Crustal Isostasy Indicates Anomalous Densities Beneath Iceland
William Menke, Lamont-Doherty Earth Observatory, Palisades NY 10964 USA
Submitted to Geophys. Res. Lett., 06 January 1999
Abstract. Recent seismological determinations of crustal thickness in Iceland, made mainly through identification of wide-angle PmP and SmS reflections, indicate that the depth to the Moho deepens from 11 km beneath the low-lying Reykjanes Penninsula to 39 km beneath the central highlands. By compiling all available crustal thickness measurements, we show that this trend holds generally for Iceland, with elevation (above m.s.l.) increasing at a rate of 35.54 +/- 6.4 (1 sigma) meters for each kilometer of Moho depth (below m.s.l.). If one assumes that the topography and Moho are in local isostatic equilibrium with thickness changes in a lower crustal layer, while an upper crust maintains a constant thickness, with the part above sea level having a density of of 2500 kg/m3, then the rate of increase of depth with topographic elevation implies a density jump of 89 +/- 12 kg/m3 across the Moho. Petrological and seismological evidence indicate that the lower-crustal densities probably do not exceed about 3060 +/- 50 kg/m3. Mantle densities must therefore be low, 3150 +/- 60 kg/m3. The combined effects of chemical depletion, thermal expansion and melt entrainment in the mantle can acount for some, but not all, of the anomaly.
The high elevation of Iceland is due to the bouyancy associated with the hot, depleted, partially molten mantle produced by the Iceland plume and to its unusually thick crust (up to 40 km; Darbyshire et al., 1998). Ito et al. (1996) are able to explain the 500-1000 wide mantle swell with their Ice-1d "wide plume" model. It predicts that about 30% of the topographic anomaly is due to mantle bouyancy, with the other 70% being due to the crust (which is assumed to be in locally-compensated isostatic equilibrium with the mantle). Ito et al.'s (1996) model predicts about the right amount of melting to make the observed volume of crust. But it predicts a variation of crustal thickness with distance from the plume center that is considerably smoother than what is observed.
Recent refraction-tomography experiments have indicated that the crust of Iceland is essentially thick oceanic crust (Bjarnason et al., 1993; Staples et al. 1997; Darbyshire et al., 1998; Menke et al., 1998; Weir et al. 1998). Both the upper extrusive layer and the denser, higher velocity lower intrusive layer are thicker than in normal oceanic crust, with the intrusive layer being both very homogenous and disproportionately thick. Everywhere in Iceland the bottom of the lower crust is bounded by a sharp interface (the Moho), with the material below having mantle-like seismic velocities. These experiments provide excellent determination of crustal thickness, both because the sharp Moho produces easily detected PmP and SmS wide-angle reflections, and because the tomographic imaging provides a very accurate crustal velocity model. These measurements demonstrate that very large variations in crustal thickness occur in Iceland. Staples et al. (1997), for instance, measured a deepening of 15 km in Moho depth over a 100 km horizontal distance in northeastern Iceland, Menke et al. (1998) measured a 10 km deepening over 70 km in northern Iceland, and Weir et al. (1998) measured a 7 km deepening over 75 km in southwestern Iceland. These thickness variations occur over horizontal distances that are much shorter than the ~500 km scale that Ito et al. (1996) predict is associated with mantle bouyancy. To the extent that they are locally compensated, they can provide information on the density contrast between the crust and the mantle. The overall continuity of both genetic processes and petrological structure from normal mid-ocean ridge into the interior of a ridge-centered hotspot like Iceland facilitates such a study.
Crustal Isostasy in Iceland
We have compiled 17 published seismological measurements of Moho depth (below mean sea level) in Iceland, mostly made on the basis of PmP and SmS traveltimes ( Table 1). The Moho depth ranges from a minimum of 11 km beneath the southern end of the Reykjanes penninsula (Weir et al., 1998) to a maximum of 39 km beneath central Iceland (Darbyshire et al., 1998). Many parts of Iceland are sampled, including the zero-age neovolcanic zones and the Tertiary age portions of extreme eastern and western Iceland (Figure 1). We exclude two published measurements, Staples et al.'s (1997) measurement of 19 km beneath Krafla volcano (where there is a crustal magma chamber) and Darbyshire et al.'s (1998) measurement of 39 km beneath the Vatnajokull ice sheet, to avoid the anomalous densities associated with the magma and ice. Sensitivity studies that we have performed (not shown) indicate that the accuracy of crustal thickness measurements is typically +/- 2 km when tomographic surveys are used to constrain crustal structure (14 data) and +/- 5 km when other seismological techniques are used (3 data). We have also estimated the mean elevation of a 20 km diameter disk centered above each of these measurements, using the ETOPO5 topographic database (ETOPO5, 1988). Elevation is found to increase approximately linear with Moho depth, at a rate of 35.54 +/- 6.4 (1 sigma) meters for each kilometer of increase in Moho depth (Figure 2).
