• Broadband Seismology at the Alaska Volcano Observatory, 1993-1997
• Chemical characteristics of the Aniakchak Caldera ash-flow sheet
• Chemically and Temporally Distinct Magma Series at Aniakchak Volcano and the Role of Crustal Mixing
• Crystal Clots in the Lavas of the Makushin Volcanic Field: Implications for Cumulate Entrainment
• Deep Long-Period Events Associated with the 1992 Eruptions of Crater Peak Vent, Mount Spurr, Alaska
• Mafic Enclave Formation and Effusive Eruption at two Aleutian Arc Volcanoes
• Monitoring and Analyses of Volcanic Activity Using Remote Sensing Data at the Alaska Volcano Observatory
• Near-Real-Time Volcano Monitoring via the Internet at the Alaska Volcano Observatory
• New Geologic Map of Aniakchak Caldera, Alaska
• Real-Time Monitoring and Eruption Warning Responses by the Alaska Volcano Observatory
• Seismicity in the vicinity of the Katmai Group of volcanoes, Katmai National Park, Alaska; July 1995-March 1997
• Seismic Monitoring at Aniakchak Volcano, Alaska
• Sensitivity of Input Parameters in the "Puff" Ash-Cloud Tracking Model Using Satellite Image Observations
• Tsunami generation during the 3500 yr B.P. caldera-forming eruption of Aniakchak Volcano, Alaska
• Volcanic tremor and ground-coupled airwaves observations during the 1996 eruptions of Pavlof Volcano, Alaska, and their implications for source locations

Broadband Seismology at the Alaska Volcano Observatory, 1993-1997
S.R. McNutt, J. Benoit, D. Christensen, S. Estes, G. Tytgat, S. Stihler, S. Weimer, A. Jolly, M. Robinson, R. Hansen, K. Lindquist, M. Garces, J. Lahr, R. Hammond, J. Paskievitch, and J. Power

AVO scientists have installed broadband seismometers at five volcanoes during 1993-1997 to increase bandwidth and dynamic range. Single stations were deployed at three volcanoes for reconnaissance studies; 3 stations were used to augment a 10-station network; and 1 station was deployed for a short-term topical study. One instrument (CRB) was deployed at Mt. Spurr, Alaska, 5 km from the Crater Peak vent, from July 1993-June 1995, beginning only 3 months after Spurr's last eruption. 12-bit digital data were telemetered to Fairbanks. Numerous earthquakes and glacial events were observed, but no true long-period (T>>1 sec) volcanic events were noted. This instrument was moved to Augustine volcano, Alaska, in September 1995 and will remain as a permanent station (AUB) 1 km from the active vent. 12-bit data are telemetered to Fairbanks and recorded in both continuous and event-triggered modes. Numerous earthquakes have been recorded during this background time at Augustine, but no long-period events have been detected. A second instrument was installed at Akutan volcano, Alaska, following a strong earthquake swarm there in March 1996. This instrument (AKT) is 13 km from the vent, but only 5 km from the swarm locus. Data were telemetered in analog form from March 1996 until May 1997, and as 12-bit digital data thereafter. Many swarm earthquakes were recorded, but again no long-period events were observed. Three instruments were deployed to augment a network of 10 digitally telemetered broadband stations around the summit of Kilauea caldera, Hawaii. True long-period events (T>10 sec) occurred during the February 1, 1996 earthquake swarm and during a similar swarm in January-February 1997 (B. Chouet, pers. comm.). These events were clear on filtered seismograms but were quite small. In contrast, strong tremor and events known locally as whooshes and chugs were recorded 2.3 km from the vent during 2 weeks at Arenal volcano, Costa Rica, in April 1994. The tremor consisted of as many as 7 integer harmonics with frequencies >1.5 Hz. However, the only significant energy recorded below 0.7 Hz was due to microseisms, even though the volcano was erupting at the time. Our experience to date suggests that broadband signals are rare and small when they occur, and their value in monitoring or forecasting eruptions has yet to be optimized. However, in Alaska we are well posed to catch a long-period presursory sequence to a large explosive eruption, should one occur. Furthermore, many Alaskan volcanoes have high noise levels, so the high dynamic range of the broadband instruments is always helpful. Finally, digital receiving telemetry equipment is being installed in Anchorage, which will permit eventual upgrading of stations on Cook Inlet volcanoes, and recent analog station installations in remote areas have also been designed to permit eventual upgrading with broadband seismometers.

