A complete record of Holocene glacier variability at Austre Okstindbreen, northern Norway: an integrated approach

lable at ScienceDirect

Quaternary Science Reviews 29 (2010) 1246e1262

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

A complete record of Holocene glacier variability at Austre Okstindbreen,northern Norway: an integrated approach

Jostein Bakke a,b,*, Svein Olaf Dahl a,b, Øyvind Paasche b, Joachim Riis Simonsen a, Bjørn Kvisvik a,b,Kristina Bakke c,d, Atle Nesje b,d

aDepartment of Geography, University of Bergen, Fosswinckelsgate 6, N-5020 Bergen, NorwaybBjerknes Centre for Climate Research, Allégaten 55, N-5020 Bergen, Norwayc Statoil Research Centre, Sandslihaugen 30, N-5254 Sandsli, NorwaydDepartment of Earth Science, University of Bergen, Allégaten 41, N-5007 Bergen, Norway

a r t i c l e i n f o

Article history:Received 22 June 2009Received in revised form5 February 2010Accepted 7 February 2010

* Corresponding author. Department of Geographywinckelsgate 6, N-5020 Bergen, Norway. Tel.: þ47 5530.

E-mail address: [email protected] (J. Bakk

0277-3791/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.quascirev.2010.02.012

a b s t r a c t

Arctic glaciers are currently undergoing major changes, but accurate knowledge about how they variedduring the entire Holocene is still scarce. Here we present a new complete glacier record from AustreOkstindbreen in Nordland, northern Norway. This reconstruction is based on a number of short and longcores retrieved from several downstream basins, which have been analyzed by a suite of methodsincluding geochemical elements (XRF), rock magnetic properties, dry bulk density (DBD) and Loss-on-ignition (LOI). Lake sediment distribution was surveyed and mapped prior to coring by the use of groundpenetrating radar (GPR). Independently lichen-dated marginal moraines and historical informationabout the glacier frontal positions from old photographs and maps have made it possible to link themoraine sequence to the 210Pb dated lake sediment chronology. This integrated approach reveals thatAustre Okstindbreen is the first known glacier in Scandinavia to possibly have survived the “HoloceneThermal Optimum”. It also brackets the four largest glacier advances to c 7400e7000, 1400e1200,900e700 and 300e150 years before AD 2000 (b2k). In contrast to most reconstructed glaciers inScandinavia, the largest glacier advance was not associated with the “Little Ice Age”, but rather to anearlier period centred at 1300 b2k. Both the moraine chronology and the lacustrine records documentthis Neoglacial advance. Compared to other glacier reconstructions from the Northern Hemisphere, weidentify near-synchronous glacier advances occurring roughly at 4000 b2k; 2700 b2k; 1300 b2k andduring the “Little Ice Age”. These correlative advances across the Northern Hemisphere suggest thatthese observed centennial-scale events are a shared feature regardless of the large geographicaldistances separating them. Some minor discrepancies between different geographical areas may becaused by lack of precise dating, but local climatic conditions may play a role as well.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Alpine glaciers are commonly located in remote and high-altitude regions of the world, areas that only rarely are covered byinstrumental records. Reconstructions of glaciers have thereforeproven useful for understanding past climate dynamics on bothshorter and longer time-scales. Robust glacier reconstructions canthus be an important source of knowledge for a better under-standing of natural climate variability. Obtaining such records is

, University of Bergen, Foss-58 98 32; fax: þ47 55 58 43

e).

All rights reserved.

not straightforward and is frequently based on a blend of differentmethods, which occasionally is difficult to reconcile. One majordrawback with glacier reconstructions based solely on morainechronologies e by far the most common e is that due to selectivepreservation of moraine ridges such records do not exclude thepossibility of multiple Holocene glacier advances (e.g. Balco, 2009).To overcome this problem Karlén (1976) initially suggested thatglacial erosion and the associated production of rock flour depos-ited in downstream lakes could provide a continuous record ofglacial fluctuations, hence overcoming the problem of incompletereconstructions. The simplest and most commonly applied methodto infer glacier fluctuations examines loss-on-ignition (LOI), whichfrequently have been used as an inverse indicator of inorganicsedimentation in lake sediments (Karlén, 1976). This proxy hasbeen questioned since it simply mirrors the total amount of

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1247

minerogenic material including biogenic silica, that is depositedwithin a lake, and can not distinguish between different processesaffecting the lake sedimentation (Snowball and Sandgren, 1996;Bakke et al., 2005c). More recent studies build on multi-proxyapproaches, taking into account different physical parameters inorder to better understand the inorganic sedimentation directlyrelated to the glacier(s) in the catchment (Lie et al., 2004; Bakkeet al., 2005b,c; Osborn et al., 2007).

In Scandinavia, several studies have reconstructed glacier fluc-tuations from proglacial lake sediments (Karlén, 1976; Nesje et al.,1991, 1995; Matthews and Karlen, 1992; Dahl and Nesje, 1996;Snowball and Sandgren, 1996; Lie et al., 2004; Rosqvist et al.,2004; Bakke et al., 2005b,c). Many of these studies recognizedimportant, non-glacial sediment sources that can deliver clasticsediments to lake basins: paraglacial processes, debris flows, snowavalanches, turbidites, melt-out of permafrost etc. (Rubensdotter,2002; Rubensdotter and Rosqvist, 2003, 2009; Rosqvist et al.,2004; Bakke et al., 2005c; Jansson et al., 2005; Paasche et al., 2007).

Besides from identifying the extra-glacial components presentin any glacio-lacustrine core, there is an additional way of testingwhether or not specific sedimentary intervals can be attributed toglacial activity, namely by correlating independently datedmoraines to corresponding changes in the lake record. Establishingmoraine chronologies by means of lichenometry (e.g. Innes, 1986)cosmogenic exposure dating (e.g. Schaefer et al., 2009) or radio-carbon dating (e.g. Ivy-Ochs et al., 2009) are all commonapproaches. With an accurately dated moraine sequence, theEquilibrium-Line-Altitude (ELA) of former glaciers can be recon-structed based on moraines deposited by temperate glaciers(Andrews,1975) and subsequently tied to corresponding changes inthe lacustrine records. Numerous moraines from mid- to lateHolocene can potentially allow for a more detailed ELA recon-struction, which again is likely to improve the ELA-sedimentrelationship.

Here we present a number of lacustrine sediment records fromAustre Okstindbreen in northern Norway, based on a series of longand short cores from two downstream lakes. Through an integratedapproach, combining detailed analysis of the lacustrine sedimentsin question with independently dated moraines, we seek toimprove the current understanding for Holocene glacier variability.Being able to do so is essential if a trustworthy climate-signal is tobe extracted from such records. A suit of methods has thereforebeen invoked in order to identify the glacial signal. Hence, the firstobjective of this paper is to reconstruct Holocene glacier fluctua-tions of Austre Okstindbreen. Secondly, this reconstruction is tiedto independently dated moraines in order to produce a high-resolution reconstruction of the ELA during the Holocene. Thirdly,the findings are discussed and compared with regional trends inglacier variability across the Northern Hemisphere.

2. Study site

2.1. Glaciers, climate and bedrock

Okstindan is located in an alpine-type terrain, about 60 kmsouth of the Arctic Circle and close to the Swedish border inNordland, northern Norway (Fig. 1). There are 19 individual glaciersin the Okstindan area, covering c 46 km2 (Fig. 1). The largest glacier,Okstindbreen, constitutes an ice-field including Vestisen, VestreOkstindbreen, Austre Okstindbreen, Bessedørbreen, Steik-vassbreen, Vestre Svartfjellbreen and Oksfjellbreen. The largestoutlet glacier from Okstindbreen, Austre Okstindbreen (N66�000

E14�160), drains towards the valley Oksfjelldalen NeNE of themountain Oksskolten (1916 m a.s.l.), which is the highest mountainin northern Norway. In 1996, Austre Okstindbreen was c 14 km2,

and its sediment laden meltwater runs through a chain of lakes:Grasvatnet, Austre and Vestre Kjennsvatnet, Nedre and ØvreBleiking, Store Målvatnet and Lille Målvatnet, before it enters thehead of the fjord at Bjerka in Sørfjorden (Fig. 1).

The ELA at Austre Okstindbreen is between 1330 and 1340 m a.s.l. in years when the net balance is close to zero (Andreassen,2000) (Table 1.). The present climate is semi-continental tomaritime, and an estimate based on the meteorological stationHemnes (station no. 79110), gives a mean (1961 and 1990)summer temperature/ablation season (1 Maye30 September) atsea level (30 m a.s.l.) of 10.4 �C (Meteorologisk Institutt, 2007).Using an environmental lapse rate of 0.65 �C/100 m (Sutherland,1984), this gives a mean summer temperature (Ts) at the presentELA (1340 m) of Austre Okstindbreen of c 2.2 �C. Winter precipi-tation (Pw) (1 Octobere30 April) based on the two meteorologicalstations in Hemnes (station no. 9110, altitude 10 m; station no.9111, altitude 6 m) gives a mean (1961 to 1990) of 510 mm at sealevel (Meteorologisk Institutt, 2007). By using a suggested meanexponential increase in winter precipitation with altitude of 8%/100 m in southern Norway (Haakensen, 1989; Dahl and Nesje,1992), the precipitation is calculated to be 2235 mm at the ELAof Austre Okstindbreen. Substituting a mean ablation-seasontemperature of 2.2 �C at the ELA of Austre Okstindbreen intothe “Liestøl-equation” (Ballantyne, 1989), an estimated winterprecipitation (Pw) of c. 1930 mm is obtained.