Both volcanic origin of the Icelandic crust, and the subsequent tectonic processes that have occured there, are compatible with the assumption of isostasy. The crust has been built up over the past 16 m.y. by a complex sequence of episodes of plate tectonic spreading in the neovolcanic zones, punctuated by jumps in the position of those zones (i.e. "ridge jumps") (Palmason and Saemundsson, 1974). Some of the variability in crustal thickness appears to reflect different levels of magmatic productivity during these episodes (Staples et al., 1997). Crustal thickness also appears to decrease away from the center of the hotspot beneath Vatnajokull to where the neovolcanic zones join the Reykjanes ridge to the south and the Kolbeinsey ridge to the north (Bjarnason et al. 1993; Darbyshire et al., 1998; Menke et al., 1998). In both scenarios crustal thickness is set at the time of formation, when the lower crust and mantle are hot and ductile and have little or no strength (Menke and Sparks, 1995). We thus assume that the crust is in local isotatic equilibrium with the mantle. Seismic refraction studies indicate that the Icelandic crust thickens mostly through the addition of lower-crustal material (Bjarnason et al., 1993), with the extrusive layer being approximately 4-5 km thick in most areas of Iceland, and with the lowermost crust being very homogeneous. Thus we assume a three-layer crust: the material above sea level; the material between sea level and 11 km depth (which contains both extrusives and intrusives) and the lower crust (between 11 km and Moho). The material above sea level is mostly highly fractured basalts, and is taken to have a nominal density of 2500 kg/m3. The 0-11 km layer, which contains both extrusive and intrusive material, is assumed to be approximately laterally homogenous. Its density is not needed in the isostatic calculation. The lowermost layer (or crustal root) is also assumed homogenous. The density contrast across Moho can thus be estimated by balancing the mass excess of the topography against the mass deficit associated with the crustal root. The linear increase of elevation with Moho depth implies a density contrast of 89 +/- 12 kg/m3. Should some flexural compensation be occuring, this estimate would represent an upper bound (as the elevation of the high regions would be partially supported by the surrounding lowlands). Low values of the density jump (of 160-180 kg/m3) have also been reported by Staples et al. (1997), on the basis of 2D gravity modeling of the FIRE refraction line in northeastern Iceland.
Estimated Lower-Crustal and Mantle Density
Bulk densities of upper crustal basalts have been measured down to a depth of 2 km in eastern Iceland (Christensen and Wilkens, 1982). The deepest material is the densest, with a density of 3090 +/- 50 kg/m3 (at zero porosity and at standard temperature and pressure). In the lowermost crust, this value would decrease slightly to 3060 +/- 50 kg/m3, assuming that the composition does not vary with depth (i.e. at 1000 deg C, 30 km depth, using compressibilities and thermal expansion coefficients tabulated by Clark, 1966). this inference is consistent with the seismological evidence, which does not support major changes in chemistry with depth:
The lower crust appears very homogenous, with a small compressional velocity gradient (<0.05 s-1) and with no evidence for reflecting interfaces above the Moho. The deepest compressional velocities that have been directly measured by crustal P waves are 7.1 km/s at a depth of 24 km in northeastern Iceland (Staples et al. 1997) and 7.2 km/s at a depth of 24 km beneath central Iceland (Darbyshire et al., 1998). The compressional velocity of the base of the crust is less well-constrained. Synthetic seismogram modeling indicates that a compressional velocity jump of about 0.9 km/s is needed across the Moho in order to satisfy PmP amplitude data (Staples et al., 1998). This value, when combined with mantle velocities of 8.0 +/- 0.1 km/s measured from Moho-refracted Pn phases (Menke et al., 1996; Menke et al, 1998) indicates that the lowermost crust has a compressional velocity of 7.1-7.3 km/s. This compressonal velocity agrees well with published values for normal oceanic lower crust (i.e. Layer 3). White et al. (1992) tabulate many regions of the ocean where Layer 3 velocities reach 7.1-7.2 km/s.
These inferred lower crustal compressional velocity and density are also consistent with predictions from a mantle melting model with 25% melting (arguably reasonable for a hot spot like Iceland): 7.25 km/s and 3030 kg/m3 (McKenzie and Bickle, 1988; McKenzie, personal communication, 1993). The estimated mantle density of 3150 +/- 60 kg/m3, obtained by summing the lower-crustal density and the inferred density jump across Moho, is considerably lower than the 3330 kg/m3 value typically observed at mid-ocean ridges (i.e. a 180 +/- 60 kg/m3 discrepancy).
