cr.usgs.gov/aster/GDEM/Sum

35 maryGDEM2validationreportfinal.pdf

Tennant, C., Menounos, B., & Wheate, R. (2012). Area changeof glaciers in the Canadian Rocky Mountains, 1919 to2006. The Cryosphere, 6, 1541–1552.

Thomson, L. I., Osinski, G. R., & L. Ommanney, C. S. L.40(2011). Glacier change on Axel Heiberg Island, Nunavut,

Canada. Journal of Glaciology, 57, 1079–1086.Wolken, G. J., England, J. H., & Dyke, A. S. (2008). Changes

in late-Neoglacial perennial snow/ice extent and equilib-rium-line altitudes in the Queen Elizabeth Islands, Arctic

45Canada. The Holocene, 18, 615–627.World Glacier Monitoring Service. (2011). Glacier mass bal-

ance bulletin No. 11 (2008–2009). (M. Zemp, S. U. Nuss-baumer, I. Gärtner-Roer, M. Hoelzle, F. Paul, & W.Haeberli, Eds.). ICSU(WDS)/IUGG(IACS)/UNEP/UNE-

50SCO/WMO.Vaughan, D. G., Comiso, J. C., Allison, I., Carrasco, J., Kaser,

G., Kwok, R., … Zhang, T. (2013). Observations: Cryo-sphere. In: T. F. Stocker, D. Qin, G.-K. Plattner, M. Tignor,S. K. Allen, J. Boschung, … P. M. Midgley (Eds.), Climate

55change 2013: The physical science basis. contribution ofworking group I to the fifth assessment report of theintergovernmental panel on climate change. Cambridge:Cambridge University Press.

Yde, J. C., & Knudsen, N. T. (2005). Glaciological features in60the initial quiescent phase of Kuannersuit Glacier, Green-

land. Geografiska Annaler, Series A: Physical Geography,87, 473–485.

Yde, J. C., & Knudsen, N. T. (2007). 20th-century glacier fluc-tuations on Disko Island (Qeqertarsuaq), Greenland. Annals

65of Glaciology, 46, 209–214.

AQ29

AQ30

AQ31

AQ32

AQ33

AQ34

18 S.H. Mernild et al.

RDGS 1026917 CE: SL QA: SP13 March 2015 Initial