G51A-0047 0800h
Geodetic constraints for the mechanism of Anatahan eruption of May 2003
Anatahan Island is located at the southern end of the Mariana volcanic chain. On May 10, 2003 the first eruption occurred in its history and the eruptive activity continued to June 2003. Although volcanic activity declined in the second half of 2003, it resumed in April 2004. In order to determine crustal deformation associated with the eruption, we conducted GPS measurement on July 2003 at a benchmark (ANAT) located about 7km west-northwest of the active crater, where GPS campaign measurements had been repeated since 1992. In the period from January to July 2003 covering the eruption, a significant subsidence as large as 21cm was detected but horizontal movement was much smaller. At the same site, we started continuous GPS measurement on July 2003 to monitor transient deformation in terms of magma migration. On January 2004, we established another permanent GPS site (ANA2) at the northeastern part of the island about 3km far from the active crater to better constrain spatial pattern of the deformation. Coordinates time series at ANAT show a change of trend at around the beginning of 2004. Another subsidence of 2.9cm and a westward motion of 2.2cm were detected in the period from July to December 2003, and in turn an uplift of 5.3cm and an eastward motion of 0.8cm in the period from January to June 2004. We model three inflation/deflation sources for three different time periods, which are aligned straight and deepened westward at a dip-angle of 30 degrees from the active crater. Although the present model is very preliminary and non-unique, geodetic data will give adequate constraints on modeling the eruption mechanism.
G51A-0048 0800h
Quasi-Static Thermoelastic Deformation in an Elastic Half Space: Theory and Application to InSAR Observation at Izu-Oshima Volcano, Japan
We derive closed analytical solutions for a quasi-static thermoelastic deformation response to instantaneous point and spherical heat sources in an elastic half space, extending the result by Furuya (2004). We examine the spatial and temporal evolution of thermoelastic deformation for point and spherical heat sources. We applied the solution to a radar interferometric observation of post-eruptive deformation associated with the 1986 fissure eruption at Izu-Oshima volcano, Japan. Assuming a spherical heat source at a depth of 240~m with a volume of 1.15$\times10^{7}$ m$^{3}$ and a temperature offset $10^{3}$ K, the predicted rate of post-eruptive ground movement agrees with the observed rate within observational errors. Also, the same parameter values allow us to compute the co-eruptive ground displacement by the effect of mass intrusion, whose amplitude is consistent with the observed height (45~m) of the newly formed cone. The derived solutions can be applied to transient ground displacements observed at active volcanos, and allow us to evaluate the heat amount of magma intruded at very shallow depth.
G51A-0049 0800h
Distributed Versus Point Source Models of Volcano Deformation
Volcano deformation is the surface expression of physical processes at depth, including magma accumulation and movement, pressurization, and crystallization. Deformation models are used to extract source parameters including location, depth, and shape of the magma reservoir. Many studies assume simple predefined sources, most commonly a center of dilation or "Mogi" source, which approximates a spherical pressurized cavity in an elastic half-space. Others have employed an approximate solution for a pressurized ellipsoid due to Yang et al [1988]. Some have used more general distributions of centers of dilation, although the physical interpretation of these models is unclear. Eshelby [1957] showed that an appropriate distribution of centers of dilation and vertical double forces yields a uniform pressure on the boundary of an ellipsoid in an elastic full space. Here we explore the possibility of inverting for a distribution of such nuclei of strain that both fit geodetic data and approximately satisfy a uniform pressure boundary condition on a closed surface surrounding the sources. This would allow the data to be interpreted in terms of an internally pressurized cavity of arbitrary shape. We have tested inversions that minimize a data residual norm and regularizing functionals including spatial smoothing and compactness of the source region. We find that the resulting distributions are highly dependent on the modeling constraints chosen, and the spatial distribution of sources and the resulting stress field cannot always be easily interpreted in terms of source geometry. Based on our studies we recommend the following inversion strategy: First invert for a generalized non-isotropic point source, as in Davis [1986]. This provides a reasonable estimate of source parameters including shape (from the strength ratio of vertical and horizontal components of the point source). If the source is located deep relative to its size, a point source will adequately fit the data. If the source is shallow, finite source models approximating an ellipsoidal or sill-like structure can be tested. If simple finite source models fail to fit the data, one can invert for compact distributions about the source. In this case surfaces of constant normal stress around a source region can be used as a boundary condition to allow for non-ellipsoidal geometries.
G51A-0050 0800h
Magma Intrusion Estimated From InSAR Data: Sensitivities to Homogeneous, Isotropic, Poisson-solid, and Half-space Assumptions
Simple elastic models that simulate deformation caused by magma intrusion within a volcano are readily available and often take the form of an expansion source embedded in a homogeneous, isotropic, Poisson-solid half-space (HIPSHS). These models are commonly used in inverse methods to precisely estimate deformation source parameters from deformation data. The accuracy of these estimated deformation source parameters depends on the validity of the HIPSHS assumptions. Generally, the HIPSHS assumptions poorly represent an actual volcano. Interferometric synthetic aperture radar (InSAR) data reveal that Okmok volcano, Alaska, deflated more than a meter due to lava extrusion during the 1997 eruption of the volcano. Using InSAR-observed deformation of Okmok volcano as an example, deformation prediction sensitivities to the HIPSHS assumptions are identified via forward and inverse modeling methods. Green's functions for displacement due to an expansion source at depth are computed using a combination of analytical and finite element models that systematically relax the requirements for HIPSHS assumptions. Comparisons among the results obtained using HIPSHS versus non-HIPSHS models identify the prediction sensitivity to each of the HIPSHS assumptions. Prediction sensitivity to topographic effects is relatively subtle. Forward modeling indicates prediction sensitivities associated with each of the isotropic, Poisson-solid, and homogeneous assumptions (listed in the order of increasing severity) have magnitudes as large as several tens of centimeters. Deformation source parameters, estimated using inverse models, are particularly sensitive to lateral variations of material properties associated with a caldera. Quantitative interpretations of volcano deformation source parameters, inverted from deformation data, require careful assessments of the deformation models at the core of the inverse methods.
