Reviewed research article

A gravity study of silicic domes in the Krafla area, N-Iceland

Thorbjorg Agustsdottir, Magnús Tumi Gudmundsson and Páll EinarssonInstitute of Earth Sciences, University of Iceland, Sturlugata 7, 101 Reykjavík, Iceland

thorbag@hi.is, mtg@hi.is, palli@raunvis.hi.is

Abstract – Silicic rocks in Iceland are generally associated with central volcanoes and are often emplaced asdomes on or around caldera rims. Some of these domes were formed subglacially while others were eruptedunder ice-free conditions. A gravity survey was carried out in the area of Krafla in 2007 and 2008 to determinethe mean bulk densities of three silicic domes; essential data for meaningful modelling of the emplacementof cryptodomes and lava domes. Such data are scarce. Profiles were measured over three formations: Hlí!-arfjall, made of rhyolite, 310 m high, 2 km long and formed under ice 90 000 years BP, Hrafntinnuhryggur,made of rhyolite, 80 m high, 2.5 km long and formed subglacially 24 000 years BP and Hraunbunga madeof dacite, 125 m high and 1.8 km long, formed under ice-free conditions 10 000 years BP. Mean bulk densityfor each formation was obtained by the Nettleton method. Mean bulk density and volumes obtained were;Hlí!arfjall: 1600-1800 kg m!3, 0.143 ± 0.014 km3; Hrafntinnuhryggur: 1575–1875 kg m!3, 0.021 ± 0.002km3; Hraunbunga: 1750–1775 kg m!3, 0.040 ± 0.004 km3. The results show that all the domes have lowdensities, reflecting both low grain-density and high porosity. The density values are significantly lower thanthose of the surroundings, creating a density contrast possibly sufficient to drive the ascent of silicic magma.Furthermore, results from forward gravity modelling demonstrate that these formations are neither buried byyounger volcanic eruptives nor are any roots detected. The domes studied were therefore emplaced by a dike tothe surface.

INTRODUCTIONRhyolite magma can rise due to buoyancy forces andeither form a cryptodome in the shallow crust or riseto the surface, where it erupts. Due to its high vis-cosity and resistance to flow it often accumulates andforms a lava dome over the vent. In volcanology, adome is defined as an accumulated silicic melt of highviscosity. The term is used to describe any dome,whether it is on the surface or not. A dome can ac-tively deform adjacent strata and break through thesurface. The term coulee is used for a lava domewhere some lateral flow away from the vent has oc-curred (Fink and Anderson, 2000).

Worldwide, domes are associated with viscous,silicic magma. The occurrence of domes could there-fore be expected in areas of silicic volcanism, i.e. sub-duction zones, rather than in areas where basaltic vol-

canism is dominant, i.e. divergent rift zones and hotspots. However, dome formations are known in areasof basaltic volcanism, e.g. in Krafla, Iceland. World-wide, dikes feed many silicic domes (Fink and Ander-son, 2000). These domes tend to occur in groups, thatare either in linear or echelon arrays (Fink and Ander-son, 2000), e.g. in Long Valley, California.

Silicic rocks in oceanic crust are very rare and itseems likely that the oceanic crust has to reach a cer-tain thickness before silicic domes can develop. Thiscan be argued from studies of seismic data for Iceland(e.g. Menke, 1999) and the geographical distributionof silicic domes and nunataks in Iceland (Jónasson,2007). Geological settings in Iceland are very dif-ferent from those of a subduction zone where domesare common. Iceland is a ridge-centered hotspot is-land, situated astride the Mid-Atlantic Ridge. The

JÖKULL No. 60 135

Agustsdottir et al.

geology of Iceland is characterised by the interac-tion of spreading at the mid-ocean plate boundary andthe hotspot, centered in east-central Iceland under theVatnajökull icecap (Wolfe et al., 1997). The diver-gent plate boundary is marked by a chain of activecentral volcanoes traversing Iceland, resulting fromplate rifting and volcanism induced by upwelling ofthe hotspot. Central volcanoes are defined by fre-quent eruptions from a central area (Walker, 1963;Sæmundsson, 1978). Volcanism in the neovolcaniczone results in elongated volcanic systems, consistingof a central volcano and a transecting fissure swarm(Sæmundsson, 1978, 1979). A geothermal field isoften associated with the central volcano. Further-more, silicic rocks in Iceland are generally associ-ated with central volcanoes (Jónasson, 2007) and mostcentral volcanoes have only produced minor amountsof silicic rock in a few formations (Jóhannesson andSæmundsson, 1999). No dome eruption has beenobserved with modern monitoring equipment in Ice-land while the geological record contains many silicictephra layers, lavas and dome formations (Jóhannes-son and Sæmundsson, 1999; Larsen, 2000). Domes inIceland generally occur at or near the caldera rims ofan active central volcano. The sides of the underlyingmagma chamber are thought to represent the best con-ditions for the formation of silicic melts due to cool-ing of the hydrothermal system (Jónasson, 1994). Thecentral volcano Krafla, situated in the NVZ (NorthernVolcanic Zone), has an ice-free caldera surrounded bysubglacially formed silicic domes Hlí!arfjall, Hrafn-tinnuhryggur, Jörundur and Rani (not studied here),all located close to the caldera rim. (Sæmundsson,1991). The Krafla area is both easily accessible andone of the most intensely studied volcanic areas innorthern Iceland, due to the geothermal power plantsituated above the caldera’s shallow magma chamberand the recent volcano-tectonic episode, the Kraflafires, in 1975–1984 (e.g. Einarsson, 1991; Buck et al.,2006). For these reasons the Krafla region was cho-sen as a study area to examine Icelandic domes in anactive central volcano. The determination of the ventstructure can shed light on how these domes were em-placed, e.g. whether they were formed by a dike-federuption or by a more forceful emplacement displac-

