Pressures of Crystallization of Icelandic Magmas
Division of Earth and Planetary Dynamics, School of Earth Sciences, The Ohio State University, Columbus, OH 43210, USA
RECEIVED FEBRUARY 20, 2007; ACCEPTED DECEMBER 14, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Iceland lies astride the Mid-Atlantic Ridge and was created by seafloor spreading that began about 55 Ma. The crust is anomalously thick (
20–40 km), indicating higher melt productivity in the underlying mantle compared with normal ridge segments as a result of the presence of a mantle plume or upwelling centered beneath the northwestern edge of the Vatnajökull ice sheet. Seismic and volcanic activity is concentrated in 50 km wide neovolcanic or rift zones, which mark the subaerial Mid-Atlantic Ridge, and in three flank zones. Geodetic and geophysical studies provide evidence for magma chambers located over a range of depths (1·5–21 km) in the crust, with shallow magma chambers beneath some volcanic centers (Katla, Grimsvötn, Eyjafjallajökull), and both shallow and deep chambers beneath others (e.g. Krafla and Askja). We have compiled analyses of basalt glass with geochemical characteristics indicating crystallization of ol–plag–cpx from 28 volcanic centers in the Western, Northern and Eastern rift zones as well as from the Southern Flank Zone. Pressures of crystallization were calculated for these glasses, and confirm that Icelandic magmas crystallize over a wide range of pressures (0·001 to 1 GPa), equivalent to depths of 0–35 km. This range partly reflects crystallization of melts en route to the surface, probably in dikes and conduits, after they leave intracrustal chambers. We find no evidence for a shallow chamber beneath Katla, which probably indicates that the shallow chamber identified in other studies contains silica-rich magma rather than basalt. There is reasonably good correlation between the depths of deep chambers (> 17 km) and geophysical estimates of Moho depth, indicating that magma ponds at the crust–mantle boundary. Shallow chambers (< 7·1 km) are located in the upper crust, and probably form at a level of neutral buoyancy. There are also discrete chambers at intermediate depths (11 km beneath the rift zones), and there is strong evidence for cooling and crystallizing magma bodies or pockets throughout the middle and lower crust that might resemble a crystal mush. The results suggest that the middle and lower crust is relatively hot and porous. It is suggested that crustal accretion occurs over a range of depths similar to those in recent models for accretionary processes at mid-ocean ridges. The presence of multiple stacked chambers and hot, porous crust suggests that magma evolution is complex and involves polybaric crystallization, magma mixing, and assimilation. KEY WORDS: Iceland rift zones; cotectic crystallization; pressure; depth; magma chamber; volcanic glass
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
There is considerable interest in the depths of magma chambers beneath active volcanoes in Iceland (e.g. Soosalu & Einarsson, 2004
; Sturkell et al., 2006) for four main reasons. First, knowledge of the depths of magma chambers is important for interpreting precursory activity (seismic, deformation, gas emissions, etc.) to volcanic eruptions, and therefore for forecasting eruptions (e.g. Marti & Folch, 2005). Second, knowledge of the depths of chambers provides constraints on models for magma evolution, because phase relationships and melt compositions vary as a function of pressure (e.g. O’Hara, 1968; Thy, 1991; Grove et al., 1992; Yang et al., 1996). Third, knowledge of the distribution of magma bodies is important to understand thermal gradients, which affect variations in density and seismic velocity in the crust (e.g. Kelley et al., 2005). Fourth, knowledge of the locations and sizes of magma chambers is essential to understand mechanisms of crustal accretion and differentiation (e.g. Pan & Batiza, 2002, 2003).
Various methods have been used to estimate magma chamber depths in Icelandic crust (Table 1). Most recent studies utilize geodetic techniques (see Sturkell et al., 2006), and yield results that in many cases (e.g. Krafla, Grimsvötn, Katla, possibly Torfajökull) agree with those estimated using geophysical methods (see Table 1), although a magma chamber identified at a depth of 7 km beneath Hengill–Hrómundartindur from geodetic data has not been identified from seismic data (Soosalu & Einarsson, 2004). There are shallow chambers beneath Katla, Grimsvötn and Eyjafjallajökull, deep chambers beneath Vestmannaeyjar, both shallow and deep chambers beneath Krafla and Askja, and intermediate depth chambers beneath Hekla and Torfajökull. These data, along with evidence for lateral transport of magma along fissures and mixing of magmas from different centers (Sigurdsson & Sparks, 1978, 1981; McGarvie, 1984; Mørk, 1984; Fagents et al., 2001), suggest extremely complex magma dynamics.
|
Geodetic and geophysical methods are most useful for locating chambers beneath active volcanoes, whereas petrological methods allow the depths of magma chambers beneath both active and inactive volcanic centers to be determined. Various petrological techniques are used to determine the pressure, and hence depth, of crystallization (Table 1), but some of the results reported for Icelandic magmas (Table 1) are only qualitative. Nevertheless, petrological estimates agree well with those obtained using other methods when direct comparison is possible (e.g. Hengill, Torfajökull, and Vestmannaeyjar).
In this paper we report new estimates of the pressures of crystallization of Icelandic magmas and use these to determine the depths of magma chambers. Pressures of crystallization are determined from experimentally established phase equilibrium constraints using the method described by Yang et al. (1996). The results differ from those obtained in previous studies in two respects. First, quantitative pressure estimates were obtained using the compositions of glasses, which unambiguously represent pre-eruptive liquid compositions. Second, pressures were determined using glasses from 28 localities so that the results are applicable to a wide geographical area, and can be interpreted in terms of crustal thickness. We discuss the implications of the results for the structure and accretion of the crust, for geothermal gradients, and for magma evolution.
|
GEOLOGICAL BACKGROUND |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Iceland lies astride the Mid-Atlantic Ridge (MAR) and is characterized by crust that is
20–40 km thick (Bjarnason et al., 1993; Darbyshire et al., 1998, 2000a, b; Menke et al., 1998; Kaban et al., 2003; Foulger et al., 2003; Leftwich et al., 2005) compared with the 7·1 ± 0·9 km thick normal MAR crust (White et al., 1992; Bown & White, 1994). The anomalously thick crust indicates higher melt productivity in the underlying mantle compared with normal ridge segments (White & McKenzie, 1995). The higher melt productivity, together with geochemical differences between Icelandic basalts and normal mid-ocean ridge basalt (N-MORB; e.g. Schilling, 1973), provides evidence for a mantle plume or upwelling centred beneath the northwestern edge of the Vatnajökull ice sheet (Breddam et al., 2000**; Darbyshire et al., 2000a; Leftwich et al., 2005; Fig. 1). Alternative hypotheses to account for high melt productivity in the sub-Icelandic mantle have been discussed by Foulger & Anderson (2005) and Foulger et al. (2005).
|
The full spreading rate for Iceland is 18–20 mm/year (LaFemina et al., 2005
). The subaerial ridge forms 50 km wide neovolcanic or rift zones characterized by abundant seismic and volcanic activity (Fig. 1). The Western Volcanic Zone (WVZ) and the Northern Volcanic Zone (NVZ) formed at 7 Ma following eastward relocation of the spreading axis from the Snaefellsnes area (Hardarson et al., 1997). The WVZ can be traced across the Reykjanes Peninsula extensional leaky transform to the Reykjanes Ridge (RR), whereas the NVZ is offset from the Kolbeinsey Ridge (KR) by the 120 km long right-lateral Tjörnes Transform Fault. Propagation of the NVZ to the SW at 3 Ma (Steinthorsson et al., 1985) formed the Eastern Volcanic Zone (EVZ), so that spreading is partitioned between the WVZ and EVZ in southern Iceland. These rift zones are connected by the complex South Iceland Seismic Zone (SISZ), which accommodates left-lateral transform motion, and by the Mid-Iceland Belt (MIB), which may be a leaky transform (Oskarsson et al., 1985; LaFemina et al., 2005) or a non-transform relay zone (Sinton et al., 2005). The neovolcanic zones are mostly covered by basalt flows younger than 0·8 Ma. About 30 en echelon volcanic systems have been recognized, which mostly consist of central shield or composite volcanoes transected by fissure swarms. Eruptions from both volcanoes and fissure systems occur in the WVZ and NVZ, but central volcanoes are lacking in the EVZ. Many central volcanoes have calderas with typical dimensions of 3 km x 4 km, whereas fissure swarms are 5–20 km wide and 40–150 km long, and are often characterized by crater rows formed by scoria and spatter. Other volcanic features include small lava shields, tuff rings and maars, as well as hyaloclastite ridges, hyaloclastite cones and table mountains (tuyas) produced by sub-glacial and sub-aqueous eruptions. Olivine tholeiites and tholeiites dominate the volcanic products, but intermediate and acid volcanic rocks are also erupted, especially at central volcanoes.
Volcanic activity also occurs in three off-rift flank zones (Fig. 1): the Western Flank Zone (WFZ), the Southern Flank Zone (SFZ), and the Eastern Flank Zone (EFZ). Eruptions occur at large shield or composite volcanoes with calderas, and at small lava shields, tuff rings and scoria cones. However, the extensive fissure swarms characteristic of the rift zones are absent, and eruptions produce transitional and alkaline basalts along with intermediate and silicic compositions (Meyer et al., 1985; Oskarsson et al., 1985). Volcanic products along the NVZ, EVZ and SFZ change from tholeiitic basalts in northern and central Iceland to Fe–Ti-rich transitional basalts in southern Iceland and alkali basalts in Vestmannaeyjar.
|
METHODS AND DATA |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Method for determining pressure of crystallization
Various petrological techniques can be used to estimate the pressure (P) and hence depth (z) of crystallization of magmas (Table 1). The most appropriate method for use with a large number of samples is based on comparing the compositions of erupted melts with those of liquids lying along P-dependent phase boundaries. Many basalt magmas crystallize olivine (ol), plagioclase (plag), and clinopyroxene (cpx), and their compositions can be compared with those of liquids lying along the ol–plag–cpx cotectic boundary. The effect of pressure on the latter has been determined experimentally (e.g. O’Hara, 1968
; Grove et al., 1992), and can be seen by recasting melt compositions into normative mineral components and projecting phase relations onto pseudoternary planes in the system CaO–MgO–Al2O3–SiO2. Projection of phase relationships from plag onto the plane ol–cpx–qtz using the recalculation procedure of Walker et al. (1979) clearly shows the shift of the ol–plag–cpx cotectic towards ol with increasing P (Fig. 2). Crystallization pressure can be estimated by comparing the projected compositions of natural samples with the locations of cotectics on such diagrams, and this method has been used to estimate crystallization pressures for Hengill by Trønnes (1990), for Bláfjall Table Mountain by Schiellerup (1995), for Kistufell by Breddam (2002), and for Theistareykir by Maclennan et al. (2001).
|
The shift of the ol–plag–cpx cotectic towards ol and plag (see O’Hara, 1968
; Grove et al., 1992) reflects the different pressure dependences of cpx–liq, ol–liq and plag–liq equilibria, with higher pressure favouring earlier crystallization of cpx. This results in the development of a trend of decreasing CaO with decreasing MgO in liquids at an earlier stage of crystallization. Weaver & Langmuir (1990), Langmuir et al. (1992), Danyushevsky et al. (1996), Yang et al. (1996), and Herzberg (2004) have proposed models to quantitatively estimate the crystallization pressure based on such relationships. These models are all calibrated with experimental data, and we have elected to use that of Yang et al. (1996), who presented equations that describe the composition of liquids along the ol–plag–cpx cotectic as a function of P and T. Hence, the composition of the liquid is used to predict the P (and T) of saturation with ol, plag and cpx, as illustrated in Fig. 3 using two glass analyses from Bláfjall Table Mountain (Schiellerup, 1995). A series of liquid compositions that lie on the ol–plag–cpx cotectic have been calculated from each glass analysis at increments of 100 MPa. These predicted liquid compositions have been converted to normative mineral components assuming that 
Fe = FeO and projected from plag onto the plane ol–cpx–qtz using the procedure of Tormey et al. (1987) as modified by Grove et al. (1993). Comparison of observed glass compositions and predicted liquid compositions indicates crystallization at 100 MPa for sample 1.8.5 and 670 MPa for sample 1.8.1. These pressures agree with those estimated by Schiellerup (1995) for crystallization of units BII and BI at this locality.
|
, Predicted melt compositions at increments of 0·1 GPa for each sample; •, analyzed composition of glass in sample 1.8.1; 
, analyzed composition of glass in sample 1.8.5. Melt compositions have been converted to normative mineral components and projected from plag onto the pseudoternary plane ol–cpx–qtz using the procedure of Tormey et al. (1987Rather than use graphical methods, we have calculated pressures using the procedure described in Appendix A. An assessment of the accuracy (
110 MPa, 1
) and precision (80 MPa, 1) of the calculated pressures is given in Appendix B.
