Tectonophysics 396 (2005) 141 – 159 www.elsevier.

com/locate/tecto

Styles of surface rupture accompanying the June 17 and 21, 2000 earthquakes in the South Iceland Seismic Zone
Amy Cliftona,*, Pall Einarssonb ´
a

Nordic Volcanological Center, Institute of Earth Sciences, University of Iceland, Sturlugata 7, IS-101 Reykjavık, Iceland ´ b Institute of Earth Sciences, University of Iceland, Sturlugata 7, IS-101, Reykjavık, Iceland ´ Received 22 September 2003; accepted 22 November 2004 Available online 12 January 2005

Abstract Two large M s=6.6 earthquakes occurred on June 17 and 21, 2000 in the South Iceland Seismic Zone. The roughly E–W trending zone is undergoing left-lateral shear. However, most known surface rupture has been along north-striking, right-lateral strike-slip faults. Rupture associated with the June 2000 events follows a similar pattern. Although the two earthquakes had similar magnitude, fault plane solutions, and overall rupture lengths of 15 to 20 km, the pattern of rupture from each was notably different in character. The June 17 event ruptured along a series of NNE-trending, left-stepping segments, giving the fault as a whole an almost due north trend. At the largest scale rupture is relatively straight and continuous. At the smallest scale, rupture style seems to vary with small-scale topography and ground texture. Rupture from the June 21 event is more complex and can be divided into five discrete segments. To the north, deformation is distributed across two zones of left-stepping fractures, along which widening and subsidence have occurred. The central segment consists of right-stepping fractures defining a 2 km long, ENE trending zone. Sense of shear is clearly left-lateral strike slip. The two southernmost segments define an NNE trend. In several places along the rupture zones of both earthquakes it can be verified that the ruptures occurred along preexisting faults. The observed faulting structures are similar to those of earlier earthquakes in South Iceland, both with regard to style and spatial arrangement. However, our observations suggest that some of the historical earthquakes may have been larger than the June 2000 events. D 2004 Elsevier B.V. All rights reserved.
Keywords: Iceland; Strike-slip faults; Surface rupture; Earthquakes

1. Introduction Two large earthquakes with M s=6.6, occurred on June 17 and 21, 2000 respectively in the South Iceland Seismic Zone (SISZ). This transform-type plate boundary has been relatively quiet since 1912, but 37 destructive earthquakes have occurred in this

* Corresponding author. Fax: +354 562 9767. E-mail addresses: amy@hi.is (A. Clifton)8 palli@raunvis.hi.is (P. Einarsson). 0040-1951/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.tecto.2004.11.007

142

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

area since 1164 AD. Most of the historical accounts were compiled by Thoroddsen (1899) who also described the surface effects of the 1896 earthquakes based on reports from local priests and farmers. The surface trace of the 1912 earthquake, a single event near the eastern end of the seismic zone, was described in newspaper reports and has been studied and mapped by Einarsson and Eirıksson (1982) and ´ Bjarnason et al. (1993). However, deterioration of the rupture occurs rapidly in Iceland’s climate and it is certain that data, and therefore our understanding of the nature of these events, has been incomplete. The earthquakes of June 2000 have provided us with a first opportunity to fully document the fresh surface rupture from a major earthquake sequence in the SISZ, and to compare its distribution to geophysical models of the underlying fault planes. Angelier and Bergerat (2002) and Bergerat and Angelier (2001, 2003) have produced detailed maps and descriptions of selected small-scale features of both faults and provided analysis of the mechanical behavior of the faults based on those data. The goals of this paper are two-fold. Firstly, we wish to document, with photographs, descriptions and complete maps, the entire extent of co-seismic surface effects (e.g. ground rupture) from the June 2000 earthquakes. We have used real-time differential GPS to provide accurate maps on as fine a scale as possible. Secondly, we wish to describe the relationship between visible rupture at the surface and geophysically determined characteristics of the fault plane at depth. We use GIS software to overlay geophysical data onto our maps in order to compare aftershock distribution and satellite-based deformation data with the location and geometry of surface rupture. This data will better enable us to calibrate the magnitudes of earlier (pre-instrumentation) historical earthquakes whose damage area is known.

accumulating along the SISZ (Sigmundsson et al., 1995), but no through-going transform fault has ever been mapped here. Instead, strain is accommodated along many sub-parallel, north-trending, right-lateral strike-slip faults (Fig. 1). The term bbookshelfQ faulting has been used to denote shear deformation accommodated by an array of faults trending perpendicular to the shear direction (Sigmundsson et al., 1995). This style of deformation has been proposed to explain the faults that have been mapped in the SISZ (e.g. Einarsson and Eirıksson, 1982; Bjarnason et al., ´ 1993). Einarsson and Eirıksson (1982) suggested that ´ the lack of east–west trending left-lateral faults is due to the transient nature of the zone as it migrates southward in response to ridge propagation. Hackman et al. (1990) used a boundary element model to confirm that north-trending faults can accommodate this left-lateral strain if they are either 10 to 15 km long or bmore spatially frequent than mappedQ. Mapping conducted during the past decade (Fig. 1B) indicates that spacing of faults in the SISZ is indeed close, varying between 500 m to less than 5 km, and averaging less than 2 km. The thickness of the brittle crust decreases within the SISZ from 12 to 15 km in the east to about 5 km in the west (Stefansson et al., ´ 1993). This thinning of the brittle layer correlates with a general decrease in maximum earthquake size from east to west, and suggests that fault length may also decrease.