ing surrounding strata in the process. Buoyancy isimportant in driving magma through the crust, andtherefore density data are necessary for meaningfulmodelling of the emplacement of cryptodomes andlava domes. The aim of the present work was there-fore to determine poorly known bulk density valuesfor domes and to investigate whether the domes haveroots. Rock samples were also collected and theirdensity measured to further constrain the results.

GEOLOGICAL SETTINGSThe Krafla volcanic systemThe Krafla volcanic system (Figure 1) is located in theNorthern Volcanic Zone (NVZ). The NVZ consistsof several elongated closely spaced volcanic systems,amongst which the Krafla and Askja volcanic systemsare most active. The Krafla volcanic system consistsof a central volcano and a transecting fissure swarm.The Krafla caldera is approximately 10 km wide, tran-sected by a 100 km long and N10!E trending fis-sure swarm (Sæmundsson, 1991). Around the calderaa number of concentric fissures have erupted hyalo-clastites, lavas and rhyolite (Sæmundsson, 1991). Apowerful geothermal system lies inside the caldera,harnessed by a geothermal power station operatedsince 1977. The Krafla caldera began forming about100 000 years BP (Sæmundsson, 1991). Due to the in-tensive volcanism, stratigraphic build up has been rel-atively fast, both during periods of ice-free conditionsand periods when ice covered the land. In the geologi-cal strata of the Krafla caldera there is evidence of twomain warm periods. The silicic rocks are concentratedaround the caldera rim and away from the central partof the fissure swarm (Sæmundsson, 1991).

The Hei!arspor!ur volcanic systemHei!arspor!ur is a small volcanic system locatedsouth of Krafla. Its topographic expression is a lowand broad ridge, about 15 km long and 3 km wide,extending from the Krafla central volcano in the northtowards the Bláfjall ridge and in the south to theFremrinámur central volcano. This volcanic systemwas active for a relatively short time (1000 years) inthe early Holocene and was built up in several small

136 JÖKULL No. 60

A gravity study of silicic domes in the Krafla area, N-Iceland

2 km

Gravity station

R hyolite, 24 000 BPHolocene Dacite

R hyolite, 90 000 BP

Holocene S coria

Hrafntinnuhryggur

Hlí!arfjall

Hraunbunga

M"vatn

HB2HB1

HF1HF2

HR 2

HR 1

HR 3

65°45'

K rafla

Caldera rim

16°45'

Figure 1. Map of the Krafla and Hei!arspor!ur areas showing the main outcrops of silicic volcanic rocks to-gether with the approximate location of the inferred caldera (dot-dashed contour) of the Krafla volcanic system.The inset map shows the location of the survey area in the Northern Volcanic Zone (NVZ), with neo-volcaniczones shaded. Map redrawn from Jónasson (1994) with the geology based on maps of Sæmundsson (1991).Gravity stations of this survey are shown with black dots. – Einfalda! jar!fræ!ikort af Kröflu- og Hei!arspor!seldstö!vakerfunum. Korti! s"nir helstu súru myndanirnar á svæ!inu og Kröflu öskjuna. Litla Íslandskorti!s"nir legu mælisvæ!isins í nyr!ra gosbeltinu (NVZ); gosbeltin eru s"nd grá. #yngdarmælingarpunktar $essar-ir rannsóknar eru s"ndar me! svörtum doppum.

JÖKULL No. 60 137

Agustsdottir et al.

eruptions (Sæmundsson, 1991), including the forma-tion of Hraunbunga, a small dacitic dome or coulee.

Geological framework of studied formationsIn this study three well-defined domes were surveyed,Hlí!arfjall, Hrafntinnuhryggur and Hraunbunga (Fig-ure 1):

1. Hlí!arfjall (rhyolite, 310 m high, 2 km long)formed 90 000 years BP, parallel to the Kraflacaldera rim on the southwest side (Sæmunds-son, 1991; Sæmundsson et al., 2000).

2. Hrafntinnuhryggur (rhyolite, 80 m high, 2.5 kmlong) formed 24 000 years BP, located 0.5–1 kminside and sub-parallel with the eastern calderarim of Krafla (Sæmundsson, 1991; Sæmunds-son et al., 2000).

3. Hraunbunga (125 m high, 1.8 km long), adacitic coulee erupted from a short fissureformed 10 000 years BP in the Hei!arspor!urvolcanic system (Sæmundsson, 1991).

