- Email: [email protected]
Seismicity crisis at the Katla volcano, Iceland—signs of a cryptodome?
Journal of Volcanology and Geothermal Research 153 (2006) 177 – 186 www.elsevier.com/locate/jvolgeores
Seismicity crisis at the Katla volcano, Iceland—signs of a cryptodome? Heidi Soosalu a,*,1, Kristı´n Jo´nsdo´ttir b,2, Pa´ll Einarsson c,3 a
b
Nordic Volcanological Center, University of Iceland, Sturlugata 7, 101 Reykjavı´k, Iceland Department of Earth Sciences, Geophysics, University of Uppsala, Villava¨gen 16, 75236 Uppsala, Sweden c Institute of Earth Sciences, University of Iceland, Sturlugata 7, 101 Reykjavı´k, Iceland Received 7 March 2005; received in revised form 1 June 2005; accepted 31 October 2005 Available online 24 January 2006
Abstract Katla, one of the most active and hazardous Icelandic volcanoes, has shown signs of unrest since 1999. In July that year a small glacial flood lasting a few hours was observed, accompanied by abnormal seismicity. It was probably caused by a short-lived, shallow magma intrusion or possibly a small subglacial eruption. In 2002 seismic activity increased dramatically, particularly in the Goðabunga area west of the caldera, and earthquakes have occurred continuously ever since. A network of four portable seismometers was run at Katla in the spring–summer 2003, with one station directly at the locus of the Goðabunga activity. These observations are combined with the data of the permanent Icelandic digital seismograph network, run by the Icelandic Meteorological Office. The earthquakes at Goðabunga are concentrated in a small area with a diameter of 3 to 4 km, and the events are shallow, mainly within the uppermost 2 km. We propose that the unusual pattern of seismicity indicates an intruding hot and acidic cryptodome. Apparently this dome has been slowly rising for decades but its propagation towards the surface accelerated some time around 1999, possibly triggered by a basaltic intrusion. This may lead to a silicic dome-building eruption in the Goðabunga area, which should be taken into consideration when estimating the eruption hazard at Katla. D 2005 Elsevier B.V. All rights reserved. Keywords: Katla volcano; Iceland; My´rdalsjo¨kull; cryptodome; seismicity; silicic eruption
1. Introduction The volcanic system of Katla in Iceland is located south of the rift–transform intersection of the Eastern * Corresponding author. Tel.: +44 1223 337 188; fax: +44 1223 360 779. E-mail addresses: [email protected] (H. Soosalu), [email protected] (K. Jo´nsdo´ttir), [email protected] (P. Einarsson). 1 Present address: Bullard Laboratories, Department of Earth Sciences, University of Cambridge, Madingley Road, Cambridge CB3 0EZ, U.K. 2 Tel.: +46 18 471 1475; fax: +46 18 501 110. 3 Tel.: +354 525 4816; fax: +354 552 1347. 0377-0273/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2005.10.013
volcanic zone and South Iceland seismic zone (Fig. 1), and consists of the Katla central volcano and the Eldgja´ fissure zone north-east of it. Katla is an off-rift volcano, but the rifting activity of the Eastern volcanic zone is propagating towards it to the south-west. Katla is largely covered by the My´rdalsjo¨kull glacier and has an oval caldera with a 14-km NW–SE major axis and highest rims reaching 1380 m a.s.l. (Bjo¨rnsson et al., 2000). A two-dimensional seismic profile of Katla has revealed a travel-time anomaly inside the caldera (Guðmundsson et al., 1994), interpreted as a 5-km-wide, shallow magma chamber with a bottom at 3 km below the surface. Katla is one of the most active volcanoes in Iceland with at least 20 eruptions during the 1100-year histor-
178
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Fig. 1. Index map of the Katla area. The central volcanoes are outlined, their fissure swarms shaded grey (Einarsson and Sæmundsson, 1987) and the calderas are hatched (Jo´hannesson et al., 1990). Central volcanoes referred to in the text are E—Eyjafjallajo¨kull and K—Katla. Eg is the Eldgja´ fissure zone and JS the river Jo¨kulsa´ a´ So´lheimasandi. Dashed lines mark the glaciers, My is My´rdalsjo¨kull. Black squares are temporary seismometer stations, those of the Katla net are named, and black triangles are digital stations run by the Icelandic Meteorological Office. The small box shows the Goðabunga area. The inset shows the location of the neovolcanic zone of Iceland, divided in northern (NVZ), eastern (EVZ), and western (WVZ) section. SISZ stands for South Iceland seismic zone.
ical time (Þo´rarinsson, 1975; Larsen, 2000). Because it is covered by a glacier, its eruptions are phreato–magmatic, associated with high eruption columns and catastrophic meltwater floods. Historically large eruptions have occurred at fairly regular intervals about every 50 years. Since the most recent one was in 1918, Katla is considered long overdue. During the Holocene the volcanic system of Katla has been characterised by three types of eruptions (Larsen, 2000). Basaltic explosive eruptions inside the caldera are the most frequent type, as they have occurred approximately twice per century. The amount of material produced in an individual event is typically about 1 km3. Silicic explosive eruptions in or near the caldera are less common. Tephra layers from twelve silicic explosive eruptions in the period of about 6600–1700 yr BP have been analysed (Larsen et al., 2001) and Guðru´n Larsen (pers.comm., 2004) estimates that about twenty such eruptions have occurred in the last 6600 years. These events typically seem to produce volumes of about 0.1 km3. Large effusive basaltic fissure eruptions, producing volumes of the order of 10 km3, are the least common type; two major events have occurred during the Holocene, the most recent being the Eldgja´ fires around AD 934 (Larsen, 2000). The petrology of Katla is bipolar. Most of the eruptive products are transitional alkali basalts with a narrow compositional range (Jakobsson, 1979). The silicic eruptive material of Katla also has a narrow composi-
tional range with SiO2 content between 63% and 67% (Larsen et al., 2001). Katla has been seismically active for at least the past forty years, since seismic measurements have been conducted in the region (Einarsson and Brandsdo´ttir, 2000). There are two distinct seismic areas—currently the more active one is at the Goðabunga rise in the west; the other one is within the Katla caldera in the centre (Fig. 2). The earthquake activity is seasonal, with more events occurring during the autumn than during the spring. This pattern is especially pronounced for the Goðabunga cluster. There is virtually no seismicity at the beginning of the year, and the earthquakes start to occur in late summer. The activity is usually highest in October and drops off towards the end of the year. Einarsson and Brandsdo´ttir (2000) explain this pattern to be caused by the triggering effect of increased groundwater pore pressure in the crust beneath the glacier after the summer thaw. The deloading of the glacial ice may also contribute to the phenomenon. Seismograms of Katla earthquakes have been observed to have a characteristic low-frequency volcanic appearance, especially for those that occur in the Goðabunga area (Einarsson and Brandsdo´ttir, 2000). The Pwaves often are small and emergent, and S-wave arrivals hard to distinguish, or non-existent. The duration of the signals is long compared to the amplitude; a long coda of surface waves is observed. While there has been a general consensus about the seasonal triggering of the Goðabunga seismicity, the
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
179
Fig. 2. Epicentral map of earthquakes in the Katla region during 2003. Although this is a part of the period of unusually high activity in the Goðabunga area, it illustrates well the general pattern of the seismicity. The map is constructed using the data gathered by the Icelandic Meteorological Office (http://www.hraun.vedur.is/ja/englishweb/index.html), and the temporary Katla network, when it was available. Shown are earthquakes that were observed by at least seven stations and had at least twelve P and S readings. The box shows the closer study area. The local magnitude scale is given in the inset.
ultimate cause of the earthquakes has been a puzzle. This question gained importance in the last few years when it became clear that the Katla volcano was showing signs of reawakening. One of the signs was a great increase in the seismicity of the Goðabunga cluster. During the spring and summer of 2003 we deployed a network of portable seismometers in the region of increased earthquake activity and obtained high-precision locations for the recorded events. This paper gives initial results from this experiment. 2. Current unrest at Katla In 1999 the pattern of activity at Katla changed. A sudden glacial flood, a jo¨kulhlaup occurred on 18 July 1999 in the river Jo¨kulsa´ a´ So´lheimasandi that drains the SW part of the My´rdalsjo¨kull glacier (Fig. 1) (Sigurðsson et al., 2000). The flood was preceded by a burst of seismic tremor and a new cauldron was formed in the ice surface above the S caldera rim. This sudden melting event was most likely caused by a minor subglacial eruption (Einarsson, 2000). Subglacial geothermal activity increased in the following
weeks. A number of pre-existing ice cauldrons were observed to deepen (Guðmundsson et al., 2000). GPS measurements in the Katla area indicate uplift of the volcano, starting some time in the year 1999 (Sturkell et al., 2003a). A GPS point on the NE-rim of the caldera was uplifted 0.07 m in the period 1999–2002 with the centre of uplift in the northern part of the caldera. Unusually high earthquake activity has been observed in the Goðabunga area since at least in the autumn of 2001. Fig. 3 shows how the seasonal pattern was progressively altered. The earthquake-poor period during the first half of the year gradually gets shorter until it finally disappears in 2002. The characteristic seasonal pattern of earthquake occurrences merges into continuous activity with earthquakes occurring on a daily basis. This activity has been going on ever since. 3. The seismic data set In the spring and summer of 2003 we deployed a four-station temporary network of broadband Gu¨ralp 6TD seismometers on nunataks, bedrock knobs protruding from the ice of the My´rdalsjo¨kull glacier and
180
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Fig. 3. Time-history of earthquake activity in the Goðabunga area (63.58–63.738N, 19.22–19.448W) in 1996–2003, from the Icelandic Meteorological Office database. Events of magnitudes z 1 are plotted against the date. A seasonal pattern was obvious until the autumn of 2001, since then earthquakes have occurred on a daily basis. However, also since 2001 more events have occurred during the fall than during the summer. The apparent bipolar distribution of event magnitudes in the plot, around magnitudes of about 1.5 and 2, is an artefact caused by the difficulty of defining the magnitudes from the unusual looking signals. However, this figure qualitatively shows the real development of seismicity.