3. Methods One important goal of this study was to identify and map as much of the surface rupture as we could as quickly as possible. In the days immediately following each earthquake, cracks in main and secondary roads were reported, while we attempted to map them before they were repaired. Wherever roads were cracked, we followed their trend into farms and fields looking for rupture. Local farmers led us to fractures and sinkholes that opened up on their land which they intended to fill in order to protect livestock from injury. We also conducted reconnaissance in the entire area affected by aftershocks along strike from the epicenters of both earthquakes and along the traces of known surface faults in the immediate vicinity. Because of a shortage of time, some detail in

2. Tectonic setting The SISZ is a roughly east–west trending transform zone that connects the northeast-trending Western and the Eastern Volcanic Zones in South Iceland. The zone is slightly oblique to the 1048 direction of plate motion in south Iceland (DeMets et al., 1994). GPS measurements confirm that left-lateral strain is

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

143

A b)

66o N

NVZ North American Plate

WVZ

Hreppar Microplate

Eurasian Plate EVZ

64o N

RPVZ

SISZ

RR
22o W 16o W

64.1oN B

64oN

63.9oN

10 km
21oW 20oW

Fig. 1. (A) The plate boundary in Iceland (Einarsson and Saemundsson, 1992). RPVZ=Reykjanes Peninsula Volcanic Zone, SISZ=South Iceland Seismic Zone, WVZ=Western Volcanic Zone, EVZ=Eastern Volcanic Zone, NVZ=Northern Volcanic Zone. Arrows: direction of plate motion (DeMets et al., 1994). (B) Simplified map of surface fractures associated with strike-slip faults in the South Iceland Seismic Zone. Map includes all fractures exposed in Holocene formations, as well as faults in Plio-Pleistocene formations along which activity probably took place in the Holocene. Based on Einarsson and Eirıksson (1982), Einarsson et al. (1981), Imsland et al. (1997), Einarsson et al. (2002). Box encloses ´ area shown in Fig. 2.

measurement was sacrificed, such as the width and depth of many fractures or the height of push-up structures. However, we documented much of the rupture in photographs, only representative selections of which are presented in this paper. Mapping began on June 18, the day after the first earthquake, and was continued all summer and into the late autumn of 2000. Locations of surface rupture were mapped either as points, lines or areas using the Trimble differential GPS Pathfinder Pro XR GIS data collection system. The stated horizontal accuracy of

the unit is 50 cmF1 ppm on a second-by-second basis and submeterF2 ppm vertical accuracy on a secondby-second basis. This accuracy was sufficient for the scale of features we were mapping, which ranged from several to tens of meters in length, and was within the error involved in walking alongside a feature. The Trimble TSC1 data logger was used to store data and Trimble Pathfinder software was used for processing and exporting data into the ERDAS Imagine GIS program. Data were overlain onto a 1:50,000 scale digitised topographic base map with

144

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

100 m index contours along with road and hydrological layers.

4. Data 4.1. June 17 earthquake The June 17 earthquake affected an area of interglacial lavas with a thick soil cover. The hypocenter of the earthquake was located at 63.9758N, 20.3708W at a depth of 6.3 km (Stefansson et al., ´ 2000). A preliminary model based on the aftershock distribution indicated a 16 km long fault striking 0098, dipping 868 east and extending to 10 km depth (Stefansson et al., 2000). The best USGS (2000) fault ´ plane solution has a strike of 0058 with a dip of 838 east and a rake of 1758, indicating right-lateral strike slip with a small component of dip slip. A more recent model based on relative relocation of aftershocks
20.7oW 64.05oN

indicates an 11.5 km long fault, striking 0078 and dipping slightly towards the west (K. Vogfjfrd, personal communication). A model of uniform slip gives a value of 0.9 m of slip (Stefansson et al., 2000) ´ whereas a more detailed model of non-uniform slip gives a maximum slip of 2.6 m close to the hypocenter, tapering off towards the ends (Pedersen et al., 2003). Most surface rupture occurred along a series of NNE-trending, left-stepping fault segments, generally giving the fault an almost due N-trend. Five segments of varying length were identified and are described here from south to north (Fig. 2). 4.1.1. Pula segment The southernmost exposures of primary fracturing occur near Pulutjfrn pond (Fig. 3). The pond is bounded on the west side by a northeast-trending topographic depression which extends some distance north beyond the pond itself. A small array of leftstepping open fractures was observed in vegetated soil
20.5oW

June 21
Göltur

June 17

Hvít

á

Árnes

Hestvatn

Stufholt 64oN Hestfjall

Bitra 63.95oN
Ro ad 1

Ro

ad

30

Flatholt

Skammbeinsstadir
Thjórsá

Kvíarholt Dælarétt
Roa d1

Mykjunes
5 km

Pula

Fig. 2. Map of June 17 and June 21 earthquake areas. Mapped surface rupture shown in red. Green stars denote epicenter locations. Boxes outline rupture segments discussed in the text and shown in Figs. 3–12.

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

145

this graben-like structure during the earthquake. The eastern side of the fracture is comprised of short leftstepping segments separated by arcuate push-up structures characteristic of right-lateral strike slip and discussed above. The southern fracture has a more northeasterly trend than the rest of the fault, and is characterized by right-stepping segments typical of left-lateral strike slip. More discontinuous and irregular fractures were also found on a hill 100 m to the south. 4.1.3. Skammbeinstaðir segment Greater than 2 km length of virtually continuous rupture was mapped on the farm called SkammbeinsstaWir, directly south of the epicenter (Figs. 2 and 5). When viewed at a large scale, the rupture occurs along 18 left-stepping segments varying in length from 49 to 212 m and averaging 89 m in length. Average strike of
20.356oW 20.352oW

63.94o

4.1.2. Mykjunes segment The fault segment at Mykjunes has been described in great detail by Bergerat and Angelier (2003; see their Figs. 4 and 5). We present here a summary of our observations (Fig. 4). On the farm Mykjunes, south of the epicenter, surface rupture occurs along two conjugate fracture arrays, one trending approximately N208E and the other N558E. The northernmost fracture cuts the walls of a graben up to 15 m wide which had been previously visible only as a gentle depression in the local topography. Subsidence of approximately 50 cm occurred along both sides of

100 m

63.936

o

Fig. 4. Map of rupture along the Mykjunes segment.