Hlí!arfjall and Hrafntinnuhryggur are short, steepridges that may be regarded as subglacial equivalentsof lava domes. Hlí!arfjall together with the domesJörundur and Rani are parallel to the Krafla calderarim and may, on the basis of composition and age, beinterpreted to represent a common short-lived phase inKrafla’s history. It has been suggested that these rhy-olite eruptions were caused by the emplacement of aring dike (Jónasson, 1994) but the activity may havebeen more limited and localised, associated with oc-casional rhyolitic dike intrusions. Hrafntinnuhrygg-ur has a different composition and age and is paral-lel to the tectonic trend in the caldera and subparal-lel with the eastern caldera rim (Sæmundsson, 1991).Hraunbunga is a dasitic coulee in the Hei!arspor!urvolcanic system; one of the few silicic domes in Ice-land not associated with a central volcano (Jónasson,2007).

METHODS AND DATAIn general, silicic rocks have lower densities thanthose of a more basaltic composition. Measuring agravity profile over a rhyolite dome surrounded by

basaltic lava should demonstrate a significant den-sity contrast. The form of gravity anomalies associ-ated with outcropping surface formations can oftenbe used to determine whether they are partly buriedby younger formations or not (Figure 2). Knowl-edge of past geological processes at the observationsite also gives an idea of what can be expected underthe surface. Data from boreholes that reach to justover 2 km depth in the Krafla caldera demonstratethat bedrock in the region consists predominantly ofbasalt, with suites of layered piles of alternating lavaand units of hyaloclastite down to 2000 meters, wheregranophyre becomes dominant at the bottom (Gud-mundsson et al., 2008; Arnadottir et al., 2009).

The gravity surveys

In Iceland, gravity studies aimed at determining bulkdensity of individual topographic features have beendone on hyaloclastite tindars (hyaloclastite ridges),tuyas and lava flows but not on silicic formations.This study will attempt to fill that gap. The Nettle-ton method has proved to be useful for densitydeterminations of hyaloclastite and lava formations(Schleusener et al., 1976; Gudmundsson and Milsom,1997; Gudmundsson et al., 2001; Gudmundsson andHögnadóttir, 2004; Schopka et al., 2006), this shouldalso be the case for rhyolite domes. Density measure-ments of rock samples have been made for most typesof Icelandic rock (Pálsson et al., 1984; Gudmundssonand Högnadóttir, 2004). A knowledge of rock den-sities is necessary for applying the Bouguer anomalyand the Nettleton method (e. g. Kearey et al., 2002).

The mean bulk density of the surface bedrock inthe Krafla caldera was obtained by Johnsen (1995) byapplying the Parasnis method (Parasnis, 1973) to theentire Krafla caldera. In the Bouguer reduction we usehis value of 2500 kg m!3 as our background density.The reason is that this density best represents the sur-roundings of the domes, i.e. the bedrock in the Kraflaarea. A total of 48 gravity stations along nine profileswere measured (Figure 1) over the three dome forma-tions. Observations were carried out in the autumn of2007 and the spring of 2008, during five days.

138 JÖKULL No. 60

A gravity study of silicic domes in the Krafla area, N-Iceland

Figure 2. Three cases of expected Bouguer anomalyfrom a topographic formation. A1: Formation has nocrustal root and a density (!1), significantly less thanthe reduction density (!0). A2: Formation has a topo-graphic root and density !1 < !0. Note the differencein the form of the anomaly flanking the formation forA1 and A2. B: No root but formation density higherthan reduction density (!1 > !0). – Bouguer !yngd-arsvi" frá fjalli. A1: Fjalli" hefur ekki rót og e"lis-massi fjallsins (!1) er lægri en sá e"lismassi sem not-a"ur er til a" reikna Bouguer lei"réttingu (!0). A2:Fjalli" hefur lágan e"lismassa og rót sem nær ni"urí jar"skorpuna. B: Engin rót en e"lismassinn (!1) erhærri en vi"mi"unar e"lismassinn (!0).

The gravity field was measured along profiles,preferably perpendicular to the strike of the forma-tion, and gravity stations were placed at 0.2–0.5 kmintervals. Gravity readings were made using a Lacosteand Romberg (G-445) gravimeter. A kinematic GPSTrimble 5700 was used for elevation measurements ateach station. All surveys were carried out on foot andall equipment carried by two surveyors. A combinedBouguer and topographic correction was calculatedby direct integration of a topographic model with apixel size of 25 m (Gudmundsson and Milsom, 1997;Gudmundsson and Högnadóttir, 2004). A Bougueranomaly was then obtained by subtracting the correc-tion from the free air anomaly.

The accuracy of a gravity survey is limited bythe gravimeter, terrain corrections and the accuracyin measured elevation. The gravity meter has an accu-racy of 0.01 mGal (LaCoste and Romberg Inc., 1979).The elevation accuracy in this study is equal to or bet-ter than 0.5 m which causes a 0.1 mGal error in theBouguer anomalies. The combined accuracy of therandom uncertainty due to terrain and Bouguer cor-rections (resulting from integration of the elevationmodel) are minor and is considered to be better than0.1 mGal. Total accuracy for measurements and pro-cessing is therefore 0.2 mGal (Agustsdottir, 2009).