G51A-0051 0800h
Combined Inversion of Gravity Changes, Radial and Vertical Displacements Between 2000 and 2002 at Merapi Volcano, Indonesia Using Genetic Algorithm Techniques
A combined inversion of gravity changes, height, and radial displacements, which were measured in the period between 2000 and 2002 at Merapi volcano, Java, Indonesia, provides the possibility for the detection of potential magma/pressure sources which might be responsible for the temporal displacement and gravity changes. The combined inversion technique is based on the Navier-Stokes equation that couples elastic and gravitational effects in a mathematical halfspace. The most reliable sources found (both magma and pressure) are the results of global optimization using genetic algorithm techniques.
G51A-0052 0800h
Transient deformation on the south flank of Kilauea Volcano, Hawaii
On 30 January 1997 an intrusion on Kilauea Volcano opened a new fissure within the East Rift Zone (ERZ) at Napau Crater, 3km uprift from the ongoing eruptions Pu'u O'o. The fissure eruption lasted for two days and opened a nearly vertical dike 1.96m by 5.15km long that extended from the surface to a depth of 2.4km (Owen et al.,GRL,2000). During the eruption, the lava pond at Pu'u O'o drained pausing the eruptions there for nearly month until it eventually refilled in late February and eruptions resumed 28 March 1997. Continuous GPS data show a large transient following the January 30, 1997 dike intrusion. After lengthening 40cm during the initial eruption, the baseline between the two stations spanning the ERZ lengthened and additional 10cm over six months. The coastal station KAEP also shows the transient as it continues to move southward (5cm) over the same six months. The baseline between the two stations spanning Kilauea's summit caldera, UWEV and AHUP, contracted sharply during the eruption and gradually recovered to a little longer than it's previous length two months after the intrusion. We use the Extended Network Inversion Filter (McGuire and Segall, Geophys. J. Int., 2003) to invert the continuous GPS data for: volume change of a spherical pressure source under Kilauea's summit, opening distribution on a nearly vertical dike in the ERZ and slip distribution on a decollement 9 km beneath the south flank. We show that, following the January 30, 1997 intrusion, rift opening rate briefly increases for a few weeks to a maximum of 1.47 m/yr then gradually decreases over six months to background rates. The modeled transient rift opening patterns under Napau crater coincide with the observed behavior of Pu'u O'o's lava pond. The transient rift opening rate accelerated while the lava pond at Pu'u O'o was absent, the rate was constant after it reappeared and then slowly decelerated after eruptions resumed.
G51A-0053 0800h
Results of Seafloor Vertical Deformation Monitoring on the Submerged South Flank of Kilauea Volcano, Hawaii from 2000 - 2004
Kilauea Volcano is one of the most active volcanoes on Earth. The south flank of Kilauea undergoes large-scale deformation as a result of volcanic and tectonic processes that has been well-monitored onshore. Since the majority of the volcano is submerged, however, undersea monitoring of crustal deformation is necessary to constrain kinematic models of Kilauea. Models based on the onshore geodetic data have suggested offshore motion, but this cannot be measured with land-based approaches. By combining an acoustic technique with kinematic GPS survey methods and a seafloor survey using pressure sensors, it is possible to measure the horizontal and vertical motion of the south flank on the seafloor. In November of 2000, an array of seven precision acoustic transponders and six seafloor benchmarks was installed and measured offshore on the south flank of Kilauea. Repeat measurements were made in April of 2002 and again in August of 2004. We will present results from reductions of the pressure sensor data from 2000 - 2004 that describe the vertical motion on the seafloor during this time period. We will also present a new model of south flank deformation on Kilauea based on our seafloor geodetic results as well as the available onshore geodetic data, thus providing a better understanding of the structure and dynamics of the volcano as a whole.
G51A-0054 0800h
Gravity Changes and kinematic behaviour of the Island of Pantelleria in the Geodynamic Framework of the Sicily Channel (Southern Italy)
The island of Pantelleria is an active volcano located in the Sicily Channel (Southern Italy), in the middle of a continental rift system. Since the 1980s the island is periodically surveyed by means of geodetic techniques (levelling, EDM, GPS and high precise gravimetry) to monitor the local and regional volcanic dynamics. The analysis of the gravity data, collected during several field surveys from 1990 to 1998, have shown that the gravity field changes generally appear resulting by the superposition of short and long wavelengths, due to the combined effect of shallow and deep sources. Gravity changes fields also reflect, to some degree, the structural setting of the island as sketched by the Bouguer anomaly field which indicates that the island is constituted by a basement dislocated in two main blocks. These latter are also bordered by two lineaments, probably regional faults related to the global geodynamics of the Sicily Channel Rift Zone. Moreover, there is an inverse correlation between the gravity and height variations and their joined analysis implies that: i) Pantelleria appears to be divided in two blocks also from the kinematic behaviour; ii) the observed phenomena may be also ascribed to the influence of the geodynamics of the Sicily Channel. Moreover, a combined analysis of the time-space features of the horizontal ground deformation and gravity changes has been done to check if the characteristics previously detected are present in all the aspects of the kinematic behaviour of the island. On the basis of the previous conclusions the EDM data have been reprocessed for a proper comparison with the gravity data, the Bouguer anomaly and the altimetric data. This allows a joint interpretation taking also into account the structural setting and the geodynamics of the Sicily Channel to evaluate its influence on the local volcanic dynamics of the island.