10

on the inner flank of the depression a few hundred meters north of the pond. While no causal relationship can be demonstrated, it is an interesting coincidence that the water level in the 200 m wide and 2 m deep pond began subsiding shortly after the earthquakes and by August 2000 all the water had been drained out of the pond. At the time of writing (2003) water has not returned.

road 286

gra

Fig. 3. Map of rupture along the Pula segment. Enlargement of rupture shown in inset.

be

n

0

m

146

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

spring

a

63.97oN
100

6 28 ad Ro

b

63.96oN

a

NE
Skammbeinsstadir

400 m

b
20.35oW

20.37 W

o

Fig. 5. Map of rupture along the SkammbeinsstaWir segment and photos showing rupture style (see text for further explanation). Letters on map show photo locations. Star is epicentral location of June 17 earthquake. Filled squares show location of farmhouses which were undamaged during the earthquake. Open square denotes concrete barn that was completely destroyed.

individual segments is N068E, but the left-stepping nature of the rupture give the fault segment as a whole a slightly NNW-trend (Fig. 5). At the smaller scale, rupture style seems to vary with topography and ground texture. In fields that have been drained and mowed, rupture occurs as a series of NE to NNEtrending S-shaped gashes in the soil, with maximum opening approaching 1 m. These gashes are separated by compressional push-up structures or folds in the turf that commonly have an NW-trending axis (Fig. 5, photo b). In fields that have been recently ploughed and have no turf cover, push-up structures were not observed. Instead, gashes occur in both en echelon arrays or as relatively continuous open fractures exhibiting little or no right-lateral rotation. In fields that have been neither ploughed nor drained, rupture occurs variously in zones of short (b1 m), parallel,

narrow cracks, en echelon gashes or continuous open fractures (Fig. 5, photo a). The most severe damage at SkammbeinsstaWir was observed along the three southernmost segments shown in Fig. 3. Subsidence on the order of 1 m was observed along the longest segment close to the farmhouse. The entry road to the farm had ruptured and subsided and a concrete barn had partially collapsed during the earthquake. No surface rupture was observed for a distance of 1.3 km in the lowland between SkammbeinsstaWir and the farm Mykjunes to the south. To the north of SkammbeinsstaWir evidence of rupture disappears at a spring just north of the 100 m contour line (see Fig. 5) and appears again on the next slope approximately 600 m to the north (Stufholt segment). However, this ´ spring feeds a fairly linear, north-trending swampy depression, suggesting the presence of an underlying

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

147

pre-existing fault. South of the spring, a gentle northtrending linear depression in the topography along the strike of the surface rupture indicates that a preexisting structure failed during this earthquake. A southward kink in the 100 m contour points to the approximate trend of this depression. Surprisingly, a mapped surface fault (Johannesson et al., 1990) less ´ than 500 m to the west showed little, if any, evidence of reactivation. 4.1.4. Stufholt segment ´ North and east of Stufholt farm the surface rupture ´ follows the eastern boundary of a shallow valley for a

distance of a little over 2 km. The bottom of the valley is marshy with thick soil. This soil has in many places been compacted by the earthquake shaking and subsided. The ruptures occur where the marsh meets the hill (following the 100 m contour; Fig. 6) and therefore have the appearance of normal faulting, with water from the bog frequently accumulating on the downthrown side. Individual fractures are, however, arranged in an en echelon pattern consistent with right-lateral displacement (see inset a in Fig. 6). We conclude that the apparent normal faulting component is here caused by differential compaction across the fault trace.

a

64oN

100

roa

d

50 m

b
100

63.99oN

a b

20 m

300 m
20.37oW 20.36oW

Fig. 6. Map of rupture along the Stufholt segment (see text for further explanation). Insets showing left-stepping (a) and right-stepping (b) ´ fracture arrays.

148

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

Further to the east, a conjugate rupture segment cuts across an even field. This segment can be traced for about 350 m along the 100 m contour, and has a distinct right-stepping pattern. Shortening observed in a fence crossing the rupture, along with the geometry of the rupture pattern confirm that the sense of motion here was left-lateral (Fig. 6, inset b). The strike of this segment is near N258E, surprisingly close to the N108E strike of the adjacent right-lateral segments. This is the longest rupture with a left-lateral sense of motion observed along the surface trace of the June 17 fault.

´ 4.1.5. Arnes segment At the northern end of the June 17 fault, surface rupture is distributed along two sub-parallel fault ´ segments (Fig. 7). On the island of Arnes, fresh rupture extends northward from a north-trending stretch of the Thjorsa River at the center of the ´ ´ island. Here individual segments are generally left stepping (inset in Fig. 7), indicating right-lateral strike slip. Along strike to the south, slope failure has occurred along a high bluff above the Thjorsa ´ ´ and significant rockfall occurred from the cliffs on the western shore. On the western part of the island

Thjórsá

64.02oN

b a

a
pu pu

50 m

b
64oN
0

20 cm

Thj

10

20.34oW

´ Fig. 7. Map of rupture along the Arnes segment and photos showing rupture style (see text for further explanation). Inset showing left-stepping ´ nature of fracture array and push-ups (pu). (Photo a) Pall Einarsson standing inside reactivated portion of fault on Arnes. (Photo b) Subsidence ´ ´ of 20 cm along pre-existing fault scarp on Arnes.

