V53C-1413
Tremor Source Location at Okmok Volcano
Initial results using an amplitude-based tremor location program have located several active tremor episodes under Cone A, a vent within Okmok volcano's 10 km caldera. Okmok is an andesite volcano occupying the north-eastern half of Umnak Island, in the Aleutian islands. Okmok is defined by a ~2000 y.b.p. caldera that contains multiple cinder cones. Cone A, the youngest of these, extruded lava in 1997 covering the caldera floor. Since April 2003, continuous seismic data have been recorded from eight vertical short-period stations (L4-C's) installed at distances from Cone A ranging from 2 km to 31 km. In 2004 four additional 3- component broadband stations were added, co-located with continuous GPS stations. InSAR and GPS measurements of post-eruption deformation show that Okmok experienced several periods of rapid inflation (Mann and Freymueller, 2002), from the center of the 10 km diameter caldera. While there are few locatable VT earthquakes, there has been nearly continuous low-level tremor with stronger amplitude bursts occurring at variable rates and durations. The character of occurrence remained relatively constant over the course of days to weeks until the signal ceased in mid 2005. Within any day, tremor behavior remains fairly consistent, with bursts closely resembling each other, suggesting a single main process or source location. The tremor is composed of irregular waves with a broad range of frequencies, though most energy resides between ~2 Hz and 6 Hz. Attempts to locate the tremor using traditional arrival time methods fail because the signal is emergent, with envelopes too ragged to correlate on time scales that hold much hope for a location. Instead, focus was shifted to the amplitude ratios at various stations. Candidates for the tremor source include the center of inflation and Cone A, 3 km to the south-west. For all dates on record, data were band pass filtered between 1 and 5 Hz, then evaluated in 20.48 second windows (N=2048, sampling rate=100 Hz), at 20 second intervals. Root-mean- square (rms) values were then calculated for each window of data. The ratios of these RMS amplitudes were used to investigate the tremor behavior. The ratio changes between tremor and non-tremor events suggest that the sources for episodes were closer to Cone A (and station OKCF) than they were to other locales in the caldera. Methods from Battaglia's PhD thesis (2001) were used as guidelines for a tremor location program based on amplitude decay. Written in MATLAB®, this program can be run in near-real time to estimate the tremor source location and strength. Further refinement is underway, as is an examination of all other days that have suitable data .
V53C-1414
Monitoring and imaging Alaska volcanoes using seismic noise
The utility of ambient seismic noise for monitoring and imaging the subsurface has implications for the understanding of volcanic systems and processes. Recent studies have demonstrated that the cross-correlation of continuous seismic noise recordings at two stations yields portions of the impulse response, or Green's function. The impulse response is the data which would have been recorded had a controlled seismic source been activated at one of the stations and the resulting seismic waves measured at the other station. The idea of cross-correlating seismic noise can be traced back to the spatial auto-correlation (SPAC) method first introduced by Keiiti Aki in 1957. In contrast to the SPAC method, contemporary studies have popularized the use of temporal cross-correlations between pairs of seismic stations. We have conducted an initial study into the use of seismic noise for imaging at Mount Spurr volcano, located 100 km west of Anchorage, Alaska. In part because of its proximity to Anchorage, Mount Spurr is one of the most densely instrumented volcanoes in the network run by the Alaska Volcano Observatory, with three permanent broadband and thirteen short period stations. In addition, data from eight broadband stations exist from a temporary deployment during three months in the summer of 2005. The complete data set, including permanent/temporary and broadband/short period stations, provides good station coverage and makes surface wave tomography using cross-correlated seismic noise recordings feasible at Mount Spurr. Preliminary surface wave tomograms at 2 s period give indications of an aseismic fault to the east of Mount Spurr. We have also applied the cross-correlation technique at other Alaska volcanoes, including Augustine, Iliamna, and the currently erupting Pavlof, as well as a subset of short period data from the network in and around the East Rift Zone operated by the Hawaiian Volcano Observatory. Results at Iliamna demonstrate the variability in the direction of ocean-generated noise over time due to storms in the Gulf of Alaska and the Chukchi Sea. These different data sets give indications of how the cross-correlation method performs in varied noise conditions and geologic settings.