G51A-0055 0800h
Magma dynamics and volcano geodesy in Iceland
Here we review the achievements of volcano geodesy in Iceland during the last 15 years. Extensive measurements of crustal deformation have been conducted using a variety of geodetic techniques, including levelling, electronic distance measurements, campaign and continuous GPS geodesy, and interferometric analysis of synthetic aperture radar images (InSAR). Results from these measurements provide a comprehensive view of the behavior of Icelandic volcanoes. Between inflation, intrusion, and eruption episodes, volcanoes are likely to deflate or show no sign of seismic activity. Subsidence rates are often in the range of a few millimeters to a few centimeters a year, reducing progressively with time since the last eruption or intrusion at the volcano. Subsidence can be caused by cooling and contraction of magma, outflow of magma, or it can related to plate spreading.Volcano subsidence or lack of deformation is often interrupted by episodic magma flow towards near-surface locations. Such magma recharge has been observed geodetically at Hengill, Hekla, Eyjafjallajokull, Katla, Grimsvotn, and Krafla volcanoes, with inflow inferred to last from a few months up to two decades. In the last 15 years, four volcanic eruptions, three intrusive events and two $>$M6 earthquakes have occurred. In recent years, the Grimsvotn and Katla volcanoes have exhibited inflation at a rate of 1-2 cm per year, while the Hekla and Torfajokull volcanoes have inflated at rates an order-of-magnitude less. Subsidence is occurring presently at the Askja and Krafla volcanoes. Within the period of geodetic measurement, signals consistent with no deformation are typical for most of the 35 active volcanoes in Iceland.
G51A-0056 0800h
One and a Half Year of INSAR Monitoring of Volcanic Activity at Piton de la Fournaise, R\'{e}union Island, With ASAR-ENVISAT Data
The activity of Piton de la Fournaise volcano (R\'{e}union Island) was marked by a significant increase during the last one and a half year (April 2003 to September 2004) with eight eruptive episodes. This period coincides with the beginning of systematic data acquisition on Piton de la Fournaise by the radar ASAR on board of the ENVISAT satellite. We have used 80 ASAR images to perform an INSAR monitoring of the Piton de la Fournaise activity. We produced a dataset including more than 150 interferograms which provide an exceptional record of the ground displacements associated with each individual eruption. Thanks to the various angles of view of ASAR, we were able to characterize accurately the EW and vertical components of the displacements. Most of the eruptions exhibit an asymmetric fringe pattern with strong uplift (5 to 15 cm) and horizontal eastward displacement (15 to 45 cm) to the East of the eruptive fissures and with moderate subsidence to the West. These displacements are compatible with the opening of a series of eastward-dipping dikes in the N10$\deg$E and S10$\deg$E trending fracture zones. In contrast, the interferograms spanning the January 2004 eruption show an impressive fringe pattern (40 fringes) extending from the central cone to the Plaine des Osmondes. It could be due either to relatively extended, sill like, magma source or to a decollement. We modeled the displacements associated with the August 2003, September 2003 and May 2004 eruptions using a 3D mixed boundary element method for elastic media. The best-fit models converged towards a set of dikes dipping 45 to 55$\deg$ eastward whose bottoms lie at around 1500 m a.s.l. (depth 800 - 1000 m). The repetitive dike intrusions may have weakened the flank stability by increasing the local strains and then induced the larger displacements of January 2004.
G51A-0057 0800h
Deformation of the Aniakchak Caldera, Alaska, Mapped by InSAR
Aniakchak Volcano, Alaska, a stratovolcano with a summit caldera about 10 km in diameter, is located 670 km southwest of Anchorage on the Alaska Peninsula. The caldera was formed by collapse about 3,500 years ago during an eruption of more than 50 km\^3 of andesite-dacite pyroclastic debris. Eruptions since the caldera-forming event have produced numerous vents and cones. Aniakchak's most recent eruption occurred in 1931 from a crater located near the base of the west caldera wall. We investigate the deformation of Aniakchak Caldera using 19 ERS-1/2 interferometric synthetic aperture radar (InSAR) images acquired from 1992 through 2002. InSAR images through this time period reveal that the 10-km-wide caldera has been subsiding. The pattern of subsidence does not reflect the distribution of pyroclastic deposits from the last eruption of the caldera in 1931. The maximum subsidence is identified near the center of the caldera, with a rate of as much as 13 mm/yr. Deformation outside the caldera is insignificant. Least-squares inversion of the multi-temporal deformation maps indicates that the subsidence rate has been relatively constant over the time of investigation. Modeling the deformation with a Mogi point source locates the source of subsidence at about 4 km beneath the caldera center, which is consistent with past reports on the depths of magma storage based on geochemical analysis. In addition, recent field surveys observed CO2 coming out of a lake inside the caldera and identified several fumaroles within the caldera. These observations suggest that the causes of the subsidence are probably due to cooling or degassing of magma body, and/or reduction of the pore fluid pressure of a cooling hydrothermal system.
G51A-0058 0800h
Volcanic Deformation Interpretations From Campaign-Mode GPS Measurements of the Commonwealth of Dominica, Lesser Antilles, 2001-2003 Epoch I Data
The Commonwealth of Dominica is located on the eastern margin of the Caribbean plate midway along the Lesser Antilles Arc. It is the most mountainous of the Lesser Antilles, primarily due to its seven potentially active volcanic centers. A shallow seismic swarm was recorded in southern Dominica from September 20, 1998 to June 30, 2001. This seismicity prompted geodetic investigation due to its relative shallow nature and arc-shaped distribution around the peaks of Morne Anglais, Morne Plat Paye, and Morne Patates. A second seismic swarm took place in 2003. This seismicity clustered in the north of the Island in the vicinity of Morne Aux Diable. Campaign GPS observations were done using Ashtech Z-12 dual frequency, code-phase GPS receivers, with choke ring antennae. Three types of mounting systems were used depending on differing site restrictions. A total of 18 established monuments now exist throughout Dominica: nine were constructed in 2001 after the first seismic swarm occurred and three in 2003 after a second seismic swarm. Six additional sites were installed in June 2004. This presentation will focus on data collected from the 2001 and 2003 campaigns. Absolute point positions were obtained and velocity vectors were resolved in June 2003 for the initial nine benchmarks. Each data point was then compared with previously measured locations in 2001. Seven of the nine sites are located in the south where much of the seismic activity had been concentrated. The remaining two sites are located in the north. The preliminary velocity field is complex. All of the sites showed some measure of local deformation over a 2-year period. In the southern portion of the island, a set of vectors located in the Gommier Estate (GOMM) and Wotten Waven School (WOTT) point toward each other as well as towards the largest concentration of 2001 seismic epicenters. The vector for GOMM exhibited a velocity of 4.21 mm/a trending N68.1°E. WOTT displayed a velocity of 5.38 mm/a trending S16.8°W. A similar occurrence is observed in the northern region of Dominica, where another set of vectors located in Cabrits National Park (CABR) and in Concord (CNCD) point toward the main collection of shallow earthquakes from the more recent seismic swarm in 2003. CABR velocity is 3.89 mm/a trending S68.8°E, while CNCD velocity is 3.70 mm/a directed in a N25.8°W trend. Both seismic episodes are arcuate in shape. The overall shape and depth of the two sets of seismicity on Dominica, when joined with the geometry of local deformation suggest reactivation of old caldera collapse structures. However, the possibility of a shallow magmatic source or hydrothermal activity as grounds for seismicity is not out of the question. Continued annual GPS monitoring is imperative in order to better constrain the nature of deformation on the island of Dominica.