The bedrock beneath Austre Okstindbreen is of Lower Palae-ozoic age, and is associated with the Beiaren cover consisting ofmetamorphic rocks dominated by gneissic schists and schists withsome elements of calcareous biotite schists. In a narrow band e tothe north of Austre Kjennsvatnet e there is an area with gneissicgranite rich in quartz (Hoel, 1910; Fosslie, 1941).

2.2. Catchment lakes

Sediment cores were retrieved from Vestre- and AustreKjennsvatnet (Figs. 1 and 3). At present, the meltwater from AustreOkstindbreen is routed through Grasvatnet 595 m a.s.l. This isa rather large lake (c 22.6 km2) with several sub-basins (Fig. 3). Thisimplies that only sediments in suspension are transported furtherdownstream before entering Austre- and Vestre Kjennsvatnet. Theterrain surrounding lake Grasvatnet consists of several mountains(e.g. Grasfjellet and Skullen) with the highest elevation close to1300 m a.s.l., but with a gentle relief.

The connected Austre- and Vestre Kjennsvatnet (5 km2) at527 m a.s.l. is situated in a glacially eroded bedrock-basin. Themain river inlet comes through Austre Kjennsvatnet and theoutlet is to the north from Vestre Kjennsvatnet. A narrow and onemetre deep channel separates the two lakes. The bathymetry ofAustre Kjennsvatnet shows a circular-shaped-basin witha maximum depth of c 10 m. The bathymetry of lake VestreKjennsvatnet reveals that the lake consists of two large sub-basinswith a maximum depth of c 35 m separated by a 20 m deep sill.Altogether 16 small streams enter Austre- and Vestre Kjennsvat-net in addition to the main river from lake Grasvatnet. In general,most of small streams have modest influence on the sedimenta-tion environment in Austre- and Vestre Kjennsvatnet, as theyhave relatively small drainage areas (0.1e0.3 km2). However, twolarge non-glacial rivers enter Vestre Kjennsvatnet; Nordskard-bekken with a drainage area of c 7.5 km2 and Storskardbekkenwith a drainage area of c 10.5 km2. These two river catchments arelarge enough to potentially deliver clastic sediments to the lakebasin during heavy snowmelt and/or heavy precipitation. Theslopes surrounding Austre- and Vestre Kjennsvatnet are gentlewith a scattered cover of superficial sediments, mainly consistingof erratic boulders and a thin cover of till.

Fig. 1. Study area at Okstindan in Nordland, northern Norway. The eastern North Atlantic region is shown in the lower right square. The red square marks the study area atOkstindan. A 3D visualization of the summit area of Okstindan with the ice field centred on Oksskolten is shown. Austre Okstindbreen drains to the east of the ice fieldOkstindbreen. Blue arrows indicate present meltwater routes from Austre Okstindbreen. The red arrow indicates a former meltwater route, which was active whenever theelevation of the glacier surface was higher than the 815 m a.s.l. Red dots mark coring sites used in this study.

Table 1Glacier front positions at Austre Okstindbreen based on historical sources, historicalmaps and lichenometry

Front positions Year (AD) CorrespondingELA (m a.s.l.)

Reference:

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621248

3. Methods

The reconstruction of Holocene glacier fluctuations at AustreOkstindbreen is based upon a combination of geomorphologicalmapping (aerial photos; 1:30 000 by Fjellanger Widerøe 1998,series 12326) in field, lake coring and laboratory analyses. A firmchronology has been obtained by AMS radiocarbon dating and210Pb (see below). In addition, we have used historical data (oldphotos and maps) in combination with an earlier study of lichengrowth on marginal moraines to the east of Austre Okstindbreen(Winkler, 2003) in order to constrain whenever the glacier waslarger than at present.

0 2005 1340 This study�100 1998 1323 NVE�200 1981 1316 NVE�370 1962 1300 Knudsen and

Theakstone (1988)M6* 1950� 5 1291 Winkler (2003)�630 1944 1288 Worsley (1974)M5* 1936� 8 1285 Winkler (2003)�475 1908 1237 Hoel (1910)M4* 1874� 14 1233 Winkler (2003)M3* 1842� 16 1243 Winkler (2003)M2* 1757� 26 1184 Winkler (2003)M1 Older than “LIA” 1153 Winkler (2003)Melt water

overthreshold

<1190 This study

*¼ 10% error (Innes, 1986).

3.1. ELA calculations

The hypsometry of the glacier Austre Okstindbreen is recon-structed for different periods based on lateral moraines whereespecially the LIA moraines are well preserved. The ELAs (presentand past) of the individually dated glacier advances are recon-structed using different approaches, including an accumulation-area ratio (AAR) of 0.6 (Porter, 1975) and an ablation-accumulationbalance ratio (BR) that takes the glacier hypsometry into account(Furbish and Andrews,1984). The approach has been used at severalother places where it is also compared to AccumulationeArea Ratio(AAR) reconstructions in order to check the validity (Ballantyne,2002; Bakke et al., 2005b). In this study the ELA estimates are

given by an Area-Altitude Balance Ratio (AABR) of 1.8 (Osmaston,2005). This is based on the discussion in Rea (2009) where glaciersinWesternNorwayare given avalue of 1.5� 0.4 andglaciers at high-latitude is given values of 2.24� 0.85. In addition we have usedmass-balance data produced by NVE (Andreassen, 2000) and ELAestimates obtained by remote sensing (König et al., 2000).

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1249

3.2. Lake sediments

Six long cores were retrieved using a 110 mm diameter pistoncorer (Nesje, 1992) from a raft. The short cores were collected usingan HTH-gravity corer (Renberg-type) allowing sampling of up to50 cm of undisturbed surface sediments. Prior to coring, both lakeswere surveyed with an echo sounder in order to map thebathymetry (Fig. 3). We collected c. 14 km of ground penetratingradar (GPR) profiles penetrating through the soft sediments inorder to map the palaeo-basins (Fig. 3). For this purpose we useda RAMAC GPR fromMäla with a 25 Mhz RTA antenna inside a 10 mlong PVC tube, towed after a rubber boat that was connected toa motorized boat with a 25 m long rope.

3.2.1. Laboratory analysesThecores fromAustre-andVestreKjennsvatnetwere subjected to

laboratory analysis for every half centimeters. The methods appliedincludedmagnetic susceptibility (MS) (measuredon split cores usinga Bartington equipped with a MS2E surface sensor), weight loss-on-ignition (LOI) (Dean, 1974), dry bulk density (g/cm3) (DBD), watercontent and grain-size analysis using a Micromeretics Sedigraph5100 (X-ray determination) (Webb and Orr, 1997). Grain-size statis-tics were obtained using Gradistat 4.0 (Blott and Pye, 2001).

Material used for magnetic analyses on core VKJP-103 wassampled with cubic plastic boxes (2� 2� 6 cm3) for every 1 cmthroughout the core (n¼ 328). Initial magnetic susceptibility (cbulk)was determined on a KLY-2 induction bridge (sensitivity: 4�10�8 SI)on both wet and freeze-dried samples. The wet samples were cor-rected for the diamagnetic effect of water (�0.9�10�8 m3 kg�1).Frequency-dependant susceptibility was measured on a BartingtonMS2B dual-frequency sensor. Anhysteretic remanent magnetisation(ARM) was imposed with a 0.1mT DC field and a 100mT AC-field ina 2 G af-demagnetiser, whereas remanent coercivity was obtained byimposing selected samples to progressively higher magnetic fields.A solenoid was used to 175mT, succeeded by a pulse magnetiser(Redcliffe, maximum field 4.2 T). Imposed remanent magnetisation(IRM) was measured either on a Digico spinner (noise level:5�10�7 Am�1) or a 3-axes Cryogenic magnetometer (CCL 350/450,noise level 3�10�8 Am�1). Finally, susceptibility on HTH gravitycores was measured on plastic bags for every 0.5 cm at the KLY-2induction bridge with the exception of the dated intervals.

Table 2Radiocarbon dates from the different cores in Austre and Vestre Kjennsvatnet.

Core Lab. No. Depth(cm)

Type ofmaterial

Raag

AKJP-103 POZ8095 34.5 Macro 2AKJP-103 POZ 8101 64.5 Macro 7AKJP-103 POZ 8094 117 Macro 17AKJP-103 POZ 8098 135 Macro 17AKJP-103 POZ 8131 165 Macro 24AKJP-103 POZ 8099 200 Macro 30AKJP-103 POZ 8132 208 Macro 33AKJP-103 POZ 8339 298 Macro 69AKJP-103 POZ 8096 309 Macro 82VKJP-204 POZ12404 20 Macro 5VKJP-204 POZ12405 41 Macro 26VKJP-204 POZ12406 58 Macro 30VKJP-204 POZ12408 67 Macro 41VKJP-204 POZ12409 87 Macro 69VKJP-204 POZ12410 103.5 Macro 60VKJP-204 POZ12412 112 Macro 46VKJP-204 POZ12413 141 Macro 83VKJP-304 POZ12414 123 Macro 61VKJP-404 POZ12415 145 Macro 39LMP 204 POZ12416 74 Macro 35LMP 204 POZ12417 161 Macro 62

The XRF analyses was done with a AVAATECH X-Ray Fluores-cence Core Scanner, a non-destructive logging instrument applyingenergy dispersive X-ray fluorescence spectrometry for the deter-mination of major element concentrations in split sediment-coresamples (Richter et al., 2006).

3.2.2. Radiocarbon datingThe radiocarbon dates shown in Table 2, were produced by the

Pozna�n Radiocarbon Laboratory and calibrated (cal yr BP) accordingto OXCAL 4.0 (Bronk, 2001) and recalculated to b2k years. Wecollected macrofossils from lake sediments by wet-sieving freshsediment samples from the cores. Macrofossils were oven-dried at50 �C and placed in sealed glass vials prior to submission for accel-erator mass spectrometry (AMS) dating. All radiocarbon dates areobtained from terrestrial plant macrofossils. The ageedepth rela-tionship is based on linear extrapolation between individual dates.