V53C-1415
Multiplets and Detection of Seismic Velocity Changes During the 1998-99 Seismic Series at Deception Island Volcano, Antarctica
During the 1998-1999 Antarctic summer, the pattern of seismicity at Deception Island Volcano changed significantly with respect to previous years. This was characterized by the occurrence of an intense swarm of local earthquakes. More than 2000 local earthquakes were recorded in the period January-February 1999. The average moment magnitude was around 0.5, with the exception of two earthquakes of magnitude 2.3 and 3.4; array analysis was used to establish the hypocentral distribution of the earthquakes. A regional origin of the seismicity, or direct involvement of magmatic fluids, was demonstrated to be unlikely. Most earthquakes in the series were found to be related to small (<0.4 bar) stress changes on small faults which can be explained by lubrication by pressurized fluids (Ibanez et al. 2003). The presence of fluids in the source area could explain the fracturing process and is supported by the observation of hybrid events in the same source region. The study of the spatial distribution of the first motion of the P waves suggests that different source mechanisms acted in a very small volume, and often at the same time. Many distinct families of events with very similar waveforms were identified. When recorded at the same station, the similarity of repeating earthquakes indicates that source and wavepath are the same, and therefore any observed difference in waveforms is related to a change in the medium. Employing Coda Wave Interferometry (Snieder et al., 2002), the families of events identified in the series show a drop in the velocity over a short time period around the two events with bigger magnitudes. These variations may be related to cracking and/or fluid influx. After the seismic crisis, a change in the composition of the fumarolic gases was observed with the appearance of sulfur deposits around the fumaroles, and a clear uplift and inflation of a few centimeters was detected (Garcia et al., 2001). The seismic characteristics and the spatial and temporal behavior of the series, associated with other geochemical, bathymetric and geodetic observations, favor the hypothesis that the seismic series was caused by the stress generated by uplift of the source area due to magmatic injection in depth (Ibanez et al., 2003). The correlation between the observed velocity variations, and their timing with seismological, geochemical and geodetical evidence, supports the influx hypothesis as a causative mechanism.
V53C-1416
Exponential Acceleration of VT Seismicity in the Years Prior to Major Eruptions of Basaltic Volcanoes
The evolution of the seismicity at three basaltic volcanoes (Kilauea, Mauna-Loa and Piton de la Fournaise) is analysed during phases of magma accumulation. We show that the VT seismicity during these time-periods is characterized by an exponential increase at long-time scale (years). Such an exponential acceleration can be explained by a model of seismicity forced by the replenishment of a magmatic reservoir. The increase in stress in the edifice caused by this replenishment is modeled. This stress history leads to a cumulative number of damage, ie VT earthquakes, following the same exponential increase as found for seismicity. A long-term seismicity precursor is thus detected at basaltic volcanoes. Although this precursory signal is not able to predict the onset times of futures eruptions (as no diverging point is present in the model), it may help mitigating volcanic hazards.
V53C-1417
How much is the volcano seismicity directly driven by volcano processes?
To investigate how much of volcano seismicity is related to fluid movement within a volcanic edifice, we analysed seismicity patterns associated to different volcano activities: Vesuvius (1972-2006, dormant), Piton de la Fournaise (2005-2006, 3 eruptions), Etna (1988-2001, 9 eruptions and the 2002-2003 long lasting eruption), 7 dyke intrusions at Piton de la Fournaise (1988-1992). First, we homogenize the different datasets and purify them from local network bias by normalizing the seismicity rate and energy by the catalogue magnitude spreads and the local seismogenic volume. Second, using different techniques, we quantify seismic energy release and foreshock and aftershock rates from volcano seismicity catalogues. Analysis of foreshocks and aftershocks rates allows to distinguish and to quantify the amount of seismicity generated by earthquake interactions from the one directly driven by volcanic processes. Preliminary results show that (i) there is no significant change between the seismicity of a dormant volcano, the inter-eruption seismicity on active volcanoes and the tectonic seismicity: the earthquake interaction levels, as measured by aftershock rates, are comparable in the different cases; (ii) during the last phase of magma propagation within the volcanic edifice, i.e. few hours from the surface lava flow, there is a huge increase in seismicity rate which is dominated by uncorrelated seismicity, i.e. there is a lack of aftershocks â"foreshocks patterns during the dyke propagation phase. This results suggests that for volcano seismicity, as well as for tectonic seismicity, an increase in the external forcing (i.e. magma injection in the volcanic edifice in the first case and the occurrence of a tectonic earthquake in the other one) is followed by an increase in the seismicity rate. Whereas in the tectonic case the observed increase in seismicity rate is driven by events that are correlated in time, i.e. aftershocks, for the volcano driven seismicity the increase in rate is solely due to the uncorrelated component of seismicity. These are positive evidences for the loading rate or/and the fluid-temperature to drive the peculiarity of volcano seismicity rate.