Rock samplesRock samples were gathered from each formation.Usually five representative samples for density deter-mination were collected at important gravity stationsalong the survey profiles. A total of 95 samples werecollected at 15 sites. The dry and wet densities ofthese samples were obtained in the laboratory fromthe difference in their weight in air and when sub-merged in water.

Nettleton’s methodA two-step approach is adopted here to determine thebest density and study the roots of the domes:

1. Nettleton’s method, used to determine the mostlikely bulk density of a formation and/or to indi-cate whether a dome is partly buried in youngerformations.

JÖKULL No. 60 139

Agustsdottir et al.

2. Forward modelling, where details of dome formare examined, including the possible existenceof a root extending into the crust, and whetherthe assumption used in Nettleton’s method of asingle bulk density of a formation is valid.

Nettleton’s method (Nettleton, 1976; Keareyet al., 2002) of density determination involves tak-ing gravity observations over an isolated topographicprominence. Field data is reduced using a seriesof different densities for the Bouguer and terraincorrections. The density that provides a Bougueranomaly with the least correlation with the topogra-phy is taken to represent the mean bulk density of thetopographic formation (Kearey et al., 2002). The Net-tleton method works where formations are not buriedor where formation and surroundings have the samedensity. A Nettleton profile determines the mean bulkdensity of an entire mountain, but can give a mislead-ing result if the formation is partly buried in lava orsediments.

It is common that the density obtained from Nett-leton’s method is less than the mean density of rocksamples from the same formation, especially in frac-tured and porous rock. This discrepancy arises be-cause 5–10 cm diameter rock samples do not includethe volume occupied by large fractures or voids. Thus,a systematic difference is to be expected between thetwo methods. However, the overestimation of bulkdensity found using the mean of the densities of rocksamples, is likely to be similar for porous and frac-tured rocks, regardless of composition. This is sup-ported by comparisons of Nettleton profiles and rocksamples (Agustsdottir, 2009). The accuracy of esti-mating the bulk density in this study by Nettleton’smethod is considered to be ±100 kg m!3 (Agusts-dottir, 2009). Figure 3 shows the Nettleton’s profileover Hlí!arfjall. We note a slight local rise in Bougueranomaly immediately adjacent to the formation (Fig-ure 2). Comparing Figure 2 with Figure 3 we con-clude that this dome has insignificant roots.

Gravity forward modellingForward, 2.5-D gravity models are generated usingthe GravMag sofware (Pedley et al., 1997), using thebackground density 2500 kg m!3 (Johnsen, 1995) and

the density values obtained for each formation withthe Nettleton method. The model profiles are con-structed by subtracting a regional field obtained as thelinear fit (since all the profiles are short) that best rep-resents the mean trend of of the Bouguer anomalyat the location of the profile. Bodies are assumedto strike perpendicular to the survey line and a finitestrike length can be assigned, i.e. the true length ofthe formation perpendicular to strike can be used asthe length of the modelled body.

RESULTSDensity valuesThe main results are that all the domes are of low den-sity, reflecting both low grain-density and high poros-ity (Table 1). Table 1 also shows that the dome’s den-sity values are significantly smaller than those of thesurroundings. Table 2 shows the volume and mass ofthe formations.

Table 1. Mean bulk densities (!̄) for each profile (Fig-ure 1), determined by the Nettleton method (!N ) androck samples (!s). The density of the surroundingsare from Johnsen (1995). – Me!ale!lismassi ákvar!-a!ur me! a!fer! Nettletons fyrir hvert sni! á 1. myndog út frá grjóts"num (!s). E!lismassi umhverfisins(bakgrunns-e!lismassi) samkvæmt Johnsen (1995).

! [kg m!3]Location profile !N !s !̄

Hlí!arfjall HF1 1600 2060 1700HF2 1800 2060

Hrafntinnuhr. HR1 1575 1750 1692HR2 1875 1750HR3 1625 1750

Hraunbunga HB1 1775 1950 1763HB2 1750 1950

Surroundings 2500

Gravity modelsGravity models for all the formations show consistentresults. Therefore only one selected profile from eachformation is presented. Three models are presentedfor Hlí!arfjall and Hrafntinnuhryggur, four for Hraun-bunga. The three models are defined as:

140 JÖKULL No. 60

A gravity study of silicic domes in the Krafla area, N-Iceland

Figure 3. Nettleton’s method used to determine the bulk density of Hlí!arfjall (survey line HF1). Gravity reduc-tions have been performed using densities ranging from 1500–3000 kg m!3 (a). The dot-dash line representsthe density determined by the rock samples. The fine dashed line show the best fit density determined by eye.The lower figure (b) represents the topography along the profile, where the dots represent observed gravity sta-tions. – A!fer! Nettletons notu! til a! ákvar!a me!ale!lismassa Hlí!arfjalls (sni! HF1). "yngdargögnin hafaveri! lei!rétt fyrir e!lismassa á bilinu 1500–3000 kg m!3 (a). Punkta-brotalínan s#nir e!lismassa grjóts#na.Fínger!a brotalínan s#nir e!lismassann sem hefur minnsta fylgni vi! lögun fjallsins. Ne!ri myndin (b) s#nirlandslag eftir sni!inu $ar sem punktarnir s#na hvar mælt var.

