Geomorphological evidence towards a de-glacial control on volcanism

EARTH SURFACE PROCESSES AND LANDFORMS Earth Surf. Process. Landforms 34, 1164–1178 (2009) Copyright © 2009 John Wiley & Sons, Ltd. Published online 8 April 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/esp.1811

Geomorphological evidence towards a de-glacial control on volcanism Chichester, ESP EARTH The 1096-9837 0197-9337 Earth ESP1811 9999 Research Copyright John 2006 Journal Wiley Science Surface Surf. SURFACE Article Articles © Process. & UK of 2006 Sons, Processes thePROCESSES John British Ltd. Landforms Wiley and Geomorphological Landforms AND & Sons, LANDFORMS Ltd. Research Group

Geomorphological evidence towards a de-glacial control on volcanism

Jonathan L. Carrivick,1* Andrew J. Russell,2 E. Lucy Rushmer,3 Fiona S. Tweed,4 Phillip M. Marren,5 Hugh Deeming6 and Oliver J. Lowe7 1 School of Geography, University of Leeds, Leeds, UK 2 School of Geography, Politics and Sociology, University of Newcastle, Newcastle Upon Tyne, UK 3 Jacobs, Leeds, UK 4 Geography Department, Staffordshire University, Stoke on Trent, UK 5 Department of Forest Resources Management, University of British Columbia, Vancouver, BC, Canada 6 Department of Geography, Lancaster University, Lancaster, UK 7 Environment Agency, Sentinel House, Lichfield, UK Received 16 July 2008; Revised 29 January 2009; Accepted 9 February 2009 * Correspondence to: Jonathan L. Carrivick, School of Geography, University of Leeds, Leeds, West Yorkshire, LS2 9JT. UK. E-mail: [email protected]

ABSTRACT: A number of theoretical, conceptual and numerical models exist for de-glacial controls on volcanism, but geological evidence is scarce. We describe and explain a regional topographic and geomorphic expression of sub-glacial volcanism, namely that at Kverkfjöll, Iceland. This area comprises a series of parallel sub-glacially-erupted volcanic edifices, which together give excellent 3D geological exposure. These ridges are orientated north-south or along the most probable line of LGM ice margin retreat. We combine topographic and geomorphic observations to explain the changing style of volcanism in space, and attribute this to the recession and downwasting of the LGM ice sheet. Specifically, we observe that there is no spatial pattern to pillow lava edifice heights or volumes, indicating that fissure eruptions beneath the LGM ice sheet were of similar dynamics in space. However, hyaloclastite and hyalotuff deposits are restricted to proximal and high elevation positions. Furthermore, lithofacies are split by erosional contacts and hyalotuffs are faulted and intruded. These observations together suggest that as the overlying ice sheet became thinner volcanic activity became more explosive. Volcanic activity also appears to have ‘retreated’ towards the Kverkfjöll central volcano. Glacial outburst floods or ‘jökulhlaups’ were prevalent during deglaciation, partly because sub-glacial meltwater could not be impounded due to the high gradient bedrock topography. For northern Iceland, our proposed sequence of landscape development suggests two major glacial advances during the Holocene, at least one of which at Kverkfjöll probably coincided with volcanic activity and a jökulhlaup. Future work should look to establish an absolute dating control and/or chronology. Copyright © 2009 John Wiley & Sons, Ltd. KEYWORDS: sub-glacial volcanism; volcanic lithofacies; sediment-landform associations; ice-volcano interactions; jökulhlaups; glacial outburst floods; landscape development

Introduction A prescient paper by Hall (1982) in Earth Surface Processes and Landforms first suggested that rapid de-glaciation could stimulate volcanic activity. Twenty-five years later, this concept is still worthy of considerable attention, not least because a growing range of interdisciplinary studies are exploring the topographic expression of coupling between endogenic and exogenic processes, but also because of the obvious consequences of enhanced volcanism due to rapid ice-mass loss in modern glaciated volcanic regions such as Antarctica, Alaska and Patagonia (e.g. Smellie, 2002; Larsen et al., 2005). As wasting ice sheets and ice caps unload the solid Earth, stress releases both deform the Earths surface (e.g. Pagli et al., 2007), and decompress the Earth’s mantle (Sigvaldason et al.,

1992). This combined flexure of the Earths crust and stress changes within magma chambers is likely to stimulate volcanic activity, which in Iceland has been suggested to have been up to several orders of magnitude greater during de-glaciation from the Last Glacial Maximum (LGM) than at present (Jull and McKenzie, 1996). This peak in volcanism apparently finished less than 2000 years after the end of de-glaciation (Maclennan et al., 2002). In northern Iceland, volcanism was ~20–30 times greater during de-glaciation (Slater et al., 1998; Maclennan et al., 2002). Since magma chambers exist beneath both central volcanoes and the most active fissure swarms tend to be those associated with central volcanoes (e.g. Björnsson, 1985; Guðmundsson, 2000), we consider that a de-glacial control on volcanism could exist for extensional tectonic landscapes as well as for glaciated volcanoes.