Comparison with the mid-Atlantic ridge and Ontong Java Plateau
Along-axis crustal thickness variations are common on mid-ocean ridges, and are inferred to give rise to the gravity bulls-eye effect. Tolstoy et al. (1993) has used seismological techniques to measure crustal thickness variations along a segment of the mid-Atlantic ridge at 33 deg S. The observed gravity anomaly can be fit with a density jump of 330 kg/m3 across the Moho (i.e. from 3000 to 3330 kg/m3; the "expected" jump for mid-ocean ridges (e.g. Wang and Cochran, 1995), except for a part of the segment where the crust is very thin (and presumably lower density), where an even larger jump is required. The density jump inferred for Iceland is thus about a factor of 3.7 less than for a normal ridge. The highlands of central Iceland would be about 4 km tall were the density contrast the same as for a normal ridge.
The Ontong Java Plateau is a Cretaceous age basaltic plateau located in the western Pacific that, like Iceland, may also have been formed by a ridge-centered hotspot. It has a crustal thickness of 35-42 km (Araki et al., 1998; Araki, personal communication, 1998), seismologically determined from clear wide-angle PmP reflections. Ito and Cliff (1998) report a sediment-unloaded depth of 3200 m at ODP Site 807, compared to 5600 m in the surrounding abyssal plain. A density jump of 112 kg/m3 is implied (assuming isostasy and a normal 8 km thick crust for the abyssal plain). The similarity of this value with the one inferred for Iceland, despite the large difference in age, suggests that the density contrast is compositionally, and not thermally, controlled. In the case of Ontong Java (at least), partial melt is not present in the uppermost mantle (although entrained solid basalt might be).
Possible Sources of Low Mantle Density
Thermal expansion, entrained basalt (either as a melt or as a solid), and depletion are all possible mechanisms for reducing the density of the Icelandic mantle below that inferred for the mid-Atlantic ridge. Thermal models for Iceland predict that the temperature of the Moho is 800-1000 deg C, not significantly higher than those postulated for the normal mid-Atlantic ridge (Menke and Sparks, 1995). However, temperature does not effect mantle density strongly, and even a 250 deg C temperature difference would yield only a 20 kg/m3 density difference (Oxburgh and Parmentier, 1977). The amount of entrainment is limited by the measured seismic velocity: a maximum bound of about 20% is achieved by mixing a high velocity (solid) basalt (say 7.2 km/s) with an unusually fast peridotite (say 8.2 km/s, the upper range of White et al.'s 1993 tabulation of Atlantic mantle velocities). Even this maximum bound on entrainment does not decrease the mantle density sufficiently (50 kg/m3. Predictions of the density change due to depletion of the mantle vary. The mantle melting model mentioned above predicts a mantle density of 3250 kg/m3, equivalent to a density reduction of 80 kg/m3. However, for the same degree of melting, Oxburgh and Parmentier (1977) predict a much smaller density reduction, only 36 kg/m3. Even taking upper bounds, no single mechanism can acount for the inferred density difference. Acting together, they can satisfy it only marginally: 150 compared to the inferred 180 +/- 60 kg/m3.
Summary and Conclusions
Crustal isostasy indicates that the density contrast across the Icelandic Moho is 89 +/- 12 kg/m3, much less than the 330 kg/m3 inferred for normal parts of the mid-Atlantic ridge, but similar to the amount inferred for Ontong Java, a tectonically similar oceanic plateau. Either the lower crust must be anomalously dense, or the uppermost mantle must be anomalously light. Petrological and seismic data indicate that the lower crust of Iceland and of the Mid-Atlantic ridge are very similar, suggesting that the density descrepancy between the plateaus and normal ridges arises from a difference between their respective mantles. Chemical depletion, acting alone, does not seem to be sufficient to acount for the inferred low mantle density of 3150 +/- 60 kg/m3. Other sources, such as thermal expansion and entrainment of basaltic material, seem only marginally able to make up the difference, suggesting that another, unmodeled process is occuring. This process appears to have no signature in the mantle seismic velocities, which although based on limited observations, are normal.