G51A-0059 0800h
Non-Double-Couple Source Models of Volcanic Seismicity
The source mechanisms of many earthquakes from the 2000 Miyakejima, Japan, eruption sequence are unusual in that the seismic data clearly indicate the existence of non-double-couple components. The period of volcanic unrest was marked by substantial crustal deformation as well as an earthquake swarm which included many magnitude five and larger earthquakes. These earthquakes were well observed by local and regional seismic and GPS networks. Thus the Miyakejima eruption provides an excellent opportunity to study non-double-couple earthquake source processes. Earthquakes related to volcanic unrest may be expected to have non-double-couple mechanisms or to have mechanisms which include significant volume changes. These types of earthquake source mechanisms have been successfully determined through moment tensor inversions. However, unconstrained moment tensor inversions can produce moment tensors which are indicative of non-physical event parameters or which contain multiple compensating volume changes. Also, the additional model parameters in an unconstrained moment tensor inversion can lead to instabilities. By imposing certain constraints on the seismic source, we can develop models which are physically realistic, such as source models related to extension and tensile faulting. We will present a new moment tensor decomposition to describe the faulting mechanics of non-double-couple earthquakes. In this moment tensor decomposition, non-deviatoric moment tensors are described as the sum of a double-couple mechanism and a tensile component. GPS-measured coseismic surface displacements will be used to provide additional constraints on the event kinematics. Such a source model is both mechanically realistic and well constrained. This model will be compared to other possible models for the source mechanisms of these volcanic earthquakes, including both conventional models and more exotic non-double-couple models. We will then use these mechanisms to infer the volcanic processes which led to this seismicity.
G51A-0060 0800h
Constraining the Source of Unrest at Campi Flegrei (Italy) Using Geodetic And Micro-Gravity Data.
Campi Flegrei is an active volcanic caldera, 12-14 km in diameter, immediately west of Naples, Southern Italy. Geological and historical records indicate that Campi Flegrei has been the site of intense uplift and subsidence phenomena. A historical period of uplift has occurred within Campi Flegrei from mid-1982 to 1984. Peak surface deformation reached about 1.8 m and the caldera experienced significant seismic activity. Since January 1985, the caldera floor is slowly sinking. Several mechanisms have been proposed to explain the caldera unrest: hot fluid migration, intense magma degassing, or the intrusion of a magma body followed by fluid removal. Combined geodesy and gravity measurements can be used to infer the density of the intrusive fluids, and can better constrain the deformation source. Given the significant density difference between silicate melts ($\sim$2500 kg/m$^{3}$) and hydrothermal fluids ($\sim$1000 kg/m$^{3}$), it is reasonable to use density estimates from gravity to distinguish between these two possible sources of caldera unrest. We determine the location and geometry of the inflation source inverting leveling and trilateration measurements collected between 1980 and 1983, assuming a vertical prolate ellipsoidal source, a spherical source and a penny shape crack source. The two sources that best fit the geodetic data are a penny shape crack (\chi$^{2}$=64, 4.2 km deep beneath the town of Pozzuoli) or a point source (\chi$^{2}$=70, 3.1 km deep beneath the town of Pozzuoli). To resolve the density of the intrusion, we invert leveling and micro-gravity data from 01/1982 to 06/1984. Preliminary results, using a spherical source, indicate an inflation source 3.0 km deep, with a volume change of 0.066 km$^{3}$ and a density of 2350 kg/m$^{3}$. On the other hand, a point source 3.1 km deep beneath the town of Pozzuoli fits geodetic and gravity measurements from the 1989-91 subsidence period (volume decrease = 0.005 km$^{3}$), but provides an unrealistic estimate of the body density. Future work will include the use of sill-like sources to improve the density estimates.
G51A-0061 0800h
InSAR as a volcano monitoring tool for Auckland City, New Zealand.
Auckland City, New Zealand, with a population exceeding 1 million, is built directly on top of an active monogenetic volcanic field, which last erupted several hundred years ago. The field contains around 50 discrete volcanoes that have erupted over the last 250ka. The eruptions are typically small volume (<0.1km3), although the most recent event produced a total volume of 2.4km3 or 59% of the total volume of the field as a whole. It is important therefore to have an effective and robust early warning system in place to detect geodetic changes that may be precursors to such eruptions. However, because each new eruption forms a new vent anywhere within the 25 by 15 km volcanic field, it is extremely difficult to use field-based techniques to monitor the entire area at an adequate spatial and temporal density. We therefore looked to orbital InSAR measurements of deformation as a potential solution to this problem. The viability of using C-band InSAR as a monitoring tool was established using ERS data during the 1990s, and the largely urbanised nature of the Auckland region means that coherence between acquisitions persists for years, allowing long-term baselines to be established. Auckland magma is thought to ascend exceptionally rapidly. We undertook Mogi modelling of ascending volumes at different depths in the crust to simulate a rising diapir, and ascertained whether the precursive deformation signals are likely to be (a) within the detection limits of C-band InSAR, and (b) discernible above any atmospheric noise. As a result of our research, a program of regular radar data collection from the Envisat satellite has been established for monitoring purposes, with data collected every 3 months, ensuring that an adequate archive of suitable data is available for inter-comparison with future Envisat acquisitions, should a volcanic crisis occur. Once near-real-time data supply issues are totally overcome, the InSAR monitoring capability at Auckland will be fully operational.
