V52B-01 INVITED
The Consequences of Increased Magma Supply to Kilauea Volcano, Hawai`i
The summer of 2007 was a time of intense activity at Kilauea. By mid-2007, ~4 years of summit inflation had uplifted and extended the caldera by 30 cm and 55 cm, respectively. Lava continued to erupt from the Pu`u `O`o vent on the east rift zone (ERZ) during the inflation. On May 24, 2007, two M4+ normal-faulting earthquakes occurred on caldera-bounding faults southeast of the summit. The seismicity did not affect summit inflation, which continued until June 17 when a dike intruded the upper and middle ERZ, causing a pause in the eruption, collapse of Pu`u `O`o's floor, and a small eruption 6 km uprift of Pu`u `O`o. The inflated state of the summit, relative timing of summit deflation and east rift zone extension, and abundant co-intrusive earthquake activity suggest forcible intrusion of magma. Lava returned to Pu`u `O`o by July 2, forming a lake that gradually refilled much of the collapsed crater. Early on July 21, the lake drained suddenly, the cone began to collapse, and a 2-km-long series of discontinuous eruptive fissures opened on and beyond the east flank of Pu`u `O`o. Sesimicity in Kilauea's south flank has been elevated since June and several M3+ earthquakes have occurred there, including a M5.4 on August 13. An increase in magma supply to Kilauea's shallow magmatic system is the probable cause for the events of summer 2007. Summit inflation since 2003 occurred during a period of constant or increasing magma supply to Pu`u `O`o, based on SO2 emissions from the ERZ. The rate of inflation increased markedly in early 2006, and uplift also began in the southwest rift zone. CO2 emissions at the summit, indicative of the quantity of magma degassing beneath Kilauea's caldera, more than doubled between 2003 and 2006. Also since 2003, the ERZ immediately downrift of Pu`u `O`o extended, and subsidence in the lower ERZ ceased. Together, these factors suggest that the magma supply rate to Kilauea's shallow magmatic system (the summit and rift zones above about 5 km depth) approximately doubled between 2003 and 2006. Subsequent volcanic and earthquake activity, including the events of mid-2007, are probably a result (either directly or indirectly) of this increase.
V52B-02
Deep crustal storage of large volume of magma prior to catastrophic eruptions: The role of visco-elastic response to magma accumulation
Volumes of eruption products in individual volcano-tectonic episodes in Iceland span more than three orders of magnitude (less than 0.2 to over 20 km3), suggesting widely different volumes of magma may accumulate in the crust prior to eruptions. Yet, elastic deformation models suggest that only volumes near the low end of this range can accumulate in magma chambers within a medium assumed to behave in an elastic manner, as the uppermost elastic layer of the crust. An overpressure in a magma storage volume within an elastic medium, created e.g. by inflow of new magma, will cause sudden displacement of the magma chamber walls in accordance with Hookes law. No further deformation takes place if the overpressure remains constant. A completely different situation arises if the magma accumulates in a ductile medium, e.g. in a Maxwell material. If overpressure is sustained in magma chamber within a visco-elastic material, e.g. by inflow of new magma, then deformation will by sustained over long times. Crust will flow away from the magma accumulation site, at a speed scaling inversely with the viscosity. For a lower crustal viscosity near magma accumulation site of 1x1018 Pa s (about five times lower than recently inferred average viscosity in Iceland's lower crust/mantle), a crustal magma accumulation time scale of 1-200 years would reproduce three orders of magnitude range in accumulated magma volumes prior to eruptions, if Maxwell rheology is appropriate. A one-dimensional model of Maxwell material demonstrates this effect. Assume the Earth behaves as a spring and dashpot connected in series. Application of sudden pressure causes immediate elastic strain. If pressure is kept constant over time , t, the strain will grow linearly with time. The total strain can be represented as the elastic strain scaled by a factor of (1+t/τ), where τ is the Maxwell relaxation time, the ratio of viscosity over the elastic rigidity. For viscosity about 1x1018 Pa s and rigidity of about 16 GPa (rigidity inferred from annual cycle in land elevation in Iceland attributed to snow load on Iceland's ice caps), the Maxwell time is about 2 years. This model suggests that sustained overpressure in a magma storage volume over time span up to about 200 years, can be associated with accumulation of 100 times more magma than in the elastic limit, providing one explanation of how large volumes of magma may accumulate in the mid- and lower crust prior to catastrophic eruptions.