V53C-1418
Melt flow in a conduit and seismic signals time evolution: a laboratory study
A variety of seismic signals have been related to fracturing and magma transport in the volcanic edifice. Previous studies have provided a first experimental support to this association by reproducing fluid generation and migration while recording seismic signals in a layered sample comprising an olivine-MORB-olivine sandwich. Here, we developed a new experimental set up, consisting in a melt reservoir beneath a volcanic conduit, which allowed a much better control on the physical mechanisms taking place and the related seismic signals. Experiments up to 1373K and 300MPa confining pressure were carried out using an internally-heated Paterson gas apparatus especially designed for the measurements of physical properties of rocks. Acoustic emissions were measured during heating of cold pressed synthetic aggregate of MORB powder till complete melting. High frequency events were recorded at 750-960K, corresponding to glass transitions in the MORB glass (Giordano and Dingwell, 2003). A long lasting low frequency event took place at 1200K, corresponding to melt migration in the conduit, followed by high frequency events, related to brittle intrusive mechanisms due to the magma intrusion in the brittle medium at the top of the conduit. Experiments were also carried out on higher (ξ2 order of magnitudes) viscosities aggregates of Fish canyon tuffs, and varying the length of the conduit (from 3 to 15mm). Results highlight the seismic features related to the intrusion of the silicic magma in the conduit. Repeated occurrence both of swarms of high frequency events and long lasting signals at 500-1000K took place, followed from episodes of steady long lasting events at 1000-1273K. A direct relationship between seismic waveforms + spectrograms and physical phenomena can be assessed, by scaling length and frequency. These evidences provide solid and well constrained new experimental insight into magma migration in the lithosphere and the mechanism of dyke emplacement in volcanic edifices
V53C-1419
Gravity Data Confirm Addition of Mass Beneath Mauna Loa Volcano, Hawaii.
We combine deformation data from radar interferometry studies and measurements of changes in gravity at Mauna Loa over time to constrain the nature and time-dependent geometry of deep rift intrusions. We re-occupied 13 existing geodetic sites with a portable gravimeter (LR EG026) in the area of Mauna Loa's summit for two years. These sites are divided into two loops: a) Five sites of Mauna Loa's leveling network that have been occupied for gravity measurements several times since early 1984 by the USGS-Hawaiian Volcano Observatory. b) Eight sites of Mauna Loa's permanent GPS network. To check the quality of temporal gravity changes along the Mauna Loa gravity network, we established a control network of 5 stations from the NOAA Mauna Loa Observatory to BM6697 on the slope of Mauna Kea Volcano. Preliminary results show a gravity change of up to 0.210 ± 0.010 mGal at the Mauna Loa summit (benchmark ML1) between 2002 and 2006. The gravity changes posses a radial symmetry and becomes negligible at about 15 km from the summit. It is possible to fit the data with a point source with a depth of 4.7 km and a radius of 1.1 km (Amelung et al., 2007). The mass of the intrusion is about 0.9 MU (1 MU = 1012 kg).
V53C-1420
Distribution and geometry of magma bodies within Hawaiian volcanic edifices inferred from 3-D seismic velocity and density models
Magmatic intrusions within active Hawaiian volcanoes, e.g., Kilauea and Mauna Loa, often result in measurable surface deformation. However, constraining the source of such deformation is often difficult, and dependent upon assumptions regarding source mechanism, geometry and depth. Estimates for these parameters can be improved by independent constraints on the distribution and geometry of magma bodies within the volcanic edifices. Here, we present seismic P-wave velocity and density models of the onshore and offshore regions around the Island of Hawaii, including parts of Hualalai, Mauna Kea, Mauna Loa, and Kilauea volcanoes, and Loihi seamount. The velocity and density models suggest that the distribution and geometry of magma bodies within the volcanic edifices, indicated by high-velocity and high-density anomalies, might be deeply related to the surface deformation. The velocity model was determined by tomographic inversion of ~200,000 first-arrival traveltime picks of earthquakes and airgun shots recorded by the Hawaiian Volcano Observatory (HVO). The summits of Mauna Loa and Kilauea are underlain by localized high-velocity anomalies of 4.0-4.3 km/s just a few km below the surface, probably indicating shallow summit magma reservoirs responsible for localized inflation and deflation cycles. More extensive high-velocity anomalies of 6.5-7.0 km/s occur beneath the active rift zones of Kilauea, Mauna Loa, and Loihi, and are attributed to intrusive complexes comprising both dense dikes and olivine cumulates precipitated from long-lived deep magma chambers. These deeper bodies are responsible for the outward creep of the volcanic edifices, and probably also upper flank subsidence. Interestingly, the high-velocity bodies are not continuous, but commonly occur as discrete features, which may account for deformation partitioning along the volcano flanks. Kilauea's east rift zone (ERZ) shows two zones of high velocities, one near the summit and upper ERZ and one beneath the lower ERZ, with a region of low velocities beneath the central ERZ. Mauna Loa's lower southwest rift zone (SWRZ) is marked by a prominent high-velocity zone, lacking along the central and upper SWRZ. Mauna Loa's northeast rift zone shows high-velocity materials offset to the south of the current vent trend. High-velocity regions also occur without obvious surface expression, for example, beneath the south flanks of Hualalai, Mauna Kea, and Mauna Loa. These point to buried rift zones, unrelated to the current trends of surface vents on these volcanoes. The density structure of the island and its surroundings is obtained by converting the seismic velocity model to a density model using an empirical relationship. This relationship defines an initial/reference model for the 3-D inversion of onshore and offshore gravity data. The final inversion yields density anomalies that provide additional constraints on material properties. The largest positive density anomalies are observed beneath the summits and upper rift zones of Mauna Loa, Kilauea and Loihi, possibly due to the molten condition of the underlying magma cumulates. In contrast, the high-velocity features beneath the south flank of Mauna Loa yield densities consistent with their velocities, and are thus interpreted to reflect solidified magma cumulates and dike swarms along the old rift zone, resisting flank deformation.
