Eruption Site Research Articles

Direct observation of a submarine volcanic eruption from a sea-floor instrument caught in a lava flow.

Our understanding of submarine volcanic eruptions has improved substantially in the past decade owing to the recent ability to remotely detect such events and to then respond rapidly with synoptic surveys and sampling at the eruption site. But these data are necessarily limited to observations after the event. In contrast, the 1998 eruption of Axial volcano on the Juan de Fuca ridge was monitored by in situ sea-floor instruments. One of these instruments, which measured bottom pressure as a proxy for vertical deformation of the sea floor, was overrun and entrapped by the 1998 lava flow. The instrument survived-being insulated from the molten lava by the solidified crust-and was later recovered. The data serendipitously recorded by this instrument reveal the duration, character and effusion rate of a sheet flow eruption on a mid-ocean ridge, and document over three metres of lava-flow inflation and subsequent drain-back. After the brief two-hour eruption, the instrument also measured gradual subsidence of 1.4 metres over the next several days, reflecting deflation of the entire volcano summit as magma moved into the adjacent rift zone. These findings are consistent with our understanding of submarine lava effusion, as previously inferred from seafloor observations, terrestrial analogues, and laboratory simulations.

Interpretation of magnetic anomalies over the Waimangu geothermal area, Taupo volcanic zone, New Zealand

The study area is located on the eastern side of the Taupo volcanic zone in central North Island of New Zealand. It lies a few kilometres to the southwest of Mt Tarawera, the site of the biggest New Zealand volcanic eruption in historical times (the June 1886 Tarawera eruption). The study area includes the Waimangu geothermal field and a small part of Waiotapu and Waikite fields. The extensive surface thermal expressions (boiling springs, hot lakes, craters, and sinter terraces) occurring at the Waimangu field were all formed following the 1886 Tarawera eruption. Another large area of less intense thermal manifestations (thermal ground and hydrothermally altered rocks) exists about 5 km southwest of Waimangu, extending towards the Waiotapu field in the south. In 1993 an aeromagnetic survey was conducted over the study area at an average altitude of about 350 m above the ground. The results show a subdued negative residual anomaly (about −100 nT) over the Waimangu field, which can be interpreted by near-surface hydrothermal demagnetisation of rhyolitic host rocks. The lateral distribution of the demagnetised rocks is much greater than the thermal area of Waimangu, and is consistent with the extent of low resistivity rocks across the study area. The magnetic interpretation also shows that two-high standing dacite domes situated about 5 and 7 km to the southwest of Waimangu have been affected by hydrothermal demagnetisation. There are negative residual anomalies outside the low resistivity zone that could be associated with reversely magnetised rocks (age >0.78 Ma). A strong positive residual anomaly (up to 450 nT) occurs to the east of the Waimangu field. Results from 3-D magnetic interpretation indicate some alternative models for this positive anomaly: (1) southwest–northeast trending, vertical basalt dykes (magnetisation 10 A/m), tops between −0.1 and −0.65 km RL (reduced level=relative to sea level), (2) a thick (⩾ 1 km) sequence of rhyolites (magnetisation 2.5 A/m) extending from the surface down to about −0.8 km RL, and (3) a rather thin (0.35 km) sequence of rhyolites (from surface to sea level) underlain by basalt bodies similar to those of model (1).

Comparison of Mount Etna, Kilauea, and Piton de la Fournaise by a quantitative modeling of their eruption histories