Herzberg (2004) noted that pressures obtained with his method can differ significantly (up to 300 MPa) from those obtained using the method of Yang et al. (1996). The methods were calibrated with different sets of experimental data, but even so the reason for this discrepancy is not clear. We believe that the method used in this paper yields reliable results for the following reasons: (1) Yang et al. (1996) obtained results for MORB and Hawaiian samples that are consistent with those obtained by other methods; (2) pressures estimated by Maclennan et al. (2001) for basalts from Krafla and Theistareykir using the Yang et al. (1996) method agree with those obtained from clinopyroxene geobarometry; (3) pressures calculated for samples from Midfell in this work agree to better than ±60 MPa with results obtained by Gurenko & Sobolev (2006) using the method of Danyushevsky et al. (1996), and to better than ±80 MPa with results obtained by these workers using clinopyroxene geobarometry; (4) the results obtained in the present study yield estimates of the depths of magma chambers that are consistent with those obtained using geodetic, geophysical, or other petrological methods. In addition, we find that Herzberg's (2004) method yields negative and therefore unrealistic pressures for 22% of the samples used to constrain the depth of magma crystallization in this paper. Nevertheless, it is apparent that more experimental data are needed to refine petrological methods used for the geobarometry of magmas. In particular, there is need for additional experiments to more closely establish the composition of liquids along the ol–plag–cpx cotectics for a range of basalt compositions over the pressure range 100–1000 MPa. The implications of using the Herzberg (2004) method to calculate crystallization pressures for Icelandic magmas are briefly discussed below in the section ‘Interpretation of pressure’.
Samples
Volcanic activity on Iceland produces about 0·12 km3/year of fresh lava, hyaloclastite, scoria and spatter. Many recently erupted samples contain glass; especially those erupted in sub-glacial and sub-aqueous environments. Glass analyses are preferable to whole-rock analyses for calculating the crystallization pressure, because glasses represent samples of quenched melts. Therefore, glasses formed from liquids in equilibrium with ol, plag and cpx should have compositions that lie exactly on the cotectic at the pressure of crystallization. Some whole-rock samples represent melts, but others represent mixtures of crystals and melt. It cannot be assumed that the latter formed by closed-system crystallization, because the crystals may be of accumulative origin or may represent xenocrysts (e.g. Trønnes, 1990; Hansteen, 1991; Révillon et al., 1999; Hansen & Grönvold, 2000). In such cases, the whole-rock samples do not represent melts and this explains why many whole-rock compositions are displaced from glass compositions in pseudoternary projections (Fig. 4). The erroneous assumption that whole-rock samples represent melts can lead to large errors in pressure estimates (up to 1000 MPa for the examples shown in Fig. 4).
|
We have compiled 524 published glass analyses from localities listed in the Supplementary Data (available for downloading at http://www.petrology.oxfordjournals.org) and shown in Fig. 1. Most glasses are from localities in the NVZ, WVZ, and EVZ, but 44 are from Hekla and Katla in the SFZ.
We also compiled 201 analyses of glasses in melt inclusions. The compositions of melt inclusions can be modified by post-entrapment crystallization and diffusive re-equilibration with the host mineral (e.g. Danyushevsky et al., 2002). Therefore, we used only those inclusion glasses with compositions that plot along and within arrays defined by groundmass glasses on variation diagrams. Using this criterion, 64 glass inclusion analyses were selected and used to supplement data for groundmass glasses.
Compositional data for samples from each locality are summarized in Table 1 of the Supplementary Data. Detailed discussion of chemical variations shown by the glasses is beyond the scope of this paper, and we describe only the general compositional characteristics along with evidence that these characteristics are consistent with crystallization of ol, plag and cpx.
Based on their SiO2 to Na2O + K2O ratios, 584 out of 588 samples can be classified as basalts (Fig. 5a). The remaining samples are basaltic andesites (two from Hengill and two from Askja). In addition, most glasses (553) are tholeiitic (subalkaline) according to the criterion proposed by Macdonald (1968) for Hawaiian lavas (Fig. 5b), and range in composition from olivine tholeiites (normative Ol up to 18 wt %) to quartz tholeiites (normative Q up to 8 wt %). The remaining 35 glasses plot in the alkaline field but only two of these (from Hekla) contain normative nepheline [CIPW norms were calculated with Fe2O3 and FeO contents fixed assuming log fO2 = FMQ – 1 (McCann & Barton, 2004), where FMQ is the fayalite–magnetite–quartz buffer] and we therefore consider all of these samples to represent transitional basalts. All glasses show strong enrichment in FeO as MgO decreases (Fig. 5c).
|
Chemical variations shown by the glasses can be qualitatively explained by crystallization of ol–plag–cpx (±spinel) (Figs 5 and 6). Quantitative mass-balance models confirm that removal of these three phases accounts for major-oxide variations at many localities (e.g. Meyer et al., 1985
; Schiellerup, 1995; Maclennan et al., 2001, 2003), but this does not necessarily mean that all glasses represent liquids lying along ol–plag–cpx cotectics. The scatter shown on variation diagrams (Figs 5c and 6) indicates that the basalts do not evolve along a single liquid line of descent (LLD). This might indicate crystallization along ol–plag–cpx cotectics at different pressures (polybaric crystallization), but other explanations are also possible, as described in a later section of this paper (see also Herzberg, 2004). This requires that results for the various localities are evaluated in the context of geochemical and petrographic data. Some samples (24) from the Hengill–Hrómundartindur volcanic complex in the WVZ, as well as from Hrimalda, Gigoldur, and Sprengisandur in the NVZ, have anomalously high CaO contents as a result of clinopyroxene assimilation (CaO/Al2O3 > 1; see Fig. 6) as discussed by Trønnes (1990). These compositions lie outside the range of those used by Yang et al. (1996) to calibrate the method for pressure calculation. Therefore, we do not report pressures for glasses with anomalously high CaO/Al2O3.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Equilibration pressures
Pressures were calculated for all glasses, but some results are unrealistic and others are considered unreliable. The average uncertainty in P calculated for nearly all samples is <120 MPa, and is similar to that estimated from experimental data (Appendix B). In contrast, uncertainties calculated for 24 of the samples are substantially greater (up to 180 MPa). We consider these pressures to be unreliable, and exclude most of them from further consideration. This has no effect on the conclusions, because similar but more reliable estimates of P are obtained for other samples from the same localities. We made an exception for results obtained for three samples from Kverkfjoll (Hoskuldsson et al., 2006
). Pressures calculated for these samples (790–860 ± 160 MPa) are much higher than those obtained for other samples from this locality (
490 MPa), and may provide insight into the evolution of magmas at this volcanic center.
Pressures calculated for 55 samples lie between 0 and –100 MPa and can be interpreted as indicating crystallization at 0·1 MPa if uncertainties are taken into account, and indeed positive pressures close to 0·1 MPa are obtained for other samples from these localities. Nevertheless, we exclude all samples that yield negative pressures from further consideration.
The results demonstrate that Icelandic magmas crystallize over a wide range of P, from 0·1 MPa to 1000 MPa (Table 2), indicating that magmas evolve over a range of depths in the crust and, perhaps, upper mantle. Few pressures exceed 600 MPa, and the majority of magmas appear to have crystallized at P 
300–400 MPa (Fig. 7a). It is important to note that results obtained using the method of Herzberg (2004) also indicate that Icelandic magmas crystallize over a wide range of P, from 0·1 MPa to 750 MPa, although the majority of magmas are predicted to crystallize at 200 MPa rather than the 300–400 MPa found here.
|
|
There is no correlation between calculated P and MgO (Fig. 7b), and this is unusual in light of the positive correlation between P and MgO demonstrated for MORB glasses by Michael & Cornell (1998
). Again, it is important to note that there is also no correlation between MgO and values of P calculated using the method of Herzberg (2004). This indicates that the lack of correlation between P and MgO is not an artifact of our method of calculation. However, for glasses from most localities there is a positive correlation between P and MgO, whereas for glasses from the Hengill complex there is a negative correlation between P and MgO. The unusual negative correlation for the Hengill complex samples is attributed to the effects of crustal assimilation (see discussion below). The apparent lack of correlation between P and MgO in Fig. 7b is therefore the consequence of plotting results for a relatively large number of samples (120) from the Hengill complex together with those for samples from other localities on the same diagram.
Magma temperatures were calculated using the method of Yang et al. (1996) (Table 2). Yang et al. (1996) noted that their geothermometer reproduces temperatures to better than 20°C for most samples. The range of temperatures obtained for Iceland glasses is 1232–1134°C, and is similar to that shown by MORB glasses. A similar range in T (1254–1099°C) is obtained using the olivine–melt geothermometer of Sugawara (2000). As expected, there are positive correlations between T and P (Fig. 7c) and, with the exception of Hengill (see discussion below), between T and MgO.
Examples of results from four localities
Representative results obtained for samples from four localities are shown in Fig. 8. Results obtained for some localities suggest magma evolution at one pressure, for example, 140 ± 80 MPa at Gigoldur and 750 ± 40 MPa at Kalfstindar (Fig. 8a). Results for other localities such as Bláfjall Table Mountain (Fig. 8b) indicate magma evolution at two distinct pressures (590 ± 70 MPa and 100 ± 50 MPa), whereas results for most localities indicate magma evolution over a relatively wide range of pressure rather than at one or two distinct pressures. This is illustrated by results from Laki (380 ± 100 MPa) and Hengill (260 ± 170 MPa) shown in Figs 8c and 9d. The results for Hengill are unusual in that many calculated pressures are close to 0·1 MPa (negative pressures were calculated for 36% of glasses from this locality) and, as described above, there is a negative correlation between P and MgO.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Interpretation of pressure
The calculated pressures reflect the pressure of crystallization only for glasses that represent liquids lying along ol–plag–cpx cotectics. Glass analyses from some localities show compositional variations (e.g. increasing CaO and CaO/Al2O3 with decreasing MgO) that are consistent with crystallization of ol–plag rather than of ol–plag–cpx. Glasses from other localities have compositions consistent with ol–plag–cpx crystallization but occur in samples that lack phenocrysts or microphenocrysts of cpx. In other words, petrographic data provide no evidence for ol–plag–cpx crystallization for these glasses. This is the ‘pyroxene paradox’ recognized for many MORB and other lava suites (e.g. Dungan & Rhodes, 1978
; Fisk et al., 1982; Grove et al., 1992; Elthon et al., 1995). There are several possible explanations for this paradox, including: cotectic crystallization of ol–plag–cpx followed by crystallization of ol and plag (with dissolution of cpx) during ascent, magma mixing and crystallization accompanied by assimilation of gabbroic crust (Meyer et al., 1985; Trønnes, 1990; Hansteen, 1991; Schiellerup, 1995; Hansen & Gronvöld, 2000; Breddam, 2002; Maclennan et al., 2003; Gurenko & Sobolev, 2006). In principle, the results for each locality can be filtered to identify those melts in equilibrium with ol ± plag ± spinel. Most of these provide no constraints on crystallization pressure, but some can be used to place limits on the pressure(s) of ol–plag–cpx cotectic crystallization as illustrated in Fig. 9. (see also Michael & Cornell, 1998; Herzberg, 2004). Melts lying along path a–b in Fig. 9a crystallize ol + plag and have compositions that do not carry a signature of cpx crystallization. Nominal pressures calculated for these melts do not accurately reflect the pressure of magma evolution, but the lowest calculated pressure (for melts with compositions close to b) represents an upper limit for cotectic crystallization.