V53C-1421
Surface loading effects complicate the derivation of internal pressure source characteristics from volcano deformation signals
An increasing number of crustal deformation studies relate volcano deformation to internal pressure sources (e.g., magma chambers). However, surface loads such as lava flows provide an additional source of deformation. The initial elastic response due to a load on the surface of the Earth is followed by a visco-elastic response of the ductile crust below the uppermost elastic layer. Thus, a deformation signal recorded in the vicinity of a volcano is often composed of at least two contributors: an internal pressure source (the magma chamber) and a surface load (a composition of previously erupted lava flows) at the extreme the volcano edifice itself. A test case for a circular lava flow on top of a deflating magma chamber shows that the crust will adjust to the load towards final relaxed response. During this relaxation process gradual subsidence occurs that may mistakenly be interpreted as due to pressure decrease in a magma chamber since the deformation pattern of both processes are very similar. This poses a problem when characteristics of a magma chamber are to be derived. We suggest that additional factors posed by the surface load must be considered and constrained by careful measurements and interpretations of observations and eruptions histories to correct the recorded data for composed signal sources. We conclude that neglecting the surface loads (elastic and visco-elastic responses) may lead to incorrect estimates of magma chamber depths (i.e., too shallow) in magma chamber studies. We conducted a case study in which Green's functions are utilized to estimate the final relaxed response due to recent lava flows of the Icelandic volcano Mt. Hekla. We compare the model results to InSAR data and to the deformation pattern induced by pressure changes in a hypothetical shallow magma chamber. The magma chamber is modeled as a point source of pressure utilizing the so-called Mogi model. The final relaxed response fits the deformation due to the point source very well. In fact, the modeling results show almost identical deformation signals in the vertical displacement field directly under the load. In the far field, however, the response due to the Mogi model is up to 16 % below the final relaxed response. Very significant differences show up the horizontal displacement fields with the horizontal displacement in response to the pressure change in the magma chamber being and order of magnitude higher than the final relaxed response. Thus, to identify whether elastic chamber pressure response or visco-elastic deformation due to surface loads is recorded in the data we suggest to carefully look at the horizontal displacement in addition to vertical displacement.
V53C-1422
Inflation of Sierra Negra Volcano Since the 2005 Eruption
Sierra Negra volcano, an active volcano in the western Galapagos Islands, erupted in Oct. 2005, following prolonged and rapid uplift of the caldera floor and several episodes of trapdoor faulting. Deformation data indicate that a shallow (~ 2 km) sill underlies the caldera floor, and the pre-eruptive intrusion rate between the trapdoor faulting and the eruption has been calculated to be 64 x 106 m3/y. Immediately after the eruption, the inflation rate was very fast, averaging ~ 212 cm/y during November, 2005 to February 2006. Since then, inflation of the caldera center has decelerated to a rate of 73 cm/y in 2006 and 44 cm/y 2007. The caldera rim is inflating at only 1.8 cm/y, indicating that the pressure source continues to be very shallow. We interpret the decelerating inflation as being caused by a decreased pressure gradient between the deep magma source and the shallow subcaldera magma chamber; this was driven mostly by depressurization of the shallow body during the 2005 eruption and continuous repressurization as the shallow magma body refills. Seismic activity in February, 2007 coincided with a change to deflation at a rate of 22 cm/y in the southwestern part of the caldera area, near an active geothermal area. This deflation is likely related to the depressurization of the geothermal system, but it is unclear how that is related to the measured earthquake. More recently, the southwestern caldera floor has renewed inflation at 14 to 18 cm/y.