From published data we found characteristic relations between the amount V of erupted lava and the duration d of eruption for Mount Etna, Kilauea, and Piton de la Fournaise. The relation is similar between Mount Etna and Kilauea, where the increase of V with increasing d is slow, showing a trend of a lower flow rate for a larger eruption. For Piton de la Fournaise, however, the trend is distinctly different, showing a higher flow rate for a larger eruption. We constructed quantitative models of a magma system with reservoirs at various levels and tested hypotheses about the existence of large reservoirs under these volcanoes using the observed V‐d relations. We found that the observed V‐d relation is consistent with the presence of a large reservoir at a shallow depth under Kilauea, and with the presence of a large reservoir near the bottom of the volcanic edifice under Piton de la Fournaise. The above models for Kilauea and Piton de la Fournaise are characterized by a simple path from the mantle reservoir which leads to the shallowest reservoir connected to a hierarchy of channels with varying resistance to eruption sites. We obtained less satisfactory agreement between the observed V‐d relation and the predicted using our model for Mount Etna. Time histories of pressures in the reservoirs at various levels obtained by the modeling explain why inflation‐deflation cycles observed at Kilauea have not been reported for Piton de la Fournaise. The absence of volcano‐tectonic earthquakes with magnitude greater than ∼2.5 under Piton de la Fournaise is attributed to the simplicity of the magma path from the mantle to the shallowest reservoir and the underdeveloped rift zone, which result in a stress concentration localized only beneath the summit area.

Glacier–volcano interactions deduced by SAR interferometry

AbstractGlacier-surface displacements produced by geothermal and volcanic activity beneath Vatnajökull ice cap in Iceland are described by field surveys of the surface topography combined with interferograms acquired from repeat-pass synthetic aperture radar images. A simple ice-flow model serves well to confirm the basic interpretation of the observations. The observations cover the period October 1996–January 1999 and comprise: (a) the ice-flow field during the infilling of the depressions created by the subglacial Gjálp eruption of October 1996, (b) the extent and displacement of the floating ice cover of the subglacier lakes of Grímsvötn and the Skaftá cauldrons, (c) surface displacements above the subglacier pathways of the jökulhlaups from the Gjálp eruption site and the Grímsvötn lake, (d) detection of areas of increased basal sliding due to lubrication by water, and (e) detection of spots of temporal displacement that may be related to altering subglacial volcanic activity. At the depression created by the Gjálp eruption, the maximum surface displacement rate away from the radar decreased from 27 cm d−1to 2 cm d−1over the period January 1997–January 1999. The observed vertical displacement of the ice cover of Grímsvötn changed from an uplift rate of 50 cm d−1to sinking of 48 cm d−1, and for Skaftá cauldrons from 2 cm d to 25 cm d−1.

Glaciological and chemical studies on ice cores from Hans Tausen Iskappe, Greenland

The paper presents studies of various chemical and isotopical parameters from ice cores drilled in the northernmost located ice cap, Hans Tausen Iskappe, Pearyland, Greenland (HT). The 346 m main ice core (MC95) was drilled to bedrock in 1995 as well as a 35 m shallow core (SC95). A 60 m shallow core (SC75) and a 51 m shallow core (SC76) was drilled at two different positions in1975 and 1976, respectively. A 60 m shallow core (SC94) was drilled in 1994. Continuous stable isotope records exist for all of these cores, total β-activity only from SC75 and SC76. Continuous ECM inferred acidity records exist along the 1995 cores (MC95 and SC95) and finally detailed records of dust and water soluble ion concentrations exist on selected parts of MC95.
 To determine a time scale for the ice core is an important prerequisite for the interpretation of other records. The age scale is based on acid layers, caused by known volcanic eruptions, and by comparison of the chemical composition of these layers to that found in ice cores from other arctic locations. The total β-activity data from SC75 and SC76 provide fixed points to the time scale because a pronounced increase in total β-activity is related to the thermo-nuclear tests in the atmosphere in the early 1960'ies. Many of the investigated parameters exhibit seasonal variations e.g. δ18O, acidity, Cl– and dust, therefore the study of the accumulation history of the ice cap improves our knowledge of the question if the mass balance of the Hans Tausen Iskappe is in equilibrium or not.
 In the upper half part of the ice core the MC95 ECM record reveals several events of high acidity which can be connected to volcanic events known from other Greenland ice core records. Among the identified volcanic events are AD 1912 (Katmai, Alaska), AD 1815 (Tambora, Indonesia), AD 1783 (Laki, Iceland), AD 934 (Eldgjà, Iceland) and e.g. the signals of AD 1259 and 49 BC. The two latter signals originate from major volcanic eruptions of unknown eruption sites with a probable location close to the Equator in the case of the AD 1259 event. Some of the volcanic events are selected for an analysis of dust and water soluble chemical components, including F–, CH3SO2–, Cl–, NO3–, SO42–, Na+,NH4+, K+, Mg2+ and Ca2+. Coulter counter technique was used for the dust measurements and the chemical analysis were carried out by ion chromatography.

