Formation and growth of normal faults at the divergent plate boundary in Iceland

The divergent plate boundary in Iceland is characterized by 40–80 km long and 5–10 km wide swarms of tension fractures (∼10² m long) and normal faults (∼10³ m long). The upper part of the crust is mainly composed of lava flows, with abundant columnar joints that are mostly perpendicular to the lava contacts. The lava flows are horizontal at the surface of the rift zone but become tilted at the rate of 1° for every 150 m depth in the crust. At the surface of the rift zone the joints are vertical and parallel to the vertical principal stress. Because of tilting of the lava pile, the columnar joints become oblique to this stress, hence becoming potential shear fractures, and form echelon sets at greater depths in the crust. Theoretical considerations suggest that normal faults start to nucleate on sets of en echelon columnar joints and/or large-scale tension fractures at crustal depths of 0.5–1.5 km. The width (depth) must be the smallest (controlling) dimension of many faults. Nevertheless, there is a positive linear relation (r = 0.91) between maximum throw and length of the Holocene faults. If the faults grow as self-similar structures, the throw-length relationship can be explained by a similar relation between fault length and width.     

On the structure and formation of fracture zones

Studies of the oceanic fracture zones, as well as field observations of the on-land parts of a fracture zone in Iceland, show that there are numerous tension fractures, normal faults, small-scale grabens and dykes within, and trending subparallel with, the fracture zones. These structures indicate that, in addition to the shear displacement, there is considerable extension associated with the development of fracture zones and that many of them may be regarded as complex graben structures. The ridges surrounding the continents of Africa and Antarctica are examples of mid-ocean ridges that are moving away from the continental margins where they originated and therefore expanding. Here it is suggested that as the perimeter of an expanding ridge increases, the tensile stresses associated with the ridge extension may contribute to the formation of fracture zones.     

Stratigraphy and paleomagnetism of a 2.9-km composite lava section in Eyjafjördur,Northern Iceland: a reconnaissance study

Eight mountainside profiles in lava flows south of the fjord Eyjafjördur, Northern Iceland, were sampled for paleomagnetic studies. The sampling was concurrent with pilot stratigraphic mapping of the lava sequences in these and several supplementary profiles. The eight profiles are correlated with minor overlaps so that they form a composite section of 2.9 km thickness, estimated to cover the age range between about 9 and 5 million years ago (Hardarson et al. 1999). Paleomagnetic measurements made on 319 lavas generally yield primary remanence directions of high stability and within-unit consistency. Evidence for at least 17 reversals of the geomagnetic field is seen, as well as numerous field excursions. Frequent clustering of directions in successive lavas indicates that the volcanism was episodic.     

The effects of layering and local stresses in composite volcanoes on dyke emplacement and volcanic hazards

Since most volcanic eruptions are fed by dykes, any assessment of volcanic hazards in an area must include an evaluation of the probability of injected dykes either reaching the surface or becoming arrested. Composite volcanoes are normally composed of alternating stiff (high Young's modulus) and soft (low Young's modulus) layers. Numerical models indicate that during unrest periods with magma-chamber inflation, the local stresses in composite volcanoes commonly prevent dyke-fed eruptions: while the stresses in the stiff layers may favour dyke propagation and seismogenic faulting, the local stresses in the soft layer remain seismically quiet and favour dyke arrest. Geodetic and field studies also indicate that most dykes never reach the surface, and that only a small fraction of the magma volume injected from a chamber erupts at the surface. I propose that for a dyke-fed eruption to occur, all the layers along the potential pathway of the dyke must have local stresses that favour magma-driven extension-fracture propagation. Thus, the stress field along the pathway must be homogenised. To cite this article: A. Gudmundsson, C. R. Geoscience 337 (2005).     