V53C-1423
Ground deformation in La Palma (Canary Islands) detected using Stacking Radar Interferometry
Ground deformation in ocean volcanic islands can reveal deep accommodation of strain at depth due to very different mechanisms. Nowadays, volcano-tectonics and its relation to volcanic activity is a very open research field. Using InSAR measurements, we are able to detect transient and/or persistent patterns of deformation in this kind of volcanic environments. We have studied an 8-year period (1992-2000) using ERS-1/2 SAR images over the island of La Palma, Canary Islands. First, single interferograms have not shown significant deformation over several cycles (or fringes), and second, La Palma is located in a tropical area so interferometric phase measurements would be greatly affected with the existence of tropospheric water vapour and dense vegetation. In order to mitigate those effects, we use a total number of 48 interferogram to obtain a stacked (average) velocity map of the surface deformation covering most of the island, where coherence is preserved. We found two areas of deformation. First, a subsidence signal around 5-6 mm/yr is located in the southern tip of the island in the area of 1971-Teneguia volcano (a recent cinder cone and lava flows). Second, a broader subsidence area around 3-4 mm/yr is located covering the western of the Cumbre Vieja volcano. We present preliminar modelling results about the causative sources. Ground deformation using InSAR is an important tool to reveal the actual state of strain at the surface in an oceanic volcanic island and using jointly with mathematical modelling is able to infer useful information about deep processes in and below the volcanic edifice.
V53C-1424
Magma plumbing system of Akutan volcano, Alaska inferred from InSAR images and numerical modeling
Akutan, one of the most active volcanoes in the Aleutian arc, has experienced over 27 eruptive episodes in the last two centuries. The most recent eruptive activity was a series of small steam and ash emissions from March to May 1992. Between 11 and 16 March 1996, more than 3000 earthquakes occurred in Akutan Island. This increase of seismicity was accompanied by significant surface deformation. We analyze interferograms from ERS-1/-2 and Envisat satellites which use C-band signal and an interferogram from JERS-1 satellite which uses L-band signal. The L-band signal, whose wavelength is about four times as long as the C-band signal, is less prone to vegetation-induced temporal decorrelation. Thus, the JERS-1 interferogram has better coherence and wider coverage. The ERS-1/2 interferograms consist of a number of isolated patches across which the relative offset cannot be determined. We explore the possibility of constraining the relative offsets using the JERS-1 interferogram. To explain the interferograms we model the surface deformation associated with the seismic swarm using a dike subject to a uniform pressure and a uniformly depressurized shallow dipping ellipsoidal spheroid. Combining interferograms before and after the 1996 seismic swarm, we produce a mechanical model to explain deformation observed from 1992 to 2006.
V53C-1425
Continuing Inflation at Three Sisters Volcanic Center, Central Oregon Cascade Range, USA, From GPS, InSAR, and Leveling Observations
Uplift of a broad area centered ~5 km west of South Sister volcano in central Oregon started sometime after fall 1996, accelerated after fall 1998, and was continuing when last surveyed with GPS and leveling in fall 2006. Surface displacements were measured whenever possible since 1992 with satellite radar interferometry (InSAR), annually since 2001 with GPS and leveling campaigns, and with a continuous GPS station since 2001. The average maximum displacement rate from InSAR was 3 to 5 cm/yr during 1998-2001 and ~1.4 cm/yr during 2004-2006. The other three datasets show a similar pattern, i.e., surface dilation and uplift rates decreased over time but deformation continued through 2006. Our best-fit model is a spherical point pressure (Mogi) source located 6.0--6.5 km below the surface and 4.5-5 km west-southwest of the summit of South Sister volcano. Any marginal improvement gained by using a more complicated source shape is not constrained by the data. This same model fits the deformation data for 2001-2003 and 2003--2006 equally well, so there is no indication that the location or shape of the source has changed. However, the source inflation rate has decreased exponentially since 2001 with a 1/e decay time of about 4 years. The net increase in source volume from the beginning of the episode (~1997) through 2006 was 60 × 106 m3 ± 10 × 106 m3. The only unusual seismicity near the deforming area was a swarm of about 300 small earthquakes on March 23- 26, 2004 -the first notable seismicity for at least two decades. Timing of the swarm generally coincides with slowing of surface deformation, but any link between the two, if one exists, is not understood. Similar episodes in the past probably would have gone unnoticed if, as we suspect, most were small intrusions that do not culminate in eruptions.