4. Results

4.1. Mapping of marginal moraines

Marginal moraines deposited by Austre Okstindbreen aremapped to the east and south of the mountain Ridaren, and indi-cate six glacier advances or periods of slowed recession (Fig. 2). Themoraine ridges to the east of Ridaren was mapped and dated byWinkler (2003) (see Table 1). The marginal moraines south ofRidaren were mapped in this study and correlated, based on thenumber of moraines, with the lichen-dated moraines east ofRidaren (see Fig. 2). The marginal moraines are marked with sitenames and numbers in Fig. 2, and all moraines with numbersM2eM6were formed during the LIA or later (c. AD1750,1870,1920,1936 and 1950, respectively). The lichenometric dating of themoraines indicate slow retreat from the LIA maximum until themid- nineteenth century (Winkler, 2003). The moraines named M1are older than the LIA maximum (see discussion in section 5.2).

4.2. Lacustrine sediments

The initial coring strategy at Austre Okstindbreen was to followthe glacial sediments from source to sink through a chain of distal-fed glacial lakes. During the summers of 2003 and 2004 all together

diocarbone� 1. sigma

�1 sigma(cal yr BP)

� 2 sigma(cal yr BP)

Mean(b2k)

60 �25 285e315 280e320 35040 �30 665e690 660e725 73030 �30 1605e1695 1560e1710 170075 �30 1620e1730 1610e1745 173080 �30 2365e2690 2450e2720 263590 �30 3265e3360 3240e3380 339030 �35 3485e3615 3470e3640 340560 �50 7720e7845 7685e7870 782040 �40 9135e9280 9085e9320 925590� 30 590e640 540e650 66560� 30 2750e2780 2740e2845 281540� 35 3215e3330 3160e3360 331000� 35 4530e4630 4450e4815 461090� 40 7700e7925 7720e7930 786520� 40 6795e6905 6750e6960 690550� 35 5315e5450 5310e5470 544030� 40 9305e9420 9150e9465 941070� 40 7010e7130 6950e7170 711000� 40 4290e4415 4230e4430 437040� 35 3730e3885 3705e3910 386060� 50 7035e7265 7010e7275 7195

Fig. 2. Map showing the northern and eastern terminus of the glacier Austre Okstindbreen. The northern part of the moraine chronology was mapped and dated byWinkler (2003).The arrow in the upper left corner indicates the former meltwater channel draining directly into Austre Kjennsvatnet as depicted in Fig. 1.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621250

14 cores were retrieved at different locations in Austre- and VestreKjennsvatnet (Figs. 1 and 3). In general, the cores can be dividedinto three units (based on the MS stratigraphy) (Fig. 4). A detaileddescription of the cores and the different core parameters arepresented in the following sections.

4.2.1. Austre KjennsvatnetLaboratory analyses on AKJP-103 (328 cm long) and AKJP-203

(286 cm long) included LOI, DBD, WC and MS. Both cores weretaken close to each other and show good lithostratigraphicalcorrelation (Fig. 4). Since AKJP-103 has higher sedimentation ratewe used this core for radiocarbon dating, XRF analyses, grain-sizeanalyses and analyses of different rock magnetic properties.

4.2.1.1. AKJP-103. Regression analyses show that SIRM, DBD andmedium silt co-vary at a statistical significant level throughout thecore (see Table 3) (Fig. 5). The regression between SIRM and DBDshows that the core can be divided into three statistically differentunits. The lowermost unit from 328 to 287.5 (unit C) cm consists ofgrey clayey silt and the percentage of silt is declining from 22 to 15%whereas MS declines from 10�6� 40e20 (SI) towards the end ofthe unit. The magnetic analyses indicate a negative trend fromrelatively high values towards lower values in all measuredparameters (Fig. 6). Between 297 and 283 cm there is a markedpeak in DBD (0.8e0.9 g/cm3), S-ratio and ARM/SIRM (indicative ofmagnetic grain size) (Fig. 6). Silicon count-rates show a decliningtrend indicative of reduced clastic input to the lake.

From 283 to 120 cm (unit B) the colour of the sediments isbrownish grey, intercalated by layers of grey silt and very fine sand.DBD varies with a high frequency between 0.6 and 0.8 g/cm3,whereas medium silt shows an average content of 12%. In twointervals (from 159 to 169 cm and 143e132 cm) the medium siltcontent peaks at about 20%, indicating periodically higher transportcapacity in the meltwater stream entering the lake. The measure-ments of magnetic properties show low values, indicating a stable

sedimentation except during the intervals 210e191 cm and181e150 cm when the SIRM value peaks. Silicon count-rates showtwo marked excursions (107 and 253 cm) indicating a differentsource area for the sediments during these periods.

The uppermost 120 cm (Unit A) is light grey with some evenlighter grey bands consisting of fine silt. The first part of the unit(from 120 to 100 cm) shows a gradually rise in SIRM and DBD. Alsothe LOI values drop to between 1.5 and 3%, whereas medium siltvaries around 12% and the cbulk shows a rapid increase with twoprominent peaks between 86e62 and 36e11 cm. These two peaksare distinct in nearly all sediment parameters. The silicon count-rates show high values as in phasewith the other physical sedimentparameters except the mirrored structure between 98 and 60 cm assilicon reach a higher level earlier than for example surfacesusceptibility.

4.2.2. Vestre KjennsvatnetThe cores in Vestre Kjennsvatnet are taken along a transect

stretching throughout the lake, starting close to the inlet (exceptcore VKJP-104 which is taken in the western basin). The GPRanalyses indicate that lithostratigraphical units should be consis-tent throughout the lake based on the uniform distribution of softsediments (see Fig. 5).

4.2.2.1. VKJP-104, 204, 304 and 404. Core VKJP-104was retrieved inthe western part of the lake (Fig. 3) and is divided into three mainsedimentary units. The lower unit (466e78 cm) consists of massiveand weakly laminated clayey silt. The middle unit (77e19 cm)consists of grey brownish clayey gyttja. The upper unit (19e0 cm) isdominated by grey clay (Fig. 7).

Core VKJP-204 and VKJP-304 was retrieved in a distal position ofthe lake in a flat areawithin the deepest part of Vestre Kjennsvatnetand show a very similar lithostratigraphy (Figs. 1, 3 and 4).Regression analyses on VKJP-204 between the different sedimentparameters show no clear relationships, indicating that the coring

Fig. 3. Bathymetric map of Austre- and Vestre Kjennsvatnet. All coring sites, and also the GPR profiles, are outlined. Austre Kjennsvatnet consists of one shallow and circular basin,whereas Vestre Kjennsvatnet consists of two sub-basins. At present the lake is 2 m higher than its natural elevation as the lake is used for water storage and hydroelectric powerproduction.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1251

site is rather insensitive to changes in the transport capacity of themain river entering the lake from Austre Kjennsvatnet (Table 3).However, DBD and MS indicates two different sedimentationregimes distinguishing unit A from units B and C. Unit C(166e114 cm) contains grey sediments with fine silt, LOI valuesbetween 1 and 4%, DBD values between 1.4 and 1.6 g/cm3 and MSvalues between 10�6� 25e40 (SI). Unit B (114e68 cm) contains LOIvalues from 4 to 12%. Grain-size analyses show increasing claycontent and decreasing medium silt, whereas the remaining grain-size fractions are relatively stable. LOI in unit A (68e0 cm) is steadilydecreasing from 8 to 2% towards the top of the core. At 40 cm depth,a distinct peak in DBD and a similar marked drop in MS are sug-gested to represent a flood event, which is supported by severalin-washed plant macrofossils in the layer. The grain-size analysesshow some more variability in clay content peaking close to 50% at17 cm depth. Core VKJP-304 is closely correlated to VKJP-204, butwith a slightly higher sedimentation rate as it is 20 cm longer.

Core VKJP-404 in Vestre Kjennsvatnet was retrieved in the basinjust north of the shallow sill dividing Austre- and Vestre Kjenns-vatnet (Fig. 3), and is expected to be very sensitive to any change inthe current through the narrow passage from Austre Kjennsvatnet.

The regression analyses between the different core parametersindicate that very coarse silt controls both the MS and DBD vari-ability (Table 3). The core is in total 160 cm long and contains onlyunit A and B, where unit A is the most interesting. UnitA (43.5e4 cm) indicates a rapid shift in the sedimentation envi-ronment in the lake were MS peaks at close to 10�6� 40 (SI) andDBD is steady on 1.4 g/cm3 and LOI drops to below 1%. Grain-sizeanalyses indicate an abrupt shift in the sedimentation environmentat 42.5 cm as the finest fractions (finer than medium silt) dropabruptly, possible as a response to a stronger current through thepassage from Austre Kjennsvatnet towards Vestre Kjennsvatnet asa consequence of direct meltwater supply from the glacier AustreOkstindbreen. The uppermost 4 cm in the core shows a decline inDBD and MS and an increase in LOI. This is indicating a shift in thesedimentation environment similar to unit B in the core.

4.2.3. Surface samples in Austre and Vestre KjennsvatnetIn order to get undisturbed surface sediments from Austre and

Vestre Kjennsvatnet, altogether eight HTH gravity cores from theuppermost c 50 cm of the sediments were retrieved. MS wasmeasured on all the HTH cores and correlated with MS obtained

Fig. 4. Vertical view of Austre and Vestre Kjennsvatnet showing the thresholds dividing the two lakes into three different sub-basins in addition to coring sites. In the lower panelthe magnetic susceptibility (MS) stratigraphy is shownwith implied correlation (dotted) between the different sedimentary units in the cores retrieved from the Austre- and VestreKjennsvatnet.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621252

from the piston cores. This was done to validate the uppermost partof the piston cores. The correlations show that VKJP-104 and VKJP-404 are disturbed in the uppermost part, whereas the remainingcores are undisturbed as they contain sediments from AD 2004.