Progressive cooling of the hyaloclastite ridge at Gjálp, Iceland, 1996–2005

In the subglacial eruption at Gjálp in October 1996 a 6 km long and 500 m high subglacial hyaloclastite ridge was formed while large volumes of ice were melted by extremely fast heat transfer from magma to ice. Repeated surveying of ice surface geometry, measurement of inflow of ice, and a full Stokes 2-D ice flow model have been combined to estimate the heat output from Gjálp for the period 1996–2005. The very high heat output of order 10⁶ MW during the eruption was followed by rapid decline, dropping to  2500 MW by mid 1997. It remained similar until mid 1999 but declined to 700 MW in 1999–2001. Since 2001 heat output has been insignificant, probably of order 10 MW. The total heat carried with the 1.2 × 10¹² kg of basaltic andesite erupted (0.45 km³ DRE) is estimated to have been 1.5 × 10¹⁸ J. About two thirds of the thermal energy released from the 0.7 km³ edifice in Gjálp occurred during the 13-day long eruption, 20% was released from end of eruption until mid 1997, a further 10% in 1997–2001, and from mid 2001 to present, only a small fraction remained. The post-eruption heat output history can be reconciled with the gradual release of 5 × 10¹⁷ J thermal energy remaining in the Gjálp ridge after the eruption, assuming single phase liquid convection in the cooling edifice. The average temperature of the edifice is found to have been approximately 240 °C at the end of the eruption, dropping to  110 °C after 9 months and reaching  40 °C in 2001. Although an initial period of several months of very high permeability is possible, the most probable value of the permeability from 1997 onwards is of order 10^− 12 m². This is consistent with consolidated/palagonitized hyaloclastite but incompatible with unconsolidated tephra. This may indicate that palagonitization had advanced sufficiently in the first 1–2 years to form a consolidated hyaloclastite ridge, resistant to erosion. No ice flow traversing the Gjálp ridge has been observed, suggesting that it has effectively been shielded from glacial erosion in its first 10 years of existence.     

Formation of dykes,feeder-dykes,and the intrusion of dykes from magma Chambers

The paper discussed the formation of dykes, and applies the results to Iceland. It is postulated that dykes follow the pathway of least work and of least tensile strength and thereby intrude the subvertical joints in lava flows. It is suggested that dykes form in magma pulses, where each dyke is split in two by the next magma pulse and so on. In Iceland the estimated time between successive magma pulses is of the order of several hundred days. Statistical considerations indicate that in Iceland the probability of seeing dykes end upwards is only 1–2%, which agrees with observations. It is concluded that many and probably the majority of dykes are non-feeders. This, together with the low probability of finding the connection between feeder and lava flow, explains the scarcity of observed feeder-dykes. It is concluded that overpressure in shallow magma chambers needed to drive magma through crustal fractures in Iceland is usually smaller than the tensile strength of the host rock (several MPa), which thereby is the critical factor in dyke intrusion.     

Effects of sedimentary interfaces on fracture pattern,linkage, and cluster formation in peritidal carbonate rocks

The permeability of a reservoir is particularly dependent upon the proportion of its fractures that penetrate or are arrested at interfaces such as contacts and discontinuities. Here we report on fracture penetration and fracture arrest in Lower Cretaceous peritidal deposits exposed in the Pizzicoli Quarry, Gargano Promontory, southern Italy. We measured more than 2000 fractures, in the field and using LIDAR data, of which 564 fractures from the field and 518 from LIDAR studies are the focus of this paper. Fracture arrest/deflection and penetration depend much on the effects of peritidal cycle interfaces such as paleosol horizons, laminated carbonate mudstones, and stylonodular horizons. The laminated mudstones have the greatest effect; 63–99% of the fractures are deflected or arrested at such interfaces, whereas 63–90% are deflected/arrested at paleosols, and 20–35% at stylonodular horizons. In the mudstones, many fractures are arrested at thin, internal laminae, such that few penetrate the entire laminated layer, and fewer still the boundaries between the layers. Paleosol interfaces deflect/arrest more than 60% of all fractures. However, when small-offset fractures above and below paleosols are regarded as penetrating, they are evenly spaced (non-clustered), so that fracture-related fluid transport may occur across the entire paleosol. Stylonodular horizons deflect/arrest and split some fractures, but generally have little effect compared with the other types of interfaces. We present three main mechanisms for fracture deflection and/or arrest: (1) the fracture-induced tensile stress ahead of its tip, referred to as the Cook-Gordon debonding mechanism; (2) rotation of the principal stresses at and across the interface, resulting in the formation of stress barriers; and (3) large elastic mismatch (particularly as regards Young’s moduli) between layers across an interface. All these mechanisms are likely to have operated during fracture propagation and arrest in the carbonate rocks of the Pizzicoli Quarry.