at edges of the glacier (Figs. 1 and 2). The gathered data were combined with the recordings of the permanent SIL seismograph network (named after South Iceland Lowland) run by the Icelandic Meteorological Office (Bo¨ðvarsson et al., 1999) to provide a high-quality data
set of Katla earthquakes. The events were re-located with the program HYPOINVERSE (Klein, 1978) using a crustal model consisting of layers with constant velocity gradient and refined with local station corrections for the SIL stations (Soosalu and Einarsson, 1997). In this study we focus on the earthquake cluster at the Goðabunga rise. One of the stations, GODA, was located right on top of this activity and we thus had good depth control for these events. For geological interpretation of the seismicity in the Goðabunga area we accepted only events having strict location criteria. Each event was required to have root mean square travel time residual (rms) V 0.2 s, horizontal standard error (erh) V 0.6 km, vertical standard error (erz) V 0.9 km, and largest gap between observing stations V 1678. Generally these values are far smaller, in average rms = 0.14 s, erh = 0.32 km, erz = 0.37 km and gap = 938. We also required that the closest station GODA had recorded the event, that there were at least seven observing stations, and twelve picked P- or Sphases. During a hundred-day interval from March 12 to June 20 we gathered 231 such earthquakes, having local magnitudes of M L 0.3–2.4. The local network, with one station situated at the Goðabunga rise, revealed interesting features of the activity in the area. The records of the station GODA show that the seismicity was incessant during the study period. Small earthquakes, typically occurring in swarms, and often too small to be observed by more distant stations, were recorded on hourly basis. Previously, all seismic observations have shown low-frequency volcanic appearance for the Goðabunga signals. However, the records of the closest station GODA are more tectonic-like, show higher frequencies and have clear P- and S-wave arrivals. Path effects cause the low-frequency appearance in the records at the stations further away, and they vary in regard with the direction. The stations ENTA and AUST were located on nunataks near or at the caldera edge, at about 8–11 and 12–15 km from the Goðabunga area, respectively. Their records show low-frequency signals with emergent P- and unclear or absent S-phases for the same earthquakes showing high frequencies at GODA. The station SOHH was located at the south edge of the My´rdalsjo¨kull glacier, at a distance of 12–15 km from the Goðabunga rise. Typically SOHH records exhibit more high-frequency content than those of ENTA and AUST, and mostly impulsive P- and S-arrivals. An example set of records of a Goðabunga event is shown in Fig. 4. The Goðabunga events often occur not only in swarms but as QmultipleQ events. Typically there are
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
181
Fig. 4. Example seismograms from the four Katla net stations. This event was located at a depth of 0.9 km (shown with a black dot in Figs. 5 and 6) and had a M L magnitude of 1.2. The stations are shown in an increasing order of distance from the source. The amplitudes are not to scale between different stations. Z is the vertical component, N horizontal north and E horizontal east. The picked P and S arrivals are shown.
1–3 small events very close in time preceding a larger (M L~2) earthquake. Event records frequently overlap and reading onsets may be difficult. In some cases the GODA records start with a high-frequency onset, which is followed by low-frequency coda with considerably larger amplitude, apparently consisting of surface waves. Their appearance resembles hybrid events discussed by Chouet (1996).
Fig. 5. Horizontal slices of the Goðabunga earthquake cluster, only well-located (see criteria in the text) events are plotted. (A) All events. (B–D) 0.8 km thick slices from the surface down to a depth of 2.4 km (E) the deepest events at 2.4–4 km depth. The M L magnitude scale is in the inset. Error bars of one standard error are shown in B–E.
182
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
This study focuses on the cluster of the Goðabunga events (see the small box in Figs. 1 and 2). We analysed their spatial and temporal distribution during the 100day interval data were collected and suggest a geological interpretation. Horizontal and vertical cross-sections of the earthquake cluster are shown in Figs. 5 and 6. This bseismicity imagingQ of the Goðabunga volume reveals a well-delineated hypocentral distribution. Most events occurred in the uppermost 2 km; no events were observed at depths greater than 4 km. A view from the south (Fig. 6) shows that the cluster is inclined: the shallower events occurred in the west part and the deeper events in the east. The inclination of the cluster is towards the Katla caldera (its rim is located about 1 km east of the area shown in Fig. 5). The shallower events tend to be smaller in size than the deeper events. A small aseismic volume is found at depth of about 2 km depth, with horizontal dimensions of the order of 1 km, outlined by a bcapQ of fairly large earthquakes (M L V 2). During the 100-day operation period of the local net no spatial migration of seismicity was evident. Events occurred quite evenly throughout the whole volume of active seismicity at this time. 4. Geological interpretation and discussion
Fig. 6. Vertical cross-section of the Goðabunga earthquake cluster, seen from the south. Both the horizontal and the vertical scale are the same. (A) All events (B–E) 0.68-km-thick slices, progressively from south to north. The latitude limits of the slice are given in each plot. The M L magnitude scale is in the inset. Error bars of one standard error are shown in B–E.
Our observations of the Goðabunga seismicity in the spring and early summer 2003 suggest an inclined, stock-like body extending from the surface, downwards and eastwards to a depth of about 4 km. The largest earthquakes (M L~2) are clustered at about 1.5 km depth. Beneath this cluster there are few events and none deeper than at 4 km. The cluster sketches a convex lens-like object beneath which there is an aseismic area with the diameter of 1 km. We interpret the tightly-clustered shallow seismicity to reflect an intruding hot cryptodome, by definition a dome-shaped structure created by accumulation and ascent of viscous magma at shallow depth. It increases stress in the bedrock above and around its head, and this is released in the form of numerous earthquakes. The concentration of larger events at about 1.5 km depth is taken as a sign of intense deformation, and it outlines the head of the cryptodome. The cryptodome itself, however, is in a ductile state and thus aseismic. It is not possible to sketch the lower structure of this feature on the basis of seismic observations. The 1-km diameter is a rough estimate for the size of the head of the cryptodome and remains tentative with our event location accuracy. Its dimensions could be refined by calculating relative locations for the Goðabunga events. Extruded domes have diameters from a
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
few metres to several kilometres (Fink and Anderson, 2000). Our size estimate for the Goðabunga cryptodome fits within these dimensions. It is possible that the cryptodome itself is smaller in size. It may heat the rock in the immediate surroundings so that it too is in a ductile state. At least for the last three decades, i.e., the period of accurate epicentral determination, the seismicity distribution at Katla has shown a remarkably persistent pattern. Epicentres have been clustered in two areas, at Goðabunga rise by the caldera rim and within the Katla caldera (Einarsson and Brandsdo´ttir, 2000). The tight clustering as well as the eccentric position of the Goðabunga cluster have been the subject of speculation for some time. According to Fink and Anderson (2000) domes may have steady or episodic growth, with emplacement times ranging from a few hours to many decades. Our suggestion of a cryptodome implies that the Goðabunga dome has existed for a long time, at least some decades, being the ultimate cause of the seismicity in the area. Until about 1999 the stress loading due to the dome ascent was moderate and comparable to the rate of the seasonal triggering effect. The triggering effect was thus sufficient to limit the seismic activity to the summer and autumn months and turn the earthquakes off during the winter months. The dome ascent rate appears to have increased around 1999, contemporaneously with the beginning of inflation in the caldera (Sturkell et al., 2003a,b), and in the years 2001–2004 the seismicity at Goðabunga has been continuous. The seasonal triggering is active throughout this period also, however. The seismicity is clearly lower during the winter months although it does not turn off. The difference in the appearance of the seismograms for a single event at different stations is important for determining the nature of the Goðabunga earthquakes, and it should be studied further. Before installing the station GODA very close to the origins of these events it was not known if they are intrinsically low-frequency events or not. At an active volcano it is possible that nearby magma volumes can affect the seismic signal and cause misleading interpretations of the earthquake source. A similar observation was noted at the highly active Colima volcano in Mexico. Nu´n˜ez-Cornu´ et al. (1994) observed seismicity related to the 1991 eruption of Colima and state that all the seismicity should be classified as B-type events according to Minakami (1974), that is earthquakes with small magnitudes, unclear S-waves and low frequencies. However, they note that the seismometer closest to the volcano’s summit frequently observed the same events with Minakami’s
183
A-type features, i.e., the seismograms look similar to those of shallow tectonic events. Since the unrest began in 1999, the Katla area has been monitored closely. Observations of increased seismicity, inflation of the volcano and increased geothermal activity clearly show that eruptive activity may occur at any time. Initially the most likely scenario was expected to be a btraditionalQ Katla eruption, i.e., another basaltic eruption within the Katla caldera. Indeed, a minor basaltic eruption may already have occurred on July 18, 1999, when a glacial flood was observed and a depression formed on the ice surface in the south part of the caldera. However, the increased Goðabunga seismicity, tightly clustered outside the caldera, indicates that a wider range of scenarios should be considered. The feature that we interpret as a cryptodome may make its way to the surface and lead to volcanic phenomena associated with a silicic, domeforming eruption. The Goðabunga area has experienced abnormally high seismicity since the autumn of 2001. Apparently around this time, and possibly as early as 1999, our hypothesized cryptodome became more active and its slow propagation towards the surface accelerated. It is possible that the unrest beginning in the summer of 1999 spurred the propagation of the cryptodome, and in the autumn of 2001 it reached the seismogenic crust, which initiated continuous earthquake activity. The inclination of the seismically active volume towards the caldera speaks for a connection in that direction. Larsen (2000) states that silicic eruptions have only recently been recognized as a distinct phase in the Holocene activity of the Katla volcanic system. Although such events are infrequent and have not occurred during Historic times, they must be reckoned with in the future. Silicic Katla eruptions are not known to have occurred since 1700 yr BP. It has been suggested that the major effusive basaltic Eldgja´ eruption (ca. AD 934) changed the magma plumbing system of Katla and broke the formerly regular pattern of silicic eruptions (Larsen, 2000). However, a new pattern of silicic Katla eruptions may emerge. There are several examples of seismicity related to doming activity at volcanoes elsewhere in the world that can be compared with the observed Goðabunga activity. The Usu volcano in Japan, for instance, is a classical example of a volcano with dacitic dome growth. Okada et al. (1981) studied the seismicity related to the magma intrusion process of Usu in the period of 1978–1980. They detected a circular distribution of numerous shallow (b2 km) earthquake hypocentres beneath the volcanic edifice. Within this