Table 2. The volume of the formations, estimatedfrom the digital elevation model using Surfer 8.0 R!.Mass of each formation is calculated using thesevolumes and the densities obtained from Nettleton’smethod (Table 1). – Rúmmál jar!myndana var meti!út frá hæ!arlíkani me! Surfer 8.0 R!. Massi hverrarmyndunar var reikna!ur me! $ví a! nota rúmmálinog svo e!lismassann sem fékkst me! a!fer! Nettle-tons (tafla 1).

Locality volume mass min mass max(km3) (109kg) (109kg)

Hlí!arfjall 0.143±0.014 229±23 272±27Hrafntinnuhr. 0.021±0.002 34±3 40±4Hraunbunga 0.040±0.004 72±7 76±8

JÖKULL No. 60 141

Agustsdottir et al.

a. The dome is strictly a surface formation withouta root extending down into the crust.

b. Same as a but a narrow dyke extends down-wards into the crust.

c. The dome is modeled as the top of a partlyburied formation.

All gravity models of type b are shown with themaximum possible width of the feeder dike, 20–25m. Insertion of a wider feeder results in a significanteffect on the shape of the calculated anomaly and di-vergence between model and data.

Hlí!arfjall – profile HF1The HF1 profile is 3137 meters long from SW-NE,and measured across the subglacially formed rhyolitedome Hlí!arfjall (Sæmundsson, 1991).

Hlí!arfjall is clearly a surface structure (Figures 2and 4a), and is not buried by younger volcanic erup-tives. This model fits very well to the data. Further-more, Hlí!arfjall is likely to be a vent-formed struc-ture (Figure 4a,b), i.e. formed by a dike or vent to thesurface. It is not possible to distinguish between mod-els a and b for Hlí!arfjall. On the other hand, model cshows that Hlí!arfjall does not have significant roots(Figure 4c).

Hrafntinnuhryggur – profile HR2The HR2 profile is 2264 meters long and is the mid-dle profile, measured over Hrafntinnuhryggur fromWest to East. Models a, b and c are constructed inthe same way as for Hlí!arfjall. Model a in Figure 5shows Hrafntinnuhryggur and a hyaloclastite ridge toits east. The model suggests that Hrafntinnuhryggur islikely to be a surface formation like Hlí!arfjall (Figure4), i.e. it is not buried by younger volcanic eruptives(Figure 5), nor does it have gravitationally significantroots (Figure 5c). Hrafntinnuhryggur is likely to beemplaced as a vent-forming dome but the vent or anyunderlying dike cannot be distinguished, neither bymodel a nor model b.

It is difficult to determine the depth of the hyalo-clastite ridge to the west of Hrafntinnuhryggur due tothe sparsity of gravity stations around it. Furthermore,

the density of the hyaloclastite ridge is quite low com-pared to values from elsewhere (e.g. Gudmundssonand Högnadóttir, 2004). The density value chosen(1500 kg m!3) gives a plausible depth for the ridge,a higher and more conventional value demands largerdepths of the ridge than the data set allows.

Hraunbunga - profile HB2

The HB2 profile is 2965 meters long and is measuredfrom NE to SW over the coulee Hraunbunga. Thesimplest model is shown in Figure 6a. A slightly bet-ter fit is obtained when density consistent with a silicicscoria cone is used for the top of Hraunbunga (Figure6b). The scoria was observed in the field and is consis-tent with geological maps (e.g. Sæmundsson, 1991).

Hraunbunga does not have gravitationally signif-icant roots (Figure 6a), nor is it buried by youngervolcanic eruptives (Figure 6c). Consequently, Hraun-bunga is a surface formation and most likely a vent-formed dome/coulee. It is not possible to distinguishbetween model b (Figure 6b) and model c (Figure 6c),but they demonstrate clearly that Hraunbunga couldhave formed from a dike to the surface. Consequently,the results from Hraunbunga suggest similar emplace-ment mechanisms as for Hrafntinnuhryggur and Hlí!-arfjall.

The density of the flanks of Hraunbunga is pos-sibly underestimated by the Nettleton method. Theyare dacite lavas, and the models (Figure 6) show devi-ations from the observed anomaly over the shouldersof Hraunbunga.