Melts plotting along the path c–d–e in Fig. 9b isobarically crystallize ol–plag–cpx along the cotectic and then crystallize ol + plag during ascent. Melts saturated in ol + plag with compositions between d and e carry a signature of cpx crystallization, but calculated nominal pressures do not accurately reflect the pressure of magma evolution. In this case, the highest pressure (for melts with compositions close to d) represents a lower limit for cotectic crystallization.
Mixing of ol ± plag saturated melts with ol–plag–cpx saturated cotectic melts (Fig. 9c, path f–g–h) will produce hybrids (e.g. path f–h) with the compositional signature of cpx crystallization (Langmuir, 1989). The lowest pressure calculated for the hybrid melts represents an upper limit for cotectic crystallization.
There is petrographic evidence that some Icelandic magmas, notably those from the Hengill complex, evolve by crystallization of ol + plag combined with assimilation of cpx rather than by cotectic crystallization of ol–plag–cpx (Trønnes, 1990; Hansteen, 1991; Gurenko & Sobolev, 2006). If assimilation occurs after melts leave the ol–plag–cpx cotectic and ascend towards the surface (Fig. 9d, path j–k), as appears to be the case at Hengill, the highest calculated pressure represents a lower limit for cotectic crystallization. It should be noted, however, that it is extremely difficult to discriminate between magmas that evolve via ol–plag–cpx cotectic crystallization and those that evolve via ol + plag crystallization accompanied by assimilation of cpx if petrographic or other (e.g. isotopic) evidence for assimilation is lacking. Indeed, it is possible that the high CaO contents and high CaO/Al2O3 ratios of some glasses reflect assimilation of cpx. Consequently, the estimated pressure of cotectic crystallization may be too low if all samples from one locality (e.g. Hrimalda) have these characteristics (see Fig. 9d). However, the glasses from most localities show a range of CaO and CaO/Al2O3, and it is likely that the highest pressures calculated for these glasses provide a reasonably accurate estimate of the pressure of cotectic crystallization.
Plots of P vs MgO for samples from four localities are shown in Fig. 10 to illustrate interpretation of the results in light of the relationships described above. For each locality, plots of CaO and CaO/Al2O3 vs MgO, together with available petrographic data, were used to discriminate between ol + plag + cpx saturated melts and ol ± plag ± spinel saturated melts. For both Herdubreid and Hlodufell (Fig. 10a and b), there is a positive correlation between P and MgO for glasses in equilibrium with ol ± plag ± spinel whereas there is no correlation between P and MgO for glasses in equilibrium with ol + plag + cpx.
|
Calculated pressures for ol–plag–cpx saturated melts from Herdubreid (Fig. 10a) define two arrays parallel to the MgO axis that are interpreted to indicate cotectic crystallization at 480 ± 30 MPa and 310 ± 20 MPa. The low pressure calculated for one sample could be an aberrant result, possibly caused by analytical error, or could indicate crystallization along an even lower pressure cotectic (
170 MPa). This interpretation is preferred because results obtained for other localities (Hlodufell, Grimsvötn, Efstadalsfjall, Askja, Hrimalda, Kverkfjoll, Theistareykir, Thingvellir) also provide evidence for low-pressure (90–200 MPa) cotectic crystallization.
Pressures calculated for ol–plag–cpx saturated melts from Hlodufell (Fig. 10b) define a single broad array parallel to the MgO axis, and could be interpreted to indicate cotectic crystallization between maximum (PMax = 380 MPa) and minimum (PMin = 290 MPa) pressures. However, the range in pressure is about equal to the expected precision of the results, and therefore, the results are interpreted as indicating cotectic crystallization at 340 ± 40 MPa. As with Herdubreid, the low pressure (170 MPa) calculated for one glass is interpreted to indicate crystallization along a lower pressure cotectic.
Calculated pressures for ol–plag–cpx saturated melts from Grimsvötn (Fig. 10c) show more scatter than those from Hlodufell, and the range is greater than estimated precision. The results are interpreted to indicate cotectic crystallization between maximum (PMax = 510 MPa) and minimum (PMin 100 MPa) pressures, but there is clear evidence for cotectic crystallization at intermediate pressures (see below). A more realistic estimate of the pressure of low-pressure cotectic crystallization is obtained by averaging the results for samples that crystallized at similar pressures. Four samples yield pressures between 100 and 200 MPa, whereas other samples yield pressures >300 MPa (Fig. 11c). The average pressure for the four samples is 150 ± 60 MPa and is the preferred value for low-pressure cotectic crystallization. Preferred values for low-pressure cotectic crystallization have been calculated for Theistreykir, Bardabunga, Grimsvötn, Veidivötn, Katla, Hekla, and Thingvellir. Likewise, results for samples that crystallized at similar high pressures can be averaged and used to calculate preferred values for high-pressure cotectic crystallization at some localities (e.g. Hengill).
|
Finally, glasses from Kalfstindar represent melts saturated with ol + plag, and the lowest calculated pressure provides an upper limit (PUL = 650 MPa) for cotectic crystallization (Fig. 10d). It should be noted that pressures calculated for some ol–plag saturated melts from Herdubreid and Hlodufell are also higher than those calculated for ol–plag–cpx saturated melts from these localities (Fig. 11a and b).
We have noted that results from the Hengill complex are anomalous compared with those obtained from other volcanic centers. Plots of CaO and CaO/Al2O3 vs MgO are consistent with crystallization of ol + plag + cpx. Nominal pressures calculated for 120 glasses range from 0·4 to 754 MPa (Table 2), and the simplest interpretation is that cotectic crystallization occurred over a range of pressures between these values. However, detailed studies by Trønnes (1990), Hansteen (1991), and Gurenko & Sobolev (2006) indicate that this interpretation is oversimplified. Trønnes (1990) has divided Hengill glasses into four groups based primarily on MgO and CaO content. We have followed his subdivision for Groups I and II, but have combined glass analyses from his two other groups into a single group (Group III). Nominal pressures for MgO-rich samples from Group I range from 0·1 to 311 MPa (Fig. 11). These samples contain Al- and Cr-rich endiopside as resorbed phenocrysts and as components of partly disaggregated gabbroic nodules, suggesting that the MgO-rich Group I magmas have assimilated crustal material. The highest calculated pressure for these samples (311 MPa) represents the lower limit for cotectic crystallization (e.g. Fig. 9d). On the other hand, nominal pressures for MgO-poor Group III samples range from 340 to 613 MPa (Fig. 11). Some of these samples contain augite microphenocrysts, suggesting that the MgO-poor Group III magmas have crystallized ol + plag + cpx along cotectics between a maximum (613 MPa) and minimum (340 MPa) pressure. The lower limit of cotectic crystallization for Group I glasses and the minimum pressure of cotectic crystallization for Group III glasses are virtually identical (Fig. 11) and provide a tight constraint on low-pressure cotectic crystallization (325 MPa). Pressures calculated for Group II glasses encompass values obtained for Group I and Group III glasses (Fig. 11). Crystallization of MgO-rich Group I melts at lower pressure than the MgO-poor Group III melts produces the unusual negative correlations between P and MgO and P and T observed for Hengill samples. The strong evidence that Group I magmas have interacted with gabbroic crust suggests that the negative correlation between P and MgO for Hengill samples results from assimilation combined with crystallization as magmas ascend beneath this volcanic complex.
In summary, melts that lie along ol–plag–cpx cotectics can be identified allowing crystallization pressures to be established. Similarly, melts in equilibrium with ol ± plag can be identified and used to place upper limits (PUL) and/or lower limits (PLL) on crystallization pressures. Finally, melts that provide no constraints on crystallization pressure can be identified and filtered out of the results.
Pressures of ol–plag–cpx cotectic crystallization for all localities are listed in Table 3. A striking feature of the results (Fig. 12) is the large number of samples (85%) that have crystallized along either low-P (40–220 MPa) or relatively high-P (430–1030 MPa) cotectics. This near-bimodal distribution of pressures contrasts strongly with the distribution of nominal pressures obtained using unfiltered results (Fig. 7a), and illustrates the importance of evaluating geochemical and petrographic data for samples from the various localities to identify melts that lie along ol–plag–cpx cotectics. It is also clear that some melts crystallized along ol–plag–cpx cotectics at pressures between 220 and 430 MPa (Table 3).
|
|
Effect of H2O
The method used to calculate pressure is based on experiments carried out under nominally anhydrous conditions, and it is important to assess the possible effect of water on the results. Experimental studies show that addition of H2O leads to expansion of the stability field of olivine and contraction of the stability field of plagioclase (Kushiro, 1969
; Nicholls & Ringwood, 1973; Baker & Eggler, 1987; Sisson & Grove, 1993), so that the location of the ol–plag–cpx cotectic will shift with increasing water content at constant pressure. However, relatively high water contents are required to produce large shifts in the position of the cotectic, and the water contents of Icelandic magmas are probably too low to have much effect. Measured water contents range from 0·1 to 1 wt % (Jamtveit et al., 2001; Nichols et al., 2002), and agree with water contents calculated for 88 Icelandic glasses using the method of Danyushevsky et al. (1996) and Danyushevsky (2001) (average 0·18 wt %, range 0–0·94 wt %). Herzberg (2004) found no correlation between calculated pressures and H2O for MORB with 0·1–0·8 wt % H2O. Additional evidence that calculated pressures have not been significantly affected by water is provided by the close agreement (40 ± 40 MPa) between pressures calculated for single samples using two projections (see Appendix A), one from plag onto the plane ol–cpx–qtz, the other from ol onto the plane plag–cpx–qtz. Addition of water causes the ol–plag–cpx cotectic to shift towards cpx (away from ol) in the plag projection and away from cpx (towards plag) in the ol projection. In other words, water affects the stability fields of ol, plag and cpx differently and has a different effect on the positions of the ol–plag–cpx cotectic in these two projections. Accordingly, pressures calculated for hydrous melts will be associated with large uncertainties reflecting large differences between pressures calculated from the two projections (see Appendix A, Table A1). We noted in a preceding section that the uncertainty in calculated pressure for most samples is similar to that estimated from anhydrous experimental data, and it is concluded that the results reported in Tables 2 and 3 closely reflect the pressures of crystallization. Indeed, pressures associated with large uncertainties have been filtered out of the results. It is unlikely that these samples crystallized at higher water contents than most Icelandic magmas, because water contents calculated using the method of Danyushevsky et al. (1996) and Danyshevsky (2001) (average 0·28 wt %, range 0–0·85 wt %) are similar to those obtained for other Icelandic glasses. Use of this method also indicates that samples that yield nominally anhydrous pressures between 0 and –100 MPa did not crystallize from magmas containing unusually high water contents, supporting the interpretation that these samples crystallized at 0·1 MPa taking uncertainties into account.