184
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
activity there was an aseismic zone, which they interpreted as an intruding dome. The seismicity preceding and related to the 1990–1995 eruption of Unzen volcano in Japan also shows interesting patterns of migrating seismicity, starting about one year before the onset of the eruption, and continuing during the dome growth activity (Nakada et al., 1999; Umakoshi et al., 2001). Another current example of seismicity preceding the onset of a silicic eruption is the Soufrie`re Hills volcano in Montserrat, West Indies, where a domebuilding eruption started in 1995 and is still ongoing (Aspinall et al., 1998). Milia et al. (2000) made seismic studies at Campi Flegrei in Italy and found an offshore seismic unit, which they interpret as a cryptodome. They remark that the location of this cryptodome corresponds to an area of shallow (b 0.6 km) seismicity during a seismic crisis in 1970–1972. Dome-forming activity has occurred earlier in the Katla system, as silicic domes are found as nunataks around its caldera, such as Enta and Austmannsbunga (AUST), the sites of our seismic stations (Sturkell et al., 2003a). Some of them are thought to be of late-glacial age according to their chemical characteristics (Larsen, 2000). A comparable Icelandic example of a central volcano with acidic dome forming and bipolar geochemistry of basalts and rhyolites is Krafla in the Northern volcanic zone. Jo´nasson (1994) has studied the spatial distribution and geochemistry of rhyolites at Krafla. The bulk of the Krafla products are basaltic, but rhyolite domes are found at the caldera rim of Krafla or around it. Jo´nasson suggests that the rhyolites are formed by partial melting of crustal, hydrothermally altered basaltic rock. The rhyolites are generated on the peripheries of an active basaltic magma chamber of intrusive domain, where the temperature range is favourable for the genesis of rhyolite, i.e., not too hot, as would be expected above the chamber. Thus the result is a bhaloQ of rhyolites around the volcano. A similar pattern may well be valid at Katla. Larsen (2000) deduces that at least some rhyolitic eruptions have an origin inside the Katla caldera, according to the tephra layer observations. This conclusion may be based on insufficient data, and remains speculative (Guðru´n Larsen, pers. comm., 2004). Because of the glacial cover the caldera bottom geology of Katla cannot be directly observed. Acidic dome, or cryptodome growth is a feature that characteristically has a very localized stress regime. It may cause only little deformation in the surrounding bedrock, and the deformation signal decreases quickly with distance. To be able to detect Goðabunga deformation geodetically, measurements within a short dis-
tance are needed. GPS measurements at Katla and in the surroundings have been conducted since 1986 (Sturkell et al., 2003b) and a GPS point at Goðabunga (located at the same site as the temporary seismic station GODA) has recently been installed. Measurements in January and June 2004 reveal 2.6 cm uplift of the site during that period, pointing to very localised stress regime, distinctive for a protruding dome (Einarsson et al., 2005; Erik Sturkell, pers. comm., 2005). GPS points close to or at the sites AUST, ENTA and SOHH have been measured for a longer period, and they show uplift in recent years, with a best-fitting point source in the north part of the Katla caldera at 4–5 km depth, interpreted as the result of magma accumulation (Sturkell et al., 2004; Einarsson et al., 2005). Signals resulting from possible deformation in the Goðabunga area are not detected at these observation sites. The geothermal activity in the Katla area has been observed to have increased since the start of the unrest in 1999 (Guðmundsson et al., 2000). However, no distinct increase has been observed precisely at Goðabunga yet. It is to be expected that geothermal activity will intensify in this area in the future if the cryptodome continues propagating towards the surface. If the cryptodome is still moving upwards it may be possible to detect it in the development of the seismic activity, even though precise hypocentral locations for the whole period are not available. When a cryptodome propagates upwards, the brittle crust above it thins with time and the deformation field becomes more localised. This leads to a smaller volume in which cryptodomeinduced earthquakes occur, with smaller and fewer events. Our 100-day period of well-constrained hypocentral locations of Goðabunga events provides only a snapshot of the cryptodome-related seismicity. We do not have as good accuracy for event locations during the whole period since the fall of 2001, and thus cannot follow their temporal development in detail. Seismicity in the Goðabunga area from the autumn of 2001 until end of 2004 has been incessant. A decrease in the rate of earthquake activity, or decrease in the sizes of the events, may be either a good or a bad sign. It is possible that the cryptodome may freeze and remain at depth, and thus the seismicity related to it would cease. But if the cryptodome continues its propagation towards the surface, it may be expressed by a similar pattern of decreasing seismicity, when the brittle, seismogenic volume above it gets smaller and smaller. In such a case an eruption may be imminent. Other indications, such as ground uplift and increased geothermal activity, will then be expected to provide additional warning.
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Katla is generally considered the most dangerous of the Icelandic volcanoes. Until quite recently the concerns of the civil protection agencies have been directed towards the eastern side of the volcano where almost all catastrophic floods in Historic times have occurred. The Goðabunga activity in recent years and attempts to explain it have demonstrated that a wider spectrum of scenarios has to be considered. It appears conceivable that the next large eruption of Katla will be different from eruptions of the last centuries. In particular, the possible consequences of a silicic eruption in the western part of the volcano have to be looked into. The recent activity has led to re-evaluation of hazards due to volcanic activity in western part of Katla and its neighbouring volcano Eyjafjallajo¨kull (Guðmundsson and Gylfason, 2005). A Katla eruption becomes more likely as time passes. Possible scenarios include the following (Einarsson et al., 2005): 1. A basaltic, phreato-magmatic eruption within the caldera, following the typical pattern of Katla’s historic eruptive activity. 2. If the hypothesized cryptodome reaches the surface, a silicic, explosive dome-forming eruption at Goðabunga. The hazard from such an event in its full extent may be difficult to estimate, as there is no historical experience of such an event. It is also important to bear in mind that Goðabunga is located close to popular tourist and hiking areas, thus posing a great risk to people, especially if an eruption starts during the summer season. 3. A major debris avalanche triggered by an intruding dome at Goðabunga, in a same manner as the case of Mt. St. Helens eruption in 1980 (e.g., Fink and Anderson, 2000). 4. A basaltic eruption within the Katla caldera, or a basaltic intrusion, which may induce a silicic eruption in the Goðabunga area. The eruptive activity may even be more widespread, with an accompanying eruption of the neighbouring Eyjafjallajo¨kull volcano to the west (Fig. 1). Eyjafjallajo¨kull showed signs of unrest in 1994 and 1999 (Sturkell et al., 2003b), with magma intrusions beneath its southern flank (Pedersen and Sigmundsson, 2004; 2005). Both its documented eruptions in Historic times, in 1612 and 1821, were associated with Katla eruptions. 5. Conclusions Due to the improved accuracy of locations of Goðabunga earthquakes during the operation of our field instruments in the spring and summer of 2003 we are
185
able to conclude that the earthquake cluster at Goðabunga is even denser than has been suggested before. This concentrated source of seismicity has been persistently active for at least four decades and is associated with very local deformation. We offer an explanation for this activity and suggest it is an expression of a hot, acidic cryptodome that has been intruding towards the surface during the last decades. At present its top appears to be at the depth of 1.5 km. Its rate of ascent has increased during the last years. The interpretation is supported by geodetic measurements near the centre of the cluster, and the existence of older acidic domes around the caldera rim of Katla. Acknowledgements We are grateful to Bob White and the SEIS-UK instrument pool for kindly lending the portable seismometers, which made the whole study possible. Icelandic Meteorological Office gathered the additional data. Erik Sturkell suggested the logistically great idea to unite the forces in a joint field campaign with his GPS measurements. Heartfelt thanks to the brave field ´ lafsson, Gre´tar people Haukur Brynjo´lfsson, Halldo´r O Einarsson, Jo´nas Erlendsson, Halldo´r Geirsson, Karl Gro¨nvold, Dorthe Holm, Francien Peterse, Rune Selbekk, the Rescue Team in Vı´k, and Benedikt Bragason with his men from the Arcanum Adventure Tours for the competent snow scooter guiding. The late Katrı´n Brynjo´lfsdo´ttir and Guðgeir Guðmundsson from the village Vı´k at the roots of Katla are warmly remembered for homelike accommodation for the team during the field trips. Guðru´n Larsen provided an interesting discussion upon the silicic Katla volcanism. Bjo¨rn Lund made good comments on the manuscript. The public domain GMT software (Wessel and Smith, 1998) was used to draw the figures. Constructive criticism by Silvio De Angelis, Alejandro Nava and the editor Peggy Hellweg improved the manuscript. References Aspinall, W.P., Miller, A.D., Lynch, l.L., Latchman, J.L., Stewart, R.C., White, R.A., Power, J.A., 1998. Soufrie`re Hills eruption, Montserrat, 1995–1997: volcanic earthquake locations and fault plane solutions. Geophys. Res. Lett. 25, 3397 – 3400. Bjo¨rnsson, H., Pa´lsson, F., Guðmundsson, M.T., 2000. Surface and bedrock topography of the My´rdalsjo¨kull ice cap, Iceland: the Katla caldera, eruption sites and routes of jo¨kulhlaups. Jo¨kull 49, 29 – 46. Bo¨ðvarsson, R., Ro¨gnvaldsson, S.Th., Slunga, R., Kjartansson, E., 1999. The SIL data acquisition system—at present and beyond year 2000. Phys. Earth Planet. Inter. 113, 89 – 101.