GEOMORPHOLOGICAL EVIDENCE TOWARDS A DE-GLACIAL CONTROL ON VOLCANISM

1165

Whilst a number of theoretical/conceptual and numerical models exist for such de-glacial controls on volcanism (e.g. Jull and McKenzie, 1996; Höskuldsson and Sparks, 1997; Maclennan et al., 2002; Pagli and Sigmundsson, 2008), geological evidence is scarce. In the southwest of Iceland Vilmundardóttir and Larsen (1986) have inferred a change in eruptive productivity over time from geochemical observations in the Veiδivötn region. In central Iceland Sigvaldason et al. (1992) have presented geochemical data on the post-glacial volcanic production rate of the Dyngjufjöll area. Aside from these studies, inferences from geological data used to reconstruct sub-glacial eruptive style and environment have been largely restricted to observations on individual edifices and isolated outcrops (e.g. Lescinsky and Fink, 2000; Tuffen et al., 2001; Skilling, 2002; Loughlin, 2002; Mee et al., 2006; Höskuldsson et al., 2006; Smellie, 2002, 2006, 2008). If models of a de-glacial control on volcanism are to be tested, there first needs to be a comprehensive regional assessment of the character of sub-glacial volcanic products. This is because these eruptive products are controlled by local ice/water, overburden pressure and magma chemistry (e.g. Wilson and Head, 2002). They would therefore be expected to systematically change in space and time as de-glaciation proceeded. The aim of this paper is therefore to describe and explain a regional topographic and geomorphic expression of sub-glacial volcanism. We place this study aside from previous descriptions of volcaniclastic sequences in Iceland (e.g. Lescinsky and Fink, 2000; Tuffen et al., 2001; Skilling, 2002; Loughlin, 2002) because we are considering a much larger area and because we are specifically interested in geomorphic evidence of a de-glacial control on volcanism.

Study Area Kverkfjöll is in central Iceland on the northern margin of Vatnajökull and is the southernmost and glaciated part of the Kverkfjöll Volcanic System (KVS) (Figure 1). Kverkfjöll rises to 1920 m above sea level (a.s.l.) and comprises two sub-glacial calderas, a supra-glacial lake and a sub-glacial lake (Björnsson and Einarsson, 1991; Björnsson, 2002). Kverkjökull breaches the rim of the most northerly of these calderas and descends to ~900 m a.s.l within 6 km (Figure 1). The northern pro-glacial part of the KVS is known as Kverkfjallarani (Figure 1). We specifically target Kverkfjöll for this study because: (1) It was close to the centre of the LGM ice sheet during the last major glaciation in Iceland (Hubbard, 2006) and because it is above a zone of enhanced magmatic upwelling. The former means that ice flow velocities and hence sub-glacial erosion of volcanic edifices would have been minimal during the LGM, and the latter suggests that Kverkfjöll should be highly volcanically productive. (2) The de-glaciation of central Iceland coincided with a period of enhanced volcanic and geothermal activity (Bourgeois et al., 2000); most likely because crustal stresses were released due to rapid ice sheet wasting. (3) Kverkfjallarani comprises a series of parallel sub-glacially-erupted volcanic edifices, which together give excellent three-dimensional (3D) geological exposure. These ridges are orientated north– south or along the most probable line of LGM ice margin retreat (Hubbard, 2006). (4) Kverkfjöll is the only alpine area in northcentral Iceland and has moraines that are stratigraphically related to sub-glacial volcanic edifices and Holocene lava flows.

Methods Geomorphological and sedimentological field observations in 2002–2004 and 2007 focused on volcanic edifices and Copyright © 2009 John Wiley & Sons, Ltd.

Figure 1. Location and topographic overview of the Kverkfjöll Volcanic System (KVS) within Iceland, highlighting other active volcanic zones by black shaded areas. The hillshaded 10 m grid digital elevation model (DEM) was derived from photogrammetry analysis. The dashed box delimits the extent of a hillshaded 1 m grid DEM derived from a 2007 LiDAR survey. White dashed lines A–F mark sections plotted in Figure 4. Transparent circles denote location of maps in Figure 3.

particularly on volcanic lithofacies descriptions (Table I) and follow the codes of Loughlin (2002) and Skilling (2002) (Table I). We distinguish fresh from altered vitric tephra because basaltic tephra glass (sideromelane) is black and when altered through subsequent contact with meteoric water or hydrothermal fluids becomes yellow. We refer to that alteration as palagonitization (Schiffman et al., 2002) and the resultant material as either palagonitized glass; the hydration and alteration product of sideromelane, or palagonitized tuff; a consolidated rock Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

1166 Table I.

EARTH SURFACE PROCESSES AND LANDFORMS Volcaniclastic lithofacies descriptions and interpretations, based on Loughlin (2002) and Skilling (2002) Description

Interpretation

Volcaniclastic lithofacies assemblages Lc

Compound flows on valley floors with a massive or thick (>50 cm) columnar structure, commonly with reddened surfaces.