Depths of magma chambers and magma plumbing systems
The pressures reported in Table 3 are those at which melts are last saturated with ol, plag, and cpx. Ascending magmas must pause and crystallize for a sufficiently long period of time to reach multiple saturation, and this most probably occurs in magma chambers. The melt is then rapidly erupted from the chamber and quenched to form glass. The depths of magma chambers may therefore be calculated from pressures of cotectic crystallization using an appropriate value for crustal density. We have used a value of 2900 kg/m3 to calculate the depths listed in Table 3. Preferred values for the pressure of cotectic crystallization were used in the calculations wherever possible.
Comparison with results obtained using geodetic and/or geophysical methods (see Table 1) reveals agreement for the depths of chambers beneath Askja (6·5 and 18·9 km vs 1·3–3·5 and 16–20 km) and Grimsvötn (5·4 km vs 1·5–4 km), and reasonable agreement for the depth of a chamber beneath Hengill (11·5 km vs 7 km). Our estimate of a minimum depth of 17·5 km for a chamber beneath Hekla is greater than the recent estimate of 11 km based on strain data (see Sturkell et al., 2006), but is consistent with recent analysis of seismic data (Soosalu & Einarsson, 2004), which provide no evidence for a significant magma body at depths <14 km. However, we find no evidence for the shallow chamber (2–5 km) identified beneath Katla in geophysical and geodetic studies. This probably indicates that this chamber does not contain basalt, because Soosalu et al. (2006b) suggested that a cryptodome containing relatively viscous (silica-rich) magma occurs at shallow (2 km) depths beneath Katla.
Comparison with results obtained in other petrological studies (see Table 1) reveals excellent agreement in the case of Bláfjall Table Mountain (6·2 and 18·6 km vs 2–7 and 14–26 km) and the Hengill volcanic complex (11–12 km vs 8–12 km). However, the depths obtained for crystallization at Kistufell (<18 km) are much lower than those obtained by Breddam (2002) (>30 km), and the reason for this discrepancy is unclear.
Our results clearly suggest that magma chambers are located at different depths in the Icelandic crust (Figs 13–15), and this is corroborated by geodetic and geophysical studies. There is evidence for only one chamber located in the shallow (Hrimalda and Gigoldur in the NVZ), middle (Burfell in the NVZ, Raudafell in the WVZ), or deep (Kalfstindar in the WVZ) crust beneath some centers, but there is clear evidence for two or more stacked chambers beneath most centers (Figs 13 and 15). Shallow and deep chambers occur beneath Askja, Bláfjall Table Mountain (including Halar and Seljahalli; see Schiellerup, 1995) and Langjökull (Fig. 15a), whereas shallow, intermediate and deep chambers occur beneath Herdubreid and Efstadalsfjall (Fig. 15b). Dikes presumably connect chambers located at different depths, and provide conduits for magmas to reach the surface.
|
|
|
Results for Theistareykir, Sprengisandur, Kverkfjoll, Bardabunga–Veidivötn, Grimsvötn–Laki, Langjökull, Thingvellir, and Hengill are consistent with the presence of both shallow and deep chambers, and magmas also appear to have crystallized at intermediate depths although there is no evidence that this occurred in a discrete chamber (see Fig. 10c). As discussed above, it is possible that these magmas evolved by high-pressure cotectic crystallization of ol + plag + cpx followed by crystallization of ol + plag (with partial or complete resorption of cpx) during ascent. However, petrographic data indicate that at least some glasses are from magmas that evolved via ol + plag + cpx crystallization, which suggests the presence of cooling and crystallizing magma bodies or pockets throughout the middle and lower crust. Multiple, stacked, discrete chambers might exist beneath some volcanoes as shown in Fig. 15c. However, it is also possible that crystallization occurs in extensive crystal mush zones as suggested by Hansen & Grönvold (2000
). In addition, crystallization probably occurs at various depths in feeder dikes and conduits during lulls in eruptive activity.
Recent work suggests that MORBs partially crystallize over a range of pressure from 1 to 1000 MPa (Michael & Cornell, 1998; Le Roex et al., 2002; Herzberg, 2004). The average crustal thickness along mid-ocean ridges is 7·1 ± 0·9 km (White et al., 1992; Bown & White, 1994), so that crystallization of MORB must begin in the upper mantle. Some Icelandic magmas may also crystallize in the mantle. The glasses have chemical characteristics of evolved magmas and must have formed from primary magmas by crystallization in the deep crust or mantle. Conclusive evidence for crystallization in the mantle is lacking, but the high pressures (up to 800 MPa) obtained for ol ± plag saturated melts from Herdubreid, Efstadalsfjall, and Kalfstindar (Table 2) are consistent with crystallization at upper mantle depths.
Thickness and structure of Icelandic crust
The factors that determine the depth at which magma chambers form are complex and are poorly understood, but include the buoyancy force (reflecting the density contrast between melt and surrounding rocks), any excess pressure over the buoyancy force, and local variations in stress conditions in the crust (Gudmundsson, 2000). It can be assumed to a first approximation that variations in the buoyancy force, excess pressure and stress conditions are negligible for magmas entering the base of the crust along the rift zones. Therefore, the main control of the location of deep chambers is likely to be a lithological boundary associated with a density discontinuity. These chambers occur over a remarkably small range of depths for volcanic centers in the NVZ (19·5 ± 2·6 km, excluding Kverkvjoll) and WVZ (20·9 ± 2·3 km, including samples from Geitafell) (Fig. 13b). The small range of depths for the rift zones (20·1 ± 2·5 km) suggests that the deep chambers occur at or near the crust–mantle boundary (Kelley et al., 2004; Kelley & Barton, 2005). There is excellent agreement between magma chamber depth beneath Theistareykir and crustal thickness inferred from seismic data (Brandsdóttir et al., 1997; Staples et al., 1997) beneath nearby Krafla (19·5 vs 19 km). Likewise, there is very good agreement between magma chamber depths and crustal thickness inferred from seismic data (Bjarnason et al., 1993; Weir et al., 2001) for Geitafell (17·8 ± 0·9 vs 16 km), Hengill (18·9 ± 1·8 vs 17·5–24 km), Hekla (34 ± 2·6 vs 30–35 km), and Katla (23·1 ± 1·6 vs 20–25 km).
Seismic data do not provide reliable estimates of the depth of the crust–mantle transition beneath other centers, and crustal thickesses must be estimated from gravity data (Darbyshire et al., 1998, 2000a; Allen et al., 2002; Kaban et al., 2002; Fedorova et al., 2005; Leftwich et al., 2005) and from the relationship between Moho depth and height above sea level given by Gudmundsson (2003). These methods do not allow small-scale variations in crustal thickness to be resolved, and there is considerable uncertainty in the estimated crustal thickness beneath most volcanoes. In addition, the crust–mantle boundary may represent a transition zone 5 ± 3 km thick (Foulger et al., 2003). Despite this, there is reasonable agreement between estimated magma chamber depth and the base of the crust beneath Kverkfjoll (29 vs 30–35 km) and beneath Bardabunga–Veidivötn and Grimsvötn–Laki (27–29 km vs 30–40 km) given that absolute uncertainties in calculated magma chamber depths are ±3·9 km. These volcanic centers are located on or near the Vatnajökull ice cap (Fig. 1) in a region of thicker crust above the thermal and/or compositional anomaly in the mantle (Darbyshire et al., 2000a; Kaban et al., 2002; Fedorova et al., 2005; Leftwich et al., 2005). Also, depths estimated for magma chambers and the base of the crust are similar for the Bláfjall complex (25 km), Thingvellir (22 km vs 20–25 km), and Kalfstindar (23 km vs 20–25 km).
The agreement between magma chamber depths and Moho depths provides strong evidence that magmas pond at the base of the crust beneath most volcanic centers in Iceland. However, there is poor correlation between Moho depth and magma chamber depth for some centers, and it is possible that the true, maximum depth of chambers beneath these centers is greater than that reported in Table 3. As noted above, some ol–plag-saturated melts from Herdubreid and Efstadalsfjall appear to have crystallized at higher pressure than the maximum value calculated for cotectic melts. Also, we are not confident that the maximum depth of chambers has been established for Langjökull, Sprengisandur and Askja, because of the relatively small number of samples available for study.
The depth of shallow chambers is also relatively constant for volcanic centers along the rift zones (Fig. 13a). The average depths are 5·2 ± 1·6 km for the NVZ, 6·1 ± 2·1 km for the WVZ (including samples from the Reykjanes Peninsula), and 5·4 ± 0·6 km for the EVZ. The average depth for all three rift zones is 5·5 ± 1·6 km, indicating that the chambers are located in upper crust that has an average thickness of 5–7 km (Darbyshire et al., 2000b; Foulger et al., 2003). [Various workers use different definitions of upper, middle and lower crust for Iceland (see Foulger et al., 2003). For this paper, we arbitrarily define upper crust as that <7 km, middle crust as that from 7 to 15 km, and lower crust as that >15 km.] The upper crust is a heterogeneous mixture of flows, hyaloclastites, and intrusive rocks that have been metamorphosed under greenschist- to amphibolite-facies conditions. The observed rapid increase in seismic velocity with depth probably reflects a decrease in the proportion of hyaloclastites and decrease in porosity and permeability (Foulger et al., 2003, and references therein). It seems likely that shallow chambers form at a level of neutral buoyancy at or near the base of the upper crust, although the exact depth may be controlled by local variations in lithology and/or stress conditions (Gudmundsson, 2000).
Additional petrological studies to refine estimates of magma chamber depth are desirable to provide tighter constraints on the thickness of the whole crust and of the upper crust, as well as to define regional variations in these thicknesses. The occurrence of discrete chambers located at intermediate depths (Fig. 14) also has important implications for models of crustal structure. Those in the NVZ and WVZ occur at a relatively constant depth (11·4 ± 0·4 km). There is no convincing evidence for a seismic discontinuity at this depth, and it seems likely that these chambers are located at a phase transition (amphibolite–granulite facies?), and/or at a rheological boundary. Fedorova et al. (2005) suggested that the base of the elastic lithosphere beneath Vatnajökull is 15 km deep, and is underlain by relatively ‘soft’ lower crust. The ‘intermediate depth’ chambers beneath Katla and Hekla are significantly deeper (16–18 km) than the intermediate depth chambers in the active rift zones, which may indicate a different lithological structure or stress regime in the crust of the SFZ.