186
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Chouet, B., 1996. Long-period volcano seismicity: its source and use in eruption forecasting. Nature 380, 309 – 316. Einarsson, P., 2000. Course of events connected with the flood in Jo¨kulsa´ a´ So´lheimasandi in July 1999. Geoscience Society of Iceland, February Meeting, 2000, p. 14 (in Icelandic). Einarsson, P., Sæmundsson, K., 1987. Earthquake epicenters 1982– 1985 and volcanic systems in Iceland (map). In: Sigfu´sson, Þ.I. (Ed.), I´ hlutarins eðli, Festschrift for Þorbjo¨rn Sigurgeirsson. Menningarsjo´ður, Reykjavı´k. Einarsson, P., Brandsdo´ttir, B., 2000. Earthquakes in the My´rdalsjo¨kull area, Iceland, 1978–1985: seasonal correlation and connection with volcanoes. Jo¨kull 49, 49 – 73. Einarsson, P., Soosalu, H., Sturkell, E., Sigmundsson, F., Geirsson, H., 2005. The activity at the central volcano Katla and its vicinity since 1999 and possible scenarios. In: Guðmundsson, M.T., Gyl´ .G. (Eds.), Evaluation of Risk Due to Volcanic Eruption fason, A in Western My´rdalsjo¨kull and Eyjafjallajo¨kull. University Press, Reykjavı´k, Iceland, pp. 151 – 158 (in Icelandic). Fink, J.F., Anderson, S.W., 2000. Lava domes and coulees. In: Sigurðsson, H. (Ed.), Encyclopedia of Volcanoes. Academic Press, pp. 307 – 319. ´ .G. (Eds.), 2005. Evaluation of Guðmundsson, M.T., Gylfason, A Risk Due to Volcanic Eruption in Western My´rdalsjo¨kull and Eyjafjallajo¨kull. University Press, Reykjavı´k, Iceland, 210 pp. (in Icelandic). ´ ., Brandsdo´ttir, B., Menke, W., Sigvaldason, G.E., Guðmundsson, O 1994. The crustal magma chamber of the Katla volcano in South Iceland revealed by 2-D seismic undershooting. Geophys. J. Int. 119, 277 – 296. Guðmundsson, M.T., Ho¨gnado´ttir, Þ., Bjo¨rnsson, H., Pa´lsson, F., Guðmundsson, G., 2000. Geothermal activity at My´rdsalsjo¨kull and events in the summer 1999. Geoscience Society of Iceland, February Meeting, 2000, p. 13 (in Icelandic). Jakobsson, S.P., 1979. Petrology of recent basalts of the Eastern Volcanic Zone, Iceland. Acta Nat. Isl. 26 103 pp. Jo´hannesson, H., Jakobsson, S.P., Sæmundsson, K., 1990. Geological map of Iceland, sheet 6, South-Iceland, third edition. Icelandic Museum of Natural History and Iceland Geodetic Survey, Reykjavı´k. Jo´nasson, K., 1994. Rhyolite volcanism in the Krafla central volcano, North-East Iceland. Bull. Volcanol. 56, 516 – 528. Klein, F.W., 1978. Hypocenter location program HYPOINVERSE. US Geol. Surv. Open-file Rep., vol. 78-694, 113 pp. Larsen, G., 2000. Holocene eruptions within the Katla volcanic system, South Iceland: characteristics and environmental impact. Jo¨kull 49, 1 – 28. Larsen, G., Newton, A.J., Dugmore, A.J., Vilmundardo´ttir, E.G., 2001. Geochemistry, dispersal, volumes and chronology of Holocene silicic tephra layers from the Katla volcanic system, Iceland. J. Quat. Sci. 16, 119 – 132.
Milia, A., Torrente, M.M., Giordano, F., 2000. Active deformation and volcanism offshore Campi Flegrei, Italy: new data from highresolution seismic reflection profiles. Mar. Geol. 171, 61 – 73. Minakami, T., 1974. Seismology of volcanoes in Japan. In: Civetta, L., Gasparini, P., Luongo, G., Rapolla, A. (Eds.), Physical Volcanology, Developments in Solid Earth geophysics, vol. 6. Elsevier Scientific Publishing, Amsterdam, pp. 1 – 27. Nakada, S., Shimizu, H., Ohta, K., 1999. Overview of the 1990–1995 eruption at Unzen volcano. J. Volcanol. Geotherm. Res. 89, 1 – 22. Nu´n˜ez-Cornu´, F., Nava, F.A., De la Cruz-Reyna, S., Jime´nez, Z., Valencia, C., Garcı´a-Arthur, R., 1994. Seismic activity related to the 1991 eruption of Colima volcano, Mexico. Bull. Volcanol. 56, 228 – 237. Okada, Hm., Watanabe, H., Yamashita, H., Yokoyama, I., 1981. Seismological significance of the 1977–1978 eruptions and the magma intrusion process of Usu volcano, Hokkaido. J. Volcanol. Geotherm. Res. 9, 311 – 334. Pedersen, R., Sigmundsson, F., 2004. InSAR based sill model links spatially offset areas of deformation and seismicity for the 1994 unrest episode at Eyjafjallajo¨kull volcano, Iceland. Geophys. Res. Lett. 31, L14610. doi:10.1029/2004GL020368. Pedersen, R., Sigmundsson, F., 2005. Temporal development of the 1999 intrusive episode in the Eyjafjallajo¨kull volcano, Iceland, derived from InSar images. Bull. Volcanol. doi:10.1007/s00445005-0020-y. Sigurðsson, O., Zo´phonı´asson, S., I´sleifsson, E., 2000. The jo¨kulhlaup from So´lheimajo¨kull, July 18th 1999 (in Icelandic with English summary). Jo¨kull 49, 75 – 80. Soosalu, H., Einarsson, P., 1997. Seismicity around the Hekla and Torfajo¨kull volcanoes, Iceland, during a volcanically quiet period, 1991–1995. Bull. Volcanol. 59, 36 – 48. ´ lafsson, Sturkell, E., Einarsson, P., Sigmundsson, F., Geirsson, H., O ´ lafsdo´ttir, R., Guðmundsson, G.B., 2003a. The pressure H., O increases under Katla. Na´ttu´rufræðingurinn 71, 80 – 86 (in Icelandic with English summary). Sturkell, E., Sigmundsson, F., Einarsson, P., 2003b. Recent unrest and magma movements at Eyjafjallajo¨kull and Katla volcanoes, Iceland. J. Geophys. Res. 108, 2369. doi:10.1029/2001JB000917. ´ rnado´ttir, Þ., Pedersen, Sturkell, E., Geirsson, H., Sigmundsson, F., A R., Pagli, C., Einarsson, P., 2004. Crustal deformation and volcano dynamics in Iceland. The 26th Nordic Geological Winter Meeting, vol. 126. GFF, Usala, p. 51. Umakoshi, K., Shimizu, H., Matsuwo, N., 2001. Volcano-tectonic seismicity at Unzen volcano, Japan, 1985–1999. J. Volcanol. Geotherm. Res. 112, 117 – 131. ´ rbo´k Þo´rarinsson, S., 1975. Katla and annal of Katla eruptions. A Ferðafe´lags I´slands 1975 (The Iceland Touring Association Yearbook 1975), pp. 125 – 149 (in Icelandic). Wessel, P., Smith, W.H.F., 1998. New, improved version of Generic Mapping Tools released. Eos 79, 579.