V52B-03
Gravity Variations at a Dynamic Basaltic Caldera: Before and After the 2005 Eruption of Sierra Negra Volcano, Galapagos Islands
Sierra Negra volcano, an active basaltic volcano in the western Galapagos, last erupted in October 2005 following a period of accelerated uplift of the central caldera floor that started in April 2003. Deformation data indicate that a shallow (~ 2 km) sill underlies the caldera floor, and an intrusion rate of 64 x 106 m3/y for the 6 months prior to eruption was calculated from a continuous GPS network installed in 2002. Micro-gravity measurements were conducted in 2005, 2006, and 2007 at three stations in the center of the caldera and one station on the outer flank of the volcano and referenced to a base station on the NE rim of the caldera. From June 2005 to June 2006, residual gravity measured in the caldera increased by 1500 microgals at the center of the caldera to 184 microgals halfway to the northern edge of the caldera. This increase in residual gravity (height corrected) was accompanied by an uplift rate of ~ 212 cm/y until February 2006 after which the uplift rate decreased to 73 cm/y in 2006 and 44 cm/y in 2007. Similarly, from June 2006 to June 2007 gravity increased less dramaticaly than in 2005-2006 with an average increase of 11 microgals at the center of the caldera and 132 microgals at the more northern part of the caldera. Interestingly, the center of maximum gravity change shifted from the center of the caldera to the northern part sometime between June 2006 and June 2007. Gravity measurements on the outer rim of the caldera showed a 300 and a 200 microgal decrease from June 2005 to June 2006 and June 2006 to June 2007, respectively accompanied by low rates of inflation (1.8 cm/y). The coupling of gravity and deformation change supports the hypothesis of significant mass increase in the central-northern part of the caldera.
V52B-04
Factors Determining the Propagation and Arrest of Magmatic Fissures
Volcanic eruptions are typically preceded by seismic swarms and ground deformations that result from the propagation of magma-filled fractures from a shallow reservoir or deeper source to the surface. However these sequences sometimes abort in the sense that a seismic crisis stops without an eruption taking place. The long term evolution of volcanic edifices is partly conditioned by whether magma erupts at the surface or stops as intrusions in the subsurface. The interpretation of these observations in terms of physical processes is an important and challenging scientific problem as well as one of potentially considerable operational value for observatories on active volcanoes. Analysis of several recent seismic crises at Piton de la Fournaise volcano using continuously recordings show that different patterns of seismic energy release can be recognised. We propose possible interpretations of these data based on the results of laboratory experiments in which constant volumes or fluxes of liquid are injected into elastic, brittle host material made with gelatin. Experimental conditions are either isothermal, or with a gelatin host temperature below the solidus of the hot liquid in the fracture: the latter causes solidification. Fracture propagation is driven by liquid buoyancy and/or elastic over- pressure, and resisted by fracture toughness, viscous stresses and or freezing. Constant volumes of liquid propagating due to buoyancy can stop even under isothermal conditions due to an effective fracture toughness of the host medium appropriate for the three-dimensional evolution of the fracture. Solidification in an advancing fracture enhances the effective toughness in the tip region and can lead to intermittent propagation in the case of constant flux and arrest of fractures in the case of decreasing flux.
V52B-05
Tectonics of the 1783 Laki Crater Row and the Associated Graben, South Iceland
One of the longest Holocene crater rows in Iceland is the one associated with the Laki (Skaftareldar) 1783 eruption in South Iceland. It trends northeast, is at least 27 km long, and has roughly 140 crater cones. It gave rise to the second largest historical (the past 1100 years) eruption in Iceland, the area of the issued lava flows during the eruption being about 565 km2, the lava volume 14.7 km3, and the tephra volume 0.4 km3. A part of the Laki Crater Row is associated with a graben. The main graben dissects the hyaloclastite Laki Mountain and extends to the northeast and southwest of the mountain. The graben is very narrow, mostly 200-300 m wide, but reaches 500 m in the Laki Mountain itself. The total visible length of the graben is 5-6 km, but parts of it may be buried by the lava flows. The vertical displacement on the boundary faults is variable, but reaches at least 6-7 m. Where the graben dissects the top of Laki Mountain, the normal faults are gaping by as much as 2-3 m and thus partly tension fractures and partly shear fractures. There is also a set of NW-trending faults dissecting the mountain. Our field data show that the crater row itself (and the associated feeder-dike) did not reach the top of the Laki Mountain, but rather ends vertically on its slopes. Volcanotectonic grabens of this kind are quite common in the rift zone of Iceland. Thus, for example, the Sveinagja Graben and the Sveinar-Randarholar Graben, in the North Volcanic Zone, and the graben cutting through the top of the hyaloclastite mountain Thorbjorn on the Reykjanes Peninsula, in the West Volcanic Zone, are well known and have been studied. The fundamental question as to the formation of such grabens is if they existed, partly or wholly, before the eruption and thus captured the associated feeder-dike, or if the feeder-dike itself was entirely responsible for the graben formation. To interpret the field data we provide new models of the interaction between volcanotectonic grabens and feeder- dikes. We show that in many cases an existing graben would be likely to capture an upward-propagating feeder- dike, as has been observed in the Sveinar-Randarholar Graben and during the 1975-1984 Krafla Fires in the North Volcanic Zone. For the Laki Graben, our data indicate that a segment of one of the graben faults was used by the feeder-dike as a channel to the surface, showing that this segment existed at the time of the eruption.