Subaerial (primary) compound basaltic lava, partly from the Biskupsfell fissure (Karhunen, 1988) but also from other unnamed fissures.

Cm

Massive or welded flow mantling slopes and crests. Commonly with agglutinated and welded surface of airfall; scoria, cowpat spatter and ash.

Subaerial (secondary) basaltic clastogenic flow deposits, produced by flowage of airfall ejecta from explosive eruptions (fire fountains) of the Biskupsfell fissure (Karhunen, 1988).

Lt

Simple thick flow with thin (>30 cm) columns with a radial pattern.

Lava infilling a tube, most likely sub-glacial.

Lp

Lobate and pillow basalts, irregularly shaped and radially jointed, some with interstitial hyaloclastite.

Subaqueous lava flow deposits of relatively low effusion rate. Piles formed in ridges erupted into ice sheet (Höskuldsson et al., 2006).

Bhm

Massive, matrix-supported, poorly-sorted, andgular to subangular clasts of basalt pillows within a palagonitized matrix.

Hyaloclastite massive breccia. Quenched stacks formed in ridges erupted into ice sheet.

Bhs

Matrix-supported bedded breccia. May be large (>5 m) scale ‘bedforms’. Matrix is more abundant than in Bhm.

Hyaloclastite bedded breccia. Quenched stacks formed in ridges erupted into ice sheet and subsequentlydeformed due to gravity, probably within a water body.

Sl

Sand sized grains in thin beds that mantle shallow topography. Indurated and semi-lithified. Well sorted with some laminae.

Pyroclastic ‘hyalotuff’ deposit derived from buoyant suspension, most likely in a sub-glacial hydrovolcanic environment.

comprising unaltered glass. We adhere to the terminology of Lescinsky and Fink (2000) for descriptions of ice-contact lava morphology. Throughout this paper volcanic lithofacies are those detailed in Table I, and facies associations are those in Table II. Further geomorphological characterization and topography was obtained via (a) georeferenced 1:30 000 aerial photographs which were obtained in 1987 by Landmælingar Islands, (b) a digital elevation model (DEM) produced using aerial photogrammetry (Carrivick and Twigg, 2004; Figure 1), and (c) georeferenced airborne LiDAR data; a hillshaded image of which is given in (Figure 1). Georeferencing firstly utilized differential Global Positioning System (dGPS) measurements of Ground Control Points (GCPs) across the study area as documented by Carrivick and Twigg (2004). Secondly, occupation of an arbitrary static control point with a dGPS ‘base’ receiver was maintained during the LiDAR overflight in August 2007. Both GCPs and the arbitrary static control point were precisely located with reference to permanent Icelandic geodetic dGPS receivers; Karahnjúkar, Höfn. The LiDAR ‘point cloud’ was interpolated using Inverse Distance Weighting (IDW) onto a regular grid. The resulting unprecedented topographic dataset for this part of Iceland has 2 m horizontal resolution and ±10 cm vertical accuracy. On the basis of our geomorphological and sedimentological field observations, aerial photograph interpretations and high-resolution DEM, we then made a semi-automated landform and surface-type extraction analysis (Table II). The automated; geographic information system (GIS), part of this method primarily used elevation and slope, the latter being calculated as the maximum gradient (rate of change) between adjacent grid cell elevations. However, also of use were surface models based on a localized ‘neighbourhood analysis’ of 25 m × 25 m to calculate local relief and surface texture/roughness. Relief defines the range of elevation values within a neighbourhood. Roughness was calculated as the neighbourhood variance within the LiDAR intensity value, although here a 5 m × 5 m neighbourhood was used. Expert judgment was used to assess the results of these automated calculations in terms of association, location and geometry. Copyright © 2009 John Wiley & Sons, Ltd.

Description of Volcanic Facies In order to understand precisely the ice–volcano interactions and thus the eruptive history represented by the landforms and sediments of Kverkfjallarani and the ice-marginal subregions, it is necessary to make a detailed examination of the constituent volcanic lithofacies assemblages (Table I). However, where pertinent we will also make a description and interpretation of sedimentary facies associations; as summarized in Table II, for these will be shown to give a wider context to understanding landscape development in the region. Subaerial lava flow deposits at Kverkfjöll are compound, have an aa surface (Figure 2A), and are rarely more than 2 m thick. The most obvious are the Kverkfoallahraun, the Lindahraun and those within Hraundalur. Flow distances range from a few hundred metres to 4–6 km. Two types of subaerial basaltic lava flow deposits can be recognized. These are older, grey lava and younger black ‘pristine’ lava. The former type originates from fissures, most notably the Biskupsfell fissure (Figure 2B) and also from some small spatter and scoria cone rows on adjacent fissures (Carrivick and Twigg, 2004). They are notable for two reasons. First, they exhibit a basal zone comprising breccia up to 0·75 m thick. This breccia is of angular, flowbanded and glassy basalt clasts of variable size, which are occasionally set in a matrix (millimetres to sub-millimetres) of lava fragments and ash. Larger clasts (10–30 cm) define jigsawfit domains. Breccias are further distinguished by welding textures and oxidation reddening. Secondly, several subaerial lava flow tops have been entirely stripped; clinker and airfall, scoria and cowpat spatter – agglutinate present on neighbouring lava flows is missing. Furthermore, the underlying massive basalt blocks are plucked, scoured and sculpted (Carrivick et al., 2004a; Carrivick, 2007a). ‘Rheomorphic’ clastogenic flow deposits are restricted the vicinity of the Biskupsfell fissure. They mantle ridge crests and cover some ridge sides (Karhunen, 1988; Carrivick and Twigg, 2004; Figure 2B) and are composed of dark, very finegrained, almost glassy material. Flow banding is common and secondary flow structures such as folding and rotated lithic fragments are present. Remnants of original clasts; cowpat Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