Seismicity crisis at the Katla volcano, Iceland—signs of a cryptodome? Heidi Soosalu a,*,1, Kristı´n Jo´nsdo´ttir b,2, Pa´ll Einarsson c,3 a
b
Nordic Volcanological Center, University of Iceland, Sturlugata 7, 101 Reykjavı´k, Iceland Department of Earth Sciences, Geophysics, University of Uppsala, Villava¨gen 16, 75236 Uppsala, Sweden c Institute of Earth Sciences, University of Iceland, Sturlugata 7, 101 Reykjavı´k, Iceland Received 7 March 2005; received in revised form 1 June 2005; accepted 31 October 2005 Available online 24 January 2006
Abstract Katla, one of the most active and hazardous Icelandic volcanoes, has shown signs of unrest since 1999. In July that year a small glacial flood lasting a few hours was observed, accompanied by abnormal seismicity. It was probably caused by a short-lived, shallow magma intrusion or possibly a small subglacial eruption. In 2002 seismic activity increased dramatically, particularly in the Goðabunga area west of the caldera, and earthquakes have occurred continuously ever since. A network of four portable seismometers was run at Katla in the spring–summer 2003, with one station directly at the locus of the Goðabunga activity. These observations are combined with the data of the permanent Icelandic digital seismograph network, run by the Icelandic Meteorological Office. The earthquakes at Goðabunga are concentrated in a small area with a diameter of 3 to 4 km, and the events are shallow, mainly within the uppermost 2 km. We propose that the unusual pattern of seismicity indicates an intruding hot and acidic cryptodome. Apparently this dome has been slowly rising for decades but its propagation towards the surface accelerated some time around 1999, possibly triggered by a basaltic intrusion. This may lead to a silicic dome-building eruption in the Goðabunga area, which should be taken into consideration when estimating the eruption hazard at Katla. D 2005 Elsevier B.V. All rights reserved. Keywords: Katla volcano; Iceland; My´rdalsjo¨kull; cryptodome; seismicity; silicic eruption
1. Introduction The volcanic system of Katla in Iceland is located south of the rift–transform intersection of the Eastern * Corresponding author. Tel.: +44 1223 337 188; fax: +44 1223 360 779. E-mail addresses: [email protected] (H. Soosalu), [email protected] (K. Jo´nsdo´ttir), [email protected] (P. Einarsson). 1 Present address: Bullard Laboratories, Department of Earth Sciences, University of Cambridge, Madingley Road, Cambridge CB3 0EZ, U.K. 2 Tel.: +46 18 471 1475; fax: +46 18 501 110. 3 Tel.: +354 525 4816; fax: +354 552 1347. 0377-0273/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2005.10.013
volcanic zone and South Iceland seismic zone (Fig. 1), and consists of the Katla central volcano and the Eldgja´ fissure zone north-east of it. Katla is an off-rift volcano, but the rifting activity of the Eastern volcanic zone is propagating towards it to the south-west. Katla is largely covered by the My´rdalsjo¨kull glacier and has an oval caldera with a 14-km NW–SE major axis and highest rims reaching 1380 m a.s.l. (Bjo¨rnsson et al., 2000). A two-dimensional seismic profile of Katla has revealed a travel-time anomaly inside the caldera (Guðmundsson et al., 1994), interpreted as a 5-km-wide, shallow magma chamber with a bottom at 3 km below the surface. Katla is one of the most active volcanoes in Iceland with at least 20 eruptions during the 1100-year histor-
178
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Fig. 1. Index map of the Katla area. The central volcanoes are outlined, their fissure swarms shaded grey (Einarsson and Sæmundsson, 1987) and the calderas are hatched (Jo´hannesson et al., 1990). Central volcanoes referred to in the text are E—Eyjafjallajo¨kull and K—Katla. Eg is the Eldgja´ fissure zone and JS the river Jo¨kulsa´ a´ So´lheimasandi. Dashed lines mark the glaciers, My is My´rdalsjo¨kull. Black squares are temporary seismometer stations, those of the Katla net are named, and black triangles are digital stations run by the Icelandic Meteorological Office. The small box shows the Goðabunga area. The inset shows the location of the neovolcanic zone of Iceland, divided in northern (NVZ), eastern (EVZ), and western (WVZ) section. SISZ stands for South Iceland seismic zone.
ical time (Þo´rarinsson, 1975; Larsen, 2000). Because it is covered by a glacier, its eruptions are phreato–magmatic, associated with high eruption columns and catastrophic meltwater floods. Historically large eruptions have occurred at fairly regular intervals about every 50 years. Since the most recent one was in 1918, Katla is considered long overdue. During the Holocene the volcanic system of Katla has been characterised by three types of eruptions (Larsen, 2000). Basaltic explosive eruptions inside the caldera are the most frequent type, as they have occurred approximately twice per century. The amount of material produced in an individual event is typically about 1 km3. Silicic explosive eruptions in or near the caldera are less common. Tephra layers from twelve silicic explosive eruptions in the period of about 6600–1700 yr BP have been analysed (Larsen et al., 2001) and Guðru´n Larsen (pers.comm., 2004) estimates that about twenty such eruptions have occurred in the last 6600 years. These events typically seem to produce volumes of about 0.1 km3. Large effusive basaltic fissure eruptions, producing volumes of the order of 10 km3, are the least common type; two major events have occurred during the Holocene, the most recent being the Eldgja´ fires around AD 934 (Larsen, 2000). The petrology of Katla is bipolar. Most of the eruptive products are transitional alkali basalts with a narrow compositional range (Jakobsson, 1979). The silicic eruptive material of Katla also has a narrow composi-
tional range with SiO2 content between 63% and 67% (Larsen et al., 2001). Katla has been seismically active for at least the past forty years, since seismic measurements have been conducted in the region (Einarsson and Brandsdo´ttir, 2000). There are two distinct seismic areas—currently the more active one is at the Goðabunga rise in the west; the other one is within the Katla caldera in the centre (Fig. 2). The earthquake activity is seasonal, with more events occurring during the autumn than during the spring. This pattern is especially pronounced for the Goðabunga cluster. There is virtually no seismicity at the beginning of the year, and the earthquakes start to occur in late summer. The activity is usually highest in October and drops off towards the end of the year. Einarsson and Brandsdo´ttir (2000) explain this pattern to be caused by the triggering effect of increased groundwater pore pressure in the crust beneath the glacier after the summer thaw. The deloading of the glacial ice may also contribute to the phenomenon. Seismograms of Katla earthquakes have been observed to have a characteristic low-frequency volcanic appearance, especially for those that occur in the Goðabunga area (Einarsson and Brandsdo´ttir, 2000). The Pwaves often are small and emergent, and S-wave arrivals hard to distinguish, or non-existent. The duration of the signals is long compared to the amplitude; a long coda of surface waves is observed. While there has been a general consensus about the seasonal triggering of the Goðabunga seismicity, the
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
179
Fig. 2. Epicentral map of earthquakes in the Katla region during 2003. Although this is a part of the period of unusually high activity in the Goðabunga area, it illustrates well the general pattern of the seismicity. The map is constructed using the data gathered by the Icelandic Meteorological Office (http://www.hraun.vedur.is/ja/englishweb/index.html), and the temporary Katla network, when it was available. Shown are earthquakes that were observed by at least seven stations and had at least twelve P and S readings. The box shows the closer study area. The local magnitude scale is given in the inset.
ultimate cause of the earthquakes has been a puzzle. This question gained importance in the last few years when it became clear that the Katla volcano was showing signs of reawakening. One of the signs was a great increase in the seismicity of the Goðabunga cluster. During the spring and summer of 2003 we deployed a network of portable seismometers in the region of increased earthquake activity and obtained high-precision locations for the recorded events. This paper gives initial results from this experiment. 2. Current unrest at Katla In 1999 the pattern of activity at Katla changed. A sudden glacial flood, a jo¨kulhlaup occurred on 18 July 1999 in the river Jo¨kulsa´ a´ So´lheimasandi that drains the SW part of the My´rdalsjo¨kull glacier (Fig. 1) (Sigurðsson et al., 2000). The flood was preceded by a burst of seismic tremor and a new cauldron was formed in the ice surface above the S caldera rim. This sudden melting event was most likely caused by a minor subglacial eruption (Einarsson, 2000). Subglacial geothermal activity increased in the following
weeks. A number of pre-existing ice cauldrons were observed to deepen (Guðmundsson et al., 2000). GPS measurements in the Katla area indicate uplift of the volcano, starting some time in the year 1999 (Sturkell et al., 2003a). A GPS point on the NE-rim of the caldera was uplifted 0.07 m in the period 1999–2002 with the centre of uplift in the northern part of the caldera. Unusually high earthquake activity has been observed in the Goðabunga area since at least in the autumn of 2001. Fig. 3 shows how the seasonal pattern was progressively altered. The earthquake-poor period during the first half of the year gradually gets shorter until it finally disappears in 2002. The characteristic seasonal pattern of earthquake occurrences merges into continuous activity with earthquakes occurring on a daily basis. This activity has been going on ever since. 3. The seismic data set In the spring and summer of 2003 we deployed a four-station temporary network of broadband Gu¨ralp 6TD seismometers on nunataks, bedrock knobs protruding from the ice of the My´rdalsjo¨kull glacier and
180
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Fig. 3. Time-history of earthquake activity in the Goðabunga area (63.58–63.738N, 19.22–19.448W) in 1996–2003, from the Icelandic Meteorological Office database. Events of magnitudes z 1 are plotted against the date. A seasonal pattern was obvious until the autumn of 2001, since then earthquakes have occurred on a daily basis. However, also since 2001 more events have occurred during the fall than during the summer. The apparent bipolar distribution of event magnitudes in the plot, around magnitudes of about 1.5 and 2, is an artefact caused by the difficulty of defining the magnitudes from the unusual looking signals. However, this figure qualitatively shows the real development of seismicity.