V52B-06
Toward Forecasting Volcanic Eruptions Using Seismic Noise
Volcanic eruptions are preceded by the pressurization of magma leading to dilatations of volcanic edifices. Monitoring this process in real-time can provide us with a useful information to forecast volcanic eruptions. Volcano inflation generates ground deformations that can be revealed by spaceborne radar interferometry or by strain and tiltmeters. In some cases, however, volcano dilatation can be localized at depth with no measurable effects at the surface and it has proved difficult to monitor changes in volcanic interiors, despite considerable effort. Here we present an observation of a new type of precursory phenomena by demonstrating that pre-eruptive dilatations of the Piton de la Fournaise volcano can be detected as decreases of seismic velocity a few weeks prior to eruptions. We are able to monitor the changes of the volcano interior continuously in time and space by measuring very small relative seismic velocity perturbations (0.05%) by using the reproducibility properties of the ambient seismic noise. This new direct observation of the dilatation of volcanic edifices should improve our ability to forecast eruptions and to a priori assess their intensity (injected magma volume and flux) and environmental impact.
V52B-07
Hidden Dykes detected on Ultra Long Period seismic signals at Piton de la Fournaise volcano ? Constraints on the upper reservoir pressure state since 1992.
Broadband seismic data enable us to test whether Ultra Long Period (ULP) signals can be used to determine the magma chamber pressure state and to forecast volcanic eruptions. In a systematic investigation of seismic signals recorded at the GEOSCOPE station RER in la Réunion Island near the Piton de la Fournaise, we show that ULP signals from the past two decades are typically associated with eruptions of Piton de la Fournaise. Some of these unusual signals were previously detected and interpreted in terms of tilt (Battaglia et al., 2000). We tentatively propose to interpret these signals as signatures of the activity of the upper magma feeding system of the Piton de la Fournaise volcano chamber. They are detected using STS-1 seismometers sensitive in the lower part of the bandwidth 10-3 to 10-2 Hz. Assuming geometrical parameters of the source and magma injection rates in the upper reservoir, the collection of all seismic events allowed us to build a time-series of absolute pressure in the upper magma chamber. The absolute pressure framework provides new constraints to parametrize a wide range of geophysical and geochemical models together. This alternative model should provide new insights in the eruptive processes that characterize Piton de la Fournaise volcano (La Réunion).
V52B-08
The 2007 Eruption of Pavlof Volcano, Alaska
Pavlof Volcano on the Alaska Peninsula began to erupt on August 15, 2007 after a 10.7 year repose. Precursor signals consisted of low-frequency earthquakes that began on August 14 and thermal anomalies that were likely coincident with the beginning of the eruption. The mainly strombolian eruptions are occurring from a new vent high on the SE flank of the volcano, separate from the NNE vent that had been active over the last several decades. Seismic activity, monitored by a network of 6 local instruments, consists of low-frequency events, explosion earthquakes, volcanic tremor, and lahar-generated signals. One station, PVV, is located only 220 m from a lahar channel, and lahars generate an easily distinguished high-frequency seismic signal. A commonly observed sequence is an increase in eruptive activity at the vent, accompanied by stronger tremor visible on all stations, and followed 12-30 minutes later by a lahar at PVV. This suggests that the eruption pulse ejects fresh hot material, which melts additional ice and snow to form new lahars. Steam and ash plumes have generally been below 15,000 ft, but rose as high as 20,000 ft on August 29 and 30. AVHRR remote sensing data showed an ash signal on these days, consistent with pilot reports. On August 30 lightning was observed in the plume from Cold Bay, 59 km SW. In response to the eruptions, AVO has been conducting 24 hr per day surveillance. Fieldwork to date has fortified seismic stations, and installed a new webcam, pressure sensor, and electric field meter. Collaborating scientists from the University of Alaska Fairbanks have installed aerosol sampling equipment at four locations, and collaborating scientists from New Mexico Tech have installed lightning detection equipment at four stations surrounding the volcano. Based on recent eruptions of Pavlof in 1981, 1986, 1996, etc., the eruptive activity is likely to last several months and may include one or more episodes of ash columns to heights of 30,000 ft or more. Additional field work is being planned as of this writing. The Alaska Volcano Observatory (AVO) is a cooperative program of the US Geological Survey, the University of Alaska Fairbanks (UAF), and the Alaska Division of Geological and Geophysical Surveys.