órs á

500 m 20.32oW

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

149

rupture occurs along pre-existing faults. Fresh opening was observed in the bottom of an old, deep, grass-covered fracture (Fig. 7, photo a). In another location, subsidence of approximately 20 cm was measured on a pre-existing fault scarp. (Fig. 7, photo

´ b). Across the river to the south of Arnes, rupture on the farm called Akbraut falls on the same trend (dashed line in Fig. 7). Short left-stepping open fractures occur in a topographic depression which is shown by the 100 m contour, suggesting the

63.935oN

20.685oW

20.68oW

Fig. 8. Map of rupture along the D&larett segment and photos showing rupture style (see text for further explanation). Dashed line represents ´ previously mapped west-dipping fault scarp. Dark lines show reactivated portions of pre-existing fault. Open areas represent sinkholes. (Photo a) Gaping fissure with component of left-lateral slip. (Photo b) Open fracture at the base of a pre-existing fault. (Photo c) Sinkhole and white residue, probably fault gauge. Letters on map correspond to photo locations.

150

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

continuation of an underlying NNE-striking structure at this location which also coincides with a bend in ´ the Thjorsa river. North of Arnes, only some minor ´ ´ rockfall and a few short opening-mode fractures were observed at the farm called Thrandarholt. 4.1.6. Kvıarholt fault ´ The mainshock on June 17 triggered a sequence of earthquakes that extended about 90 km along the plate boundary to the west (Clifton et al., 2003; Pagli et al., 2003). One of these events (m b=5) occurred on an N– S fault approximately 2 km west of the fault of the mainshock. Small surface ruptures were observed along a 1 km segment of this fault (see Fig. 2). 4.2. June 21 earthquake The June 21 event occurred 17 km west of the June 17 earthquake. Its hypocenter was at 63.988N, 20.718W and at a depth of 5.1 km. The distribution of aftershocks indicated an 18 km-long fault striking N028W, extending from the surface to 8 km depth (Stefansson et al., 2000). The most recent model ´ based on relative relocation of aftershocks indicates a fault 16.5 km long, striking N018W and dipping 888 to the west (K. Vogfjfrd, personal communication, 2004). Slip distribution along the fault, as modeled by inversion of GPS and InSAR data indicates a maximum slip of 2.9 m slightly north of the hypocenter and a second high-slip area south of the hypocenter (Pedersen et al., 2003). The earthquake affected an area of both inter- and post-glacial lavas, partly underlain by a thick sediment layer. Surface rupture from this event was more complex and can be divided into five major segments with different characteristics described below from south to north. 4.2.1. Dælarett segment ´ This southernmost fault segment (Fig. 8) consists of an array of open fractures and sinkholes which lie on either side of a pre-existing graben that had been previously mapped by Einarsson and Jonsdottir (1999, ´ ´ unpublished report). The southern part of the graben strikes N508E, but the majority of fractures in this array strike N30–358E and exhibit a large component of opening, more similar to rift-related structures than to other faults in the SISZ. On the eastern side of the graben (Fig. 8, photo a), gaping fissures up to 2 m

wide and over 4 m deep opened up a few meters west (in the hanging wall) of the west-dipping graben bounding fault (dashed line in Fig. 8). Offset along the longest fracture was measured as 10 cm of left-lateral strike slip and 38 cm of opening, although the leftstepping geometry of fractures within the graben is suggestive of right-lateral strike slip. Increased tension along the base of a barbed wire fence indicates that both extension and up to 1 m of new subsidence occurred across the graben. Many fractures and sinkholes were water-filled and several were coated in a white residue, possibly fault gouge, where water had risen and then subsided soon after the earthquake (Fig. 8, photo c). On the western side of the graben,

Road 1

63.95oN

20 m

100 m 20.69oW

Fig. 9. Map of rupture along the Flatholt segment. Inset shows blow-up example of bdouble en echelonQ pattern typical of fracture arrays separated by push-ups (dark areas).

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

151

fractures opened at the base of the graben-bounding fault (heavy lines in Fig. 8, photo b). 4.2.2. Flatholt segment The Flatholt segment extends for 750 m SSW from Road 1 (see Figs. 2 and 9). It exhibits very clearly the bdouble en echelonQ character so typical of faults in the SISZ. A series of left-stepping fracture arrays ranging from about 40 to 100 m in length and striking N30E is made up of fractures on the order of 5 m in length and striking approximately N50E. The larger order arrays are separated by push-up structures (dark areas in Fig. 9 inset) in the columnar-jointed lava that

lies under the thin soil. The segment crosses an open field, a pond and a drainage ditch. During mapping, water was observed draining out of the pond and the drainage ditch was nearly dry. 4.2.3. Bitra segment The segment at Bitra (Fig. 10) has a general trend of N778E and is approximately 2.5 km long, making it by far the longest E–W trending fault segment to be recognized in the SISZ. Four right-stepping fracture arrays were located within this segment, striking between N50 and 608E. The longest array (N500 m) at the eastern end of the segment has been

63.955 N

road 30 parking area Bitra
d1 Roa

a

b

400 m

a

b

Fig. 10. Map of rupture along the Bitra segment and photos showing rupture style (see text for further explanation). Black box shows area mapped in detail by Bergerat and Angelier (2003). Grey box shows parking area discussed in Angelier and Bergerat (2002). (Photo a) Sinkhole on the farm Bitra. (Photo b) Right-stepping en echelon open gashes indicating left-lateral strike slip.