The evidence for cooling and crystallizing magma bodies or pockets throughout the middle and lower crust is consistent with results obtained by Tryggvason (1986) for Krafla and by Maclennan et al. (2001) for Theistareykir (Table 1). The presence of multiple magma chambers at different depths beneath many volcanic centers implies high geothermal gradients in the crust (see also Meyer et al., 1985; Bjarnason et al., 1993; Maclennan et al., 2001; Kelley et al., 2005; Leftwich et al., 2005). Petrological data therefore strongly suggest that the middle and lower crust is relatively hot and porous, in contrast to the relatively cool, rigid crust proposed in some geophysical studies (e.g. Bjarnason et al., 1993; Menke & Sparks, 1995). However, receiver function data (Darbyshire et al., 2000b), combined surface- and body-wave constraints (Allen et al., 2002), combined results from explosion seismology and receiver functions (Foulger et al., 2003), and combined results from seismic, topography, and gravity data (Fedorova et al., 2005) all provide evidence for low-velocity zones (LVZs) in the crust. These may reflect variations in temperature or lithology, and/or the presence of melt. Most workers favor an explanation involving both high temperatures and the presence of small amounts of melt. Darbyshire et al. (2000b) suggested that a prominent LVZ at 10–15 km reflects the presence of partially molten sills in the lower crust beneath Krafla (see Fig. 15c), whereas Fedorova et al. (2005) proposed that crust deeper than 15 km beneath Vatnajökull contains melt. Allen et al. (2002) identified LVZs in both the upper (0–15 km) and lower (>15 km) crust. The former are thought to reflect regions of high temperature and possible melt, whereas the latter (beneath Vatnajökull) are thought to reflect the thermal halo associated with the fluxing of magma from the mantle to the upper crust. Although detailed correlation of LVZs with the occurrence of magma chambers is not possible, these results suggest that seismic and petrological data can be reconciled to develop an internally consistent model for Icelandic crust.
Crustal accretion models
There is no doubt that some crustal accretion occurs at shallow depths in Iceland by eruption of lavas and hyaloclastites that are buried beneath the products of later eruptions, and by crystallization of gabbros in shallow chambers. Shallow accretion is one of the models proposed for the formation of oceanic crust. Melt is supplied directly from the mantle to a shallow axial melt lens (Sinton & Detrick, 1992). Some of this melt is erupted whereas the rest crystallizes to form gabbros that subside to form oceanic crust. However, it is highly unlikely that the thick (20–40 km) crust in Iceland forms solely by crystallization in shallow chambers to form gabbroic crust that subsides in response to plate extension. It is probable that accretion also results from crystallization in chambers located at Moho depths (underplating), and by crystallization in dikes, melt pockets, or discrete chambers at various depths in the crust (intracrustal accretion). A model for accretion of Icelandic crust is shown in Fig. 16, and is consistent with recent seismic reflection and petrological data for mid-ocean ridges (e.g. Pan & Batiza, 2002, 2003; Maclennan et al., 2004; Singh et al., 2006). These data indicate that lower crust in axial regions is at least locally porous on an intergranular scale, and suggest that accretion occurs by crystallization of small pockets of melt over a range of depths between the shallow melt lens and the Moho.
|
Petrological implications
The results of this study are consistent with complex models of magma evolution that involve polybaric crystallization, magma mixing, and assimilation. Polybaric crystallization is expected in plumbing systems with two or more chambers located at different depths, and supporting evidence is provided by detailed mineral chemical studies of lavas that reveal the presence of different generations of minerals that formed at different pressure (e.g. Hansteen, 1991
; Hansen & Grönvold, 2000; Maclennan et al., 2001). High-pressure crystallization yields residual melts that are compositionally different from those produced by low-pressure crystallization (e.g. Kinzler & Grove, 1992), so that polybaric crystallization will produce liquids lying along different liquid lines of descent (see Figs 6c and 7). It is also likely that these magmas will mix prior to and during sequential eruptive episodes, because magmas that pond and partially crystallize in chambers, dikes, and conduits after one eruptive episode will be flushed out and mix with fresh batches of magma rising from deep chambers or from the mantle during subsequent eruptive episodes. Evidence for mixing has been described by Sigurdsson & Sparks (1981), McGarvie (1984), Mørk (1984), and Fagents et al. (2001). Moreover, crystal aggregates formed by partial crystallization of magma in the plumbing system in an early eruptive phase can be disrupted and incorporated into later batches of magma rising from deep chambers or from the mantle. This yields magmas with complex crystal cargoes derived from different magma batches, such as those described by Hansen & Grönvold (2000). Interaction between ascending melts and the crystalline products of earlier eruptive episodes can lead to complex variations in melt composition (Kvassnes et al., 2003; Danyushevsky et al., 2003), and interaction between ascending melts and clinopyroxene appears to explain the unusual chemical characteristics (e.g. high MgO, CaO, CaO/Al2O3) of some magmas erupted in the Hengill complex and elsewhere along the rift zones (e.g. Trønnes, 1990; Hansen & Grönvold, 2000; Gurenko & Sobolev, 2006).
Oxygen isotope studies provide strong evidence for assimilation of hydrothermally altered crust by Icelandic magmas (e.g. Condomines et al., 1983; Hemond et al., 1988, 1993). Assimilation may be extensive if the middle and lower crust is hot, as suggested above, because less energy is required for a fixed mass of magma to assimilate hot crustal material. Moreover, frequent replenishment with new batches of magma during mixing events leads to thermal buffering in magma chambers, which also facilitates assimilation (Cribb & Barton, 1996). Certainly, the high temperature of basaltic magma emplaced into Icelandic crust [average 1186°C using the method of Yang et al. (1996) or 1177°C using the method of Sugawara (2000)], together with the relatively small temperature range of magmas erupted at single volcanic centers (average 34°C), suggests that abundant heat is available to drive assimilation processes. High geothermal gradients coupled with high magma temperatures will also facilitate melting of crustal lithologies to form silicic magmas, which are relatively abundant on Iceland (Gunnarsson et al., 1998). Our results suggest that silicic magmas can be generated over a relatively wide depth range in the middle and lower crust, and this prediction can be tested by additional studies of appropriate compositions.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
A method to calculate pressures of crystallization of melts lying along ol–plag–cpx cotectics based on the procedure described by Yang et al. (1996
) yields results that are accurate to ±110 MPa (1) and are precise to 80 MPa (1). Pressures calculated for Icelandic glasses from 28 volcanic centers in the Western, Northern and Eastern rift zones as well as from the Southern Flank Zone indicate crystallization range from 1 to 1000 MPa, equivalent to depths of 0–35 km. Magma chamber depths estimated from these results agree well with those estimated using other methods for Askja, Bláfjall Table Mountain, Grimsvötn, Hengill, and Hekla. Deep chambers (>17 km) occur beneath most volcanic centers and appear to be located at the Moho, indicating that magma ponds at the crust–mantle boundary. Shallow chambers (<7·1 km) also occur beneath most volcanic centers. These are located in the upper crust, and probably form at a level of neutral buoyancy. There is good evidence for cooling and crystallizing bodies and pockets of magma throughout the middle and lower crust, which probably resembles a crystal mush. This strongly suggests that the middle and lower crust is relatively hot and porous, and that crustal accretion occurs over a range of depths as inferred in recent models for crustal accretion at mid-ocean ridges. The presence of multiple, stacked chambers and hot, porous crust suggests that magma evolution is complex and involves polybaric crystallization, magma mixing, and assimilation.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Supplementary data for this paper are available at Journal of Petrology online.
|
APPENDIX A: CALCULATION OF PRESSURE |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Yang et al. (1996
, table 3) presented three equations that allow direct calculation of pressure. These equations are used to calculate a series of liquid compositions (LP) lying along the ol–plag–cpx cotectic for the sample of interest at increments of 100 MPa (see Table A1). The liquid compositions are converted to normative mineral components using the procedure described by Grove et al. (1993) assuming that Fe = FeO, and projected from plag onto the plane ol–cpx–qtz and from ol onto the plane plag–cpx–qtz. The pressure dependence of each normative mineral component in the predicted liquids (LP) is found by regression, and the pressure of crystallization is found from the regression equations using the projected normative mineral components for the original sample (LS). Thus for the projection from plag, values of P are calculated from predicted and observed ol, cpx and qtz, whereas for the projection from ol, values of P are calculated from predicted and observed plag, cpx and qtz. We have used these two projections because most basalt melts are saturated with plag and ol, and obtain six values of P for each sample. The average value is taken as the pressure of crystallization, and all values are used to calculate the uncertainty (1) associated with the calculated pressure. An Excel spreadsheet to perform these calculations is available as Supplementary Data at http://www.petrology.oxfordjournals.org/.
The approach described above uses all three of the equations given by Yang et al. (1996) to calculate pressure. In contrast, Michael & Cornell (1998) used one of the equations of Yang et al. [1996, equation (2)] to calculate the pressures of crystallization of MORB. Results obtained using the two methods are compared in Appendix B.
|
APPENDIX B: ACCURACY AND PRECISION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
In principle, accuracy can be determined by comparing pressures calculated for glasses produced in experiments with the reported run pressure. A problem with this approach is that experimental glass compositions show considerable variability, much greater than that expected from analytical uncertainties, which reflects problems inherent in experimental studies (e.g. Ford et al., 1983
; Yang et al., 1996; Falloon et al., 2001). These include uncertainties in measurement of pressure and temperature, unknown amounts of volatiles in nominally anhydrous experiments, loss of Fe2+ and alkalis from the charge during the experiment, modification of melt compositions by quench crystallization, and run durations too short for the attainment of equilibrium. It appears that some experimental glasses reported to be in equilibrium with ol, plag, and cpx do not have compositions that represent liquids lying on the ol–plag–cpx cotectic. Accordingly, large differences are expected between calculated and experimental pressures, leading to poor estimates of accuracy.
The experimental glass compositions used to construct Fig. 2 are taken from the sources listed in the figure caption. Glass compositions from experiments on natural samples were combined with glass analyses from experiments by Shi (1993) in the system CaO–MgO–Al2O3–SiO2–FeO–Na2O and used to map the likely positions of the cotectic at different pressures (samples with anomalous compositions, that plot off the main data array at a given pressure, were discarded). The actual positions of the cotectic were located using a subset of these glasses that have compositions that form tightly defined arrays in projections such as those shown in Fig. 2, and therefore appear to anchor the positions of the cotectic at different pressures. The glass compositions in both datasets (excluding alkaline compositions and glasses from Shi's experiments) were used to test the accuracy of pressures calculated for sub-alkaline and transitional compositions. For the extended dataset (91 glasses), calculated pressures agree with experimental pressures to ±120 MPa (1), which is about the same as uncertainties for pressures calculated by similar methods (Herzberg, 2004). For the smaller dataset (59 glasses), calculated pressures agree with experimental pressures to ±90 MPa (1). These results suggest that reported pressures are accurate to 110 MPa.
Pressures were also calculated using the method described by Michael & Cornell (1998). For the extended dataset (91 glasses), pressures calculated using equation (2) of Yang et al. (1996) agree with experimental pressures to ±160 MPa (1), whereas for the smaller dataset (59 glasses), calculated pressures agree with experimental pressures to ±120 MPa (1). These results indicate that pressures calculated using the method used here are more accurate than those estimated using the method described by Michael & Cornell (1988).
Table A1: Example of method used to calculate pressure
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Results: average P = 6·70 kbar (1 = 0·42 kbar, range = 1·17 kbar); the average, 1, and range were calculated from six pressures obtained by regression. 
Plag–Ol = 0·30 (where Plag–Ol is the difference between average P calculated for Plag projection and average P calculated for Ol projection).
1Glass analysis from Schiellerup (1995).
2Projected mineral components for sample 1.7.4 calculated using procedure of Tormey et al. (1987).
3Projected mineral components calculated at the pressures listed in Column 1 using procedure of Tormey et al. (1987).
4Slope and intercept from regression of projected mineral components (Columns 3–8) vs P (Column 1).
5Pressures calculated from regressions and projected mineral components for sample 1.7.4.