DISCUSSIONDensity valuesThe densities determined by the Nettleton methodare essentially the same for Hlí!arfjall and Hrafn-tinnuhryggur while there is considerable difference inthe values obtained from samples. Since both for-mations are subglacially-formed rhyolitic domes andshould have similar density, we argue that the Net-tleton method is more reliable than the mean of thesamples. It appears that sampling was somewhat bi-ased towards the more coherent and less vesicular partof the rocks at Hrafntinnuhryggur. Marginally higher

142 JÖKULL No. 60

A gravity study of silicic domes in the Krafla area, N-Iceland

Figure 4. Three different models for the profile HF1, measured over Hlí!arfjall from NE to SW. For profilelocation see Figure 1. Background density is 2500 kg m−3. Maximum dike width is 20 m. – Sniðið HF1 varmælt yfir Hlíðarfjall frá NA til SV. Til glöggvunar á staðsetningu sniðsins sjá 1. mynd. Bakgrunns eðlismassinner 2500 kg m−3 og mesta breidd gangs er 20 m.

values of Nettleton’s density are found for Hraun-bunga. This observation is in accordance with whatis to be expected since the density of dacite is slightlyhigher than that of rhyolite. Moreover parts of Hraun-bunga are lava that has flowed a short distance. Hraun-

bunga was probably cooled over a longer period oftime than were the subglacial formations. Thus, thelava of Hraunbunga may have been more degassedand therefore with lower porosity and higher bulkdensity when it solidified, compared to both Hlí!ar-

JÖKULL No. 60 143

Agustsdottir et al.

Figure 5. Three different models for profile HR2, measured across Hrafntinnuhryggur from W to E. Backgrounddensity is 2500 kg m!3. Maximum dike width is 20 m. – !rjú mismunandi líkön fyrir sni"i" HR2, sem mæltvar yfir Hrafntinnuhrygg frá V til A. Bakgrunns e"lismassinn er 2500 kg m!3 og mesta breidd gangs er 20 m.

fjall and Hrafntinnuhryggur. The results obtained hereare a good addition to previous gravity studies on den-sities of Icelandic rocks (Gudmundsson et al., 2001;Gudmundsson and Högnadóttir, 2004).

All the studied domes have low densities, reflect-ing both low grain-density and high porosity. Thedomes’ density values are significantly smaller than

those of the surroundings, creating a density con-trast possibly sufficient to drive the ascent of rhyolitemagma to the surface by the action of a bouancy force.

Gravity models

A gravity survey using Nettleton’s method requireswell placed observation points. In this study obser-

144 JÖKULL No. 60

A gravity study of silicic domes in the Krafla area, N-Iceland

Figure 6. Four different models for the profile HB2, measured across Hraunbunga from NE to SW. Backgrounddensity is 2500 kg m!3. Maximum dike width is 20 m. – Fjögur líkön fyrir sni!i! HB2 sem mælt var yfirHraunbungu frá NA til SV. Bakgrunns e!lismassinn er 2500 kg m!3 og mesta breidd gangs er 20 m.

vation points where carefully placed with referenceto surface geology. For formations with steep slopeslike Hlí!arfjall and Hrafntinnuhryggur it is only pos-sible to place an observation point at the top and thenfurther points at each side of the foot of the forma-tion. Therefore, for a single profile, one observation

point controls the anomaly and thus the bulk density.However, each formation is crossed by more than oneprofile and all profiles give similar results for density,indicating that our results are robust. Moreover, theHraunbunga profile is controlled by three data pointsmeasured on the gentle slope of the formation.

JÖKULL No. 60 145

Agustsdottir et al.

The main results from gravity models are the con-straints placed on the basal structure of the studiedformations; Hlí!arfjall, Hraunbunga and Hrafntinnu-hryggur. The models show that they are neither buriedby younger volcanic eruptives nor are roots with adensity contrast with the surroundings detected. Theformations studied were therefore emplaced as dike-fed domes, i.e. by the extrusion from a dike ontothe surface or the ice-bedrock interface. This ideais in agreement with Jónasson (1994, 2005, 2007)who states that Icelandic rhyolites are rather liquidand therefore erupt more quickly than a more vis-cous magma. This rules out processes such as stop-ing, where a magma body forces its way upwards asa massive block, displacing and deforming the sur-rounding country rock in the process. When reachingthe surface, such a forcefully emplaced body is likelyto leave rotated and disrupted country rock with therhyolites extending some distance into the bedrock.Thus, the models, and as far as can be seen also thesurface geology, indicate emplacement through dikeintrusion to the surface followed by eruption. How-ever, the high viscosity of the magma and confine-ment in a glacier leads to the formation of a steepdome (Hlí!arfjall and Hrafntinnuhryggur). No struc-tural difference is seen between the domes formedduring the last glaciation and in the Holocene domeHraunbunga. We conclude therefore that the threedomes were formed in a smilar way. The domes havebeen erupted through a narrow dike (on the order 5–10 m) reaching to the surface. Maximum width ofa dike before it starts to register a gravity anomalyof its own that can be distinguished from that of theoverlying dome is 20–25 m. Our results are in broadagreement with that of Tuffen and Castro (2008) whostudied Hrafntinnuhryggur and found evidence of afeeder dike at the surface with a thickness rangingfrom 2 m to slightly more than 10 m. However, ourresults suggest that the implied structure shown ontheir Figure 13 (p. 365, Tuffen and Castro, 2008)should be modified with a narrow dike and flat orsemi-flat dome-bedrock contact, as indicated in thegravity model on Figure 5b.