Surface and bedrock topography of the Mýrdalsjökull ice cap, Iceland: The Katla caldera, eruption sites and routes of jökulhlaups

Surface and bedrock topography of the Mýrdalsjökull ice cap, Iceland: The Katla caldera, eruption sites and routes of jökulhlaups

Seismic phenomena associated with the 1996 Vatnajökull eruption, central Iceland

During late September 1996, a major eruption took place at the NW part of the Vatnajökull glacier in central Iceland. The eruption was preceded by intense seismic activity, which began with a Mw=5.6 earthquake two days previously. Two very active volcanic systems, Bárdarbunga and Grimsvötn, are situated in that area underneath the permanent ice cap. The volcanoseismic phenomena associated with the eruption were recorded on both temporary (HOTSPOT) and permanent (SIL) seismic networks, covering most parts of the country. The recorded events were categorised, according to their waveform shape and frequency content, into three groups: (1) low-frequency events; (2) mixed-frequency events; and (3) volcanic tremor. The large earthquake at Bárdarbunga volcano, which initiated the seismic activity before the eruption, was located inside the caldera and had the characteristics of a non-double couple event. The epicentres of the earthquake swarm that followed it initially delineated the caldera rim and then migrated towards Grimsvötn, possibly indicating lateral movements of magma from a shallow chamber beneath Bárdarbunga. The eruption affected an area much larger than that between these two volcanoes, since seismic activity was also observed at distances 20km away, at the Tindafjallajökull volcanic system. The spectral analysis of tremor, recorded at the nearest station to the eruption site, revealed its existence before the onset of the eruption in five narrow frequency bands (0.5–0.7, 1.6, 2.2, 2.8, 3.2Hz) representing fundamental frequencies with their half- and quarter-subharmonics. This pattern continued until the last day of the eruption. It is believed that the eruption was caused by a dyke injection that had been going on beneath the Vatnajökull area for a period of 10years.

The increasing exposure of cities to the effects of volcanic eruptions: a global survey

The most dynamic demographic process of the past 250 years has been the movement of people from rural areas to cities. For most of this period urbanisation has been concentrated in economically more developed parts of the world, but during the last 50 years the focus has shifted to economically less developed regions. Urbanisation, particularly in developing countries, has led to increasing global exposure to a variety of natural hazards, not the least of which are risks posed to large cities by volcanoes. In this paper we monitor these demographic changes and detail the various types of volcanic hazard to which cities are exposed. A major eruption affecting a city in a developing country could cause widespread loss of life and regional disruption. Effective response, however, might minimise casualties in a city within a developed nation affected by a major eruption, but the economic impact could have global consequences. We argue that global hazard exposure is often subtle and involves not only the size of a city and the types of volcanic product that may occur, but also the strategic position of the threatened city within the economy of a country and/or region and the fact that volcano-induced tsunami and other consequences of eruptions, such as climatic change, may affect cities far removed from a given eruption site. Mitigation measures informed by both specific prediction (surveillance) and general prediction (hazard mapping) are providing the potential to reduce hazard exposure. The paper concludes with a consideration of ongoing research, in particular the emphasis currently being placed on conflating hazard analysis with studies of place, economy, society and culture.