GEOMORPHOLOGICAL EVIDENCE TOWARDS A DE-GLACIAL CONTROL ON VOLCANISM Table II.

1167

Facies associations used to produce maps in Figure 3

Facies associations (Abundance)

Lithofacies

1a (4%)

(1a) Lc

1b (1%)

(1b) Cm

2a (25%)

(2a) Lt, Lp,

2b (9%)

(2b) Bhm, Bhs

2c (—)

(2c) Sm

3 (16%)

bG, Dm

4 (5%)

bG, Dm

5 (15%)

Gm, Gs, Gh

6 (5%)

Sg, Gp,

7 (0·4%)

bG

8 (0·4%)

Gp, Sm, Sh

9 (9%)

bG

10 (10%)

bG

11 (0·2%)

Sm, Sh

Landform character (predominantly from DEM-based analyses)

Landform interpretation

a) Valley floor near-horizontal in east–west direction, and in north–south direction slope length >1 km at a gradient 1°–5°. b) This could not be discriminated using GIS because it is superimposed on pillow lava ridges. Required ground-based observations of facies. a) Upstanding ridges of >75 m height above surrounding valley floor. Streamlined shape (length/ width ratio >6). b) This could not be discriminated using GIS because it is superimposed on pillow lava ridges. Required ground-based observations of facies. c) Most clearly distinguished by surface roughness contrast with surrounding moraine hummocks. Complex minor ridges and furrows; ‘drainage density’ (elevation difference in neighbouring cells gives drainage direction and continuous decline in elevation gives drainage length) > 50 m/25 m × 25 m window. Local relief (maximum – minimum elevation within a 25 m × 25 m neighbourhood) > 20 m. Could not be discriminated using GIS because same as 3. Required ground-based observations of spatial association and facies. Laterally extensive area of uniform gradient average 0·5°. Reguired subjective interpretation of situation relative to glacier to discriminate from 1a. Could not be discriminated using GIS because same as 5, although associated with contemporary surface drainage. Arcuate and curving transverse ridges, lobes and mounds, located at the foot of high-gradient free faces. Discriminated subjectively using hillshaded 1 m DEM. Could not be discriminated using GIS because superficaially very similar to 8. Required ground-based observations of facies. Subdued mounds and surfaces of highly fragmented clasts, occasionally with shallow depressions infilled by slopewash deposits. Only discriminated in field. Subdued mounds and surfaces of pillow basalts clasts and occasional hyaloclastite boulder-sized clasts. Only discriminated in field. Topographically enclosed basins with horizontal floors

Subaerial (a) effusive flows, and (b) explosive airfall, volcanic deposits (Holocene)

Sub-glacial fissure eruptions of (a) lava and (b) rapidly quenched volcanic material. (c) Formed from explosive water–volcano interaction and resultant gravitydriven plume (LGM) Ice-cored moraine (stagnant melt-out till) (e.g. Kjær and Krüger, 2001)

Contemporary moraine (active)

Palaeo outwash plain, dominated by hyperconcentrated deposits Contemporary glacifluvial deposits Debris-covered/rock glacier (active) Debris flow/slump deposits (e.g. Costa, 1988; Pierson, 1995) Highly weathered surface, predominantly of subaerial volcanic material Highly weathered surface, predominantly of subaqueous volcanic material Palaeo or contemporary lake

Note: Areal abundance is given in brackets in column one. Volcaniclastic lithofacies are explained in Table I. Geographic information system (GIS) analysis for semi-automated classification is outlined in column three. Sedimentary lithofacies follow those used by Carrivick (2005).