CONCLUSIONS

• All the studied formations, Hlí!arfjall, Hraun-bunga and Hrafntinnuhryggur, are neitherburied by younger volcanic eruptives nor wereany roots detected. The results are consistentwith a flat or semi-flat dome-bedrock contact.

• All dome formations studied are emplaced bya dike to the surface. A dike width of 5–10 mis likely and maximum possible dike width is20–25 m.

• All the domes have low densities (1600–1800kg m!3), reflecting both low grain-density andhigh porosity.

• The domes display a significant density differ-ence between the formations studied and thesurroundings.

AcknowledgementsWe thank the University of Iceland Science Fund(Rannsóknarsjó!ur Háskólans) for financing the workof this paper. Furthermore we thank Arnar Már Vil-hjálmsson and Hjalti Nönnuson for assistance in thefiled and Iceland Geosurvey (ÍSOR) for financingmost of the field work. Leó Kristjánsson, KristjánSæmundsson and an anonymous reviewer read themanuscript critically and made valuable suggestionsfor improvements.

!yngdarmælingar á súrum gúlum á Kröflusvæ"inuSúrt berg á Íslandi tengist allajafnan megineldstö!v-um og mynda gúla á e!a í kringum öskjurima. Sumir"essara gúla hafa myndast á jökulskei!um en a!rir áhl#skei!um. $yngdarmælingar voru ger!ar á Kröflu-svæ!inu 2007 og 2008, til "ess a! ákvar!a me!al-e!lismassa "riggja súrra gúla. E!lismassagögn eruforsenda líkanager!ar sem eykur skilning á mynd-un hraungúla og leynigúla. Sni! voru mæld yfir"rjá gúla. (1) Hlí!arfjall er líparítgúll, 310 m hár,2 km langur og mynda!ur undir jökli fyrir um 90"úsund árum. (2) Hrafntinnuhryggur er líparít- oghrafntinnugúll, 80 m hár og 2,5 km langur mynd-a!ur fyrir um 24 "úsund árum. (3) Hraunbunga er

146 JÖKULL No. 60

A gravity study of silicic domes in the Krafla area, N-Iceland

dasít hraungúll, 125 m hár, 1,8 km langur, mynda!-ur á íslausu landi fyrir um 10 "úsund árum. E!lis-massar og rúmmál gúlanna: 1) Hlí!arfjall: 1600–1800kg m!3, 0,143±0,014 km!3; 2) Hrafntinnuhryggur:1575–1875 kg m!3, 0,021±0,002 km!3; 3) Hraun-bunga: 1750–1775 kg m!3, 0,040±0,004 km!3. All-ir gúlarnir hafa lágan e!lismassa, sem endurspegl-ar lágan kornae!lismassa og háan poruhluta. E!l-ismassi gúlanna er mun lægri en umhverfisins, semskapar hugsanlega nógu mikinn e!lismassamun til a!súr kvika geti risi! í skorpunni. Ni!urstö!ur "yngdar-líkana s#na a! gúlarnir eru hvorki grafnir af yngri gos-efnum né hafa "eir rætur. Efni allra gúlanna barst tilyfirbor!s eftir gangi sem í mesta lagi var um 20 metra"ykkur en kann a! hafa veri! mun mjórri. Mælingarbenda ekki til "ess a! umtalsver! snörun e!a tilfærslaá a!liggjandi bergi hafi or!i! "egar gúlarnir myndu!-ust.

REFERENCESAgustsdottir, T. 2009. On the dynamics of rhyolite dome

emplacement: Densities and deformation fields. Mas-ter’s thesis, Univ. Iceland, 101 pp.

Arnadottir, S., A. K. Mortensen, A. Ingimarsdóttir, H.Tryggvason, R. B. Jónsson, and H. S. Gunnarsson2009. Krafla, IDDP-1. Drilling completion and geo-logy report for drilling stage 2. Tech. Rep. (IcelandGeosurvey) ÍSOR-2009/021, 94 pp.

Buck, W. R., P. Einarsson and B. Brandsdóttir 2006. Tec-tonic stress and magma chamber size as controls ondike propagation: Constraints from the 1975–1984Krafla rifting episode. J. Geophys. Res. 111, B12404,doi:10.1029/2005JB003879.

Einarsson, P. 1991. Kröflueldar 1975–1984 (The Kraflafires 1975–1984, in Icelandic). Náttúra M!vatns. Hi!íslenska náttúrufræ!ifélag, 61–100.

Fink, J. H. and S. W. Anderson 2000. Lava domes andcoulees. In: Sigurdsson, H. (ed.), Encyclopedia of Vol-canoes. Academic Press, 307–319.

Gudmundsson, M. T. and T. Högnadóttir 2004. Hraunog móbergsmyndanir á svæ"inu frá Brennisteinsfjöll-um a" Hengli, ni"urstö"ur #yngdarmælinga. (Lavaand hyaloclastite formations from Brennisteinsfjöll toHengill, results from gravity measurements, in Ice-landic). Tech. Rep. RH-12-2004, Jar!vísindastofnunHáskóla Íslands, 64 pp.