GOES satellite and field observations of the 1998 eruption of Volcan Cerro Azul, Galápagos Islands

The 1998 eruption of Volcan Cerro Azul, Isla Isabela, Galapagos Islands, was observed in near real-time by the Geostationary Operational Environmental Satellite-8 (GOES-8) weather satellite. Due to the remote location of the eruption site, 3.9-μm radiance values derived from GOES band 2 provide the best timing of the start and termination of the eruption, which occurred on 15 Sept. and 21 Oct., respectively. Throughout the 36-day long eruption, a total of 1335 thermal infrared images were collected, of which 851 were cloud-free and permitted the thermal anomaly to be detected. A detailed chronology including 77 separate events was assembled from the GOES data and field observations. Numerous attributes of the eruption were observed from the GOES data, including the sizes and dispersal of seven eruption plumes and the occurrence and timing of intra-caldera effusive activity. The growth of a lava flow on the SE flank, the formation of smoke and volcanic haze from the flank vent, and burning of vegetation caused by lava flows entering vegetated areas were monitored both on the ground and with the satellite data. In most cases GOES images were processed as they were received every 30 min and were then distributed over the Internet within minutes of reception. These data provided timely high-temporal information to field parties as well as enabled the documentation of the eruption. The GOES observations of Cerro Azul serve as a further example of the way in which the remote sensing community and field volcanologists can collaborate during future eruptions, and permit the temporal and spatial resolution requirements for future satellites systems to be better defined.

Seismic heterogeneity in the upper crust near the 1991 eruption site on the East Pacific Rise, 9°50′N

We report the results of a small‐aperture uppercrustal tomography experiment conducted on the East Pacific Rise near 9°50′N, the site of a volcanic eruption in 1991. The experiment geometry comprised 3 ocean bottom seismometers and 491 airgun shots fired in a 16 × 16 km² area. Two‐ and three‐dimensional inversions indicate that the average upper crustal velocities, both on and off axis, increase by ∼0.2 km/s north of 9°52′N. This location coincides with the northern limit of the 1991 eruption and of vigorous hydrothermalism. Our data can be explained by a 100 m southward increase in layer 2A thickness. On axis, previous work indicates that layer 2A thickness does not vary in this region. We infer that the sheeted dikes above the axial magma chamber are thermally segmented with the dike layer south of 9°52′N being on average at least 300°C warmer than to the north. Off axis, layer 2A may be thicker south of 9°52′N if a higher proportion of eruptions flow off axis.

Recent eruptions on the CoAxial segment of the Juan de Fuca Ridge: Implications for mid‐ocean ridge accretion processes

The 1993 seismic swarm and volcanic eruption on the CoAxial segment of the Juan de Fuca Ridge was the first verified mid‐ocean ridge accretion event monitored by the National Oceanic and Atmospheric Administration/U.S. Navy Sound Surveillance System (SOSUS). Ambiguity in the location of the initial seismicity resulted in an uncertainty as to whether the dike intruded from the summit and north rift zone of the adjacent Axial Volcano or whether it arose locally within the CoAxial segment. However, analyses of multibeam, side‐scan sonar, towed camera, submersible, and geochemical data show that the CoAxial segment is morphologically, structurally, and geochemically distinct from the north rift zone of Axial Volcano. There is no geologic or geochemical evidence that dike injections from Axial Volcano have extended north of 46°18'N in the past, whereas the 1993 eruption site is at 46°31'N. Furthermore, all seafloor manifestations of the 1993 dike injection lie along the central neovolcanic zone of the CoAxial segment. Analyses of repeat SeaBeam surveys combined with seafloor observations show that two other eruptions of approximately the same volume as the 1993 eruption occurred along the CoAxial segment in the 1981–1991 interval. These three diking events may have relieved decades of accumulated stress over ∼35 km or more of this segment. The spatial pattern and hydrothermal history of the 1993 event is consistent with a dike with a significant lateral component of injection.