G51A-0062 0800h
The Pressure Sources Beneath Unzen Volcano Inferred From Geodesic Survey
Unzen Volcano, which is located on Shimabara Peninsula, west Kyushu Island, Japan, erupted from 1990 to 1995. The ground deformations caused by volcanic activity were observed by several methods, such as Leveling survey, GPS and Tilt meters. In particular, it turned out from leveling data that the west coast area of Shimabara Peninsula sank about 8 cm since eruption had started. Joint Research Team of the national Universities suggested in 1992 that the model which has three pressure sources, could explain the ground deformation data in those days. But this model couldn_ft explain the leveling data which was observed after the eruption had stopped. In order to explain this latest leveling data, we had to add the fourth deeper pressure source beneath the Chidiwa Bay, confronted in the west seashore of the Shimabara peninsula ({\it Matsushima et al}., 2003; {\it Kohno et al}., 2003). In this study, we re-consider the source model beneath Unzen Volcano, using 1991-2001 and 2004 Leveling data along the northern flank of Unzen Volcano and the western coast of Shimabara peninsula. Also we use GPS data monitored by Kyushu University and Geographical Survey Institute. In calculation, both vertical and horizontal displacement was calculated applying the point source model ({\it e.g. Mogi}, 1958), and we get the best-fit source parameters. Parameters of the pressure sources are the location and the volume changes of pressure sources. Through the model calculation, the half-infinite surface was made for every height of each observation point, and geographical feature was reproduced in approximation. Analysis showed that after 1995 shallower source had started to deflate, on the other hand, two deeper sources still had kept expanding caused by intrusion of magma. After 1999, three shallower sources had begun to contract, and the only deepest (a depth of 15 km) source had expanded. But it is inferred from the 2004 Leveling that the deepest source turned to contract since 2001, and the all sources have started to contract.
G51A-0063 0800h
The magma chamber associated with the eruption of Miyakejima Volcano, Japan, since 2000, inferred from crustal deformation data (1983-2004).
Miyakejima Island, an active basaltic stratovolcano lying 170 km to the south of Tokyo, Japan, resumed its volcanic activity in late June 2000 after a quiescence period for 17 years since the last eruption in 1983. The activity changes its eruptive style and still continues with a significant volcanic gas emission from the summit caldera, which has forced the inhabitants of Miyakejima to evacuate from the island. In order to understand the magmatic activity and the gas emission, it is crucial to reveal the magma supply system beneath the island. For this purpose we elucidated locations and geometries of a magma chamber, intruded dikes, and a source related with the gas emission by using the crustal deformation data including GPS, tilt changes and leveling through the period after the 1983 eruption up to now. Following dike intrusions beneath the island at the initial stage of the activity (June 26 - 27, 2000), the observed crustal deformation showed a rapid contraction of the island due to a subsurface magma migration from a magma chamber to the west off the coast. To estimate the location and configuration of the magma chamber, we compared the deformation with various source models. We found that a source model consisting of a spherical deflation source about 8 km deep in the southwestern part of the island and a vertical dike-shaped deflation source on the spherical source is most appropriate for the crustal deformation (June 28 - mid September, 2000). The total amount of the contraction is ~0.4 km$^{3}$; if it is added to the amount of collapsed summit, which is considered to have penetrated into the chamber [Geshi et al., 2002], the amount of withdrew magma is estimated to be ~1.0 km$^{3}$. The amount is consistent with the amount of intruded magma at the west off the coast of Miyakejima [Nishimura et al., 2001]. An inflated source model with a similar configuration can explain the crustal deformation observed from 1983 to before the activity of 2000, suggesting the source had been accumulating magma before the activity. We infer that the source is the magma chamber of Miyakejima and the chamber consists of two parts: a shallow dike-shaped chamber ~4 km deep and a deep-seated spherical chamber ~9 km deep. Ueda et al. (2004) shows that the crustal deformation at the initial stage of the activity (June 26 - 27, 2000) can be interpreted by three dike-shaped inflation sources and one dike-shaped deflation source. They concluded that there existed two phases of magma movement; a small intrusion at a depth range of 1-3 km beneath the southwestern flank of the island from 18h30m to 21h on June 26, a large intrusion beneath the west coast, and deflation of a dike-shaped deflation source beneath the small intrusion, both of which started simultaneously at about 21h on June 26. The magma chamber model suggests that the former dike intruded from the upper bound of the dike-shaped portion of the magma chamber, and the latter intruded laterally from the chamber. Crustal deformation after October 2000 can be approximately explained by a spherical deflation source, about 3 km deep, beneath the summit. The source is located near the upper bound of the chamber, where we expect a conduit that connects the chamber with the gas vent on the summit. Since it deflates during the significant gas emission, the deflation source probably relate with the magma degassing process in the conduit.