152

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

mapped in part and discussed in detail by Bergerat and Angelier (2003). It too displays the typical bdouble en echelonQ pattern, but this time with a right-stepping geometry indicative of left-lateral strike slip. Fractures in the western part of the segment are considerably shorter and consist primar-

ily of open gashes and linear sinkholes (Fig. 10, photos a and b). The Bitra segment lies along a stretch of National Road 1 (Iceland’s Ring road). The road was severely fractured along the entire 2.5 km length of this segment, and fractures continued into fields on both sides of the road.

Fig. 11. Map of rupture at Hestfjall and photos showing rupture style (see text for further explanation). Box indicates area enlarged in Fig. 14. Letters on map corresponding to photo locations. (Photo a) Reactivated fracture. (Photo b) Shattered rock and rockfall.

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

153

4.2.4. Hestfjall segment The epicenter of the June 21 earthquake was located on the southern slopes of Hestfjall (Fig. 2). On the western side of Hestfjall, deformation is distributed across an N208E trending zone of preexisting fractures which had been mapped by Einarsson and Jonsdottir (1999, unpublished report). ´ ´ In many places freshly opened fractures are partly obscured by a thick moss cover and the only evidence of opening is small left-stepping gashes in the surface of the moss. In other places, the walls of old fractures have widened and the area between them has subsided (Fig. 11, photo a). The leftstepping nature of fracture segments indicates rightlateral strike slip. Along strike from the fractures, the fault emerges at the top of the cliffs on the south side of Hestvatn. There the entire scarp has experienced extensive shattering of rock and severe rockfall (Fig. 11, photo b).

4.2.5. Goltur segment ¨ To the northwest of Hestfjall, relatively continuous rupture was observed on the farm Gfltur (Figs. 2 and 12). This segment is approximately 3 km in length and contains the longest rupture array observed anywhere along the June 21 fault. Rupture occurs predominantly along the trace of a structure previously mapped as a normal fault (Johannesson et al., ´ 1990). With the exception of a 300 m long, northtrending set of fractures emerging from the shore of Lake Hestvatn, all of the rupture arrays strike between N20 and 308E. Most are on the order of 100–150 m in length, but the longest continuous set of fractures is over 600 m long and strikes N218E. It is comprised of closely spaced, left-stepping opening-mode fractures in thick soil (Fig. 12, inset map). Individual fractures are as much as 2 m deep and have experienced subsidence of up to 70 cm. No evidence of strike-slip displacement was observed anywhere in the valley,

20.74oW 64.04oN Eyvik

20.70oW

64.02oN 50 m

Göltur Hestvatn 500 m

Fig. 12. Map of rupture at Gfltur (see text for further explanation). Inset map shows blow-up of left-stepping en echelon fracture pattern indicating right-lateral strike slip.

154

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

but the left-stepping nature of the fractures is typical of right-lateral strike slip. Rupture on the Gfltur segment continues onto the farm Eyvik, where a 100 m length has been described by Bergerat and Angelier (2003) as resulting from normal faulting. While we are not sure precisely where their Eyvik segment is, we did not observe normal faulting. We observed two separate occurrences of rupture at the Eyvik farm. One set of fractures west of Lake Hestvatn appears to follow the topography around the base of a low hill adjacent to swampy ground. We interpreted this as subsidence due to shaking. On the eastern side of Lake Hestvatn we mapped fractures that had a consistently left-stepping nature, consistent with right-lateral strike slip. No primary surface rupture was observed north of this segment, although subsidence a few kilometers to the north caused fracturing of a road.

5. Discussion and conclusions 5.1. Distribution and geometry of rupture In general the distribution and the style of faulting associated with the June 17 and 21 earthquakes is consistent with right-lateral slip on N–S striking vertical faults as shown by focal mechanisms of the mainshocks and modeling of GPS- and InSAR data ´ (A rnado ttir et al., 2001, Pedersen et al., 2001; ´ Pedersen et al., 2003). There were, however, important deviations from a simple model. For both the June 17 and June 21 faults, displacement at the surface is distributed on several fracture arrays or segments and as deformation of the blocks between them. These fractures are sometimes parallel or sub-parallel, sometimes splay faults or overlapping en echelon segments. According to the best fit fault model from joint inversion of GPS and InSAR data (Pedersen et al., 2003), both faults extend to 10 km depth. However, over 80% of slip occurred in the middle and upper crust, above 6 km depth and above the great majority of aftershocks. GPS data for the June 21 earthquake in particular indicate high slip at ´ shallow depth (Arnadottir et al., 2001). Therefore it is ´ logical to assume that surface rupture reflects the geometry of shallow crustal structure rather than the linear fault outlined by the zone of deeper aftershocks.