In the lower part of Austre Okstindbreen, west of Bretinden,there is an ice-dammed lake. This was monitored from 1976 until1987, and most of the meltwater is suggested to have drainedthrough a subglacial meltwater channel to the snout of the glacierin Oksfjelldalen and has probably not influenced Austre Kjenns-vatnet since 1979 (Knudsen and Theakstone, 1988). Hence, theglacier surface of Austre Okstindbreen must have been below thelocal threshold necessary to route meltwater directly towardsAustre Kjennsvatnet since 1979, and thereby established thepresent meltwater route towards lake Grasvatnet. All HTH coresshow declining MS values towards the top. MS have earlier beenused as a measure of clastic sedimentation in lake sediments(Snowball, 1996). The MS values in the upper part of the HTH coresindicate lower clastic input to the lake, and is suggested to reflectthat Austre Okstindbreen had retreated behind the local threshold(815 m a.s.l.). Hence, before AD 1979 glacier meltwater from AustreOkstindbreen was routed directly towards Austre Kjennsvatnet asthe first distal-fed glacial lake (Knudsen and Theakstone, 1988).

4.3. Ageedepth relationships

The cores VKJP-204 and AKJP-103 have independent ageedepthrelationships, whereas the remaining cores fromAustre- and Vestre

Kjennsvatnet are based on scattered radiocarbon dates and corecorrelation (Figs. 4 and 8). All age-depth models are constructed bylinear interpolation between radiocarbon dates, and correlationbetween the different cores in both Austre- and Vestre Kjennsvat-net based on the MS stratigraphy. Hence, abrupt shifts to poorersorting in the minerogenic input to the lakes and anomalies in thealuminium count rate were linked to flood events. Either caused bydraining of the glacier-dammed lake at Austre Okstindbreen,abrupt temperature driven snowmelt and/or by heavy precipita-tion. These events are not relevant for the glacier reconstruction,and are regarded as noise in the record. Using this approach, fourmajor flooding events were detected and correlated between thedifferent cores in Austre- and Vestre Kjennsvatnet. The floodingevents are, in some of the cores, represented by alternating layers ofsand and plant macrofossils (Fig. 9). In addition to these majorfloods, there were several minor events observed in the sorting-mean record from Austre Kjennsvatnet. HTH core KJS-304 from thesurface sediments is tentatively correlated to the piston core VKJP-204 based on visual description and the MS stratigraphy, and this isconfirmed based on 210Pb dating in the HTH core and the radio-carbon based age-depth model from VKJP-204.

4.4. Holocene ELA variations at Austre Okstindbreen

In this study a number of different sediment parameters havebeen used to link up-valley glacier activity to down-valley lakesedimentation. A close relationship between variations in ELA and

Table 3Correlation coefficients (rs) for core parameters.

VKJP-103 DBD LOI SIRM WC VC silt C silt M silt F silt VF silt Clay

DBD 1.00LOI �0.078 1.00SIRM 0.65 �0.68 1.00WC �0.92 0.80 �0.80 1.00VC silt 0.17 �0.34 0.05 �0.09 1.00C silt 0.45 �0.59 0.63 �0.48 0.25 1.00M silt 0.57 �0.62 0.79 �0.65 0.02 0.73 1.00F silt �0.24 0.46 �0.37 0.29 �0.43 �0.52 �0.42 1.00VF silt �0.48 0.68 �0.56 0.48 �0.45 �0.75 �0.68 0.62 1.00Clay �0.47 0.53 �0.56 0.47 �0.51 �0.77 �0.66 0.22 0.54 1.00

AKJP-204 DBD LOI MS WC VC silt C silt M silt F silt VF silt Clay

DBD 1.00LOI �0.90 1.00MS 0.70 �0.80 1.00WC �0.97 0.90 �0.72 1.00VC silt 0.34 �0.20 0.05 �0.31 1.00C silt �0.14 0.31 �0.37 0.10 0.36 1.00M silt 0.06 0.08 �0.11 �0.10 0.17 0.58 1.00F silt 0.02 �0.02 0.03 �0.02 �0.37 �0.06 0.23 1.00VF silt �0.20 0.09 0.12 0.17 �0.45 �0.30 �0.48 0.24 1.00Clay �0.08 �0.09 0.15 0.12 �0.50 �0.80 �0.73 �0.30 0.20 1.00

AKJP-404 DBD LOI MS WC VC silt C silt M silt F silt VF silt Clay

DBD 1.00LOI �0.82 1.00MS 0.72 �0.71 1.00WC �0.89 0.98 �0.74 1.00VC silt 0.73 �0.42 0.28 �0.51 1.00C silt 0.52 �0.25 0.10 �0.32 0.83 1.00M silt �0.63 0.62 �0.50 0.65 �0.44 �0.07 1.00F silt �0.77 0.47 �0.35 0.55 �0.91 �0.80 0.52 1.00VF silt �0.79 0.48 �0.35 0.56 �0.92 �0.86 0.36 0.90 1.00Clay �0.22 �0.03 0.14 0.02 �0.70 �0.87 �0.26 0.51 0.62 1.00

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1253

DBD in distal-fed glacial lakes has been demonstrated (Bakke et al.,2005b). Here we use changes in geochemical composition, grain-size distribution and magnetic properties to validate the DBDvariations in core AKJP-103 (Fig. 9). A regression plot between thealtitude of the ELAs belonging to the moraines M1-M6 and thecorresponding DBD values (mean of 5 values) is shown in Fig. 10. Inadditionwe have used the DBD values from the 210Pb dated sectionin core VKJP-204, and correlated to AKJP-103, to tie DBD values tothe historical documented front positions recalculated to ELA bythe AABR method. This statistical relationship (r2¼ 0.98) is appliedin order to reconstruct continuous variations in the ELA during theHolocene. The moraines M2-M6 are independently dated by theuse of lichenometry and historical sources, whereas the age ofM1 isset to the largest peak in DBD and SIRM in core VKJP-103(110e115 cm) (Figs. 9 and 10). To make the regression modelsensitive to glacier variations when the glacier is very small, thetheoretical ELA was calculated with 2.3 �C higher ablation-seasontemperatures (Bjune and Birks, 2008), and the same precipitationas at present (Lie et al., 2003). For this scenario the ELA wascalculated to an altitude close 1600 m a.s.l. The DBD record isrepresentative for the glacier magnitude at all 15 known positionsof the glacier through a second order polynomial function(r2¼ 0.98). The chosen AABR gradient used in the BR method maybe wrong, and consequently lead to an error in the ELA estimates(Rea, 2009). Using different gradients (from 1.6 to 2), the accumu-lated error bars on the ELA reconstruction is calculated tobe�40 m, and are shown as a green shaded area in Fig. 11. Thecontinuous ELA reconstruction is based on the dry bulk densityvalues (all abrupt events are removed, see Fig. 11 upper panel) fromAKJP-103. In addition, SIRM is used in order to study the sensitivityto different physical sediment parameters. Whenever the glacier

Austre Okstindan has been behind the local threshold towardsAustre Kjennsvatnet (ELA higher than 1190 m a.s.l.), the SIRMrecord is relative insensitive. Hence, as the glacier grew larger andcrossed the local threshold, the sensitivity in SIRM was higher thanin the DBD record (Fig. 11).

5. Discussion

5.1. Approaches and interpretation of lake sediments

Using lacustrine sediments retrieved from downstream glacier-fed lakes to reconstruct past glacier activity requires careful vali-dation of the records in question. Because sediments from varioussources tend to be deposited in lakes, it implies that several sourcesof error are present whenever using such archives to reconstructglacier activity. Sediments not associated with a glacier advancemay include paraglacial reworking of old glacigenic sediments,slope processes around the lake, turbidites from the shore edge andfrom the delta, and hiati due to flooding events (Knudsen andTheakstone, 1988). The landscape relief around lakes is alsoimportant as gentle slopes can reduce the possibility of snowavalanches and rock fall, which otherwise might impact on theoverall sediment budget of the lake. However, only a thin andincomplete cover of superficial deposits is present in the catch-ment, reducing any significant influence from sediments remobi-lized by paraglacial processes.

In this study we establish the relationship between glacieractivity and lake sediments through several approaches; multiplecores from many locations within the same lake and multi-proxyanalyses including S-ratio, ARM, SIRM, MS, LOI, WC, grain-sizedistribution as well as geochemical variations (XRF). Further,

Fig. 5. Selected physical sediment properties of core AKJP-103 from Austre Kjennsvatnet (based on visual description the following subdivision includes: GS¼ grey silt, GBS¼ greybrownish silt, GSS¼ grey sandy silt, DS¼ light grey, inorganic silt). Note that all parameters except silicon and magnetic susceptibility (MS) show lower internal variability towardsthe top of core.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621254

regression analyses are used to clarify the relationship betweendifferent core properties (Table 3). An important aspect concerningthe sedimentation of glacial-derived sediments in Austre- andVestre Kjennsvatnet is a relict meltwater channel running from themoraine ridges west of Ridaren at Austre Okstindbreen towardsAustre Kjennsvatnet (see Figs. 1e3). This channel implies thatduring periods when Austre Okstindan was in an advanced posi-tion, Austre Kjennsvatnet received glacial meltwater input directlywithout first being influenced by settling of the coarsest suspendedfractions in lake Grasvatnet. The last time it was observed drainageacross the local threshold was in AD 1979 (e.g. Dahl et al., 2003).Subsequently all glacial meltwater from Austre Okstindbreen afterAD 1979 has been routed through Grasvatnet before it has enteredAustre- and Vestre Kjennsvatnet.