G51A-0064 0800h
Time Dependent Model of Magma Intrusion in and around Miyake and Kozu Islands, Central Japan in June-August, 2000
We discuss a time dependent model of magma intrusion in and around Miyake and Kozu Islands, Central Japan from GPS measurements at 28 sites in Miyake Island, Kozu Island and their surrounding islands in the period from June 27 to August 27, 2000. A dike complex model of three sheets is assumed between Miyake and Kozu Islands, suggested from the precise hypocenter distribution map (Sakai et al., 2003). Other dike intrusion models, a dike with an aseismic creep model (Nishimura et al.,2001; Furuya et al.,2003) and a dike with a deep deflation source model (Yamaoka et al., submitted) , are also discussed. Akaike's Information Criteria (AIC) value of optimal parameters of a dike complex model indicates lower than that of other two models. After fixing the geometry of three dikes using a genetic algorithm (GA), the amounts of dike openings of top, inside, and bottom of each dike are estimated by GA for seven time periods. In the period from June 27 to July 8, dike opening is concentrated in the dike near Miyake Island, and a large deflation is also estimated at a depth of 5 km of Miyake Volcano. It suggests that magma is supplied from the depths of Miyake Island. In next period until August 10, a huge dike intrusion is characterized in the dike near Kozu Island and the lower parts of dike in central and near Miyake Island. This suggests that magma is supplied from depth between Miyake and Kozu Islands. In the period of August 10 to 27, a huge deflation is estimated at a depth of 10 km under Miyake Volcano, and dike opening is limited
G51A-0065 0800h
Mount Etna 1993-1997 Inflation: Source Inference Based on Geodetic Data Modelling
Since 1993 geodetic data obtained by different techniques (GPS, EDM, SAR, levelling) detected a consistent inflation of the Mount Etna volcano. The inflation, culminated with the 1998-2001 strong explosive activity from summit craters and recent 2001 and 2002 flank eruptions, may be interpreted by magma ascent and re-filling of the volcanic plumbing system and reservoirs. We model the 1993-97 EDM and GPS data by 3D pressurized sources to infer the position and dimension of the magma reservoir. We perform analytical inversions of the observed deformation using both spheroidal and ellipsoidal sources embedded in a homogeneous elastic half-space and by applying different inversion methods. Solutions for these types of sources show evidence of a magma body located 6 km beneath the summit craters. Further, more refined and realistic models, based on the results of the analytical inversions, are developed using the Finite Element technique. This method allows to study the effect on the predicted deformation of both Mt. Etna 3D topography and its heterogeneous elastic structure as deduced by recent seismic tomography. The deformation predicted by all the models considered shows a general agreement with the 1993-1997 data, suggesting the primary role of a pressure source. However, major discrepancies are located in the SE sector; these discrepancies cannot be recovered by models characterized either by the real topography or by elastic heterogeneities. This result confirms that tectonic structures and rheological properties may control the observed deformation in the SE sector of Mt Etna. Gravitational sliding along detachment surfaces may have contributed to amplify the deformation during the inflation.
G51A-0066 0800h
Surface deformation associated with the March 1996 earthquake swarm at Akutan Island, Alaska, revealed by C-band ERS and L-band JERS radar interferometry
Akutan volcano, situated in the west-central part of Akutan Island in the eastern Aleutian volcanic arc, is a composite stratovolcano with an active cinder cone inside a circular summit caldera about 2 km across. Akutan is one of the most active volcanoes in the Aleutian arc, producing more than 27 small to moderate explosions from the active intracaldera cone in the last two centuries. The most recent eruptive activity was a series of small steam and ash emissions between March and May 1992. In March 1996, an intense volcano-tectonic earthquake swarm beneath Akutan Island was accompanied by extensive ground cracking but no eruption of Akutan volcano. Radar interferograms produced from L-band JERS-1 and C-band ERS-1/2 images show uplift associated with the swarm by as much as 60 cm on the western part of the island. The JERS interferogram has greater coherence, especially in areas with loose surface material or thick vegetation. It also shows subsidence of similar magnitude on the eastern part of the island, and displacements along faults reactivated during the swarm. The axis of uplift and subsidence strikes about N70°W, which is roughly parallel to a zone of fresh cracks on the volcano's northwest flank, to normal faults that cut the island, and to the inferred maximum compressive stress direction. A common feature of models that fit the deformation is the emplacement of a shallow dike along this trend beneath the volcano's northwest flank. Both before and after the swarm, the northwest flank was uplifted 5-20 mm/year relative to the southwest flank, probably by magma intrusion. The zone of fresh cracks subsided about 20 mm during 1996-1997 and at lesser rates thereafter, possibly because of cooling and degassing of the intrusion.
G51A-0067 0800h
2003-2004 Campaign GPS Geodetic Monitoring of Surface Deformation Proximal to Volcanic Centers, Commonwealth of Dominica, Lesser Antilles.
The Commonwealth of Dominica, located midway along the Lesser Antilles island arc, is home to several (at least eight) potentially active volcanic centers. Spurred by recent seismic crises on the island - in the south from 1998-2000 and in the north in 2003 - twelve GPS monuments were installed in two field campaigns in 2001 and 2003. All twelve sites, along with five of six newly installed sites, were occupied continuously for ~2.5 or more UTC days in 2004 using Ashtech Z-12 dual-frequency, code-phase receivers and choke ring antenna to assess the highly complex and possibly interconnected volcanic systems of Dominica. We examine data from the 2003-2004 epochs because of the highly variable, shallow seismicity preceding this period. This way one can potentially isolate the changes that occurred without the data from previous observations influencing the results. Although only two epochs have been included, data quality and reliability can be established from sites distant from volcanic centers, as such sites show consistent velocities from all three epochs of observation over the 2001-2004 period. Between 2003 and 2004, multiple sites show velocities that are inconsistent with a simple tectonic interpretation of elastic strain accumulation along the plate interface. Sites located in the vicinity of the volcanic centers in the south central part of the island are moving faster than the 3 epoch 2001-2004 average of the velocities, which is approximately 7mm/year. The four sites at which greater movement has been noted have velocities ranging from approximately 10 to 27 mm/year. We note that the largest surface deformation signal is seen in the south during the same period when the shallow seismicity was at a maximum in the north of the island. While the spatial distribution of sites remains sparse and the velocities relatively imprecise, the preliminary results may indicate shallow magmatic emplacement, geothermal fluctuations, or structural instability in that part of the island.
G51A-0068 0800h
Newly discovered subsidence at Lassen Peak, southern Cascade Range, California, from InSAR and GPS
ERS interferograms spanning the 1996-2000 time interval indicate up to 1 cm/yr of subsidence over a ~40 km-diameter area centered about 5 km southeast of Lassen Peak, California. Dry tilt and EDM measurements of the volcanic region completed in 1981, 1982, and 1984 showed no significant deformation, but this interval is too short to accumulate detectible deformation at the 1996 to 2000 rates. Since it is unclear based on InSAR results alone whether the subsidence is a recent or more long-lived process, we resurveyed one dry tilt array and measured five EDM line lengths with GPS in July 2004. Results from the three-station dry tilt array were inconclusive, likely due to an unstable benchmark in the array. Line lengths, however, shortened by up to 14 cm suggesting that the subsidence has been ongoing since at least 1981. Four mechanisms may account for the deformation observed near Lassen Peak: 1) magma withdrawal from a shallow chamber into a deeper crustal reservoir, 2) cooling and/or crystallization of a subsurface magma body, 3) loss of hydrothermal fluids and subsequent cooling, and 4) tectonic motion that is creating a pull-apart basin or other similar structure. The possible sources are similar to those invoked to explain subsidence at Medicine Lake volcano, located 100 km to the north. Geophysical and petrologic studies have found no evidence for a large magma reservoir ($>$ 5 km diameter) in the upper crust beneath the Lassen Peak area; therefore, the first two mechanisms are unlikely. The area surrounding Lassen Peak is known for numerous hydrothermal features, but the relatively shallow depth of the hydrothermal system is not consistent with the spatial extent of the subsidence. In contrast, geologic studies confirm that Basin and Range extension is occurring in the region and the Walker Lane zone of dextral shear passes through the subsiding area. Interaction between these tectonic regimes may result in the formation of a pull-apart basin or other structure that is responsible for the observed subsidence. We will present results of inverse modeling that should help distinguish between these mechanisms, as well as additional interferograms that will better constrain the behavior of the subsidence over time.