Surface rupture along the June 17 fault trace is distributed asymmetrically within a 2 km wide zone centered on the projected location of the subsurface faults delineated by the aftershock distribution (Fig. 13A) and also with respect to the best fit fault model of Pedersen et al. (2003) (Fig. 13B). Most rupture occurs on the western edge of that zone. Our data shows that the distribution and intensity of surface rupture is in excellent agreement with the distributed slip model of Pedersen et al. (2003). Rupture along the SkammbeinsstaWir segment, nearly 2.5 km of continuous fracture, falls along their zone of maximum slip. The occurrence of rupture decreases more quickly to the south than to the north of the hypocenter, also in good agreement with their model. The June 21 rupture, on the other hand, deviates considerably and in a systematic way from the surface projection of the aftershock zone, and agreement with the distributed slip model is slightly less robust (Fig. 13B). Rupture along the northern part of the fault falls well to the west of the aftershock zone, whereas rupture in the southern part is all on the eastern side of the aftershocks. The middle part is mostly occupied by the large conjugate Bitra segment. We note that, in spite of its size, the Bitra segment has the same geometric arrangement with the other large segments as most other small conjugate segments have with their adjacent segments (see discussion below). The Bitra segment does not have a clear expression in the aftershock distribution (K. VogfjfrW, 2004, personal communication), which reflects the main fault at depth. In fact the majority of aftershocks occurring west of the mainshock are located well to the south of surface rupture along the Bitra segment. The branching of the fault that is shown by the surface fractures therefore seems to be a shallow phenomenon, probably restricted to the uppermost 2 km of the crust. The segment closely parallels the geological boundary between the post-glacial Thjorsa lava and ´ ´ late Tertiary lavas and hyaloclastites, suggesting the possibility that the same underlying E–W trending structure controlled both the path of the lava and the current surface rupture. In the non-uniform slip model, the zone of maximum slip for the June 21 fault is located beneath Lake Hestvatn, where mapping could not be done. However, intense and relatively continuous rupture is found in the field both north and south of the lake,

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

155

Fig. 13. (A) Aftershock hypocenters automatically located by the South Iceland Lowland network during the period June 17 to November 22 superimposed on map of surface rupture (shown in black). Aftershocks in red from June 17 to July 1. Aftershocks in blue from July 1 to November 22. (B) Surface projection of distributed fault slip model from combined inversion of GPS and InSAR data (Pedersen et al., 2003), normalized by slip area in a vertical column beneath the surface trace. Color scale shows slip distribution in 0.5 m increments. Aftershock hypocenters shown in grey. Surface rupture mapped during this study shown in black.

along the Hestfjall and Gfltur segments, and rupture could very well have occurred beneath the lake. Surface rupture decreases to the north and south in

good agreement with the distributed slip model (Pedersen et al., 2003). The decrease in surface rupture towards the ends of both the June 17 and

156

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

June 21 faults is consistent with the conclusions of King (1986) who describes the free surface of the earth as a relaxation barrier to rupture and suggests that surface rupture will only occur when fault slip at depth is greater than 1 m. Conjugate faulting occurred along both the June 17 and June 21 faults. In most cases, conjugates are short segments of left-lateral faulting with ENE-to NE-erly trends, a few tens of meters long, located between the tips of adjacent en echelon segments of the main fault zone (see e.g. Fig. 14). They are well within the fractured zone and entirely of secondary nature. The angle between the two sets of conjugate faults is often surprisingly small. Angles as small as 108 are found. Assuming that the maximum principal stress r 1 bisects this angle, we conclude that it is almost parallel with the fractures. This implies that the friction coefficient has to be large. The aftershock pattern from the June 2000 earthquakes is consistent with the conceptual bookshelf

100 m

Fig. 14. Small scale fracture array from the Hestfjall rupture segment of the June 21 fault showing the relationship of ENE to E–W trending conjugate fractures to the predominant fractures. Length of conjugate fractures comprises 14% of total array length.

model of faulting in the juvenile transform zone of South Iceland, the South Iceland Seismic Zone. This model was proposed to explain the apparent lack of left-lateral faults conjugate to the dominating rightlateral N–S arrays of fractures (Einarsson and Eir´ksson, 1982). The conclusion of Bergerat and ı Angelier (2003) that conjugate (E–W trending) faults play a larger role in accommodating deformation in the SISZ than has been previously noted in the literature is based on their study of the Bitra segment, the single largest anomaly in the fault pattern of the whole zone. Conjugate segments clearly play an essential role in accommodating deformation in the SISZ, but our data supports the conclusion that the role they play is a subordinate one, joining together adjacent right-lateral segments or being associated with push-ups. The Bitra segment, which is by far the largest conjugate structure found so far in the SISZ, falls into this category. It comprises approximately 15% of the total length of the surface trace of the fault and bridges the gap between the northern and southern parts of the fault which are offset to the W and E, respectively, from the surface projection of the fault at depth. A 100 m long leftlateral array along the Mykjunes segment of the June 17 fault, described by Bergerat and Angelier (2003) as representative of that fault, comprises less than 1% of the entire fault length. When added to the length of the only other left-lateral array on the June 17 fault, along the Stufholt segment, less than 5% of the total fault ´ length is made up of conjugate segments. This pattern of major north-striking right-lateral strike-slip faults and minor east-striking left-lateral strike-slip conjugates was reproduced in clay models of oblique extension, where the angle, a, between the displacement direction (equivalent to the absolute spreading direction) and the deformed zone trend was 158 (Clifton et al., 2000; see Fig. 15). This highly oblique geometry well represents that of the SISZ with respect to the NUVEL-1A spreading direction (DeMets et al., 1994). Early in the evolution of the model, both sets of faults form in equal number, but with continued extension at a low strain-rate, the north–south faults grow more quickly. Their length is controlled by the width of the deforming zone, whereas the length of the conjugates is controlled by the distance between north and south faults. Earlier experiments (Withjack and Jamison, 1986) with almost identical boundary conditions, but run at a

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

157

α

R D

A
Frequency
125 100 75 50 25 0 270 300 330

α = 15o
N RαD R

α=

15o

N 358o 048o

α

D

B

C

˚

˚

Azimuth

˚ 0˚

30

˚

60

˚

90

˚

Fig. 15. (A) Fault trace map for the a=158 model from laboratory experiments of oblique rifting (Clifton et al., 2000). (B) Histogram of fault strike from model shows a primary fault set with strike centered on 3588, and secondary fault set with a more northeasterly strike (see text for discussion). (C) Schematic diagram showing the geometric relationship between N, normal to the displacement direction, D, the displacement direction, and R, the rift trend.