Yang et al. (1996) estimated the uncertainty in pressure estimates (precision) from the standard deviation of the mean of replicate electron microprobe analyses, and obtained a value of about ±50 MPa (2). The uncertainties in calculated pressures for the experimental datasets determined as described in Appendix A are somewhat larger than this value, ±60–80 MPa (1).
|
ACKNOWLEDGEMENTS |
|---|
We thank Wendy Panero, Ralph von Frese, Hal Noltimier and Tim Leftwich for their interest and for helpful discussions. Peter Meyer and Haraldur Sigurdsson kindly provided information about sample localities. Thoughtful comments by David Elliot, Wendy Panero, Ingi Bjarnason, Claude Herzberg, John Maclennan, and an anonymous reviewer significantly improved the manuscript. We are also grateful to Colin Devey and Marjorie Wilson for their editorial efforts. Financial support from the Ohio State University and from the Friends of Orton Hall is gratefully acknowledged.
*Corresponding author. Telephone: (614)-292-2721. Fax: (614)-292-7688. E-mail: kelley.196{at}osu.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDMETHODS AND DATARESULTSDISCUSSIONCONCLUSIONSSUPPLEMENTARY DATAAPPENDIX A: CALCULATION OF...APPENDIX B: ACCURACY AND...REFERENCES |
|---|
Alfaro R, Brandsdóttir B, Rowlands DP, White R, Gudmundsson MT. Structure of the Grímsvötn central volcano under the Vatnajökull icecap, Iceland. Geophysical Journal International (2007) 168:863–876.[CrossRef][Web of Science]
Allen RM, Nolet G, Morgan WJ, Vogfjord K, Nettles M, Ekstrom G, Bergsson BH, Erlendsson P, Foulger GR, Jokobsdottir S, Julian BR, Pritchard M, Ragnarsson S, Stefansson R. Plume-driven plumbing and crustal formation in Iceland. Journal of Geophysical Research—Solid Earth (2002) 107. No. B8, 2163, doi:10.1029/2001JB000584.
Arnadottir T, Delaney PT, Sigmundsson F. Sources of crustal deformation associated with the Krafla, Iceland, eruption of September 1984. Geophysical Research Letters (1998) 25:1043–1046.[CrossRef][Web of Science]
Baker DR, Eggler DH. Compositions of anhydrous and hydrous melts coexisting with plagioclase, augite, and olivine or low-Ca pyroxene from 1 atm to 8 kbar; application to the Aleutian volcanic center of Atka. American Mineralogist (1987) 72:12–28.[Abstract]
Bender JF, Bence AE, Hodges FN. Petrogenesis of basalts from the Project FAMOUS area; experimental study from 0 to 15 kbars. Earth and Planetary Science Letters (1978) 41:277–302.[CrossRef][Web of Science]
Bjarnason IT, Caress D, Flovenz OG, Menke W. Tomographic image of the Mid-Atlantic plate boundary in southwestern Iceland. Journal of Geophysical Research, B, Solid Earth and Planets (1993) 98:6607–6622.[CrossRef]
Björnsson A. Dynamics of crustal rifting in NE Iceland. Journal of Geophysical Research (1985) 90:10151–10162.[CrossRef]
Bjornsson A, Johnsen G, Sigurdsson S, Thorbergsson G, Tryggvason E. Rifting of the plate boundary in north Iceland 1975–1978. Journal of Geophysical Research (1979) 84:3029–3038.
Bown JW, White RS. Variation with spreading rate of oceanic crustal thickness and geochemistry. In: Earth and Planetary Science Letters (1994) 121. Amsterdam: Elsevier. 435–449.[CrossRef][Web of Science]
Brandsdottir B, Einarsson P, Menke W, Staples RK, White RS. Faroe–Iceland Ridge experiment; 2, Crustal structure of the Krafla central volcano. Journal of Geophysical Research, B, Solid Earth and Planets (1997) 102:7867–7886.[CrossRef]
Breddam K. Kistufell; primitive melt from the Iceland mantle plume. Journal of Petrology (2002) 43:345–373.
Breddam K, Kurz MD, Storey M. Mapping out the conduit of the Iceland mantle plume with helium isotopes. In: Earth and Planetary Science Letters (2000) 176. Amsterdam: Elsevier. 45–55.[CrossRef][Web of Science]
Condomines M, Groenvold K, Hooker PJ, Muehlenbachs K, O’Nions RK, Oskarsson N, Oxburgh ER. Helium, oxygen, strontium and neodymium isotopic relationships in Icelandic volcanics. Earth and Planetary Science Letters (1983) 66:125–136.[CrossRef][Web of Science]
Cribb JW, Barton M. Geochemical effects of decoupled fractional crystallization and crustal assimilation. Lithos (1996) 37:293–307.[CrossRef][Web of Science]
Danyushevsky LV. The effect of small amounts of H2O on crystallisation of mid-ocean ridge and backarc basin magmas. Journal of Volcanology and Geothermal Research (2001) 110:265–280.[CrossRef][Web of Science]
Danyushevsky LV, Dmitriev LV, Sobolev AV. Estimation of the pressure of crystallization and H2O content of MORB and BABB glasses. Mineralogy and Petrology (1996) 57:185–204.[CrossRef][Web of Science]
Danyushevsky LV, McNeill AW, Sobolev AV. Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas; an overview of techniques, advantages and complications. Chemical Geology (2002) 183:5–24.[CrossRef][Web of Science]
Danyushevsky LV, Eggins SM, Falloon TJ, Perfit MR. Crustal origin for coupled ‘ultra-depleted’ and ‘plagioclase’ signatures in MORB olivine-hosted melt inclusions; evidence from the Siqueiros transform fault, East Pacific Rise. Contributions to Mineralogy and Petrology (2003) 144:619–637.[Web of Science]
Darbyshire FA, Bjarnason IT, Flovenz OG, White RS. Crustal structure above the Iceland mantle plume imaged by the ICEMELT refraction profile. Geophysical Journal International (1998) 135:1131–1149.[CrossRef][Web of Science]
Darbyshire FA, Gudmundsson GB, Jakobsdottir SS, Priestley KF, Stefansson R, White RS. Central structure of central and northern Iceland from analysis of teleseismic receiver functions. Geophysical Journal International (2000a) 143:163–184.[CrossRef][Web of Science]
Darbyshire FA, Priestley KF, White RS. Structure of the crust and uppermost mantle of Iceland from a combined seismic and gravity study. Earth and Planetary Science Letters (2000b) 181:409–428.[CrossRef][Web of Science]
de Zeeuw-van Dalfsen E, Pagli C, Pedersen R, Sigmundsson F. Satellite radar interferometry 1993–1999 suggests deep accumulation of magma near the crust–mantle boundary at the Krafla volcanic system, Iceland. Geophysical Research Letters (2004) 31:5.
de Zeeuw-van Dalfsen E, Rymer H, Sigmundsson F, Sturkell E. Net gravity decrease at Askja Volcano, Iceland; constraints on processes responsible for continuous caldera deflation, 1988–2003. Journal of Volcanology and Geothermal Research (2005) 139:227–239.[CrossRef][Web of Science]
de Zeeuw-van Dalfsen E, Rymer H, Williams-Jones G, Sturkell E, Sigmundsson F. Integration of micro-gravity and geodetic data to constrain shallow system mass changes at Krafla Volcano, N. Iceland. Bulletin of Volcanology (2006) 68:420–431.[CrossRef][Web of Science]
Dungan MA, Rhodes JM. Residual glasses and melt inclusions in basalts from DSDP Legs 45 and 46; evidence for magma mixing. Contributions to Mineralogy and Petrology (1978) 67:417–431.[CrossRef][Web of Science]
Einarsson P. S-wave shadows in the Krafla Caldera in NE Iceland, evidence for a magma chamber in the crust. Bulletin Volcanologique (1978) 41:187–195.[CrossRef]
Einarsson P, Björnsson S. Earthquakes in Iceland. Jökull (1980) 29:37–43.
Elthon D, Meen JK, Ross DK. Composition variations of basaltic glasses from the Mid-Cayman Rise spreading center. Journal of Geophysical Research, B, Solid Earth and Planets (1995) 100:12497–12512.[CrossRef]
Ewart JA, Bjornsson A, Voight B. Elastic deformation models of Krafla Volcano, Iceland, for the decade 1975 through 1985. In: Bulletin of Volcanology (1991) 53:436–459.[CrossRef][Web of Science]
Eysteinsson H, Hermance JF. Magnetotelluric measurements across the Eastern Neovolcanic Zone in South Iceland. Journal of Geophysical Research, B (1985) 90:10093–10103.[CrossRef]
Fagents SA, McGarvie DW. Dynamics of the Ljotipollur phreatomagmatic eruption, S. central Iceland. EOS Transactions, American Geophysical Union (2001) 82:V52A–1044. (Fall Meeting Supplement, Abstract).
Falloon TJ, Danyushevsky LV, Green DH. Peridotite melting at 1 GPa; reversal experiments on partial melt compositions produced by peridotite–basalt sandwich experiments. Journal of Petrology (2001) 42:2363–2390.
Fedorova T, Jacoby WR, Wallner H. Crust–mantle transition and Moho model for Iceland and surroundings from seismic, topography, and gravity data. Tectonophysics (2005) 396:119–140.[CrossRef][Web of Science]
Feigl KL, Gasperi J, Rigo A, Sigmundsson F. Crustal deformation near Hengill Volcano, Iceland 1993–1998; coupling between magmatic activity and faulting inferred from elastic modeling of satellite radar interferograms. Journal of Geophysical Research, B, Solid Earth and Planets (2000) 105:25655–25670.[CrossRef]
Fisk MR, Bence AE, Schilling JG. Major element chemistry of Galapagos rift zone magmas and their phenocrysts. Earth and Planetary Science Letters (1982) 61:171–189.[CrossRef][Web of Science]
Ford CE, Craven JA, Fisk MR, Russell DG. Olivine–liquid equilibria; temperature, pressure and composition dependence of the crystal/liquid cation partition coefficients for Mg, Fe2+, Ca and Mn. Journal of Petrology (1983) 24:256–265.[Web of Science]
Foulger GR, Anderson DL. A cool model for the Iceland hotspot. Journal of Volcanology and Geothermal Research (2005) 141:1–22.[CrossRef][Web of Science]
Foulger GR, Du Z, Julian BR. Icelandic-type crust. Geophysical Journal International (2003) 155:567–590.[CrossRef][Web of Science]
Frost BR, Lindsley DH. Equilibria amount Fe–Ti oxides, pyroxenes, olivine, and quartz: Part II. Application. American Mineralogist (1992) 77:1004–1020.[Abstract]
Furman T, Frey FA, Park K. Chemical constraints on the petrogenesis of mildly alkaline lavas from Vestmannaeyjar, Iceland; the Eldfell (1973) and Surtsey (1963–1967) eruptions. Contributions to Mineralogy and Petrology (1991) 109:19–37.[CrossRef][Web of Science]
Gebrande H, Einarsson P, Miller H. Seismic structure of Iceland along RRISP-Profile I. Zeitschrift für Geophysik (1980) 47:239–249.
Grove TL, Bryan WB. Fractionation of pyroxene-phyric MORB at low pressure; an experimental study. Contributions to Mineralogy and Petrology (1983) 84:293–309.[CrossRef][Web of Science]
Grove TL, Kinzler RJ, Bryan WB. Fractionation of Mid-Ocean Ridge Basalt (MORB). In: Mantle Flow and Melt Migration at Mid-Ocean Ridges, American Geophysical Monograph—Phipps-Morgan J, et al, eds. (1993) 71:281–311.