Gudmundsson, M. T. and J. Milsom 1997. Gravity andmagnetic studies of the subglacial Grímsvötn volcano,Iceland. J. Geophys. Res. 102, 7691–7704.

Gudmundsson, M. T., T. Högnadóttir and S. P. Jakob-sson 2001. Hraun og móbergsmyndanir sunnanLangjökuls, ni"urstö"ur #yngdarmælinga. (Lava andhyaloclastite formations south of Langjökull, resultsfrom gravity measurments, in Icelandic). Tech. Rep.RH-28-2000, Raunvísindastofnun Háskólans, 37 pp.

Gudmundsson, Á., B. Steingrímsson, D. Sigursteinsson,G. Gíslason, H. Sigvaldason, J. Hólmjárn, K. H. Sig-urdsson, S. Benediktsson, T. Hauksson and V. Stefáns-son 2008. Krafla-well KG 25. drilling, geology andgeochemistry. Tech. Rep. (Iceland GeoSurvey) ÍSOR-2008/056, 30 pp.

Johnsen, G. V. 1995. $yngdarkort af Kröflusvæ!i. (Gravitymap of the Krafla area (in Icelandic)). Eyjar í eldhafi.93–100, Gott mál hf.

Jóhannesson, H. and K. Sæmundsson 1999. Geologicalmap of Iceland 1:100000. Icelandic Institute of Nat-ural History, ISBN 9979-9335-4-2.

Jónasson, K. 1994. Rhyolite volcanism in the Krafla cen-tral volcano, north-east Iceland. Bull. Volc. 56, 516–528.

Jónasson, K. 2005. Magmatic evolution of the Hei!ar-spor!ur ridge, NE-Iceland. J. Volcanol. Geotherm.Res. 147, 109–124.

Jónasson, K. 2007. Silicic volcanism in Iceland: Composi-tion and distribution within the active volcanic zones.J. Geodynamics 43, 101–117.

Kearey, P., M. Brooks and I. Hill 2002. An Introduction toGeophysical Exploration, 3rd ed. Blackwell Science,262 pp.

LaCoste and Romberg 1979. Instruction manual for La-Coste and Romberg model G geodetic gravity meterno. G-445. LaCoste and Romberg Inc., Austin, Texas,20 pp.

Larsen, G. 2000. Holocene eruptions within the Katla vol-canic system, south Iceland: Characteristics and envi-ronmental impact. Jökull 49, 1–25.

Menke, W. 1999. Crustal isostasy indicates anomalousdensities beneath Iceland. Geophys. Res. Lett. 26,1215–1218.

Nettleton, L. L. 1976. Gravity and Magnetics in OilProspecting. McGraw-Hill, New York, 464 pp.

JÖKULL No. 60 147

Agustsdottir et al.

Pálsson, S., G. I. Haraldsson and G. H. Vigfússon 1984.E!lismassi og poruhluti bergs. (Densities and rockporosities, in Icelandic). Tech. Rep. Orkustofnun OS-84048/VOD-18B, 35 pp.

Parasnis, D. S. 1973. Mining Geophysics, 2nd ed. Elsevier,Amsterdam, 395 pp.

Pedley, R., J. P. Busby and Z. K. Dabek 1997. Gravmaguser manual. British Geological Survey 73 pp.

Schleusener, A., W. Torge and W. Drewes 1976. The grav-ity field of northeastern Iceland. J. Geophys. 42, 27-45.

Schopka, H. H., M. T. Gudmundsson and H. Tuffen2006. The formation of Helgafell, southwest Iceland, amonogenetic subglacial hyaloclastite ridge: Sedimen-tology, hydrology and volcano-ice interaction. J. Vol-canol. Geotherm. Res. 152, 359–377.

Sæmundsson, K. 1978. Fissure swarms and central vol-canoes of the neovolcanic zones of Iceland. Geol. J.Spec. Iss. 10, 415–432.

Sæmundsson, K. 1979. Outline of the geology of Iceland.Jökull 29, 7–28.

Sæmundsson, K. 1991. Jar!fræ!i Kröflukerfisins (Thegeology of the Krafla volcanic system, in Icelandic).Náttúra M"vatns. Hi! íslenska náttúrufræ!ifélag, 3–60.

Sæmundsson, K., M. S. Pringle and B. S. Hardarson 2000.The age of geological strata in the Krafla volcanic sys-tem, (in Icelandic). Geoscience Soc. Iceland, springmeeting abstracts 2000, 26–27.

Tuffen, H., and J. M. Castro 2008. The emplacement ofan obsidian dyke through thin ice: Hrafntinnuhryggur,Krafla, Iceland. J. Volcanol. Geotherm. Res. In press,doi:10.1016/j.jvolgeores.2008.10.021.

Walker, G. P. L. 1963. The Brei!dalur central volcano,eastern Iceland. Quart. J. Geol. Soc. Lond. 119, 29–63.

Wolfe, C. J., I. T. Bjarnason, J. C. VanDecar and S. C.Solomon 1997. Seismic structure of the Iceland man-tle plume. Nature 385, 245–247.

148 JÖKULL No. 60