spatter fragments or of a few fragments fused together frequently appear in the dark matrix (Table II). In comparison to other lava flows, vesicles in clastogenic flow deposits are highly reduced in size and abundance and are stretched and flattened. Scoria – cowpat spatter are the dominant pyroclasts present on the uppermost part of the deposit. Lava with well-developed radially-oriented entablature jointing (Table II, Figure 2C) occurs in several of the Kverkfjallarani ridges near to Karlafjall which is ~6 km north of Biskupsfell. However, the vast majority of the Kverkfjallarani ridges are composed of pillow basalts. Most ridges comprise a single pillow basalt unit of tens to hundreds of metres thick, but some closer to the central long axis of the KVS clearly have several units that are separated by a clear near-horizontal discontinuity (Figure 2D). Pillow basalts form entirely nested sets (Figure 2D) or occur with inter-pillow hyaloclastite breccia. Inter-pillow breccia is generally vesicular. Many pillow cores in Kverkfjöll typically have 40–60% vesicles (Höskuldsson et al., 2006). Individual pillows are defined by either classic radial joints, or by irregular and blocky and occasionally concentric joints typically of 5 to 10 cm spacing. Pillow lobes at edifice margins Copyright © 2009 John Wiley & Sons, Ltd.

can have an elongated form and an apparent dip of up to 2°. These elongate lobes also have glassy margins 3–6 cm thick, which suggests spalling of glassy rims during downslope movement (e.g. Skilling, 2002; Smellie, 2006, 2008). Good examples are at the head of Hraundalur. Hyaloclastite (Table II) at Kverkfjöll is situated within the uppermost parts of the Kverkfjallarani ridges closest to, and including the massif of, Kverkfjöll, i.e. within 3 km of the present ice margin; e.g. Borgarfell, Tvihyrna, Þverfell. Hyaloclastite is always separated from the underlying pillow edifice by an erosional contact. Hyaloclastite comprises pillow basalt fragments, typically 2–10 cm diameter, in a yellowish matrix of finer-grained palagonitized glassy material (Figure 2E). We do not imply any particular fragmentation process by the term hyaloclastite, although hyaloclastites are commonly formed within relatively high-pressure water bodies (i.e. restricted vesiculation) by cooling-contraction granulation in combination with mechanical fracturing processes (e.g. Smellie, 2008). Most (~70%) hyaloclastite at Kverkfjöll is massive (lithofacies Bhm), but some hyaloclastite breccia has crudely bedded texture with subparallel alignment of clasts without glassy Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

1168

EARTH SURFACE PROCESSES AND LANDFORMS

Figure 2. Visual example of major volcanic facies associations at Kverkfjöll, as categorized in Table II. (A) Subaerial lava; note the brecciated base overlying a red-coloured tuff, and a more massive blocky upper part, (B) clastogenic flow deposits, (C) subaqueous basalt columns, (D) pillow basalts, (E) hyaloclastite, (F) hyalotuff. This figure is available in colour online at www.interscience.wiley.com/journal/espl

rims (lithofacies Bhs; Figure 2E). This bedded hyaloclastite breccia also has crude folds with concentrations of coarser clasts in fold cores. Hyalotuff is only found in two situations at Kverkfjöll. Firstly it is exposed in valley heads (Figure 2F), i.e. within 3 km of the present ice margin. Whereas hyaloclastite is upon or within ridges, hyalotuff is on exposed on valley floors. Secondly it dominates some parts of the upper walls of the northern Kverkfjöll caldera, particularly around Hveradalur, where it is predominantly of sand grade well-bedded palagonitized glass. On the eastern wall of Kverkfjöll hyalotuff is interspersed with lithic-dominated tuff and coarse hyaloclastite. Hyalotuff is a product of explosive fragmentation of volcanic glass in lowpressure (shallow) water and comprises indurated and semilithified, or palagonitized, sideromelene sand-sized grains in thin well-stratified beds and laminae (Figure 2F). Occasional lithic basalt fragments of up to 1 cm occur in these beds. Beds and laminae are distinguished by subtle grain size variations and all grains are well-rounded fine-grained palagonitized glass. We observe some hyalotuff deposits in valley heads that are cut by many minor faults and occasionally have very thin (~15 cm) basaltic dyke intrusions (Figure 2F). In all cases hyalotuff has a clear erosional contact between adjacent lithfacies. The variability in hyalotuff grain size and morphology Copyright © 2009 John Wiley & Sons, Ltd.

is possibly due to changes in the magma-water ratio, where phreatomagmatic eruptions produce fine grained and poorly sorted material because the magma-water ratio is high (Carey et al., 2007).