Grove TL, Gerlach DC, Sando TW, Baker MB. Origin of calc-alkaline series lavas at medicine lake volcano by fractionation, assimilation and mixing—corrections and clarifications (vol 82, pg 407, 1983). Contributions to Mineralogy and Petrology (1993) 114:422–424.[CrossRef]
Gudmundsson A. Effect of tensile stress concentration around magma chambers on intrusion and extrusion frequencies. Journal of Volcanology and Geothermal Research (1988) 35:179–194.[CrossRef][Web of Science]
Gudmundsson A. Dynamics of volcanic systems in Iceland: Example of tectonism and volcanism at juxtaposed hot spot and mid-ocean ridge systems. Annual Review of Earth and Planetary Sciences (2000) 28:107–140.[CrossRef][Web of Science]
Gudmundsson O. The dense root of the Iceland crust. Earth and Planetary Science Letters (2003) 206:427–440.[CrossRef][Web of Science]
Gudmundsson MT, Milsom J. Gravity and magnetic studies of the subglacial Grimsvötn Volcano, Iceland; implications for crustal and thermal structure. Journal of Geophysical Research, B, Solid Earth and Planets (1997) 102:7691–7704.[CrossRef]
Gudmundsson O, Brandsdottir B, Menke W, Sigvaldason GE. The crustal magma chamber of the Katla volcano in South Iceland revealed by 2-D seismic undershooting. Geophysical Journal International (1994) 119:277–296.[CrossRef][Web of Science]
Gunnarsson B, Marsh BD, Taylor HP Jr. Generation of Icelandic rhyolites; silicic lavas from the Torfajökull central volcano. Journal of Volcanology and Geothermal Research (1998) 83:1–45.[CrossRef][Web of Science]
Gurenko AA, Sobolev AV. Crust–primitive magma interaction beneath neovolcanic rift zone of Iceland recorded in gabbro xenoliths from Midfell, SW Iceland. Contributions to Mineralogy and Petrology (2006) 151:495–520.[CrossRef][Web of Science]
Hansen H, Grönvold K. Plagioclase ultraphyric basalts in Iceland; the mush of the rift. Journal of Volcanology and Geothermal Research (2000) 98:1–32.[CrossRef][Web of Science]
Hansteen TH. Multi-stage evolution of the picritic Maelifell rocks, SW Iceland; constraints from mineralogy and inclusions of glass and fluid in olivine. Contributions to Mineralogy and Petrology (1991) 109:225–239.[CrossRef][Web of Science]
Hardarson BS, Ellam RM, Fitton JG, Pringle MS. Rift relocation; a geochemical and geochronological investigation of a paleo-rift in Northwest Iceland. Earth and Planetary Science Letters (1997) 153:181–196.[CrossRef][Web of Science]
Hards VL, Greenwood PB, Kempton PD, Thompson RN. The magmatic evolution of the Snaefell volcanic centre; an example of volcanism during incipient rifting in Iceland. Journal of Volcanology and Geothermal Research (2000) 99:97–121.[CrossRef][Web of Science]
Hemond C, Allègre CJ, Condomines M, Fourcade S, Javoy M, Oskarsson N. Thorium, strontium and oxygen isotopic geochemistry in Recent tholeiites from Iceland; crustal influence on mantle-derived magmas. Earth and Planetary Science Letters (1988) 87:273–285.[CrossRef][Web of Science]
Hemond C, Arndt NT, Hofmann AW, Lichtenstein U, Oskarsson N, Steinthorsson S. The heterogeneous Iceland plume; Nd–Sr–O isotopes and trace element constraints. Journal of Geophysical Research, B, Solid Earth and Planets (1993) 98:15833–15850.[CrossRef]
Herzberg C. Partial crystallization of mid-ocean ridge basalts in the crust and mantle. Journal of Petrology (2004) 45:2389–2405.
Hoskuldsson A, Sparks RSJ, Carroll MR. Constraints on the dynamics of subglacial basalt eruptions from geological and geochemical observations at Kverkfjoll, NE Iceland. Bulletin of Volcanology (2006) 68:689–701.[CrossRef][Web of Science]
Jamtveit B, Brooker R, Brooks K, Larson LM, Pederson T. The water content of olivines from the North Atlantic Volcanic Province. Earth and Planetary Science Letters (2001) 186:401–415.[CrossRef][Web of Science]
Jónasson K, Holm PM, Pedersen AK. Petrogenesis of silicic rocks from the Króksfjördur central volcano, NW Iceland. Journal of Petrology (1992) 33:1345–1369.
Jónsson S, Bjornsson G, Pedersen R, Segall P. Post-earthquake ground movements correlated to pore-pressure transients. Nature (2003) 424:179–183.[CrossRef]
Juster TC, Grove TL, Perfit MR. Experimental constraints on the generation of FeTi basalts, andesites, and rhyodacites at the Galapagos spreading center, 85°W and 95°W. Journal of Geophysical Research, B, Solid Earth and Planets (1989) 94:9251–9274.[CrossRef]
Kaban MK, Flovenz OG, Palmason G. Nature of the crust–mantle transition zone and the thermal state of the upper mantle beneath Iceland from gravity modelling. Geophysical Journal International (2002) 149:281–299.[CrossRef][Web of Science]
Kaban MK, Artemieva IM, Mooney WD, Schwintzer P. Density of the continental roots; compositional and thermal contributions. Earth and Planetary Science Letters (2003) 209:53–69.[CrossRef][Web of Science]
Kelley DF, Barton M. Thickness, composition, and physical properties of crust in Iceland's neovolcanic zone. EOS Transactions, American Geophysical Union (2005) 86(52):V13E–0591. Fall Meeting Supplement.
Kelley DF, Kapostasy DD. Depths of magma chambers in the Icelandic crust. EOS Transactions, American Geophysical Union (2004) 85(17). Joint Assembly Supplement, V54A-02.
Kelley DF, Leftwich TE. Petrologic constraints on Iceland's lower crust. EOS Transactions, American Geophysical Union (2005) 86(18). Joint Assembly Supplement, V13A-02.
Kinzler RJ, Grove TL. Primary magmas of mid-ocean ridge basalts; 1, Experiments and methods. Journal of Geophysical Research, B, Solid Earth and Planets (1992) 97:6885–6906.[CrossRef]
Kjartansson E, Grönvold K. Location of a magma reservoir beneath Hekla Volcano, Iceland. Nature (1983) 301:139–141.[CrossRef]
Kushiro I. The system forsterite–diopside–silica with and without water at high pressures. American Journal of Science (1969) 267-A:269–294.[Web of Science]
Kvassnes HJ, Dick JB, Grove TL. The Atlantis Bank gabbro-suite was not a ‘normal’ magma-chamber that produced basalts. Geophysical Research Absracts (2003) 5:12203.
LaFemina PC, Arnadottir T, Dixon TH, Einarsson P, Malservisi R, Sigmundsson F, Sturkell E. Geodetic GPS measurements in south Iceland; strain accumulation and partitioning in a propagating ridge system. Journal of Geophysical Research (2005) 110:21.
Langmuir CH. Geochemical consequences of in situ crystallization. Nature (1989) 340:199–205.[CrossRef]
Langmuir CH, Klein EM, Plank T. Petrological systematics of Mid-Ocean Ridge Basalts: Constraints on melt generation beneath ocean ridges. In: Mantle Flow and Melt Migration at Mid-Ocean Ridges, American Geophysical Monograph—Phipps-Morgan J, et al, eds. (1992) 71:183–280.
Le Bas MJ, Le Maitre RW, Streckeisen A, Zanettin B. A chemical classification of volcanic rocks based on the total alkali-silica diagram. Journal of Petrology (1986) 27:745–750.
Le Roux PJ, Le Roex AP, Schilling J. Crystallization processes beneath the southern Mid-Atalantic Ridge (40–50°S), evidence for high-pressure initiation of crystallization. Contributions to Mineralogy and Petrology (2002) 142:582–602.[Web of Science]
Leftwich TE, von Frese RRB, Potts LV, Kim GR, Roman DR, Taylor PT, Barton M. Crustal modeling of the North Atlantic from spectrally correlated free-air gravity and terrain gravity. Journal of Geodynamics (2005) 40:23–50.[Web of Science][Medline]
Linde AT, Agustsson K, Sacks IS, Stefansson R. Mechanism of the 1991 eruption of Hekla from continuous borehole strain monitoring. Nature (1993) 365:737–740.[CrossRef]
Macdonald GA. Composition and origin of Hawaiian lavas. In: Studies in Volcanology: A Memoir in Honor of Howel Williams, Geological Society of America. Memoir—Coats RR, Hay RL, Anderson CA, eds. (1968) 116:477–522.
Maclennan J, Gronvold K, McKenzie D. Plume-driven upwelling under central Iceland. Earth and Planetary Science Letters (2001) 194:67–82.[CrossRef][Web of Science]
Maclennan J, Eiler JM, Gronvold K, Kitchen N, McKenzie D, Shimizu N. Melt mixing and crystallization under Theistareykir, northeast Iceland. Geochemistry, Geophysics, Geosystems (2003) 4. doi:10.1029/2003GC000558.
Maclennan J, Hulme T, Singh SC. Thermal models of oceanic crustal accretion: Linking geophysical, geological and petrological observations. Geochemistry, Geophysics, Geosystems (2004) 5. doi: 10.1029/2003GC000605.
Marti J, Folch A. Anticipating volcanic eruptions. In: Volcanoes and the Environment—Ernst GGJ, ed. (2005) Cambridge: Cambridge University Press. 90–120.
McCann, V. & Barton M. Oxygen fugacities of lavas from Iceland and implications for mantle redox states. EOS Transactions, American Geophysical Union (2004) 85(17):U53A–06. Joint Assembly Supplement.
McGarvie DW. Torfajoekull; a volcano dominated by magma mixing. Geology (1984) 12:685–688.
Menke W, Brandsdottir B, Sparks D, West M. Compressional and shear velocity structure of the lithosphere in Northern Ireland. Bulletin of the Seismological Society of America (1998) 88:1561–1571.
Menke W, Sparks D. Crustal accretion model for Iceland predicts ‘cold’ crust. Geophysical Research Letters (1995) 22:1673–1676.[CrossRef][Web of Science]
Meyer PS, Schilling J, Sigurdsson H. Petrological and geochemical variations along Iceland's neovolcanic zones. Journal of Geophysical Research, B (1985) 90:10043–10072.[CrossRef]
Michael PJ, Cornell WC. Influence of spreading rate and magma supply on crystallization and assimilation beneath mid-ocean ridges; evidence from chlorine and major element chemistry of mid-ocean ridge basalts. Journal of Geophysical Research, B, Solid Earth and Planets (1998) 103:18325–18356.[CrossRef]
Moore JG, Calk LC. Degassing and differentiation in subglacial volcanoes, Iceland. Journal of Volcanology and Geothermal Research (1991) 46:157–180.[CrossRef][Web of Science]
Mørk MBE. Magma mixing in the post-glacial Veidivoetn fissure eruption, Southeast Iceland; a microprobe study of mineral and glass variations. Lithos (1984) 17:55–75.[CrossRef][Web of Science]
Nicholls IA, Ringwood AE. Effect of water on olivine stability in tholeiites and the production of silica saturated magmas in the island arc environment. Journal of Geology (1973) 81:285–300.[Web of Science]
Nichols ARL, Carroll MR, Hoskuldsson A. Is the Iceland hot spot also wet? Evidence from the water contents of undegassed submarine and subglacial pillow basalts. Earth and Planetary Science Letters (2002) 202:77–87.[CrossRef][Web of Science]
O’Hara MJ. Are ocean floor basalts primary magma? Nature (1968) 220:683–686.[CrossRef]
Oskarsson N, Sigvaldason GE, Steinthorsson S. Iceland geochemical anomaly; origin, volcanotectonics, chemical fractionation, and isotope evolution of the crust. Journal of Geophysical Research, B (1985) 90:10011–10025.[CrossRef]
Pagli C, Arnadottir T, Einarsson P, Sigmundsson F, Sturkell E. Deflation of the Askja volcanic system; constraints on the deformation source from combined inversion of satellite radar interferograms and GPS measurements. Journal of Volcanology and Geothermal Research (2006) 152:97–108.[CrossRef][Web of Science]
Pan Y, Batiza R. Mid-ocean ridge magma chamber processes; constraints from olivine zonation in lavas from the East Pacific Rise at 9°30'N and 10°30'N. Journal of Geophysical Research, B, Solid Earth and Planets (2002) 107:13.