Sedimentary record of explosive silicic volcanism in a Cretaceous deep‐marine conglomerate succession, northern Antarctic Peninsula

Lower Cretaceous conglomeratic strata exposed on southern Sobral Peninsula were deposited on a deep‐marine apron in the back‐arc Larsen Basin close to its faulted boundary with the Antarctic Peninsula magmatic arc. The succession is dominated by amalgamated beds of clast‐supported conglomerate, which, together with minor intercalated sandstones, consist of varied, but largely basaltic to andesitic, volcanic material and clasts derived from the Palaeozoic–Triassic (meta)sedimentary basement of the arc. Most of the volcanic clasts are thought to have been derived from lithified volcanic successions or older synvolcanic deposits, rather than from sites of coeval eruption. These mixed‐provenance strata enclose a number of intervals, consisting mainly of inverse–normally graded conglomerate and graded–stratified pebbly sandstone, in which the sand fraction is dominated by crystals and vitric grains considered to have been redeposited in the immediate aftermath of explosive silicic arc volcanism. Like syneruption deposits on non‐marine volcaniclastic aprons, these intervals are more sand‐prone than the enclosing strata and appear to show evidence of unusually rapid aggradation. Plagioclase from one such interval has yielded 40Ar/39Ar ages concordant at ≈121 Ma, similar to those obtained from the non‐marine Cerro Negro Formation, deposited within the magmatic arc. It is suggested that the two successions can be viewed as counterparts, both recording a history of mainly basaltic to andesitic volcanism, punctuated by relatively infrequent, explosive silicic eruptions. Whereas the Cerro Negro Formation consists mainly of syneruption deposits, most of the volcaniclastic material delivered to the eruption‐distal, deep‐marine apron appears to have been derived by normal degradation processes. Only rare silicic eruptions were capable of supplying pyroclastic material rapidly enough and in sufficient quantities to produce compositionally distinct syneruption intervals.

Comment on “Satellite radar images capture a subglacial volcanic eruption in Iceland”] Comment: Subglacial eruptions and synthetic aperture radar images

In a news piece published in the May 4, 1999, issue of Eos,Garvin et al. present RADARSAT satellite synthetic aperture radar (SAR) images of the subglacial volcanic eruption in the Gnmsvotn caldera within the Vatnajokull ice cap, Iceland, that occurred December 18–28, 1998. As pointed out in this piece, the images show considerable changes in surface backscatter of the radar signal within the caldera during and after the eruption. The authors suggest that the dark patches with an area of approximately 4 km2, observed within Grímsvötn, are open water. On the basis of the size of the dark patches, Garvin et al. estimate the volume of ice melted to have been about 1 km3 and the energy provided for melting, E = 3 × 1017 J, to have been about 30% of that of the 1996 Gjalp eruption, which occurred 12 km north of the 1998 eruption site [Gudmundsson et al., 1997]. Our groundbased and aerial observations at Gnmsvotn, however,show that this estimate of melting and energy release is far too high.

The 1996 and 1998 subglacial eruptions beneath the Vatnajökull ice sheet in Iceland: contrasting geochemical and geophysical inferences on magma migration

Sr/86Sr (0.70322) and δ18O ( ∼2.9‰), whereas significantly lower and higher values, respectively, are found in samples from the Bardarbunga volcanic system (0.70307 and 3.8‰). These results strongly indicate that the Gjalp magma originated from the Grimsvotn magma system. The 1996 magma is of an intermediate composition, representing a basaltic icelandite formed by 50% fractional crystallization of a tholeiite magma similar in composition to that expelled by the 1998 Grimsvotn eruption. The differentiation that produced the Gjalp magma may have taken place in a subsidiary magma chamber that last erupted in 1938 and would be located directly beneath the 1996 eruption site. This chamber was ruptured when a tectonic fracture propagated southward from Bardarbunga central volcano, as indicated by the seismicity that preceded the eruption. Our geochemical results are therefore not in agreement with lateral magma migration feeding the 1996 Gjalp eruption. Moreover, the results clearly demonstrate that isotope ratios are excellent tracers for deciphering pathways of magma migration and permit a clear delineation of the volcanic systems beneath Vatnajokull ice sheet.