Interpretation of Volcanic Facies Subaerial lava flow deposits, clastogenic flow deposits and patches of airfall; scoria and cowpat spatter must have occurred during the Holocene, which in Iceland began ~13 000 BP (Guðmundsson, 1997) . However, the base of some subaerial lava flow deposits in the higher elvation valley heads are brecciated (Figure 2A), which are interpreted as snow/ ice-contact features formed when steam generated from intense melting entered tensional fractures at the flow base (White et al., 2000). In essence the geometry of the intersecting fractures is similar to ‘pseudopillow fractures’ discussed by Lescinsky and Fink (2000). In contrast, interiors and upper parts of these flows have auto-brecciated and blocky textures typical of subaerial conditions (Figure 2A), due to insulation by the underlying mobile lava. Furthermore, Carrivick et al. (2004b) document an example of subaerial lava that flowed over wet jökulhlaup sediments, and infer that this volcanic Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

GEOMORPHOLOGICAL EVIDENCE TOWARDS A DE-GLACIAL CONTROL ON VOLCANISM

and jökulhlaup activity was essentially simultaneous due to the presence of peperitic texture at the lava flow base. Clastogenic flow deposits (Cm) are the result of spatter and airfall-derived pyroclastic material that undergoes gravity-induced ductile shearing and flow, during and immediately after deposition. Pillow basalt (Lp) – hyaloclastite (Bhm/Bhs) ridges of Kverkfjöll formed as a consequence of fissure eruptions into the LGM ice sheet ( Jóhannesson and Sæmundsson, 1989). The steep edifice sides are due to banking against former ice walls. Lava with well-developed radially-oriented entablature jointing [lithofacies (Lt); Figure 2C] is also interpreted to have erupted into a relatively thick overlying ice sheet, specifically into a series of linear ice cavities, or ‘tubes’ (e.g. Mee et al., 2006). Such tubes must have been open to atmospheric pressure and not water filled, perhaps similar in character to the modern Volga Ishellið (‘ice cave’) at the snout of Kverkfjöll. Thus the ice sheet thickness was relatively thin but lava would have to have cooled slowly to form such lengthy columns (~4–6 m; Figure 2C). In contrast, lava erupting into water-filled ice cavities beneath a thick ice sheet would probably been pressurized by the confined water and steam, and thus would have formed pillow basalts (Figure 2D) and hyaloclastite (Figure 2E), depending largely on whether the eruption was effusive or explosive, respectively, and the cooling environment (Wilson and Head, 2002; Smellie, 2008). Based on water contents of the slightly vesiculated glassy rims of some pillow basalts, Höskuldsson et al. (2006) suggest that the ice was 1·2–1·6 km thick at the time of the pillow eruptions. Höskuldsson et al. (2006) also suggest that the nature of pillow vesicles at Kverkfjöll demonstrates a sudden depressurization. They attribute this to rapid water release from the ice cavity during a jökulhlaup, and note that this pressure release would have reactivated local eruptions (Höskuldsson and Sparks, 1997). The palagonitized glass hyaloclastite matrix is further evidence of a phreatomagmatic origin and thus that atmospheric pressure was reached, perhaps via meltwater release. Without recourse to trace element analysis, it is impossible to definitively say whether distinct pillow basalt units (e.g. Figure 2D) are the resultant of several eruptions, or of phases within one eruption such as reported within typical fissure-fed tindar sequences (e.g. Skilling, 2002; Smellie, 2006). However, the fact that hyaloclastite is only found within the uppermost parts of steep-sided ridges closest to, and on Kverkfjöll, indicates a spatial progression of eruptions towards Kverkfjöll, and a changing character in either eruptive style or environment. We consider that hyaloclastite is a product of a separate eruption due to a distinct erosional contact above the underlying pillow lava. Where hyaloclastite has a bedded texture (lithofacies Bhs), we infer that the deposit underwent gravitational flow and mechanical break-up; spalling, probably within water/steam filled ice cavities (Smellie, 2002, 2008; Wilson and Head, 2002). Hyalotuffs are interpreted to be products of hydromagmatic explosions and subsequent pyroclastic surges within a large but relatively low-pressure water body. Similar interpretations of deposits due to pyroclastic surges within unconfined lakes open to the atmosphere have been drawn by Mastin (1997) in a study of Kilauea volcano, Hawaii. Hyalotuff clasts are well rounded and form very thin beds that override shallow topography, which indicates turbulent lateral transport. Due to the present-day lack of any enclosed basins at Kverkfjöll, we suggest that such lakes would have been dammed by relatively thin ice. That such a lake might have drained is suggested by the propensity of faults through hyalotuff beds (Figure 2F) that imply considerable post-depositional movement. Movement would probably be due to gravitational settling and contraction on cooling but also due to changing overburden pressures as Copyright © 2009 John Wiley & Sons, Ltd.

1169

a result of ice melting or water drainage. Subsequent volcanic activity is indicated by the intrusion of thin dykes (Figure 2F) into the hyalotuff.