Hence, two ways of using multi-proxy reconstructions basedon downstream glacier-fed lakes to reconstruct former glaciermagnitude have been combined. The first utilizes proxiesreflecting transport capacity of the rivers entering the lake, andsubsequently links variability in these proxies to variations inglacier size. The critical assumption for this approach is thatlarger glaciers on average release more meltwater than smallerglaciers (e.g. Dahl et al., 2003), and that this relationship isreflected in the grain-size distribution along a length profile from

inflow to outflow through the lake. Given that the grain-sizedistributions does not change significantly with glacier size,temporal shifts in grain sizes can be interpreted to reflect varia-tions in amount of meltwater reaching Kjennsvatnet, i.e. moreclay corresponds to a small glacier and vice versa (Fig. 9). Abruptchanges in grain-size composition may indicate flooding eventsand are therefore not related to any changes in glacier magnitude.The second approach exploits the overall amount of minerogenicsediments delivered to distal-fed glacial lakes. Glacier meltwaterstreams carry large amounts of suspended silt produced by theglacier as it erodes the substrate by abrasion and plucking. Thetotal amount of silt deposited in the downstream lakes conse-quently depends on the glacier activity. This approach is based onthe same assumption which was explored by Karlén (1976) whoused loss-on-ignition as an negative indicator of inorganic sedi-mentation (i.e. glacial silt) in order to quantify variations informer glacier size. The underlying assumption is that anytemperate glacier in a catchment produces an abundant clay-siltsize fraction that will override all other potential sources ofsediments to a glacier-fed lake. Although LOI is measured in allcores presented here, more attention is devoted to other, physicalsediment proxies in terms of reconstructing past glacier activity.They include DBD, which is a sensitive sediment bulk parameter

Fig. 6. Diagram shows dry bulk density (DBD) and different rock magnetic parameters in core AKJP-103 from Austre Kjennsvatnet. The core is divided into three phases: (A) largeglacier draining over the threshold into Austre Kjennsvatnet, (B) small glacier at Austre Okstindbreen, and (C) deglaciation sediments.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1255

mainly reflecting changes in sediment grain-size composition,four different geochemical elements reflecting glacier erosion(silicon, aluminium, titanium and iron), and finally the concen-tration of magnetic particles as observed by the SIRM. The lattertwo approaches builds on the assumption that larger glaciers tendto produce more sediment and that these two methods are ableto track such changes, i.e. more magnetic particles reflects a largerglacier and vice versa (Fig. 9). The analyses of differentgeochemical elements retrieved by the use of XRF analyses canprovide detailed high-resolution information about the state ofthe glacier and its interaction with climate. Changes in sedimenttransfer rates from warm-based glaciers are related to the massturnover gradient in the glacier (Hallet et al., 1996), and thisturnover gradient is shown to be reflected in the titaniumconcentration as quantified by XRF analyses on sediments froma distal-fed glacial lake (Bakke et al., 2009).

5.2. Holocene variations in ELAloc at Austre Okstindbreen

The ELA reconstructions presented here (Fig. 11) must beregarded as a local ELA (ELAloc), as Austre Okstindbreen is stronglyinfluenced by leeward accumulation of wind-drifting snow overthe summit Oksskolten. This makes it different from the regionalELA, which is determined predominantly by summer temperatureand winter precipitation (cf. Østrem and Liestøl, 1964). The Holo-cene history of the ELAloc presented here is well constrained,building on dated moraines, historical evidence and lakesediments.

5.2.1. DeglaciationMagnetic properties in core AKJP-103 indicate that the lower-

most 28 cm contains glacial sediments of a different provenancecompared to those deposited during the Neoglacial (<4000 years

Fig. 7. Selected physical sediment properties for core VKJP-204, collected from Vestre Kjennsvatnet (based on visual description the following subdivision of the core was done;GS¼ grey silt, GBS¼ grey brown silt, BGS¼ brown grey silt and GSS¼ grey silty sand). Dark bands represent layers with homogeneous sediments, interpreted to represent abruptevents, which were consequently removed before the ageedepth relationship was established (see Fig. 9).

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621256

b2k), indicating that material from other source areas than thepresent glacier-covered area influenced the lake at that time(Fig. 5). The lowermost radiocarbon sample from core AKJP-103 at309 cm depth yielded an age of 8240� 40 14C yr BP, and is sug-gested to date the deglaciation in the area. This is in accordancewith earlier radiocarbon dating of marine molluscs from the headof several nearby fjords indicating that the retreating ice-sheetbecame land-based around 9200e9000 14C yr BP (Andersen et al.,1981). A paper by Linge et al. (2007) shows that ice remnantsfrom the Scandinavian Ice Sheet (SIS) existed in the mountainsalong the coast of Nordland as late as 9000 b2k. Observations ofdead-ice features and raised shorelines around Grasvatnet (Hoel,1910; own observations) may indicate a slow vertical down-wastage of the SIS in the area. If this was the case, it may explain theanomalous sedimentation in Austre- and Vestre Kjennsvatnetduring this early phase. Hence, Austre Okstindbreen is likely tohave become a separate glacier after 9140 b2k.

5.2.2. Glacier variations from 9140 to 7000 b2kThe first part of the reconstruction indicates a gradual rise in the

ELAloc culminating when the glacier was at its smallest c 7760 b2k.At this time the ELA was 80 m higher than at present. Within 250years however, the glacier advanced with a maximum lowering of

the ELA of 240 m adjusted for land uplift close to 7200 b2k. This isinterpreted as a short-lived event lasting only for about 50 years.The glacier advance is reflected in core AKJP-103 as an increase insilt concentration and higher IRM, S-ratio and SIRM values. In coreVKJP-204 the same event is seen as a peak in clay content, whichcorresponds to higher DBD values in VKJP-404. After 7200 b2k theELAloc rose by 400 m within a hundred years indicating a majorclimatic change allowing for decades with negative mass balance.In core AKJ-103 a time lag between the s-ratio (indicating changesin magnetic mineralogy) and the SIRM during the latter part of thisperiod is interpreted to represent input from another source areathan the bedrock underlying Austre Okstindbreen. The S-ratiovalues remain high until c 6500 b2k, whereas the SIRM concen-tration decreases rapidly immediately after the glacier advancec 7200 b2k. Several moraines are apparently located outside nowempty cirques on Kjenssvassfjellet (Hoel, 1910), which may implythat glaciers where present some time after the retreat of the SIS(w9140 b2k). As this mountain consists of a different bedrock typethan that beneath Austre Okstindbreen, it seems likely that severalsmall cirque glaciers existed at Kjenssvassfjellet during the earlypart of the Holocene. However, they did not rejuvenate during theNeoglacial phase, and this may explains the aforementioned timelag between the S-ratio and SIRM.

Fig. 8. Age-depth relationship for all piston cores retrieved from the Austre and VestreKjennsvatnet, including a 210Pb surface dating of core VKJP-204. An individual age-depth relationship, based on 9 radiocarbon dates, is constructed for Core AKJP-103relationship. Core VKJP-204 has also its own age-depth relationship based on 5radiocarbon dates, including two which were rejected. Grey shaded areas show thecorresponding error estimates. Cores VKJP-304 and VKJP-404, from Vestre Kjenns-vatnet, contains only a few radiocarbon dates. They are thus correlated with the otherchronologies using the magnetic stratigraphy presented earlier. AKJP-203 from AustreKjennsvatnet is not dated.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1257

5.2.3. ELAloc variations from 7000 to 4000 b2kFrom 7000 to 4990 b2 k the ELAloc was at the highest eleva-

tion during the entire Holocene, fluctuating around 1600e1500 ma.s.l. This prolonged period characterised by a high ELAloc waspossibly interrupted by four periods of increased glacier activityat 6580, 6115, 5270 and 4470 b2 k. The three first glacial eventswere relatively short with duration of c. 50 yrs, whereas thelatest event lasted for about 230 yrs. During this time span, thedifferent sediment parameters in Austre- and Vestre Kjennsvat-net show relatively stable sedimentation conditions. In AKJP-103IRM, S-ratio and SIRM were low, whereas the clay content varied.Core VKJP-204 shows a steady increase in clay content suggestingexpanded ice cover in the catchment from around 5000 b2k.From 4990 until 4000 b2k the ELAloc was gradually lowered andreached the present altitude (1340 m a.s.l.) close to 4000 b2k.This maximum event is recorded in core AKJP-103 to have lastedfor c. 200 yrs, and is characterised by an increase in very coarsesilt, higher ARM and S-ratio values. In core VKJP-204 it corre-sponds to a significant increase in clay content. The intervalstretching from 7000 to 4000 b2k overlaps with the so-calledHolocene Thermal Optimum, when many of the Scandinavianglaciers allegedly were melted completely away (Nesje et al.,2008; Nesje, 2009). This notion is mainly based on the assump-tion that LOI values in many distal-fed glacial lakes are high atthe time. Sediment analyses of Austre- and Vestre Kjennsvatnetindicate that, contrary to other published records, the glaciers atOkstindan may have existed continuously during the entireHolocene. This can be explained by the large altitudinal spanbetween the modern mean ELA at 1330e1340 m to above 1700 ma.s.l. that is occupied by Austre Okstindbreen. From other similarstudies in Scandinavia this period is represented with moreorganic rich sediments interpreted as a period without input ofglacial sediments (Bakke et al., 2005c; Nesje, 2009). Hence, itcannot be excluded that the sediments found in Austre- andVestre Kjennsvatnet during this time span originated from non-glacial processes.

5.2.4. ELAloc variations from 4000 to 1350From 4000 until 1350 b2k the glacier was generally small, inter-

ruptedbyfive short-livedperiodswith larger glacier advances. Theseglacier advances are bracketed between 3550e3350, 3270e3180,2960e2700, 2480e2330, 2180e1970 and 1755e1550 b2k. In coreAKJP-103 the rapid changes inELAloc is seen as large variations in claycontent and SIRM as well as in aluminium fluxes (Fig. 9).