V53C-1426
New insights on the shape, extension and volume of Stromboli magma chamber
Stromboli volcano is an "open steady-state" dynamic system. Its activity is taken as a reference to identify minor to intermediate volcanic eruptions. The so called "mild" and persistent strombolian activity may be interrupted lava effusions, major explosions and paroxysms often coeval with the generation of tsunamis. We concentrated our work on estimating, by means of fine thermobarometry, the P-T regimes of recently erupted products. This allowed as to infer the extension at depth of Stromboli magma reservoir. Thermobarometric estimates obtained by constructing a grid of selected reactions, indicate that Stromboli magmas normally equilibrate at ~300-150 MPa and progressively degas before being erupted at about 1100 °C. Conversely, the undegassed primitive magma, represented by pumices erupted during paroxysmal explosions, is ejected during nearly istantaneous isothermal decompression of a higher temperature batch (~1130-1200 °C) that equilibrated at ~300-220 MPa. In the light of these estimates, we discussed the possible shapes and volumes of Stromboli magma chamber by considering a sphere, an ellipsoid (geometrically concordant with the regional stress distribution a which approaches pure extensional failure) and a feeder dyke. An ellipspidal chamber is favored since it better explains the occurrence of abundant cumulates, crystal recycling, magma mixing and, more importantly, the viscoelastic response of the reservoir to the stress field and seismic transients. In the light of geometric assumptions based on stress distribution and consistent with the Griffith plus Coulomb criteria for the development of pure extensional fractures, we estimate the Stromboli magma chamber to have a volume of about 9-12 cubic kilometers
V53C-1427
A Possible Magma Reservoir Below Mt. Vesuvius: MT Data Vs Laboratory Measurements
Quantitative interpretation of MT anomalies in volcanic regions requires laboratory measurements of electrical conductivities of natural magma compositions. The study was applied to Mt. Vesuvius (Italy), a dangerous strato- volcano surrounded by a densely populated area, characterised by several violent eruptions and dormant since 1944. Geophysical measurements (MT as well as seismic measurements) revealed an anomaly located at a depth of 5-8 km and interpreted as a magma reservoir. We present here laboratory measurements of magma electrical conductivity allowing a discussion on the nature of the geophysical anomaly and the possibility of magma storage below Mt. Vesuvius. Experiments were performed on three compositions of eruptive products of Mt. Vesuvius, representative of the entire spectrum of differentiation: a tephrite from VIIIth century's eruption ; a phonotephrite from the eruption of Pollena (472AD) and a phonolite from the eruption of Pompei (79AD), from the less to the most evolved product. Conductivity experiments were conducted on melts using an internally heated pressure vessel connected to an impedance spectrometer. The P-T conditions overlap magma storage conditions determined by independent petrologic studies. Dry and hydrous (up to 5.6 wt% H2O) conditions were investigated. The effect of P, T, water content and composition was discriminated for this database. An empirical equation was established to calculate the electrical conductivity of Mt. Vesuvius melts ranging from tephritic to phonolitic. The modified Archie's law permitted the application of this equation to magmas. Considering pre-eruptive conditions determined by independent phase relation studies, we discuss the nature of the electrical anomaly revealed by MT data at 5-8km depth. A scenario similar to one of the three eruptions studied here is unlikely, as their electrical responses do not match. However, the present electrical anomaly is close to the one expected from a phonotephritic magma with high solid fraction. Thus, the electrical resistivities measured in MT studies would be compatible with a low temperature crystal rich magmatic system, consistent with the interpretations of seismic tomography data.
V53C-1428
Influence of surface load variations on the monitoring and behaviour of a volcanic system: Application to Katla subglacial volcano, Iceland.
Surface mass redistribution in the vicinity of volcanoes is a common process. Many volcanoes are covered by ice cap or glaciers which thickness evolves through time because of climate warming or seasonal effects. However most of changes of surface load are even directly linked to the volcano activity. The eruptive products contribute to the building of the edifice: a few millions m3 dome may rise in several months. Stress changes induced by mass redistribution can even be more important in case the edifice gets partially destroyed during the eruptive event. Such load variations around a volcanic edifice act both to induce a surface deformation signal and to produce pressure changes inside and around the storage zone. In a perspective of risk assessment, this deformation signal has to be discriminated from the effect of magma displacement at depth. Besides pressure changes act in turn to change failure conditions and exsolved gas content. We calculated the displacement induced by the variation of the Mýrdalsjökull ice cap thickness, Iceland, where an annual cycle in ice load occurs as well as a gradual ice retreat as a consequence of climate warming. Seasonal vertical displacements measured from 2000 to 2006 at two continuous GPS stations located near the edge of Mýrdalsjökull ice cap fit well to a model of an elastic response to the annual variation in ice load. A comparison of model displacements and observations provides a minimum value of 29 ± 5 GPa for the effective static local value of the Young's modulus. We infer long-term displacements induced compared to GPS measurements used to monitor the Katla volcano lying beneath the Mýrdalsjökull ice cap. A forward model considering an elastic thickness of 5 km can explain a fraction of the uplift recorded from 1999 to 2004, but it cannot account for the observed horizontal velocities. The study confirms that magma inflow is required to explain observed inflation of the Katla volcano 1999-2004. We, then, estimate the pressure change induced inside a magma storage zone by seasonal and long-term ice load variations. Results show that the seasonal effect can induce pressure changes of the order of 0.3 bar inside an underlying magma chamber.