Ice-marginal Geomorphology We split the ice-marginal part of the study site into three subregions; informally named in this study as ‘Kverkjökull’ (Figure 3A) ‘Trolldalur’ (Figure 3B), and ‘Thorbergsdalur’ (Figure 3C). These areas are far more complex than the rest of Kverkfjallarani (Figures 4–6) and the intercalation of sedimentary and volcanic surfaces is most visible in this region. For each ice-marginal sub-region we: (a) provide a full topographic and geomorphological description; (b) demonstrate that it has a distinct and unique character that reveals specific types of ice-volcano interactions; (c) highlight stratigraphical information that suggests a changing style of ice–volcano interactions and of Holocene ice-marginal landscape development. The Kverkjökull sub-region (Figure 3A) is dominated by the 4 km2 Kverkjökull outwash plain. It is bounded to the northeast by ice-cored moraine of palagonitized material; Facies 4 (Table II). The north-western edge is defined by the Jökulsá á Fjöllum and the Dyngjujökull ice margin. Portions of the Kverkjökull outwash plain can be observed beneath surgerelated moraines of Dyngjujökull. Both the Volga and Skolpa rivers are within incised channels that predominantly comprise lithofacies Gm, Gs, Gh with out-sized hyaloclastite boulders of ~1–2 m in diameter. These sediments are consistent with high-magnitude fluidal sedimentation in the area (Table II; Carrivick, 2005; Rushmer, 2006, 2007; Marren et al., 2009). The Volga channel contains many kettle holes, the largest of which are up to 8 m across and 3 m deep. Kettle holes are spatially restricted to linguoid gravel bars (Rushmer, 2006), the number and size of kettle holes increases downstream (Carrivick, 2005). These measurements indicate ice block deposition due to loss of fluid competence, as opposed to deposition due to stagnation of glacier ice. A second key feature of the Kverkjökull area is the recognition of several ice margin advances from moraines with different compositions. We also notice that these moraines overlie jökulhlaup landforms and sediments both adjacent to the Kverkjökull snout and within the high valleys of Kverkfjallarani (Carrivick et al., 2004a). Moraines adjacent to the Kverkjökull snout probably pertain to a Holocene readvance; most likely to the Little Ice Age (LIA), which in Iceland was late in the nineteenth century (Guðmundsson, 1997). Moraines in the high Kverkfjallarani valleys are most likely related to older ice margin advances, which in northern Iceland occurred at ~5000 and 3000 BP (Guðmundsson, 1997). Importantly moraines in the high valleys comprise sub-glacially-erupted volcanic material, whereas those related to the present-day Kverkjökull snout do not. An arcuate moraine (Facies 4) splits the Kverkjökull fan (Figure 3A). Inside the arc are a series of incised channels that converge through breaches in the ridge. These channels are unrelated to contemporary drainage. Sub-glacially-formed flute-type structures are exposed directly in front of the western part of the Kverkjökull snout, where it has retreated by 40 m over the last six years. This part of the snout is very thin and has previously over-ridden sediments emplaced by jökulhlaups. Jökulhlaups from Kverkfjöll have thus routed through Kverkjökull and exited from the ice margin along the entire contemporary margin. These jökulhlaups have occurred intermittently throughout the Holocene, and with various ice margin configurations. The Kverkjökull ice margin position has been both further advanced and further back than at present and this could be related to climate and/or Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

1170

EARTH SURFACE PROCESSES AND LANDFORMS

Figure 3. Geomorphological maps of the KVS ice-marginal areas: Kverkjökull (A), Trolldalur (B) and Thorbergsdalur (C). These areas are considerably more complex than the rest of Kverkfjallarani, which is represented in Figures 4 and 5.

Copyright © 2009 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

GEOMORPHOLOGICAL EVIDENCE TOWARDS A DE-GLACIAL CONTROL ON VOLCANISM

1171

Figure 4. Topography and spatial distribution of major lithofacies assemblages in Kverkfjallarani. Sections A–F are located in Figure 1. Superficial lithofacies such as clastigenic deposits (1b) and sedimentary associations such as glacifluvial outwash deposits (5 and 6) and have been omitted for clarity.

sub-glacial volcanic eruptions and enhanced meltwater production. Trolldalur (Figure 3B) is notable for a system of gorges that incise through pillow lava onto the Kverkjökull outwash fan. Each of these gorges is approximately 300 m long, 15–20 m wide, 25–35 m deep and has a floor with numerous steps of Copyright © 2009 John Wiley & Sons, Ltd.

several metres in height. Slopes immediately adjacent to the present Dyngjujökull margin and upslope from the gorges form a series of terraces, up to 10 m high and comprising lithofacies Dm, Gm and Gs (Table II). We suggest that all these gorges and terraces are a product of high-magnitude flows from a Dyngjujökull advance marked by the terraces. Trolldalur also Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

1172

EARTH SURFACE PROCESSES AND LANDFORMS

Figure 5. Generalized longitudinal change in volume of major facies associations in Kverkfjallarani with distance from the volcanic centre. Pillow lava volumes were calculated directly from edifice DEMs and hyaloclastite and hyalotuff volumes are from outcrop. Moraine and lava flow volumes were calculated using field-observed average thickness of 12 m and 3 m, respectively. Clastogenic, glacifluvial outwash and snow/ice volume are restricted to areal measurements only.