at edges of the glacier (Figs. 1 and 2). The gathered data were combined with the recordings of the permanent SIL seismograph network (named after South Iceland Lowland) run by the Icelandic Meteorological Office (Bo¨ðvarsson et al., 1999) to provide a high-quality data
set of Katla earthquakes. The events were re-located with the program HYPOINVERSE (Klein, 1978) using a crustal model consisting of layers with constant velocity gradient and refined with local station corrections for the SIL stations (Soosalu and Einarsson, 1997). In this study we focus on the earthquake cluster at the Goðabunga rise. One of the stations, GODA, was located right on top of this activity and we thus had good depth control for these events. For geological interpretation of the seismicity in the Goðabunga area we accepted only events having strict location criteria. Each event was required to have root mean square travel time residual (rms) V 0.2 s, horizontal standard error (erh) V 0.6 km, vertical standard error (erz) V 0.9 km, and largest gap between observing stations V 1678. Generally these values are far smaller, in average rms = 0.14 s, erh = 0.32 km, erz = 0.37 km and gap = 938. We also required that the closest station GODA had recorded the event, that there were at least seven observing stations, and twelve picked P- or Sphases. During a hundred-day interval from March 12 to June 20 we gathered 231 such earthquakes, having local magnitudes of M L 0.3–2.4. The local network, with one station situated at the Goðabunga rise, revealed interesting features of the activity in the area. The records of the station GODA show that the seismicity was incessant during the study period. Small earthquakes, typically occurring in swarms, and often too small to be observed by more distant stations, were recorded on hourly basis. Previously, all seismic observations have shown low-frequency volcanic appearance for the Goðabunga signals. However, the records of the closest station GODA are more tectonic-like, show higher frequencies and have clear P- and S-wave arrivals. Path effects cause the low-frequency appearance in the records at the stations further away, and they vary in regard with the direction. The stations ENTA and AUST were located on nunataks near or at the caldera edge, at about 8–11 and 12–15 km from the Goðabunga area, respectively. Their records show low-frequency signals with emergent P- and unclear or absent S-phases for the same earthquakes showing high frequencies at GODA. The station SOHH was located at the south edge of the My´rdalsjo¨kull glacier, at a distance of 12–15 km from the Goðabunga rise. Typically SOHH records exhibit more high-frequency content than those of ENTA and AUST, and mostly impulsive P- and S-arrivals. An example set of records of a Goðabunga event is shown in Fig. 4. The Goðabunga events often occur not only in swarms but as QmultipleQ events. Typically there are
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
181
Fig. 4. Example seismograms from the four Katla net stations. This event was located at a depth of 0.9 km (shown with a black dot in Figs. 5 and 6) and had a M L magnitude of 1.2. The stations are shown in an increasing order of distance from the source. The amplitudes are not to scale between different stations. Z is the vertical component, N horizontal north and E horizontal east. The picked P and S arrivals are shown.
1–3 small events very close in time preceding a larger (M L~2) earthquake. Event records frequently overlap and reading onsets may be difficult. In some cases the GODA records start with a high-frequency onset, which is followed by low-frequency coda with considerably larger amplitude, apparently consisting of surface waves. Their appearance resembles hybrid events discussed by Chouet (1996).
Fig. 5. Horizontal slices of the Goðabunga earthquake cluster, only well-located (see criteria in the text) events are plotted. (A) All events. (B–D) 0.8 km thick slices from the surface down to a depth of 2.4 km (E) the deepest events at 2.4–4 km depth. The M L magnitude scale is in the inset. Error bars of one standard error are shown in B–E.
182
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
This study focuses on the cluster of the Goðabunga events (see the small box in Figs. 1 and 2). We analysed their spatial and temporal distribution during the 100day interval data were collected and suggest a geological interpretation. Horizontal and vertical cross-sections of the earthquake cluster are shown in Figs. 5 and 6. This bseismicity imagingQ of the Goðabunga volume reveals a well-delineated hypocentral distribution. Most events occurred in the uppermost 2 km; no events were observed at depths greater than 4 km. A view from the south (Fig. 6) shows that the cluster is inclined: the shallower events occurred in the west part and the deeper events in the east. The inclination of the cluster is towards the Katla caldera (its rim is located about 1 km east of the area shown in Fig. 5). The shallower events tend to be smaller in size than the deeper events. A small aseismic volume is found at depth of about 2 km depth, with horizontal dimensions of the order of 1 km, outlined by a bcapQ of fairly large earthquakes (M L V 2). During the 100-day operation period of the local net no spatial migration of seismicity was evident. Events occurred quite evenly throughout the whole volume of active seismicity at this time. 4. Geological interpretation and discussion
Fig. 6. Vertical cross-section of the Goðabunga earthquake cluster, seen from the south. Both the horizontal and the vertical scale are the same. (A) All events (B–E) 0.68-km-thick slices, progressively from south to north. The latitude limits of the slice are given in each plot. The M L magnitude scale is in the inset. Error bars of one standard error are shown in B–E.
Our observations of the Goðabunga seismicity in the spring and early summer 2003 suggest an inclined, stock-like body extending from the surface, downwards and eastwards to a depth of about 4 km. The largest earthquakes (M L~2) are clustered at about 1.5 km depth. Beneath this cluster there are few events and none deeper than at 4 km. The cluster sketches a convex lens-like object beneath which there is an aseismic area with the diameter of 1 km. We interpret the tightly-clustered shallow seismicity to reflect an intruding hot cryptodome, by definition a dome-shaped structure created by accumulation and ascent of viscous magma at shallow depth. It increases stress in the bedrock above and around its head, and this is released in the form of numerous earthquakes. The concentration of larger events at about 1.5 km depth is taken as a sign of intense deformation, and it outlines the head of the cryptodome. The cryptodome itself, however, is in a ductile state and thus aseismic. It is not possible to sketch the lower structure of this feature on the basis of seismic observations. The 1-km diameter is a rough estimate for the size of the head of the cryptodome and remains tentative with our event location accuracy. Its dimensions could be refined by calculating relative locations for the Goðabunga events. Extruded domes have diameters from a
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
few metres to several kilometres (Fink and Anderson, 2000). Our size estimate for the Goðabunga cryptodome fits within these dimensions. It is possible that the cryptodome itself is smaller in size. It may heat the rock in the immediate surroundings so that it too is in a ductile state. At least for the last three decades, i.e., the period of accurate epicentral determination, the seismicity distribution at Katla has shown a remarkably persistent pattern. Epicentres have been clustered in two areas, at Goðabunga rise by the caldera rim and within the Katla caldera (Einarsson and Brandsdo´ttir, 2000). The tight clustering as well as the eccentric position of the Goðabunga cluster have been the subject of speculation for some time. According to Fink and Anderson (2000) domes may have steady or episodic growth, with emplacement times ranging from a few hours to many decades. Our suggestion of a cryptodome implies that the Goðabunga dome has existed for a long time, at least some decades, being the ultimate cause of the seismicity in the area. Until about 1999 the stress loading due to the dome ascent was moderate and comparable to the rate of the seasonal triggering effect. The triggering effect was thus sufficient to limit the seismic activity to the summer and autumn months and turn the earthquakes off during the winter months. The dome ascent rate appears to have increased around 1999, contemporaneously with the beginning of inflation in the caldera (Sturkell et al., 2003a,b), and in the years 2001–2004 the seismicity at Goðabunga has been continuous. The seasonal triggering is active throughout this period also, however. The seismicity is clearly lower during the winter months although it does not turn off. The difference in the appearance of the seismograms for a single event at different stations is important for determining the nature of the Goðabunga earthquakes, and it should be studied further. Before installing the station GODA very close to the origins of these events it was not known if they are intrinsically low-frequency events or not. At an active volcano it is possible that nearby magma volumes can affect the seismic signal and cause misleading interpretations of the earthquake source. A similar observation was noted at the highly active Colima volcano in Mexico. Nu´n˜ez-Cornu´ et al. (1994) observed seismicity related to the 1991 eruption of Colima and state that all the seismicity should be classified as B-type events according to Minakami (1974), that is earthquakes with small magnitudes, unclear S-waves and low frequencies. However, they note that the seismometer closest to the volcano’s summit frequently observed the same events with Minakami’s
183
A-type features, i.e., the seismograms look similar to those of shallow tectonic events. Since the unrest began in 1999, the Katla area has been monitored closely. Observations of increased seismicity, inflation of the volcano and increased geothermal activity clearly show that eruptive activity may occur at any time. Initially the most likely scenario was expected to be a btraditionalQ Katla eruption, i.e., another basaltic eruption within the Katla caldera. Indeed, a minor basaltic eruption may already have occurred on July 18, 1999, when a glacial flood was observed and a depression formed on the ice surface in the south part of the caldera. However, the increased Goðabunga seismicity, tightly clustered outside the caldera, indicates that a wider range of scenarios should be considered. The feature that we interpret as a cryptodome may make its way to the surface and lead to volcanic phenomena associated with a silicic, domeforming eruption. The Goðabunga area has experienced abnormally high seismicity since the autumn of 2001. Apparently around this time, and possibly as early as 1999, our hypothesized cryptodome became more active and its slow propagation towards the surface accelerated. It is possible that the unrest beginning in the summer of 1999 spurred the propagation of the cryptodome, and in the autumn of 2001 it reached the seismogenic crust, which initiated continuous earthquake activity. The inclination of the seismically active volume towards the caldera speaks for a connection in that direction. Larsen (2000) states that silicic eruptions have only recently been recognized as a distinct phase in the Holocene activity of the Katla volcanic system. Although such events are infrequent and have not occurred during Historic times, they must be reckoned with in the future. Silicic Katla eruptions are not known to have occurred since 1700 yr BP. It has been suggested that the major effusive basaltic Eldgja´ eruption (ca. AD 934) changed the magma plumbing system of Katla and broke the formerly regular pattern of silicic eruptions (Larsen, 2000). However, a new pattern of silicic Katla eruptions may emerge. There are several examples of seismicity related to doming activity at volcanoes elsewhere in the world that can be compared with the observed Goðabunga activity. The Usu volcano in Japan, for instance, is a classical example of a volcano with dacitic dome growth. Okada et al. (1981) studied the seismicity related to the magma intrusion process of Usu in the period of 1978–1980. They detected a circular distribution of numerous shallow (b2 km) earthquake hypocentres beneath the volcanic edifice. Within this