5.2.5. ELA variations from 1350 b2k e presentDuring the last 1350 years Austre Okstindbreen reached the

Holocene glaciermaximum. Thiswas the only time periodwhen theglacierwasbigenough to reroutemeltwaterover the local threshold,causing drainage of sediment-laden meltwater directly into AustreKjennsvatnet. The actual size of the glacier at this time is corrobo-rated by the presence of independently dated marginal moraines(Figs. 2 and 10, and Table 1). The glacier crossed this topographicthreshold for the first time between 1350 and 1240 b2k, whereaslater crossings are dated to have occurred between 850e590 and320e100 b2k. The ELAloc corresponding to the periods when theglacier obtained this extent was 1190 m a.s.l. The DBD values inAustre Kjennsvatnet indicate that the largest of these glacieradvances was bracketed between 1350 and 1240 b2k, based on theage-depth relationships in both AKJP-103 and VKJP-204. It is there-fore suggested that the undated terminalmoraineM1wasdepositedduring this time span, which is in accordance with Winkler (2003)and a maximum age of moraine formation at Austre Okstindbreenby Griffey andWorsley (1978) of around 1500 14C yr BP.

Periods with glacier magnitudes similar to that of today issuggested to have taken place during three phases: 1210e1180,1040e950 and 540e360 b2k. In Austre Kjennsvatnet (AKJP-103)these periods with glacier growth during the late Holocene arerecorded as an increase in SIRM and DBD. In core VKJP-404 they areseen as a marked increase in coarse silt and DBD (Fig. 9). I VKJP-204the periods are marked by higher clay content (Fig. 9).

5.2.6. Comparison with other glacier records along theNorwegian coast

The Holocene ELA variations from Okstindan are compared withthree other reconstructed glaciers in Norway in order to discuss thetiming of glacial advances along the coast of Norway. The recon-structed glaciers used for this purpose are Folgefonna (60�N)(Bakke et al., 2005a,c), Jostedalsbreen (62�N) (Nesje et al., 1991,2001) and Lyngen (69�N) (Bakke et al., 2005b). In Scandinavia,altogether five large glacier advances are recorded from the onset ofthe Holocene until after the 8.2-ka event (11.2, 10.5, 10.1, 9.7 and8.4 ka b2k) (Dahl and Nesje, 1996; Bakke et al., 2005a; Nesje, 2009).During the mid- and late Holocene four periods with consistentglacier advances are identified: 4e3.9, 2.8e2.7, 1.3e1.1 and0.13e0.1 ka b2k. The recurrence intervals reveal millennial scalevariability (w1000 to 1500 years), and for the late Holocene theseevents apparently occur in concert with the Bond-cycles (Bondet al., 2001).

The two first advances that are observed in all records are the4 ka event and the 2.7 ka event which apparently are part ofa millennial-scale variability also seen in other proxies around theNorth Atlantic (Griffey and Worsley, 1978; Thornalley et al., 2009).An earlier study suggested that there was a glacier advance largerthan the LIA in Okstindan, possibly as early as 3000 14C yr BP.Radiocarbon ages obtained on material collected from soil profiles,however, may be incorrect due to contamination effects. Thesecond youngest of these altogether four intervals occurredbetween 1.3 and 1.1 ka b2k, and represents the largest Holoceneglacier advance at Austre Okstindbreen. This maximum advance isalso seen in the three other glaciers (Fig. 12), but it is only atOkstindan that this advance was larger than the LIA maximum. The

Fig. 9. Different physical sediment parameters measured on cores AKJP-103, VKJP-204 and VKJP-404. Sedimentation rates (grey bars) and shifts in sorting (black bars) are depictedin the upper part of the figure. The black bars indicate abrupt events in the sediments, and were been removed from the sediment record which was used for reconstruction of thelocal ELA. Grey boxes reflect the changing sedimentation rates in core AKJP-103. Sediment parameters include clay content, SIRM, dry bulk density (DBD), Aluminium (Al)concentration and an iron/titanium-ratio. Note how the gradual increase in sedimentation rate is reflected in AKJP-103 (Al) and also in AKJP-204 (clay). Similarly, when the glaciercrosses the local meltwater threshold for drainage directly into Austre Kjennsvatnet at 1300 b2 k a sudden increase is registered in all cores, except for AKJP-204 due to its distalplacement in Vestre Kjennsvatnet.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621258

youngest of these four glacier advances coincides with the timing ofthe LIA in Europe (Grove, 1988).

5.2.7. Comparison with other glacier records in the NorthernHemisphere

During the Holocene Thermal Optimum (9e6 ka BP) manyglaciers in the Northern Hemisphere reached a minimum, althoughthe timing between the different records differs notably (Jansenet al., 2007). The onset of the Neoglacial (w4 ka BP) is marked bya gradual expansion of alpine glaciers in the Northern Hemispherewith some exceptions in e.g. North America (Menounos et al.,2009). This Neoglacial expansion culminated most places in the“Little Ice Age” advance, which frequently represents the Holocenemaximum extent. This development is evident from the compila-tion presented in Fig. 12.

This pattern has previously been attributed to the gradualweakening of summer insolation at high northern latitudes duringthe Holocene (Berger, 1978). Differences in the timing of the onsetof the Neoglacial differs not only due to insolation differences, butalso because of poor dating of sediments and moraines as well aslocal conditions promoting or counteracting the overall trends insummer insolation (Wanner et al., 2008). Even so, shared featuresare present in the regional glacier records.

Menounos et al. (2008) shows, for example, that there wasa major glacier advance c 4.2 ka b2k in western Canada. This is inaccordance with our record from northern Norway, indicatinga major advance within the same interval. The same is true forSvalbard, Iceland, the European Alps and even the Himalayas(Fig. 12). In fact, this event is arguably global in scope (Paasche andBakke, 2009).

A

B

Fig. 10. Figure shows the relationship between the DBD record from core VKJP-103 and ELA with estimated ages. (A) Red dots indicate periods when the ELA is estimated, eitherbased on historical data, lichenometric data and/or interpretation of lake sediments (only when the glacier Austre Okstindbreen crossed the local threshold towards AustreKjennsvatnet). (B) Shows the statistical relationship between periods with known ELA (m a.s.l.) and dry bulk density (DBD) values. The regression model is used to modela continuous ELAloc for the Holocene. A similar relationship was also constructed for SIRM and ELA. Both ELAloc estimates are presented in Fig. 11.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1259

The next common glacial event in the Northern Hemispheretakes place 2.7 ka b2 k and is seen as an abrupt event in theOkstindan archive, on Baffin-Island, North America and possiblyalso in the Himalayas (Fig. 12). This shift is also seen in many otherproxies from the Northern Hemisphere, including themarine realm(Jansen et al., 2004; Thornalley et al., 2009). The Neoglacialmaximum at Okstindan is reached 1.3 ka b2k and a similaradvances is also recorded in Himalaya and on Baffin Island.However, the glaciers at Svalbard were at a Neoglacial minimumduring this time span (Humlum et al., 2005). Many of the minor

Fig. 11. Local equilibrium-line-altitude reconstruction for Austre Okstindbreen based on thethe main ELAloc estimate whereas the light green area denotes the uncertainty associated wiland uplift. Black thin line is the ELA reconstruction based on SIRM. Note that the sensitivity iKjennsvatnet, whereas the sensitivity is lower whenever the glacier meltwater is routed via(lower) necessary for the glacier to cross the threshold to Austre Kjennsvatnet.

events seen in the Okstindan glacial reconstruction is withoutcounterparts in the compared archives.

The “Little Ice Age” (LIA) marks the last of the four sharedglacier advances in the Northern Hemisphere. Based on (Paascheand Bakke, submitted for publication) the LIA has been definedto have lasted from 640 to 140 yrs. b2k, which is in accordancewith data from the European Alps (Holzhauser et al., 2005). Interms of glacier variability it is possible to refine the division ofthe LIA further into three sub-phases: Early, middle, and late,similar to the chronology for the Alaskan glaciers (Mckay and

bulk density record and the regression model shown in Fig. 10. The dark green line isth the estimates. The grey line represents the reconstructed ELA prior to adjustment forn this parameter is extremely good when the glacier has crossed the threshold to Austrelake Grasvatnet. The two horizontals lines marks present ELA (upper) and ELA lowering

Fig. 12. Selected glacier reconstructions from the Northern Hemisphere. (A) Relative glacier extent from Svalbard compiled from a continuous late Holocene reconstruction(Humlum et al., 2005). (B) Relative glacier extent from Baffin Island (compiled by Briner et al., 2009). (C) Glacier reconstruction from Okstindan (this study). (D) Major periods withmoraine formation during the Neoglacial phase at Iceland (Kirkbride and Dugmore, 2008; Geirsdóttir et al., 2009). (E) Relative glacier extent based on discrete data from theEuropean Alps (Ivy-Ochs et al., 2009). (F) Major phases with moraine formation in the Himalayas (Owen, 2009).

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621260

Kaufman, 2009). The early phase of the LIA lasted from 900 to 550yrs. b2k. This phase is not consistent with any other reconstructedglacier in Scandinavia. However, it corresponds with glacieradvances in coastal areas of Alaska (Wiles et al., 2008) and in theEuropean Alps (Ivy-Ochs et al., 2009). It is therefore evident thata division of the LIA in three sub-phases is not applicable forScandinavia (Nesje, 2009).