much higher strain-rate, produced an opposite fault pattern in which east-striking left-lateral faults were predominant. Results of clay model experiments suggest that strain rate may be the controlling factor in determining which set of faults predominate. Schlische et al. (2002) concluded that, for situations where a is V308, oblique and strike-slip faults that are sub-parallel to and have the same sense of shear as the deformed zone become dominant as the deformation rate increases. In Iceland, where the spreading rate, and hence the deformation rate, is slow, the dominant faults are sub-perpendicular to the deformation zone while those sub-parallel to it are secondary. 5.2. Subsidence vs. normal faulting In many locations along the fault trace, subsidence was observed. We conclude that in most cases this

was due to differential compaction across the fault rather than by a component of normal slip along the fault. Such subsidence was observed primarily in areas of thick soil, and particularly where the fault trace follows the foot of a hill where boggy soil meets bedrock. Similar structures were also found far from the fault trace where no fault is present. For example, much of the cracking of roads was due to shaking and compaction of poorly consolidated roadbed material. In most cases, the orientation of these cracks had no tectonic significance, but instead appeared either parallel to or perpendicular to the road itself. The one location where actual normal slip may have occurred is along the D&larett segment, where open´ ing occurred along both bounding faults of a preexisting graben, and tension in a fence indicated subsidence of the graben floor and extension across the graben. In a stress system consisting of an E–W

158

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159

oriented left-lateral shear couple, structures with an NE strike, such as the graben at D&larett, could well ´ experience extension and normal slip. 5.3. Fault reactivation It has been a matter of some controversy whether SIZS earthquakes occurred on new faults each time or were caused by reactivation of pre-existing structures. Bjarnason et al. (1993) found some evidence that the fault active in the 1912 earthquake had ruptured earlier. Other evidence has not been found for repeated rupture on the same fault. The June 2000 events offered a possibility to check this. Some of the fault segments active in June 2000 had been identified and mapped previously. Rupture on the island of ´ Arnes and most of the rupture on the farm Gfltur utilized fractures which appear on the geological map of South Iceland (Johannesson et al., 1990). Rupture ´ at D&larett and some of the rupture on Hestfjall ´ occurred along faults that had been previously mapped by Einarsson and Jonsdottir (1999, unpub´ ´ lished report). In other cases, it became apparent during our mapping that the surface rupture was associated with pre-existing structures. Primary surface rupture along the graben at Mykjunes, the northern part of the Skammbeinsstadir segment and ´ the Arnes segment on the farm Akbraut followed small-scale topographic lineaments that clearly outline the trace of a buried fault. 5.4. Comparison with historical earthquakes Comparing the observed surface faults of the 2000 earthquakes with those of earlier events in the SISZ, such as the 1630 event, it must be concluded that the earlier, pre-instrumental, event was significantly larger. In the short account of the 1630 event, it is specifically stated that bnew fractures were formed near the farm Minnivellir where there were none beforeQ (Thoroddsen, 1899). The prominent fractures that are still exposed in the area, and were mapped first by Einarsson and Eirıksson (1982), were there´ fore by all probability formed in the 1630 event. They include structures, both fissures and push-ups, that are an order of magnitude larger than anything mapped from the 2000 earthquakes. The size of the 1630 structures is, however, comparable to similar struc-

tures formed in the 1912 earthquake (Einarsson and Eirıksson, 1982; Bjarnason et al., 1993), which has an ´ instrumentally determined magnitude of M s=7.0. It can be concluded that the 1630 event was of magnitude close to 7. Similar arguments cannot be used for other historic events or faults because it cannot be ascertained whether the fractures were formed in one or more events. With few exceptions, the scale of newly mapped rupture from the June 2000 earthquakes is such that it is not likely to be preserved at the surface for long. This fact highlights the difficulty encountered in determining the slip history of faults in the SISZ from surface appearance alone.

Acknowledgements This work was partially funded by European Union projects RETINA (EVG1-CT-2001-0044) and PREPARED (EVG1-CT-2002-00073). We gratefully acknowledge the contributions of Kristın VogfjfrW ´ and her colleagues at the Icelandic Meteorological office, who provided the seismic data presented. We also thank Rikke Pederson for providing original data on slip distribution. The authors thank two anonymous reviewers for critical remarks.

References
Angelier, J., Bergerat, F., 2002. Behaviour of a rupture of the 21 June 2000 earthquake in South Iceland as revealed in an asphalted car park. J. Struct. Geol. 24, 1925 – 1936. ´ Arnadottir, Th., Hreinsdottir, S., GuWmundsson, G., Einarsson, P., ´ ´ Heinert, M., Vflksen, C., 2001. Crustal deformation measured by GPS in the South Iceland Seismic Zone due to two large earthquakes in June 2000. Geophys. Res. Lett. 28, 4031 – 4034. Bergerat, F., Angelier, J., 2001. Mecanismes des failles des seismes ´ ´ des 17 et 21 Juin 2000 dan la Zone Sismique Sud-Islandaise d’ ´ apres les traces de surface des failles d’Arnes et des l’Hestfjall. Comptes Rendus de l’Academie des Sciences, Paris, Sci. Terre ´ & Planetes 333, 35 – 44. ´ Bergerat, F., Angelier, J., 2003. Mechanical behavior of the Arnes ND Hestfjall Faults of the June 2000 earthquakes in Southern Iceland: inferences from surface traces and tectonic model. J. Struct. Geol. 25, 1507 – 1523. Bjarnason, I.Th., Cowie, P.A., Anders, M.H., Seeber, L., Scholz, C.H., 1993. The 1912 Iceland Earthquake rupture: growth and development of a nascent transform system. Bull. Seismol. Soc. Am. 83, 416 – 435. Clifton, A.E., Schlische, R.W., Withjack, M.O., Ackermann, R.V., 2000. Influence of rift obliquity on fault-population system-