184
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
activity there was an aseismic zone, which they interpreted as an intruding dome. The seismicity preceding and related to the 1990–1995 eruption of Unzen volcano in Japan also shows interesting patterns of migrating seismicity, starting about one year before the onset of the eruption, and continuing during the dome growth activity (Nakada et al., 1999; Umakoshi et al., 2001). Another current example of seismicity preceding the onset of a silicic eruption is the Soufrie`re Hills volcano in Montserrat, West Indies, where a domebuilding eruption started in 1995 and is still ongoing (Aspinall et al., 1998). Milia et al. (2000) made seismic studies at Campi Flegrei in Italy and found an offshore seismic unit, which they interpret as a cryptodome. They remark that the location of this cryptodome corresponds to an area of shallow (b 0.6 km) seismicity during a seismic crisis in 1970–1972. Dome-forming activity has occurred earlier in the Katla system, as silicic domes are found as nunataks around its caldera, such as Enta and Austmannsbunga (AUST), the sites of our seismic stations (Sturkell et al., 2003a). Some of them are thought to be of late-glacial age according to their chemical characteristics (Larsen, 2000). A comparable Icelandic example of a central volcano with acidic dome forming and bipolar geochemistry of basalts and rhyolites is Krafla in the Northern volcanic zone. Jo´nasson (1994) has studied the spatial distribution and geochemistry of rhyolites at Krafla. The bulk of the Krafla products are basaltic, but rhyolite domes are found at the caldera rim of Krafla or around it. Jo´nasson suggests that the rhyolites are formed by partial melting of crustal, hydrothermally altered basaltic rock. The rhyolites are generated on the peripheries of an active basaltic magma chamber of intrusive domain, where the temperature range is favourable for the genesis of rhyolite, i.e., not too hot, as would be expected above the chamber. Thus the result is a bhaloQ of rhyolites around the volcano. A similar pattern may well be valid at Katla. Larsen (2000) deduces that at least some rhyolitic eruptions have an origin inside the Katla caldera, according to the tephra layer observations. This conclusion may be based on insufficient data, and remains speculative (Guðru´n Larsen, pers. comm., 2004). Because of the glacial cover the caldera bottom geology of Katla cannot be directly observed. Acidic dome, or cryptodome growth is a feature that characteristically has a very localized stress regime. It may cause only little deformation in the surrounding bedrock, and the deformation signal decreases quickly with distance. To be able to detect Goðabunga deformation geodetically, measurements within a short dis-
tance are needed. GPS measurements at Katla and in the surroundings have been conducted since 1986 (Sturkell et al., 2003b) and a GPS point at Goðabunga (located at the same site as the temporary seismic station GODA) has recently been installed. Measurements in January and June 2004 reveal 2.6 cm uplift of the site during that period, pointing to very localised stress regime, distinctive for a protruding dome (Einarsson et al., 2005; Erik Sturkell, pers. comm., 2005). GPS points close to or at the sites AUST, ENTA and SOHH have been measured for a longer period, and they show uplift in recent years, with a best-fitting point source in the north part of the Katla caldera at 4–5 km depth, interpreted as the result of magma accumulation (Sturkell et al., 2004; Einarsson et al., 2005). Signals resulting from possible deformation in the Goðabunga area are not detected at these observation sites. The geothermal activity in the Katla area has been observed to have increased since the start of the unrest in 1999 (Guðmundsson et al., 2000). However, no distinct increase has been observed precisely at Goðabunga yet. It is to be expected that geothermal activity will intensify in this area in the future if the cryptodome continues propagating towards the surface. If the cryptodome is still moving upwards it may be possible to detect it in the development of the seismic activity, even though precise hypocentral locations for the whole period are not available. When a cryptodome propagates upwards, the brittle crust above it thins with time and the deformation field becomes more localised. This leads to a smaller volume in which cryptodomeinduced earthquakes occur, with smaller and fewer events. Our 100-day period of well-constrained hypocentral locations of Goðabunga events provides only a snapshot of the cryptodome-related seismicity. We do not have as good accuracy for event locations during the whole period since the fall of 2001, and thus cannot follow their temporal development in detail. Seismicity in the Goðabunga area from the autumn of 2001 until end of 2004 has been incessant. A decrease in the rate of earthquake activity, or decrease in the sizes of the events, may be either a good or a bad sign. It is possible that the cryptodome may freeze and remain at depth, and thus the seismicity related to it would cease. But if the cryptodome continues its propagation towards the surface, it may be expressed by a similar pattern of decreasing seismicity, when the brittle, seismogenic volume above it gets smaller and smaller. In such a case an eruption may be imminent. Other indications, such as ground uplift and increased geothermal activity, will then be expected to provide additional warning.
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Katla is generally considered the most dangerous of the Icelandic volcanoes. Until quite recently the concerns of the civil protection agencies have been directed towards the eastern side of the volcano where almost all catastrophic floods in Historic times have occurred. The Goðabunga activity in recent years and attempts to explain it have demonstrated that a wider spectrum of scenarios has to be considered. It appears conceivable that the next large eruption of Katla will be different from eruptions of the last centuries. In particular, the possible consequences of a silicic eruption in the western part of the volcano have to be looked into. The recent activity has led to re-evaluation of hazards due to volcanic activity in western part of Katla and its neighbouring volcano Eyjafjallajo¨kull (Guðmundsson and Gylfason, 2005). A Katla eruption becomes more likely as time passes. Possible scenarios include the following (Einarsson et al., 2005): 1. A basaltic, phreato-magmatic eruption within the caldera, following the typical pattern of Katla’s historic eruptive activity. 2. If the hypothesized cryptodome reaches the surface, a silicic, explosive dome-forming eruption at Goðabunga. The hazard from such an event in its full extent may be difficult to estimate, as there is no historical experience of such an event. It is also important to bear in mind that Goðabunga is located close to popular tourist and hiking areas, thus posing a great risk to people, especially if an eruption starts during the summer season. 3. A major debris avalanche triggered by an intruding dome at Goðabunga, in a same manner as the case of Mt. St. Helens eruption in 1980 (e.g., Fink and Anderson, 2000). 4. A basaltic eruption within the Katla caldera, or a basaltic intrusion, which may induce a silicic eruption in the Goðabunga area. The eruptive activity may even be more widespread, with an accompanying eruption of the neighbouring Eyjafjallajo¨kull volcano to the west (Fig. 1). Eyjafjallajo¨kull showed signs of unrest in 1994 and 1999 (Sturkell et al., 2003b), with magma intrusions beneath its southern flank (Pedersen and Sigmundsson, 2004; 2005). Both its documented eruptions in Historic times, in 1612 and 1821, were associated with Katla eruptions. 5. Conclusions Due to the improved accuracy of locations of Goðabunga earthquakes during the operation of our field instruments in the spring and summer of 2003 we are
185
able to conclude that the earthquake cluster at Goðabunga is even denser than has been suggested before. This concentrated source of seismicity has been persistently active for at least four decades and is associated with very local deformation. We offer an explanation for this activity and suggest it is an expression of a hot, acidic cryptodome that has been intruding towards the surface during the last decades. At present its top appears to be at the depth of 1.5 km. Its rate of ascent has increased during the last years. The interpretation is supported by geodetic measurements near the centre of the cluster, and the existence of older acidic domes around the caldera rim of Katla. Acknowledgements We are grateful to Bob White and the SEIS-UK instrument pool for kindly lending the portable seismometers, which made the whole study possible. Icelandic Meteorological Office gathered the additional data. Erik Sturkell suggested the logistically great idea to unite the forces in a joint field campaign with his GPS measurements. Heartfelt thanks to the brave field ´ lafsson, Gre´tar people Haukur Brynjo´lfsson, Halldo´r O Einarsson, Jo´nas Erlendsson, Halldo´r Geirsson, Karl Gro¨nvold, Dorthe Holm, Francien Peterse, Rune Selbekk, the Rescue Team in Vı´k, and Benedikt Bragason with his men from the Arcanum Adventure Tours for the competent snow scooter guiding. The late Katrı´n Brynjo´lfsdo´ttir and Guðgeir Guðmundsson from the village Vı´k at the roots of Katla are warmly remembered for homelike accommodation for the team during the field trips. Guðru´n Larsen provided an interesting discussion upon the silicic Katla volcanism. Bjo¨rn Lund made good comments on the manuscript. The public domain GMT software (Wessel and Smith, 1998) was used to draw the figures. Constructive criticism by Silvio De Angelis, Alejandro Nava and the editor Peggy Hellweg improved the manuscript. References Aspinall, W.P., Miller, A.D., Lynch, l.L., Latchman, J.L., Stewart, R.C., White, R.A., Power, J.A., 1998. Soufrie`re Hills eruption, Montserrat, 1995–1997: volcanic earthquake locations and fault plane solutions. Geophys. Res. Lett. 25, 3397 – 3400. Bjo¨rnsson, H., Pa´lsson, F., Guðmundsson, M.T., 2000. Surface and bedrock topography of the My´rdalsjo¨kull ice cap, Iceland: the Katla caldera, eruption sites and routes of jo¨kulhlaups. Jo¨kull 49, 29 – 46. Bo¨ðvarsson, R., Ro¨gnvaldsson, S.Th., Slunga, R., Kjartansson, E., 1999. The SIL data acquisition system—at present and beyond year 2000. Phys. Earth Planet. Inter. 113, 89 – 101.