6. Conclusions

Sediments in distal-fed glacial lakes and independent datingof marginal moraines are complementary data sets which whencombined can yield detailed information about past glacier fluc-tuations. Connecting exact positions in time and space of glacierextent with corresponding changes in the sedimentary recordcontained in down-valley lakes are likely to improve the accuracyof the glacier reconstruction. In order to obtain reliable results it isnecessary to carefully select coring sites, avoiding those areas thatreceive significant amounts of extra-glacial sediments. Well-con-strained ageedepth models are of critical importance when

validating the modern analogue were lake sediments overlap withhistorically observations and chronologies of terrestrial deposits.

The following general conclusions can be drawn from thisstudy: (1) terrestrial deposits that have been dated and comparedwith sediments from distal-fed glacial lakes providing direct andunequivocal evidence for glacier fluctuations; (2) regressionmodels between known glacier extent in time and space andsedimentary DBD values may provide continuous records ofup-valley glacier fluctuations; (3) long-term changes in the prop-erties of the distal-fed glacial lake sediments may reflect changesin glacier extent, both indicating long trends but also abruptchanges in glacier mass balance; (4) analyses of LOI, DBD, WC, MS,rock magnetic properties and grain-size distribution are valuabletools to validate and understand the complex sedimentationpatterns in distal-fed glacial lakes; (5) analyses on geochemicalcomposition may supplement the more traditionally analyses ofphysical sediment properties, and also resolve source area for thedeposited sediments; (7) the complex history of Holocene glacierfluctuations is best understood using a multi-proxy approach; and(8) based on our comparison we conclude that late Holoceneglacier variations are both synchronous and asynchronous over the

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e1262 1261

Northern Hemisphere with synchronous advances 4.2 ka, 2.7 kaand, 1.3 ka b2k, and during the “Little Ice Age”.

Acknowledgments

This is a contribution from NORPAST2 a NORKLIMA projectheaded by Prof. Morten Hald and funded by the NorwegianResearch Council (NFR). We are grateful to Trygve Snøtun andTordbjørn Eidsheim Bø for help during the fieldwork at Okstindanand to Herbjørn Heggen for help with the laboratory analyses. Weare also grateful to Prof. Gerald H. Haug who provided the XRFfacilities at Climate Geology group at Geological Institute, ETHZurich. Magnetic measurements were carried out at the Paleo-magnetic Laboratory at the University of Bergen, which is admin-istrated by Prof. Reidar Løvlie. This is publication no A 283 from theBjerknes Centre for Climate Research.

References

Andersen, B., Boen, F., Nvdal, R., Rasmussen, A., Vallevik, P.A., 1981. Radiocarbondates of marginal moraines in Nordland, North Norway. Geografiska Annaler63A (34), 155e160.

Andreassen, L.M., 2000. Regional Changes of Glaciers in Northern Norway. NVE.Andrews, J.T., 1975. Glacial Systems e An Approach to Glaciers and Their Envi-

ronments. Duxberry Press, Massachusetts, 199 pp.Bakke, J., Dahl, S.O., Nesje, A., 2005a. Lateglacial and early-Holocene palaeoclimatic

implications based on reconstructed glacier fluctuations and equilibrium-linealtitudes at northern Folgefonna, Hardanger, western Norway. Journal ofQuaternary Science 20 (2), 179e198.

Bakke, J., Dahl, S.O., Paasche, Ø., Løvlie, R., Nesje, A., 2005b. Glacier fluctuations,equilibrium-line altitudes and palaeoclimate in Lyngen, northern Norway,during the Lateglacial and Holocene. The Holocene 15 (4), 387e409.

Bakke, J., Lie, Ø., Nesje, A., Dahl, S.O., Paasche, Ø, 2005c. Utilizing physical sedimentvariability in glacier-fed lakes for continous glacier reconstructions during theHolocene, northern Folgefonna, western Norway. The Holocene 15 (2), 161e176.

Bakke, J., Lie, Ø., Heegaard, E., Dokken, T., Haug, G.H., Birks, H.H., Dulski, P., Nilsen, T.,2009. Rapid oceanic and atmospheric changes during the Younger Dryas coldperiod. Nature Geoscience 2 (3), 202e205.

Balco, G., 2009. The geographic footprint of glacier change. Science 234, 599e600.Ballantyne, C., 1989. The Loch Lomond readvance on the Island of Skye, Scotland:

glacier reconstruction and paleoclimate implications. Journal of QuaternaryScience 4, 95e108.

Ballantyne, C., 2002. The Loch Lomond Readvance on the Isle of Mull, Scotland:glacier reconstruction and palaoclimatic implications. Journal of QuaternaryScience 17, 759e771.

Berger, A.L., 1978. Long-term variations of daily insolation and quaternary climaticchanges. Journal of the Atmospheric Sciences 35 (12), 2362e2367.

Bjune, A.E., Birks, H.J.B., 2008. Holocene vegetation dynamics and inferred climatechanges at Svanå vatnet, Mo i Rana, northern Norway. Boreas 37 (1), 146e156.

Blott, S.J., Pye, K., 2001. Gradistat: a grain size distribution and statistics package forthe analysis of unconsolidated sediments. Earth Surface Processes and Land-forms 26, 1237e1248.

Bond, G., Kromer, B., Beer, J., Muscheler, R., Evans, M.N., Showers, W., Hoffmann, S.,Lotti-Bond, R., Hajdas, I., Bonani, G., 2001. Persistent solar influence on NorthAtlantic climate during the Holocene. Science 294, 2130e2136.

Briner, J.P., Thompson Davis, P., Miller, G.H., 2009. Latest Pleistocene and Holoceneglaciation of Baffin Island, Arctic Canada: key patterns and chronologies.Quaternary Science Reviews 28 (21e22), 2075e2087.

Bronk, R.C., 2001. Development of the radiocarbon program OxCal. Radiocarbon 43(2A), 355e363.

Dahl, S.O., Bakke, J., Lie, O., Nesje, A., 2003. Reconstruction of former glacier equi-librium-line altitudes based on proglacial sites: an evaluation of approachesand selection of sites. Quaternary Science Reviews 22 (2e4), 275e287.

Dahl, S.O., Nesje, A., 1992. Palaeoclimatic implications based on equilibrium-linealtitude depressions of reconstructed Younger Dryas and Holocene cirqueglaciers in inner Nordfjord, Western Norway. Palaeogeography, Palae-oclimatology, Palaeoecology 94 (1e4), 87e97.

Dahl, S.O., Nesje, A., 1996. A new approach to calculating Holocene winter precip-itation by combining glacier equilibrium-line altitudes and pine-tree limits:a case study from Hardangerjøkulen, central southern Norway. Holocene 6 (4),381e398.

Dean, W.E., 1974. Determination of carbonate and organic matter in calcareoussediments and sedimentary rocks by loss on ignition: comparison with othermethods. Journal of Sedimentary Petrology 44, 242e258.

Fosslie, S., 1941. Tysfjords geologi. Beskrivelse til det geologiske gradteigskartTysfjord. Norges geologiske undersøkelser, NGU, Oslo.

Furbish, D.J., Andrews, J.T., 1984. The use of hypsometry to indicate long-termstability and response of valley glaciers to changes in mass transfer. Journal ofGlaciology 30, 199e211.

Geirsdóttir, Á., Miller, G.H., Axford, Y., Ólafsdóttir, S., 2009. Holocene and latestPleistocene climate and glacier fluctuations in Iceland. Quaternary ScienceReviews 28 (21e22), 2107e2118.

Griffey, N.J., Worsley, P., 1978. The pattern of neoglacial glacier variations in theOkstindan region of northern Norway during the last three millennia. Boreas 7(1), 1e17.

Grove, J.M., 1988. The Little Ice Age. Methuen, London.Hallet, B., Hunter, L., Bogen, J., 1996. Rates of erosion and sediment evacuation by

glaciers: a review of field data and their implications. Global and PlanetaryChange 12 (1e4), 213e235.

Hoel, A., 1910. Okstinderne e fjeldgrunden og bræerne. Norges geologiske under-søkelser 57 (2).

Holzhauser, H., Magny, M., Zumbuhl, H.J., 2005. Glacier and lake-level variations inwest-central Europe over the last 3500 years. Holocene 15 (6), 789e801.

Humlum, O., Elberling, B., Hormes, A., Fjordheim, K., Hansen, O.H., Heinemeier, J.,2005. Late-Holocene glacier growth in Svalbard, documented by subglacialrelict vegetation and living soil microbes. Holocene 15 (3), 396e407.

Haakensen, N., 1989. Akkumulasjon på breene i Norge vinteren 1988e89. Været 13,91e94.

Innes, J.L., 1986. Dating exposed rock surfaces in the Arctic by lichenometry e theproblem of thallus circularity and its effect on measurement errors. Arctic 39(3), 253e259.

Ivy-Ochs, S., Kerschner, H., Maisch, M., Christl, M., Kubik, P.W., Schlüchter, C., 2009.Latest Pleistocene and Holocene glacier variations in the European Alps.Quaternary Science Reviews 28 (21e22), 2137e2149.

Jansen, E., deMenocal, P., Grousset, F., 2004. Holocene climate variability e a marineperspective. Quaternary Science Review 23/20, 2061e2268 (Special Issue)Elsevier Ltd, Oxford.

Jansen, E., Overpeck, J., Briffa, K.R., Duplessy, J.-C., Joos, F., Masson-Delmotte, V.,Olago, D., Otto-Bliesner, B., Peltier, W.R., Rahmstorf, S., Ramesh, R., Raynaud, D.,Rind, D., Solomina, O., Villalba, R., Zhang, D., 2007. Palaeoclimate. ClimateChange 2007: the Physical Science Basis. Contribution of Working Group I to theFourth Assessment Report of the Intergovernmental Panel on Climate Change.Cambridge University Press, Cambridge, United Kingdom and New York, NY,USA.