G51A-0069 0800h
Volcano-tectonic deformation at Mount Shasta and Medicine Lake volcanoes, northern California, from GPS: 1996-2004
Mount Shasta and Medicine Lake volcanoes are two of the three Cascade volcanoes targeted for dense GPS and strainmeter deployments by the magmatic systems component of Earthscope's Plate Boundary Observatory (PBO). Leveling surveys indicate an average subsidence rate of $\sim$9 mm/yr at Medicine Lake volcano since at least 1954, which could result from draining of a magma reservoir, cooling/crystallization of a subsurface body of magma or hot rock, loading by the volcano and dense intrusions, crustal thinning due to regional extension, or some combination of these mechanisms. Displacements from GPS surveys in 1996 and 1999 revealed regional block rotation and contraction across the summit of the volcano, but the time interval was too short to distinguish between possible mechanisms. On Mount Shasta, a 21-line, 12-km aperture EDM network was measured in 1981, 1982, and 1984 with no significant deformation detected, nor was there significant length change in three EDM lines recovered with GPS in 2000. We present results from GPS surveys completed in June and July 2004 of the region surrounding both Mount Shasta and Medicine Lake volcanoes. We find regional deformation to be dominated by a block rotation about a pole in southeast Oregon, similar to but generally south of poles determined by other workers using GPS in western Oregon and Washington. No significant residual deformation remains in the four GPS stations located on Mount Shasta, which were previously measured in 2000. In contrast, GPS results from six stations on the upper flanks of Medicine Lake volcano confirm the known subsidence and are consistent with elastic half-space models of volume loss that fit the leveling data. No significant residual regional strain was detected. As a result, we believe that subsidence at Medicine Lake does not likely result from crustal thinning due to regional extension. A more detailed examination of Medicine Lake subsidence sources, Mount Shasta edifice deformation, and possible local and regional temporal deformation changes will be available after installation of continuous GPS stations and strainmeters by the Plate Boundary Observatory. In addition, we have begun annual microgravity measurements that in the future should help to distinguish between possible deformation mechanisms for Medicine Lake volcano.
G51A-0070 0800h
Monitoring vertical deformation at Axial Seamount since its 1998 eruption using deep-sea pressure sensors
Axial Seamount is an active volcano on the Juan de Fuca Ridge in the northeast Pacific Ocean. During a 1998 eruption, the center of the caldera subsided up to 3.4 m. Since that time, seafloor pressure measurements made within Axial caldera show evidence that the volcano has been re-inflating. Data from long-term bottom pressure recorders show that the rate of inflation was highest in the months immediately following the eruption (20 cm/mo) and has since declined to a steady rate of ~15 cm/yr. An independent data set was obtained by making repeated measurements at an array of seafloor benchmarks with a mobile pressure recorder from a remotely operated vehicle each year since 2000. These data, which are not susceptible to pressure gauge drift ambiguities that affect the long-term bottom pressure recorders, also indicate uplift is occurring in the caldera at a rate up to 19 ± 3.6 cm/yr relative to a point outside the caldera. These observations suggest that the center of the caldera has re-inflated about 1.4 m (or about 40%) since the 1998 eruption, assuming uplift has been continuous since the 1998 eruption. This translates to a change in the volume of the magma reservoir of 85 x 106 m3 (assuming a spherical reservoir), or an average magma supply rate of 13 x 106 m3/yr. For comparison, this is 4-7 times less than long-term magma supply rates calculated at Kilauea Volcano, Hawaii. The rate of inflation at Axial Seamount suggests a recurrence interval of 16 ± 2 years between eruptions, assuming that it will be ready to erupt again when it has re-inflated to 1998 levels.
G51A-0071 0800h
SAR interferometry analysis of volcanoes in Kamchatka using ENVISAT and RADARSAT-1 data
The Kamchatka Peninsula, Russia, is one of the most volcanically active areas in the world. Large andesitic volcanoes such as Sheveluch, Kluchevskoy, Bezymianny, and Karymsky, are highly active and pose a significant risk, especially to commercial air traffic in the northwestern Pacific. We will present preliminary analysis of synthetic aperture radar (SAR) interferometry (InSAR) from the European Space Agency's ENVISAT and the Canadian Space Agency's RADARSAT-1 satellites for the Kamchatka Peninsula. The significant eruptive activity of these volcanoes in recent years suggests that significant deformation may be expected despite the relatively limited time intervals over which RADARSAT-1 (since 1999) and ENVISAT (since 2003) have been well suited and/or available for deformation studies. Initial results suggest reasonable coherence over less vegetated areas near volcanoes, although significant snow and ice on the higher volcanoes and the long winters in Kamchatka limit the space-time coverage of InSAR for these volcanoes. We anticipate a significant assessment of InSAR as applied to the May 10, 2004 eruption of Sheveluch volcano. If significant volcano deformation is found we will present preliminary numerical solutions for its source.