Pan Y, Batiza R. Magmatic processes under mid-ocean ridges; a detailed mineralogic study of lavas from East Pacific Rise 9°30'N, 10°30'N, and 11°20'N. Geochemistry, Geophysics, Geosystems (2003) 4. doi:10.1029/2002GC000309.
Pedersen R, Sigmundsson F. InSAR based sill model links spatially offset area of deformation and seismicity for the 1994 unrest episode at Ejyafjallajökull Volcano, Iceland. Geophysical Research Letters (2004) 31:5.
Pedersen R, Sigmundsson F. Temporal developement of the 1999 intrusive episode in the Eyjafjallajökull volcano, Iceland, derived from InSAR images. Bulletin of Volcanology (2006) 68:377–393.[CrossRef][Web of Science]
Révillon S, Arndt NT, Hallot E, Kerr AC, Tarney J. Petrogenesis of picrites from the Caribbean Plateau and the North Atlantic magmatic province. Lithos (1999) 49:1–21.[CrossRef][Web of Science]
Rymer H, Tryggvason E. Gravity and elevation changes at Askja, Iceland. Bulletin of Volcanology (1993) 55:361–371.
Rymer H, Cassidy J, Locke CA, Sigmundsson F. Post-eruptive gravity changes from 1990 to 1996 at Krafla Volcano, Iceland. Journal of Volcanology and Geothermal Research (1998) 87:141–149.[CrossRef][Web of Science]
Sack RO, Carmichael ISE, Walker D. Experimental petrology of alkalic lavas; constraints on cotectics of multiple saturation in natural basic liquids. Contributions to Mineralogy and Petrology (1987) 96:1–23.[CrossRef][Web of Science]
Schiellerup H. Generation and equilibration of olivine tholeiites in the northern rift zone of Iceland; a petrogenetic study of the Bláfjall table mountain. Journal of Volcanology and Geothermal Research (1995) 65:161–179.[CrossRef][Web of Science]
Schilling JG. Iceland mantle plume. Nature (1973) 246:141–143.[CrossRef]
Shi P. Low-pressure phase relationships in the system Na2O–CaO–FeO–MgO–Al2O3–SiO2 at 1100°C, with implications for the differentiation of basaltic magmas. Journal of Petrology (1993) 34:743–762.
Sigmundsson F, Bilham R, Einarsson P. Magma chamber deflation recorded by the Global Positioning System; the Hekla 1991 eruption. Geophysical Research Letters (1992) 19:1483–1486.[CrossRef][Web of Science]
Sigmundsson F, Einarsson P, Foulger GR, Hodgkinson KM, Rognvaldsson ST, Thorbergsson G. The 1994–1995 seismicity and deformation at the Hengill triple junction, Iceland; triggering of earthquakes by minor magma injection in a zone of horizontal shear stress. Journal of Geophysical Research, B, Solid Earth and Planets (1997) 102:15151–15161.[CrossRef]
Sigurdsson H, Sparks RSJ. Rifting episode in North Iceland in 1874–1875 and the eruptions of Askja and Sveinagja. Bulletin Volcanologique (1978) 41:149–167.
Sigurdsson H, Sparks RSJ. Petrology of rhyolitic and mixed magma ejecta from the 1875 eruption of Askja, Iceland. Journal of Petrology (1981) 22:41–84.
Singh SC, Harding AJ, Kent GM, Sinha MC, Combier V, Bazin S, Tong CH, Pye JW, Barton PJ, Hobbs RW, White RS, Orcutt JA. Seismic reflection images of the Moho underlying melt sills at the East Pacific Rise. Nature (2006) 442:287–290.[CrossRef][Medline]
Sinton JM, Detrick RS. Mid-ocean ridge magma chambers. Journal of Geophysical Research, B, Solid Earth and Planets (1992) 97:197–216.[CrossRef]
Sinton J, Gronvold K, Saemundsson K. Postglacial eruptive history of the Western volcanic zone, Iceland. Geochemistry, Geophysics, Geosystems (2005) 6. doi:10.1029/2005GC001021.
Sisson TW, Grove TL. Temperatures and H2O contents of low-MgO high-alumina basalts. Contributions to Mineralogy and Petrology (1993) 113:167–184.[CrossRef][Web of Science]
Slater L, Gronvold K, McKenzie D, Shimizu N. Melt generation and movement beneath Theistareykir, NE Iceland. Journal of Petrology (2001) 42:321–354.
Soosalu H, Einarsson P. Seismicity around the Hekla and Torfajökull volcanoes, Iceland, during a volcanically quiet period, 1991–1995. Bulletin of Volcanology (1997) 59:36–48.[CrossRef][Web of Science]
Soosalu H, Einarsson P. Seismic constraints on magma chambers at Hekla and Torfajökull volcanoes, Iceland. Bulletin of Volcanology (2004) 66:276–286.[CrossRef][Web of Science]
Soosalu H, Jonsdottir K, Einarsson P. Seismicity crisis at the Katla volcano, Iceland—signs of a cryptodome? Journal of Volcanology and Geothermal Research (2006a) 153:177–186.[CrossRef][Web of Science]
Soosalu H, Lippitsch R, Einarsson P. Low-frequency earthquakes at the Torfajökull volcano, south Iceland. Journal of Volcanology and Geothermal Research (2006b) 153:187–199.[CrossRef][Web of Science]
Spulber SD, Rutherford MJ. The origin of rhyolite and plagiogranite in oceanic crust; an experimental study. Journal of Petrology (1983) 24:1–25.[Web of Science]
Staples RK, Brandsdottir B, Maguire PKH, McBride JH, Menke W, White RS. Faroe–Iceland Ridge experiment; 1, Crustal structure of northeastern Iceland. Journal of Geophysical Research, B, Solid Earth and Planets (1997) 102:7849–7866.[CrossRef]
Steinthorsson S, Oskarsson N, Sigvaldason GE. Origin of alkali basalts in Iceland; a plate tectonic model. Journal of Geophysical Research, B (1985) 90:10027–10042.[CrossRef]
Sturkell E, Sigmundsson F. Continuous deflation of the Askja Caldera, Iceland, during the 1983–1998 noneruptive period. Journal of Geophysical Research, B, Solid Earth and Planets (2000) 105:25671–25684.[CrossRef]
Sturkell E, Einarsson P, Sigmundsson F. Recent unrest and magma movements at Eyjafjallajökull and Katla Volcanoes, Iceland. Journal of Geophysical Research, B, Solid Earth and Planets (2003) 108:13.
Sturkell E, Einarsson P, Freysteinn S, Holldor G, Olafsson O, Pedersen R, de Zeeuw-van Dalfsen E, Linde AT, Sacks IS, Stefansson R. Volcano geodesy and magma dynamics in Iceland. Journal of Volcanology and Geothermal Research (2006) 150:14–34.[CrossRef][Web of Science]
Sugawara T. Empirical relationships between temperature, pressure, and MgO content in olivine and pyroxene saturated liquid. Journal of Geophysical Research, B, Solid Earth and Planets (2000) 105:8457–8472.[CrossRef]
Thy P. High and low pressure phase equilibria of a mildly alkalic lava from the 1965 Surtsey eruption; experimental results. Lithos (1991) 26:223–243.[CrossRef][Web of Science]
Thy P, Lofgren GE. Experimental constraints on the low-pressure evolution of transitional and mildly alkalic basalts; the effect of Fe–Ti oxide minerals and the origin of basaltic andesites. Contributions to Mineralogy and Petrology (1994) 116:340–351.[CrossRef][Web of Science]
Tormey DR, Bryan WB, Grove TL. Experimental petrology of normal MORB near the Kane fracture zone; 22°–25°N, Mid-Atlantic Ridge. Contributions to Mineralogy and Petrology (1987) 96:121–139.[CrossRef][Web of Science]
Trønnes R. Basaltic melt evolution of the Hengill Volcanic System, SW Iceland, and evidence for clinopyroxene assimilation in primitive tholeiitic magmas. Journal of Geophysical Research (1990) 95:15893–15910.[CrossRef]
Tryggvason E. Observed ground deformation during the Krafla eruption of March 16, 1980. (1980) Nordic Volcanological Institute, University of Iceland. Series: Nordic Volcanological Institute 0258–7718, 80–05.
Tryggvason E. Multiple magma reservoirs in a rift zone volcano; ground deformation and magma transport during the September 1984 eruption of Krafla, Iceland. Journal of Volcanology and Geothermal Research (1986) 28:1–44.[CrossRef][Web of Science]
Tryggvason E. Ground deformation in Askja, Iceland; its source and possible relation to flow of the mantle plume. Journal of Volcanology and Geothermal Research (1989) 39:61–71.[CrossRef][Web of Science]
Tryggvason E. Surface deformation at the Krafla Volcano, North Iceland, 1982–1992. Bulletin of Volcanology (1994a) 56:98–107.[Web of Science]
Tryggvason E. Observed ground deformation at Hekla, Iceland prior to and during the eruptions of 1970, 1980–1981 and 1991. Journal of Volcanology and Geothermal Research (1994b) 61:281–291.[CrossRef][Web of Science]
Walker D, DeLong SE, Shibata T. Abyssal tholeiites from the Oceanographer fracture zone; II, Phase equilibria and mixing. Contributions to Mineralogy and Petrology (1979) 70:111–125.[CrossRef][Web of Science]
Weaver JS, Langmuir CH. Calculation of phase equilibrium in mineral–melt systems. Computers & Geosciences (1990) 16:1–19.[CrossRef][Web of Science]
Weir NRW, Brandsdottir B, Darbyshire F, Einarsson P, Gunnarsson G, Hjorleifsdottir V, Jonsdottir K, Mochizuki M, Nakanishi A, Shimamura H, Shiobara H, Smallwood J, Staples R, White RS. Crustal structure of the northern Reykjanes Ridge and Reykjanes Peninsula, Southwest Iceland. Journal of Geophysical Research, B, Solid Earth and Planets (2001) 106:6347–6368.[CrossRef]
White RS, McKenzie D. Mantle plumes and flood basalts. Journal of Geophysical Research, B, Solid Earth and Planets (1995) 100:17543–17585.[CrossRef]
White RS, McKenzie D, O’Nions RK. Oceanic crustal thickness from seismic measurements and rare earth element inversions. Journal of Geophysical Research, B, Solid Earth and Planets (1992) 97:19683–19715.[CrossRef]
Yang HJ, Grove TL, Kinzler RJ. Experiments and models of anhydrous, basaltic olivine–plagioclase–augite saturated melts from 0·001 to 10 kbar. Contributions to Mineralogy and Petrology (1996) 124:1–18.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||