Jansson, P., Rosqvist, G., Schneider, T., 2005. Glacier fluctuations, suspendedsediment flux and glacio-lacustrine sediments. Geografiska Annaler 87A (1),37e50.

Karlén, W., 1976. Lacustrine sediments and tree-line variations as indikators ofclimatic fluctuations in Lappland, northern Sweden. Geografiska Annaler 58A,1e34.

Kirkbride, M.R., Dugmore, A.J., 2008. Two millennia of glacier advances fromsouthern Iceland dated by tephrochronology. Quaternary Research 70 (3),398e411.

Knudsen, N.T., Theakstone, W.H., 1988. Drainage of the Austre-Okstindbreen Ice-Dammed Lake, Okstindan, Norway. Journal of Glaciology 34 (116), 87e94.

König, M., Winther, J.G., Knudsen, N.T., Guneriussen, T., 2000. Equilibrium- and firn-line detection with multi-polarization SAR e first results. In: Proceedings ofEARSeL-SIG-Workshop Land Ice and Snow. Dresden/FRG, pp. 273e280.

Lie, O., Dahl, S.O., Nesje, A., 2003. A theoretical approach to glacier equilibrium-linealtitudes using meteorological data and glacier mass-balance records fromsouthern Norway. Holocene 13 (3), 365e372.

Lie, O., Dahl, S.O., Nesje, A., Matthews, J.A., Sandvold, S., 2004. Holocene fluctuationsof a polythermal glacier in high-alpine eastern Jotunheimen, central-southernNorway. Quaternary Science Reviews 23 (18e19), 1925e1945.

Linge, H., Olsen, L., Brook, E.J., Darter, J.R., Mickelson, D.M., Raisbeck, G.M., Yiou, F.,2007. Cosmogenic nuclide surface exposure ages from Nordland, northernNorway: implications for deglaciation in a coast to inland transect. NorwegianJournal of Geology 87, 269e280.

Matthews, J.A., Karlen, W., 1992. Asynchronous neoglaciation and HOLOCENEclimatic-change reconstructed from Norwegian glaciolacustrine sedimentarysequences. Geology 20 (11), 991e994.

Mckay, N.P., Kaufman, D.S., 2009. Holocene climate and glacier variability at Halletand Greyling Lakes, Chugach Mountains, south-central Alaska. Journal ofPaleolimnology 41 (1), 143e159.

Menounos, B., Clague, J.J., Osborn, G., Luckman, B.H., Lakeman, T.R., Minkus, R.,2008. Western Canadian glaciers advance in concert with climate change circa4.2 ka. Geophysical Research Letters 35 (7) Article number: L07501.

Menounos, B., Osborn, G., Clague, J.J., Luckman, B.H., 2009. Latest Pleistocene andHolocene glacier fluctuations in western Canada. Quaternary Science Reviews28 (21e22), 2049e2074.

Meteorologisk Institutt, 2007. eKlima.met.no, Oslo.Nesje, A., 1992. A Piston Corer for Lacustrine and Marine-Sediments. Arctic and

Alpine Research 24 (3), 257e259.Nesje, A., 2009. Latest Pleistocene and Holocene alpine glacier fluctuations in

Scandinavia. Quaternary Science Reviews 28 (21e22), 2119e2136.Nesje, A., Bakke, J., Dahl, S.O., Lie, O., Matthews, J.A., 2008. Norwegian mountain

glaciers in the past, present and future. Global and Planetary Change 60 (1e2),10e27.

Nesje, A., Dahl, S.O., Løvlie, R., 1995. Late Holocene glacier and avalanche activity inthe Ålfotbreen area, western Norway: evidence from a lacustrine sedimentaryrecord. Norsk Geologisk Tidsskrift 75, 120e126.

Nesje, A., Kvamme, M., Rye, N., Løvlie, R., 1991. Holocene glacial and climate historyof the Jostedalsbreen region, western Norway; evidence from lake sedimentsand terrestrial deposits. Quaternary Science Reviews 10, 87e114.

J. Bakke et al. / Quaternary Science Reviews 29 (2010) 1246e12621262

Nesje, A., Matthews, J.A., Dahl, S.O., Berrisford, M.S., Andersson, C., 2001. Holoceneglacier fluctuations of Flatebreen and winter-precipitation changes in the Jos-tedalsbreen region, western Norway: evidence from pro-glacial lacustrinesediment records. Holocene 11 (3), 267e280.

Osborn, G., Menounos, B., Koch, J., Clague, J.J., Vallis, V., 2007. Multi-proxy record ofHolocene glacial history of the Spearhead and Fitzsimmons ranges, southernCoast Mountains, British Columbia. Quaternary Science Reviews 26 (3e4),479e493.

Osmaston, H., 2005. Estimates of glacier equilibrium line altitudes by the Area -�Altitude, the Area�Altitude Balance Ratio and the Area�Altitude BalanceIndex methods and their validation. Quaternary International 138, 22e31.

Owen, L.A., 2009. Latest Pleistocene and Holocene glacier fluctuations in theHimalaya and Tibet. Quaternary Science Reviews 28 (21e22), 2150e2164.

Østrem, G., Liestøl, O., 1964. Glasiologiske undersøkelser i Norge i 1963. NorskGeografisk Tidsskrift, Bd. 18, h7e8, 281e340.

Porter, S.C., 1975. Equilibrium-line altitudes of late Quaternary glaciers in theSouthern Alps, New Zeland. Quaternary Research 5, 27e47.

Paasche, O., Dahl, S.O., Bakke, J., Lovlie, R., Nesje, A., 2007. Cirque glacier activity inarctic Norwayduring the last deglaciation. Quaternary Research 68 (3), 387e399.

Paasche, Ø, Bakke, J., 2009. The Holocene Turnover e a Global Climatic Shift at w4ka. EGU, Vienna. 2009e14017.

Paasche, Ø., Bakke, J. Defining the “Little Ice Age”. Proceedings of the NationalAcademy of Sciences, submitted for publication.

Rea, B.R., 2009. Defining modern day areaealtitude balance ratios (AABRs) and theiruse in glacier-climate reconstructions. Quaternary Science Reviews 28 (3-4),237e248.

Richter, T.O., Van der Gaast, S., Koster, B., Vaars, A., Gieles, R., De Stigter, H.C., DeHaas, H., Van Weering, T.C.E., 2006. The Avaatech XRF core scanner: technicaldescription and applications to NE Atlantic sediments. In: Rothwell, R.G. (Ed.),New Techniques in Sediment Core Analysis. Geological Society, London, pp.39e50.

Rosqvist, G., Jonsson, C., Yam, R., Karlen, W., Shemesh, A., 2004. Diatom oxygenisotopes in pro-glacial lake sediments from northern Sweden: a 5000 yearrecord of atmospheric circulation. Quaternary Science Reviews 23 (7-8),851e859.

Rubensdotter, L., 2002. Detailed geomorphological survey of a small mountaindrainage area, Abisko, northern Swedish Lapland. Geografiska Annaler Series84A (3e4), 267e273.

Rubensdotter, L., Rosqvist, G., 2003. The effect of geomorphological setting onHolocene lake sediment variability, northern Swedish Lapland. Journal ofQuaternary Science 18 (8), 757e767.

Rubensdotter, L., Rosqvist, G., 2009. Influence of geomorphological setting, fluvial-,glaciofluvial- and mass-movement processes on sedimentation in alpine lakes.Holocene 19 (4), 665e678.

Schaefer, J.M., Denton, G.H., Kaplan, M., Putnam, A., Finkel, R.C., Barrell, D.J.A.,Andersen, B.G., Schwartz, R., Mackintosh, A., Chinn, T., Schluchter, C., 2009.High-Frequency Holocene glacier fluctuations in New Zealand differ from theNorthern Signature. Science 324 (5927), 622e625.

Snowball, I.F., 1996. Holocene environmental change in the Abisko region ofnorthern Sweden recorded by the mineral magnetic stratigraphy of lake sedi-ments. GFF 118, 9e17.

Snowball, I.F., Sandgren, P., 1996. Lake sediment studies of Holocene glacial activityin the Kårsa valley, northern Sweden: contrasts in interpretation. The Holocene6, 367e372.

Sutherland, D., 1984. Modern glacier characteristics as a basis for inferring formerclimates with particular reference to the Loch Lomond stadial. QuaternaryScience Reviews 3, 291e309.

Thornalley, D.J.R., Elderfield, H., McCave, N.I., 2009. Holocene oscillations intemperature and salinity of the sufrace subpolar North Atlantic. Nature 457(7230), 711e714.

Wanner, H., Beer, J., Bütikofer, J., Crowley, T.J., Cubasch, U., Flückiger, J., Goosse, H.,Grosjean, M., Joos, J., Kaplan, J.O., Küttel, M., Müller, S.A., Prentice, C.,Solomina, O., Stocker, T.F., Tarasov, P., Wagner, M., Widmannm, M., 2008. Mid-to Late Holocene climate change: an overview. Quaternary Science Reviews 27(19e20), 1791e1828.

Webb, P.W., Orr, C., 1997. Analytical Methods in Fine Particle Technology. Micro-meritics Instrument Corporation, Norcross, GS, USA.

Wiles, G.C., Barclay, D.J., Calkin, P.E., Lowell, T.V., 2008. Century to millennial-scaletemperature variations for the last two thousand years indicated from glacialgeologic records of Southern Alaska. Global and Planetary Change 60 (1e2),115e125.

Winkler, S., 2003. A new interpretation of the date of the ‘Little Ice Age’ glaciermaximum at Svartisen and Okstindan, northern Norway. The Holocene 13 (1),83e95.

Worsley, P., 1974. Recent “annual” moraine ridges at Austre Okstindbreen, Okstin-dan, north Norway. Journal of Glaciology 13 (68), 265e277.