V53C-1429
Unscented Kalman Filter: A Tool for Time-Dependent Non-linear Modeling
The continuous records of permanent GPS sites provide a favorable data set for tracking magma migration. Campaign observations usually provide a better picture of the overall deformation field, at the expense of an unbroken temporal record. Combining these observations provides the best of both worlds. Changes beneath a volcano can be observed through position changes in a GPS network, but distinguishing the source of site motion is not always straight forward. A Kalman filter provides a means for integrating discrete and continuous measurements and for interpreting subtle signals. The Unscented Kalman Filter (UKF) is a non-linear method for time-dependent observations. We demonstrate the application of this technique to deformation data by applying it to GPS data collected at Okmok volcano. The Extended Kalman Filter (EKF) is a common non-linear filtering method that produces results by linearizing the transformation from model parameters to observations. The UKF approaches the problem from a different perspective; it uses a set of "points" that capture the true mean and covariance of model parameters. The points are propagated through the non-linear transformation and the statistics of the model parameters are calculated from the distribution of the transformed points. The result is that the mean and covariance estimates are accurate to second order, while the EKF only produces first order accuracy. This method is useful for both determining model parameters and tracking changes in the model. Seven years of GPS observations at Okmok are analyzed using a Mogi source model and the UKF. The deformation source at Okmok is relatively stable at 3km depth beneath the center of the caldera, which means the surface deformation is caused by changes in the strength of the source. During the seven yeas of GPS observations more than 0.5m of uplift has occurred, a majority of that during the time period, January 2003 to July 2004. The UKF allows us to solve simultaneously for the time-dependence of the source strength and for the location without a priori information about the source.
V53C-1430
Magma Flow and the Redistribution of Crystals in Shallow Intrusions at Sinker Butte Volcano, Western Snake River Plain, Idaho
Sinker Butte is the erosional remnant of one of the largest tholeiitic volcanoes in the western Snake River Plain. Because the edifice was dissected by the Snake River, a nearly complete record of its eruptions is exposed in the walls and alcoves of the canyon. The stratigraphy indicates that initial eruptions produced a shield cone composed of many thin pahoehoe flows. The effusive phase was followed by a series of phreatomagmatic explosions that built a very large tuff cone with a summit crater complex up to 1 km across. A final series of magmatic eruptions filled the tuff cone crater with lava and produced at least 20 radial dikes and other small intrusions, some of which appear to have fed late stage flank eruptions. About 80 oriented drill cores collected from 7 of the radial dikes were analyzed for anisotropy of magnetic susceptibility (AMS). Average magnetic lineations, which are assumed to reflect flow directions, are nearly horizontal for two of the dikes; the others plunge downward and away from the crater area at angles of 45o to 70o. The AMS data are consistent with the propagation of radial dikes outward from a column of magma in the upper part of the edifice, possibly beneath a lava lake. Steeply dipping flow directions in some of the dikes may indicate late downward flow in response to draining of magma during breakouts on the lower flanks of the tuff cone. Crystal clots composed of plagioclase and olivine are abundant in many of the lavas and intrusive sheets, and chemical variations throughout the suite can be attributed to the redistribution of these minerals. Microprobe analyses of olivines in samples in known stratigraphic order indicate at least one recharge event took place during the life of the volcano. It is suggested here that pulses of ascending magma disrupted mushy cumulates along the walls and floors of the shallow plumbing system. Crystal clots in the lavas and dikes are interpreted to be fragments of these cumulates. High abundances of clots in the interiors of some of the radial dikes suggest they were concentrated by flowage differentiation. This process, operating in small conduits throughout the edifice, may have contributed to the overall chemical diversity observed in the suite of lavas and tephras at Sinker Butte.