Nasolacrimal duct obstruction secondary to ectopic teeth

ObjectiveTo describe two patients with nasolacrimal duct obstruction (NLDO) caused by ectopic eruption of teeth. The literature concerning nasal and other unusual ectopic sites of tooth eruption is reviewed. DesignTwo interventional case reports and literature review. ParticipantsA 3-year-old girl with epiphora and recurrent dacryocystitis of the right eye. Previous medical and surgical management was unsuccessful. A 32-year-old female with a long history of right eye discomfort and epiphora. Previous examinations and workup were negative. InterventionA computed tomographic (CT) scan of the orbits and sinuses was performed in both patients. The ectopic teeth were surgically removed. Main outcome measuresNasolacrimal system function and response to treatment at the last follow-up were recorded. ResultsIn the first patient, CT imaging disclosed two teeth within the right inferior meatus compressing the nasolacrimal duct. In the second patient, CT revealed a large dental structure in the maxillary sinus compressing the nasolacrimal duct. Endoscopic tooth extraction and nasolacrimal duct probing in the first patient and surgical removal of the dental structure in the second patient effected complete resolution of symptoms. Both patients were symptom free at last follow-up. ConclusionsThese cases suggest that ectopic eruption of teeth should be added to the differential diagnosis of NLDO. Surgical removal of the ectopic teeth compressing the nasolacrimal duct results in resolution of the lacrimal drainage obstruction.

Evidence for iron and sulfur enrichments in hydrothermal plumes at Axial Volcano following the January–February 1998 eruption

In response to 12 days of seismic activity at Axial Volcano, Juan de Fuca Ridge in January–February 1998, the NSF/NOAA‐sponsored Axial Response Team conducted three response cruises in February, August, and August/September to map and sample hydrothermal plumes over the region. Vertical profiles of particulate Fe and S over the eruption site show high concentrations from about 1400 m to the bottom. Chemical and scanning electron microscope analysis of the February plume samples revealed anhydrite, Fe‐ferrihydrites, elemental sulfur, and angular glassy basalt shards up to 190 µm in the longest dimension. Many of these shards had halite coatings, which is consistent with subseafloor basalt seawater reactions at temperatures >450°C at 1500 m depth. In August and September cruises the basalt shards were no longer present in the hydrothermal plumes. Instead, the plumes were predominantly composed of Fe‐ferrihydrites, elemental sulfur, and sulfur filaments. A unique feature of this data set is the high concentrations of elemental sulfur in the lower 60 m of the water column. The sulfur results are suggestive of a significant enrichment of the bacterial biomass in the water column over the eruption site with time. Within the Axial caldera, approximately 10–20% of the total sulfur is present as sulfur filaments. These post‐eruption particle compositional changes have strong similarities to the results from the 1993 CoAxial eruption on the Juan de Fuca Ridge.

In situ observations of the onset of hydrothermal discharge during the 1998 Submarine Eruption of Axial Volcano, Juan de Fuca Ridge

A volcanic eruption at the summit of Axial Volcano on January 25,1998, instantaneously created extensive and vigorous hydrothermal discharge. Moorings 2 km apart along the eruption fissure recorded temperature increases of ∼0.6°C up to 115 m above bottom within hours of initial seismic activity. Water temperatures at the mooring sites remained high for about 5 days, then declined steadily over the next 2 weeks. A response cruise 18 days after the eruption found hydrothermal temperature anomalies of ∼0.1°C over the eruption site, and a more intense and much thicker plume 20 km downstream of the eruption. We estimate the steady‐state heat flux required to produce this distal plume, evidence of discharge conditions perhaps 1–13 days after the eruption, as 60–230 GW. The Axial eruption thus produced the largest vent field heat flux yet measured, but these high levels lasted less than 3 weeks.

The January 1998 Earthquake swarm at Axial Volcano, Juan de Fuca Ridge: Hydroacoustic evidence of seafloor volcanic activity

On January 25, 1998 at 11:33Z, an intense swarm of earthquakes began within the summit caldera of Axial Volcano, central Juan de Fuca Ridge (JdFR). The earthquake swarm was detected using the U.S. Navy hydrophones in the northeast Pacific Ocean. The earthquake swarm lasted 11 days and produced 8247 detected (1037 located) earthquakes. During the first two days of the swarm, earthquake activity migrated from the summit caldera ∼50 km along the south rift zone of Axial Volcano at rates of 0.92±0.13 and 0.23±0.09 m s−1. Earthquake epicenter migration is characteristic of a lateral magma dike injection. The largest three swarm earthquakes occurred within the caldera. Their timing and mechanisms are consistent with adjustment of the caldera floor as magma is removed from beneath the summit. Earthquakes with short T‐wave rise times (shallow focal depths) cluster at two spots along the summit and south rift zone, and are considered possible seafloor eruption sites. In situ ground deformation and hydrothermal plume monitors, and later shipboard observations, confirmed the occurrence of a seafloor volcanic eruption at the volcano's summit.