A. Clifton, P. Einarsson / Tectonophysics 396 (2005) 141–159 atics: results of experimental clay models. J. Struct. Geol. (22), 1491 – 1509. Clifton, A.E., Pagli, C., Jonsdottir, J.F., Eythorsdottir, K., VogfjfrW, ´ ´ ´ K., 2003. Surface effects of triggered fault slip on Reykjanes Peninsula, SW Iceland. Tectonophysics 369, 145 – 154. DeMets, C., Gordon, R., Argus, D., Stein, S., 1994. Effect of recent revisions to the geomagnetic reversal time scaled on estimates of current plate motions. Geophys. Res. Lett. (21), 2191 – 2194. Einarsson, P., Eirıksson, J., 1982. Earthquake fractures in the ´ districts Land and Rangarvellir in the South Iceland Seismic ´ Zone. Jfkull 32, 113 – 120. Einarsson, P., Saemundsson, K., 1992. Earthquake epicenters 1982– 1985 and volcanic systems in Iceland (map). In: Sigfusson, Th. ´ ´ (Ed.), I Hlutarsins EWli: Festschrift for Thorbjorn Sigurgeirsson. MenningarsjoWur, Reykjavık. ´ ´ Einarsson, P., Bjfrnsson, S., Foulger, G., Stefansson, R., Skaftadottir, ´ ´ Th., 1981. Seismicity pattern in the South Iceland seismic zone. In: Simpson, D., Richards, P. (Eds.), Earthquake Prediction— An International Review, Am. Geophys. Union, Maurice Ewing Ser., vol. 4, pp. 141 – 151. Einarsson, P., Bfttger, M, Thorbjarnarson, S., 2002. Faults and fractures of the South Iceland Seismic Zone near Thjorsa. The ´ ´ Icelandic Power, Report LV-2002/090, 8 p. Hackman, M.C., King, G.C.P., Bilham, R., 1990. The mechanics of the south Iceland seismic zone. J. Geophys. Res. 95, 17,339 – 17,351. Imsland, P., Einarsson, P., Thorbjfrnsson, D., Jonsson, S., Bjarna´ son, P., 1997. Mapping of earthquake fractures in Selfoss town for planning purposes, (Report in Icelandic). Science Institute, University of Iceland, Report RH-05-97, 52 p. Johannesson, H., Jakobsson, S.P., S&mundsson, K., 1990. Geo´ logical map of Iceland, sheet 6, South-Iceland. (third edition) Icelandic Museum of Natural History and Iceland Geodetic Survey, Reykjavık. ´ King, G.C.P., 1986. Speculations on the geometry of the initiation and termination processes of earthquake rupture and its relation

159

to morphology and geologic structure. Pure Appl. Geophys. 124, 567 – 585. Pagli, C., Pedersen, R., Sigmundsson, S., Feigl, K.L., 2003. Triggered fault slip on June 17, 2000 on the Reykjanes Peninsula, SW-Iceland captured by radar interferometry. Geophys. Res. Lett. 30 (6), 1273m, doi:10.1029/2002GL015310. ´ Pedersen, R., Sigmundsson, F., Feigl, K.L., Arnadottir, Th., 2001. ´ Co-seismic interferograms of two M s=6.6 earthquakes in the South Iceland Seismic Zone, June 2000. Geophys. Res. Lett. 28, 3341 – 3344. ´ Pedersen, R., Jonsson, S., Arnadottir, Th., Sigmundsson, F., Feigl, ´ ´ K.L., 2003. Fault slip distribution of two June 2000 Mw6.5 earthquakes in South Iceland estimated from joint inversion of InSAR and GPS measurements. Earth Planet. Sci. Lett. 6713, 1 – 16. Schlische, R.W., Withjack, M.O., Clifton, A.E., 2002. Effect of deformation rate on fault populations. Geological Society of America Annual Meeting, Abstracts with programs, pp. 110 – 112. Sigmundsson, F., Einarsson, P., Bilham, R., Sturkell, E., 1995. Rifttransform kinematics in south Iceland: deformation from Global Positioning System measurements, 1986 to 1992. J. Geophys. Res. 100, 6235 – 6248. Stefansson, R., BfWvarsson, R., Slunga, R., Einarsson, P., Jakobs´ dottir, S.S., Bungum, H., Gregersen, S., Havskov, J., Hjelme, J., ´ Korhonen, H., 1993. Earthquake prediction research in the South Iceland seismic zone and the SIL project. Bull. Seismol. Soc. Am. 83, 696 – 716. Stefansson, R., GuWmundsson, G., Halldorsson, P., 2000. The two large ´ ´ earthquakes in the South Iceland Seismic Zone on June 17 and 21, 2000. http://hraun.vedur.is/ja/skyrslur/June17and21_2000/ index.html. Thoroddsen, T., 1899. JarWskjalftar a SuWurlandı (Earthquakes in ´ ´ ´ ´ South Iceland). HiW Islenska Bokmenntafelag, Copenhagen. ´ ´ Withjack, M.O., Jamison, W.R., 1986. Deformation produced by oblique rifting. Tectonophysics 126, 99 – 124.