V53C-1431 [WITHDRAWN]
Upper Mantle Magma Storage and Transport Beneath the Miocene Teno Volcano, Tenerife (Canary Islands)
The nature and dynamics of magma plumbing systems are key variables to the understanding of overlying volcanic edifices. With the exception of a few intensely studied localities, these variables are typically unconstrained at volcanoes around the world. Where attempted, studies of magma storage and transport reveal complex plumbing geometries, for a range of geological settings, indicating that assumptions of shallow, spherical-elliptical magma chambers are often oversimplified. At the highly active, basaltic shield-volcanoes, geophysical monitoring is an effective tool to investigate plumbing system geometries. In the Canary Islands, however, the low eruption frequency results in poor deformation and volcano-seismic data sets and, hence, volcanologists have to rely on alternative methods to study the magma plumbing system of Canarian volcanoes. We use clinopyroxene-liquid thermobarometry [Putirka et al. 1996, Contrib. Mineral. Petrol.], aided by petrography and mineral major element chemistry, to reconstruct the magma plumbing system of the late Miocene Teno shield-volcano, Tenerife. Thin section observations show that the numerous clinopyroxene phenocrysts display darker coloured outer rims, which commonly host acicular apatite microcrystals and sometimes form dentritic protrusions. This is coupled with steep normal Fe-Mg zoning and drastic TiO2 enrichment. Supported by similar Fe-Mg zonations in olivine, this suggests that these rims formed due to decompression induced crystallisation upon rapid magma ascent and accompanying degassing and undercooling. This process took place under disequilibrium conditions, implying that clinopyroxene rim compositions may not always be suitable for thermobarometric investigations. On the other hand, clinopyroxene compositions excluding the outer rims generally appear to be in chemical equilibrium with the melt. Thermobarometry indicates that clinopyroxene crystallisation occurred in the uppermost mantle, mostly from 20 to 40 km depth. The interval between the brittle- ductile transition and the crust-mantle boundary thus appears to provide an efficient magma trap, supporting the concept of progressive magmatic underplating in the Canary Archipelago. The combination of careful sampling strategy, petrography, mineral chemistry and clinopyroxene thermobarometry is a promising tool for investigations of magma plumbing systems at basalt-erupting volcanoes.
V53C-1432
Liquidus tracking by vigorous convection in ascending magma
Basaltic magmas commonly erupt at or near their liquidi and have never been observed to be superheated. In the light of the steep P-T slope of magma adiabats relative to liquidi, superheated magmas should be common. That they are not may reflect a fundamental feature of rapid convective heat transfer in ascending magmas, and that they seem to adhere to the liquidus may also reflect this process. Moreover, this may alleviate the well-known thermal entry length enigma pointed out by Delaney and Pollard that magma under laminar flow in dikes should solidify after a relatively short transit distance. (This is, in essence, because the flow velocity is normal to the thermal gradient and their vector product vanishes, leaving the sheet to progressively solidify by conduction regardless of flow rate.) Key insight on the meaning of the lack of superheat comes from thermal convection studies involving crystallizing fluids. In experiments intended to simulate thermal convection in magmas using analog crystallizing fluids (paraffin, isopropanol-water), a number of studies have found thermal convection to be vigorous only when the magma' is superheated (Marsh, 89'; Brandeis & Marsh, 89'; 90'; Hort et al., 99'). Convection ceases once the superheat is evicted and further cooling is by conduction. Because of the relatively low viscosity and significant length scales of basaltic magmas, the governing Rayleigh number (Ra) for thermal convection is large for almost any appreciable superheat. All the physical features associated with convection can be related to Ra. The rate of convective heat transfer relative to conduction is measured by the Nusselt number (Nu) and, for example, Nu is proportional to Ra to the 1/3. We report here on analytical and numerical results that model this cooling process during magma ascent. The thermal history is a function of two dimensionless numbers: Rao based on the temperature difference between the liquidus at the initial depth and the surface, and the Fourier modulus, which is a dimensionless time measuring the rate of ascent or rate of superheat production. When superheat is available, thermal convection is rapid and cooling is rapid. With approach to the liquidus, convection and cooling wane, but continued ascent attempts to follow the adiabat, which initiates new superheat and the cycle repeats itself. Because the rise time for convection is short and heat transfer highly efficient, for a constant ascent velocity an equilibrium is established between the rates of superheat production and loss due to cooling. The pattern of cooling does not oscillate about the liquidus, but instead tracks the liquidus in a slightly superheated state. Because convection ceases at the liquidus, the cooling trajectory cannot reach or cross the liquidus unless conductive heat loss is also included. That the final temperature of erupting magmas is so often near the liquidus probably reflects the slow rate of conductive heat loss once the liquidus is crossed and the contribution of latent heat with the onset of nucleation, which acts as an internal heat source or an enhanced heat capacity. The faster the ascent rate, the more vigorous the rate of convection. The not uncommon presence in alkali basalts of anorthoclase megacrysts, which may be several cm in size, may reflect this process of large crystals growing and being suspended in vigorous convection, perhaps akin to hailstones, in rapidly ascending magmas.