Variations in magma supply and magma partitioning: the role of tectonic settings

Magma supply rates for 200 years at Krafla and Lakagigar, Iceland, and those for 150 years at Kilauea and Mauna Loa, Hawaii, are estimated roughly, based on their geophysical and geological observations. A diagram that relates erupted volumes to eruption intervals at volcanoes under various tectonic settings is represented. These results lead to a new model that a large volume (1–10 km 3) of magma is supplied intermittently at a long interval (10 2–10 4 years) beneath volcanoes in rift zones, while magma is supplied continuously with oscillations or fluctuations beneath intraplate volcanoes. Chemical data such as the MgO wt.% of lava may be one indicator in evaluating the magma supply rates of Hawaiian volcanoes. Systematic variation with time in magma partitioning within a volcano or to the surface is obtained in comparisons between among migration patterns of eruption sites, cumulative supplied volumes, and the volume ratios of erupted to supplied magma at Krafla and Kilauea. The variations suggest that a magma plumbing system may act under self-control (regulating) system through stress as one system. In response to a change in magma supply rate, the system partitions magma horizontally into dikes or vertically toward the surface. A large magma supply rate promotes the vertical extent of a crack to result in an eruption with a large volume ratio of erupted to supplied magma. This tendency is supported by field observations of flood basalts. The partitioned magma as dike intrusions suppresses magma supply partially in the shallow crust. Using analog experiments on liquid-filled cracks in gelatin, this paper demonstrates fundamental processes for magma partitioning on the effect of magma supply and stress change by the partitioned magma. A dynamical system of two differential equations on magma supply rate and stress around a magma plumbing system is proposed, to understand the qualitative variations in magma supply rate imposed by tectonic settings.

The fault nature of the Ms 5.4 volcanic earthquake preceding the 1996 subglacial eruption of Grimsvötn volcano, Iceland

The seismic activity preceding and accompanying the September 30, 1996 subglacial eruption of the Grimsvötn volcano, Iceland, developed in three stages. (1) The first stage (first 19 h) included the main shock of magnitude M s 5.4 and its aftershocks along the northern slope of the Bardarbunga volcano, situated about 20 km NW from Grimsvötn. (2) Seismic foci during the next 17 h marked a line connecting two volcanoes, Bardarbunga and Grimsvötn. This stage culminated in the opening of a fissure and the beginning of eruption. (3) The third stage was observed during the eruption (midnight of September 30–October 7), and included continuous seismic activity along the northern slope of the Bardarbunga volcano as well as more western distributed epicenters, which together had clustered the ring-fault epicentral zone. A finite-fault, broadband teleseismic P waveform inversion was applied to the main shock of September 29. The inversion showed that the rupture had developed downdip from the hypocenter in about 3 s, and the main asperity was ruptured at depths between 4 and 6 km. A comparison with the deep distribution of aftershocks showed that their hypocenters were situated above the destroyed asperities at the depths between 0 and 4 km. These space-time seismic patterns and the results of the finite-fault inversion may be interpreted in the following way. There was a magma chamber under Bardarbunga volcano covered by the strong rocks, or asperities. The main shock of September 29 broke the asperity and formed a path for magma transport. Then a fissure was formed connecting the opened magma chamber with the feeding structures of Grimsvötn volcano, and magma was injected laterally to the site of eruption. The continuous seismic activity during the third period was related to the collapse of rocks around the released parts of the magma chamber. The role of the M s 5.4 earthquake in this sequence was decisive.