One could step back and question the underlying assumptions under which our estimate of mantle density was derived. Perhaps the crust is not in isostatic equilibrium with the mantle. If so, some force must be acting to make the thickest parts of Iceland topographically lower (by 3 km) than if they were isostatically compensated. This force must operate on both the Tertiary areas extreme eastern and western Iceland and on the zero-age region in central Iceland. If it were exerted by the neighboring, thinner crust, the stresses in the thin brittle part of the upper crust would exceed its strength (i.e. stresses exceeding 5 x 108 N/m2 in the upper 6 km). There is no reason to suppose that such a downward force might originate in the mantle: if anything, the deep (>100 km) partial melt detected by Wolfe et al. (1997) suggests that an upward force might be present. Alternatively, perhaps the lower crust is 100 kg/m3 denser than the the 3060 +/- 50 kg/m3 that we have assumed. Yet this densification must occur without significantly increasing the seismic velocity above what is normal for oceanic Layer 3. Simply increasing its olivine content would not work, because the required 30% would raise the compressional velocity to 7.5 km/s, far higher than what is observed. As we cannot find plausible mechanisms for either of these senarios, we are left with the sense that the density structure of the Icelandic crust and uppermost mantle contains an anomaly that we as yet do not understand.
AcknowledgementsWe thank Paul Asimow, Emilie Hooft, Marc Spiegelman, and Maya Tolstoy for helpful discussion. Lamont-Doherty Contribution number 0000.
Araki, E, K Mochizuki, K Suyehiro, A Taira, S Yoneshima, S Miura and R Hino, Seismic structure of the Ontong Java plateau (Abs), EOS Trans. AGU, 1998 Falling Meeting Supplement, F869, 1998.
Bjarnason, I, W Menke, O Flovenz and D Caress, Tomographic image of the spreading center in southern Iceland, J. Geophys. Res. 98, 6607-6622, 1993.
Christensen, N and RH Wilkens, Seismic properties, density and composition of the Icelandic crust near Reydarfjordur, J. Geophys. Res. 87, 6389-6395, 1982.
Clark, SP (ed), Handbook of Physical Constants, 2nd edition, Memoir 97, Geological Society of America, 1966. Wang, J. and J. Cochran, Along-axis gravity gradients at mid-coean ridges: implications for mantle flow and axial morphology, Geology 23, 29-32, 1995.
Darbyshire, I Bjarnason, RS White and OG Flovenz, Crustal structure above the Iceland mantle plume images by the ICEMELT experiment, Geophys. J. Int., 135, 1131-1149, 1998.
ETOPO5: Digital relief of the surface of the earth - Bathymetry / topography data, Rep. 88-M66-2, Natl. Ocelanic and Atmos. Admin, Boulder, Colo., 1988.
Ito, G, J Lin and CW Gable, Dynamics of mantle flow and melting at a ridge-centered hotspot: Iceland and the mid-Atlantic ridge, Earth. Planet. Science Lett. 144, 53-74, 1996.
Ito, G and PD Clift, Subsidence of Pacific Cretaceous plateaus, Earth and Planetary Science Letters 161, 85-100, 1998.
McKenzie, D and MJ Bickle, The volume and composition of melt generated by extension of the lithosphere, J. Petrology 29, 625-679, 1988.
Menke, W, M West, B Brandsdottir and D. Sparks, Compressional and shear velocity structure of the lithosphere in northern Iceland, Bull. Seism. Soc. Am, 88, 1561-1571, 1998.
Menke, W and D Sparks, Crustal accretion model for Iceland predicts 'cold' crust, Geophys. Res. Lett., 1673-1676, 22, 1995.
Menke, W, B Brandsdottir, P Einarsson and I Bjarnasson, Reinterpretation of the RRISP-77 Iceland shear wave profiles, Geophys. J. Int. 126, 166-172, 1996.
Oxburgh, ER and EM Parmentier, Compositional and density stratification in oceanic lithosphere - causes and consequences, J. Geolo. Soc. London 133, 343-355, 1997.
Palmasson, G and K Saemundsson, Iceland in relation to the mid-Atlantic ridge, Annu. Rev. Earth. Planet. Sci. 2, 25-50, 1974.
R Staples, R White, B Brandsdottir, W Menke, P Maguire, J McBride and J Smallwood, Faeroe-Iceland Ridge Experiment -- I. The crustal structure of north-eastern Iceland, J. Geophys. Res. 102, 7849-7866, 1997.
Tolstoy, M, A Harding and J Orcutt, Crustal thickness on the mid-Atlantic ridge: bulls-eye gravity anomalies and focused accretion, Science 262, 726-729, 1993.
Weir, NRW, RS White, B Brandsdottir, P Einarsson, H Shimamura, and H Shiobara, Crustal structure of the mid-Atlantic ridge: The Reykjanes Iceland Seismic Experiment, (Abs), EOS Trans. AGU, 1998 Falling Meeting Supplement, F805, 1998.
White, RS, D McKenzie and RK O'Nions, Oceanic crustal thickness from seismic measurements and rare earth element inversions, J. Geophys. Res. 97, 19,683-19,715, 1993.
Wolfe, C., I. Bjarnason, J. VanDecar and S. Solomon, Seismic structure of the Iceland mantle plume, Nature 385, 245-247, 1997.