186
H. Soosalu et al. / Journal of Volcanology and Geothermal Research 153 (2006) 177–186
Chouet, B., 1996. Long-period volcano seismicity: its source and use in eruption forecasting. Nature 380, 309 – 316. Einarsson, P., 2000. Course of events connected with the flood in Jo¨kulsa´ a´ So´lheimasandi in July 1999. Geoscience Society of Iceland, February Meeting, 2000, p. 14 (in Icelandic). Einarsson, P., Sæmundsson, K., 1987. Earthquake epicenters 1982– 1985 and volcanic systems in Iceland (map). In: Sigfu´sson, Þ.I. (Ed.), I´ hlutarins eðli, Festschrift for Þorbjo¨rn Sigurgeirsson. Menningarsjo´ður, Reykjavı´k. Einarsson, P., Brandsdo´ttir, B., 2000. Earthquakes in the My´rdalsjo¨kull area, Iceland, 1978–1985: seasonal correlation and connection with volcanoes. Jo¨kull 49, 49 – 73. Einarsson, P., Soosalu, H., Sturkell, E., Sigmundsson, F., Geirsson, H., 2005. The activity at the central volcano Katla and its vicinity since 1999 and possible scenarios. In: Guðmundsson, M.T., Gyl´ .G. (Eds.), Evaluation of Risk Due to Volcanic Eruption fason, A in Western My´rdalsjo¨kull and Eyjafjallajo¨kull. University Press, Reykjavı´k, Iceland, pp. 151 – 158 (in Icelandic). Fink, J.F., Anderson, S.W., 2000. Lava domes and coulees. In: Sigurðsson, H. (Ed.), Encyclopedia of Volcanoes. Academic Press, pp. 307 – 319. ´ .G. (Eds.), 2005. Evaluation of Guðmundsson, M.T., Gylfason, A Risk Due to Volcanic Eruption in Western My´rdalsjo¨kull and Eyjafjallajo¨kull. University Press, Reykjavı´k, Iceland, 210 pp. (in Icelandic). ´ ., Brandsdo´ttir, B., Menke, W., Sigvaldason, G.E., Guðmundsson, O 1994. The crustal magma chamber of the Katla volcano in South Iceland revealed by 2-D seismic undershooting. Geophys. J. Int. 119, 277 – 296. Guðmundsson, M.T., Ho¨gnado´ttir, Þ., Bjo¨rnsson, H., Pa´lsson, F., Guðmundsson, G., 2000. Geothermal activity at My´rdsalsjo¨kull and events in the summer 1999. Geoscience Society of Iceland, February Meeting, 2000, p. 13 (in Icelandic). Jakobsson, S.P., 1979. Petrology of recent basalts of the Eastern Volcanic Zone, Iceland. Acta Nat. Isl. 26 103 pp. Jo´hannesson, H., Jakobsson, S.P., Sæmundsson, K., 1990. Geological map of Iceland, sheet 6, South-Iceland, third edition. Icelandic Museum of Natural History and Iceland Geodetic Survey, Reykjavı´k. Jo´nasson, K., 1994. Rhyolite volcanism in the Krafla central volcano, North-East Iceland. Bull. Volcanol. 56, 516 – 528. Klein, F.W., 1978. Hypocenter location program HYPOINVERSE. US Geol. Surv. Open-file Rep., vol. 78-694, 113 pp. Larsen, G., 2000. Holocene eruptions within the Katla volcanic system, South Iceland: characteristics and environmental impact. Jo¨kull 49, 1 – 28. Larsen, G., Newton, A.J., Dugmore, A.J., Vilmundardo´ttir, E.G., 2001. Geochemistry, dispersal, volumes and chronology of Holocene silicic tephra layers from the Katla volcanic system, Iceland. J. Quat. Sci. 16, 119 – 132.
Milia, A., Torrente, M.M., Giordano, F., 2000. Active deformation and volcanism offshore Campi Flegrei, Italy: new data from highresolution seismic reflection profiles. Mar. Geol. 171, 61 – 73. Minakami, T., 1974. Seismology of volcanoes in Japan. In: Civetta, L., Gasparini, P., Luongo, G., Rapolla, A. (Eds.), Physical Volcanology, Developments in Solid Earth geophysics, vol. 6. Elsevier Scientific Publishing, Amsterdam, pp. 1 – 27. Nakada, S., Shimizu, H., Ohta, K., 1999. Overview of the 1990–1995 eruption at Unzen volcano. J. Volcanol. Geotherm. Res. 89, 1 – 22. Nu´n˜ez-Cornu´, F., Nava, F.A., De la Cruz-Reyna, S., Jime´nez, Z., Valencia, C., Garcı´a-Arthur, R., 1994. Seismic activity related to the 1991 eruption of Colima volcano, Mexico. Bull. Volcanol. 56, 228 – 237. Okada, Hm., Watanabe, H., Yamashita, H., Yokoyama, I., 1981. Seismological significance of the 1977–1978 eruptions and the magma intrusion process of Usu volcano, Hokkaido. J. Volcanol. Geotherm. Res. 9, 311 – 334. Pedersen, R., Sigmundsson, F., 2004. InSAR based sill model links spatially offset areas of deformation and seismicity for the 1994 unrest episode at Eyjafjallajo¨kull volcano, Iceland. Geophys. Res. Lett. 31, L14610. doi:10.1029/2004GL020368. Pedersen, R., Sigmundsson, F., 2005. Temporal development of the 1999 intrusive episode in the Eyjafjallajo¨kull volcano, Iceland, derived from InSar images. Bull. Volcanol. doi:10.1007/s00445005-0020-y. Sigurðsson, O., Zo´phonı´asson, S., I´sleifsson, E., 2000. The jo¨kulhlaup from So´lheimajo¨kull, July 18th 1999 (in Icelandic with English summary). Jo¨kull 49, 75 – 80. Soosalu, H., Einarsson, P., 1997. Seismicity around the Hekla and Torfajo¨kull volcanoes, Iceland, during a volcanically quiet period, 1991–1995. Bull. Volcanol. 59, 36 – 48. ´ lafsson, Sturkell, E., Einarsson, P., Sigmundsson, F., Geirsson, H., O ´ lafsdo´ttir, R., Guðmundsson, G.B., 2003a. The pressure H., O increases under Katla. Na´ttu´rufræðingurinn 71, 80 – 86 (in Icelandic with English summary). Sturkell, E., Sigmundsson, F., Einarsson, P., 2003b. Recent unrest and magma movements at Eyjafjallajo¨kull and Katla volcanoes, Iceland. J. Geophys. Res. 108, 2369. doi:10.1029/2001JB000917. ´ rnado´ttir, Þ., Pedersen, Sturkell, E., Geirsson, H., Sigmundsson, F., A R., Pagli, C., Einarsson, P., 2004. Crustal deformation and volcano dynamics in Iceland. The 26th Nordic Geological Winter Meeting, vol. 126. GFF, Usala, p. 51. Umakoshi, K., Shimizu, H., Matsuwo, N., 2001. Volcano-tectonic seismicity at Unzen volcano, Japan, 1985–1999. J. Volcanol. Geotherm. Res. 112, 117 – 131. ´ rbo´k Þo´rarinsson, S., 1975. Katla and annal of Katla eruptions. A Ferðafe´lags I´slands 1975 (The Iceland Touring Association Yearbook 1975), pp. 125 – 149 (in Icelandic). Wessel, P., Smith, W.H.F., 1998. New, improved version of Generic Mapping Tools released. Eos 79, 579.