contains a series of ridges of a uniform height 15–20 m and 30–40 m breadth that are orientated southwest-northeast, i.e. parallel to the present margin of unnamed glaciers and snowfields that descend from slopes on either side of Hveradalur (Figure 3B). Each ridge has been incised by a debris flow, and by small streams emanating from Langafönn (Figure 3B). These reveal that the ridges comprise lithofacies Dm, with some bG and Sm, which we interpret as moraine (Table II). Dyngjujökull, which is known to surge periodically (Björnsson et al., 2003) could have produced the ridges whilst in a position advanced from that of today. Thorbergsdalur is bounded to north by Hveragil; ‘hot spring gorge’, Krokagil; ‘winding gorge’, and Kvislargil; ‘forked gorge’ (Figure 3C). All three gorges incise into pillow basalts. Hveragil contains tufa precipitates in the upper gorge and these are laminated, which could provide seasonal palaeoclimatic records and could also indicate the length of time that this gorge has been stable. All three gorges are up to 40 m deep and 50 m wide. Below 20 m depth the gorges are usually box-shaped, and above 20 m the gorges are usually v-shaped, due to pillow basalt scree ingress. We interpret the Thorbergsdalur gorges to have been excavated by high magnitude meltwater releases due to sub-glacial volcanic eruptions. This is in contrast to the likely subaerial origin of the Trolldalur gorges, or the gorges of Kverkfjallarani (Carrivick, 2005; Carrivick et al., 2004a). We highlight three reasons for our sub-glacial interpretation. (1) Thorbergsdalur gorges are far longer, narrower and deeper than any other gorge in Kverkfjallarani (Carrivick et al., 2004b). Copyright © 2009 John Wiley & Sons, Ltd.

(2) The gorges have a planform that is not related to fissures or topography (Figure 3C) and are clearly unrelated to present hydrological conditions. (3) Depositional surfaces situated above gorge walls are ambiguous in origin, but could be due to subglacial sheetflow. The majority of the remaining area of Thorbergsdalur comprises incoherent ridges hummocks that feature depressions, slumps and tensional surface cracking; i.e. collapse moraine ridges and hummocky topography (Figure 3C). As in Trolldalur, these ridges and hummocks are a result of lowering relief due to gradual thawing of ice-cored ground (e.g. Table II; Kjær and Krüger, 2001). The Thorbergsdalur ice-cored moraine fields can be split into two types; those comprising grey subaerial basalt and those comprising palagonitized material. These different moraines appear to reflect differing provenances because the eastern rim of Kverkfjöll contains distinct zones of palagonitized tuff and hyaloclastite.

Kverkfjallarani Topography and Geomorphology An overview of the topography of Kverkfjallarani is given in Figure 1, which clearly indicates how Kverkfjallarani is dominated by a series of parallel volcanic pillow-hyaloclastite ridges 1–6 km long. These ridges mark fissures that most likely erupted beneath the LGM ice sheet (Jóhannesson and Sæmundsson, 1989) or ~12 000 BP (Guðmundsson, 1997), although there is no absolute age dating. We determine here Earth Surf. Process. Landforms 34, 1164–1178 (2009) DOI: 10.1002/esp

GEOMORPHOLOGICAL EVIDENCE TOWARDS A DE-GLACIAL CONTROL ON VOLCANISM

Figure 6. Areal changes in major facies associations within the KVS as a proportion of each 5 km2 boxed area.

from our topographic data that the height of edifices above local valley floors is similar across the whole of Kverkfjallarani, being in the range 75–125 m (Figures 4A and 4B). There is no spatial pattern to ridge height above valley floors (Figures 4A, 4B, 4D and 4E). All ridges have east- and west-facing slopes of 27º–45º, and north- and south-facing slopes that are much shallower, usually 11º–20º. Ridges are streamlined, presumably by glacial ice although they would also have had steep sides originally due to the eruptive products banking against former ice walls (e.g. Smellie, 2006). Since the retreat of the LGM ice sheet volcanic activity at Kverkfjöll has continued (Table III, Figure 8) with the most notable product being post-glacial subaerial lava flow deposits and clastogenic flow deposits that occupy Kverkfjallarani valley floors and walls, respectively. Thus valley floors such as that of Hraundalur appear relatively flat (Figure 1, Figure 4B). Holocene jökulhlaups have routed along these valleys and have produced distinctive meso-scale landforms and sediments (Carrivick et al., 2004a, 2004b, 2007; Carrivick, 2005, 2006, 2007a, 2007b). A volumetric (Figure 5) and areal (Figure 6) analysis of facies associations clearly indicates how Kverkfjallarani geomorphology generally becomes simpler with increasing distance from Copyright © 2009 John Wiley & Sons, Ltd.

1173

the volcanic centre of Kverkfjöll. Pillow lava volumes were calculated directly from edifice DEMs and hyaloclastite and hyalotuff volumes are from outcrop. A mean volume of pillow lava is 0·035 km3 for a 1 km breadth transect across Kverkfjallarani. Moraine and lava flow volumes were calculated using field-observed average thickness of 12 m and 3 m, respectively. Lava flow volumes reach 0·08 km3 for a 1 km breadth transect. Pillow lava, hyaloclastite, hyalotuff and moraine facies associations are all most abundant at