G51A-0072 0800h
Tilt Recorded by a Portable Broadband Seismograph: The 2003 eruption of Anatahan, Northern Mariana Islands
The horizontal components of broadband seismographs are highly sensitive to tilt, suggesting that widely deployed portable broadband seismic sensors may record important tilt information associated with volcanic eruptions. We report on a tilt episode that coincides with the first historical eruption of Anatahan volcano on May 10, 2003. The tilt was recorded by a portable PASSCAL STS-2 seismograph fortuitously deployed four days prior to the eruption as part of the Mariana Subduction Factory Imaging Experiment. The seismograph, located in an underground insulated chamber about 6 km west of the active vent, recorded continuously throughout the eruption sequence. A long-period signal with a dominant period of several hours was recorded on the EW component beginning at 06:30 GMT on May 10, which coincides with the onset of continuous volcano-tectonic (VT) seismicity and is one hour prior to the eruption time estimated by the Volcanic Ash Advisory Center based on satellite photos. The signal was much larger than a long period diurnal signal presumably resulting from temperature perturbations. A much smaller signal was recorded on the NS component, and the signal was totally absent on the vertical component, suggesting it results from tilt that is radial with respect to the active vent. An estimate of the tilt as a function of time was recovered by deconvolving to acceleration within a passband of 500-50,000 seconds, and dividing by g. The tilt signal records an initial episode of tilt down away from the volcanic center from 06:30 - 09:30 GMT, which we interpret as inflation of the shallow volcanic source. The tilt reverses and records deflation from 09:30 until 17:50, after which any large tilt signal ceases. The period of inflation corresponds to a period of numerous VT events, whereas fewer events were recorded during the deflation episode, and the VT events resumed again after the end of the deflationary tilt. The maximum tilt is about 2 microradians, although it is possible the tilt magnitude is poorly estimated since the signal is far outside the passband of the STS-2 instrument. A scenario relating the tilt to inflation volume and depth will be presented. The tilt signal is not obvious in the raw seismic records, so careful examination of broadband records of other eruptions may disclose further unrecognized tilt signals.
G51A-0073 0800h
Bi-directional volcano-earthquake interaction at Mauna Loa Volcano, Hawaii
At Mauna Loa volcano, Hawaii, large-magnitude earthquakes occur mostly at the west flank (Kona area), at the southeast flank (Hilea area), and at the east flank (Kaoiki area). Eruptions at Mauna Loa occur mostly at the summit region and along fissures at the southwest rift zone (SWRZ), or at the northeast rift zone (NERZ). Although historic earthquakes and eruptions at these zones appear to correlate in space and time, the mechanisms and implications of an eruption-earthquake interaction was not cleared. Our analysis of available factual data reveals the highly statistical significance of eruption-earthquake pairs, with a random probability of 5-to-15 percent. We clarify this correlation with the help of elastic stress-field models, where (i) we simulate earthquakes and calculate the resulting normal stress change at volcanic active zones of Mauna Loa, and (ii) we simulate intrusions in Mauna Loa and calculate the Coulomb stress change at the active fault zones. Our models suggest that Hilea earthquakes encourage dike intrusion in the SWRZ, Kona earthquakes encourage dike intrusion at the summit and in the SWRZ, and Kaoiki earthquakes encourage dike intrusion in the NERZ. Moreover, a dike in the SWRZ encourages earthquakes in the Hilea and Kona areas. A dike in the NERZ may encourage and discourage earthquakes in the Hilea and Kaoiki areas. The modeled stress change patterns coincide remarkably with the patterns of several historic eruption-earthquake pairs, clarifying the mechanisms of bi-directional volcano-earthquake interaction for Mauna Loa. The results imply that at Mauna Loa volcanic activity influences the timing and location of earthquakes, and that earthquakes influence the timing, location and the volume of eruptions. In combination with near real-time geodetic and seismic monitoring, these findings may improve volcano-tectonic risk assessment.
http://www.geodesy.miami.edu
G51A-0074 0800h
Constraints on the geometry of the magma chamber of Mauna Loa volcano, Hawaii, from InSAR data
A period of inflation of Mauna Loa volcano has started in May 2002. Since the beginning of the inflation period the GPS network of the Hawaiian Volcano Observatory has detected a total shortening of ~25 cm of a baseline across the summit caldera. The inflation of the volcanic edifice is commonly explained by the intrusion of new magma into a shallow magma chamber. InSAR data derived from Radarsat imagery provide new constraints on the shallow magmatic system of the volcano. The InSAR data show changes in distance between the ground and the satellite at rates of ~5 cm/yr, depending on the SAR viewing geometry. We present models of the magmatic system to explain the observed surface deformation. The data can not be explained using commonly used models for volcanic inflation, such as point or ellipsoidal source models in an elastic half-space. We discuss alternative models using finite magma chambers that include the effects of the volcano topography.
G51A-0075 0800h
Finding Realistic Dike Models from InSAR Data: Applications to 1998 - 2000 Eruptions at Piton de la Fournaise
Dike intrusions often cause surface displacements that cannot sufficiently be explained by simple analytical models. We developped a method to find and appraise realistic dike models from InSAR data based on the combination of a three-dimensional mixed boundary element method and a neighbourhood algorithm inversion method. Dikes are represented by a quadrangle-like shape with their top reaching the ground surface. The inversion consists of two stages: searching and appraising. In the searching stage, data correlation is taken into account. The appraising stage involves calculations of marginal probability density functions using the samples collected in the searching stage. Synthetic tests confirmed that the method well retrieves the true model with small uncertainties, and showed that all the tested data-subsampling methods are relevant to our problem. In the years between 1998 and 2000, Piton de la Fournaise volcano (Reunion Island) experienced five eruptions, which had been recorded by the RADARSAT-1 satellite. The data set for the February 2000 eruption on the northern flank, made from two ascending and two descending orbits, showed asymmetric displacements indicating an eastward slip of the east side of the eruptive fissures. The acceptable dike models inverted for this eruption share common characteristics such as a moderate seaward dip, a north-south strike, and a subhorizontal bottom line at a shallow depth of 800 to 1000 m, passing beneath the summit Dolomieu crater. The data set for the September 1999 eruption on the summit and southern flank showed, on the contrary, displacements that are symmetric with respet to the eruptive fissures. Preliminary analysis found a very shallow (less than 500 m for the deepest point) and vertical dike model for this eruption.