The Aniakchak caldera ash-flow sheet, which erupted about 3400 BP, is one of spectacular mobility and chemical zonation (Miller and Smith, 1978) and may be implicated in global climate anomalies of about the same period (Beget' et al., 1992). Both the magmatic processes in the source chamber and the volatile contribution of the eruption to the atmosphere are therefore of considerable interest.

Through much of its extent, the sheet is matrix supported, poorly sorted, and unstratified. Induration ranges from nonwelded to incipiently welded. The deposit contains at least two discrete compositions of pumice. The lower section contains mostly more silicic (~65-71 wt.% SiO2), lighter-colored clasts. Darker, more mafic (~58-60 wt. % SiO2) pumice blocks are present throughout much of the lower unit; dark-and-light banded pumice is also present. The upper portion of the tuff sheet, where it has been preserved, is nearly entirely composed of the darker, more mafic pumice. The change from light to dark pumice is abrupt but appears to be time-continuous, implying a single, stratified magma body immediately prior to eruption. However, the unsorted, matrix-supported nature of the sheet indicates a turbulent depositional environment. In light of this apparent surface chaos and the possibility of subsurface mixing as well (Spera, 1984), it is remarkable that such a distinct separation of dacitic and andesitic pumice could be maintained.

Preliminary electron microbe data from hand-picked andesine phenocrysts in silicic pumice indicate SiO2 contents of the melt inclusions of 70-71 wt. %. Except for slight enrichment in Na2O and depletion in FeO, they are remarkably similar to bulk compositions of the most silicic pumice. Initial estimates for the pre-eruptive water (by difference) and chlorine contents are 3.3 to 5.9 and 0.18 to 0.23 wt. % respectively. These volatile values are thus at least broadly similar to neighboring 1912 Novarupta rhyolite, although the melts are dramatically less evolved than even that of Novarupta dacite. Infrared spectroscopy will provide CO2 contents and improved H2O analyses for both melt inclusions and degassed glasses.

Chemically and Temporally Distinct Magma Series at Aniakchak Volcano and the Role of Crustal Mixing
C.J. Nye, T.P. Miller, and P.W. Layer

Two dramatically different magma series have erupted at Aniakchak Volcano in the eastern Aleutian arc. The calcalkaline (CA) series is characterized by a 2.5x increase in REE and HFSE (e.g. La, Sm, Nb, Hf, Y) and FeOt/MgO, and a 3x increase in LILE (K, Rb, Cs), Th and Pb, over a SiO2 increase from 52 to 70 wt.%. CA magmas have been erupted over at least the last 80 ka, and include all caldera-forming and post-caldera units (there was an older CA phase from more than 850 to less than 600 ka). The tholeiitic series (TH) is also characterized by a 2.5x increase in REE, HFSE, and FeOt/MgO, as well as LILE, Th and Pb, but over a SiO2 increase of only 52 to 54 wt.%. TH magmatism dominated volcanism from more than 450 to less than 240 ka (between the CA episodes). Mafic members of the two magma series are compositionally similar. However, evolved members are quite distinct, and few evolved samples are transitional between the two magma series, nor is there temporal overlap.

Trace-element systematics (e.g. K/Rb vs Rb) includethat both series are heavily contaminated. The TH series contaminant is andesitic while the CA series contaminant is rhyolitic. In CA magmas Ni drops to 0 ppm at 57% SiO2, then increases to 10 ppm at 70% SiO2. The rate of increase of Ni in the CA series is much greater than that of highly incompatible elements such as Rb, thus the CA trend cannot be produced by fractional crystallization. Ni concentrations of Aleutian anatectic hi-Si rhyolites are appropriate for a mixing endmember. The contaminant in the TH series is less-well understood. It cannot be average Aleutian andesite, but may be an intermediate partial melt of mafic to ultramafic rocks in the lower crust.

Crystal Clots in the Lavas of the Makushin Volcanic Field: Implications for Cumulate Entrainment
A L Roach and J C Eichelberger

Makushin volcano is located on Unalaska Island in the central portion of the Aleutian island arc. The predominantly andesitic Makushin cone is surrounded by several mafic flank vents. Lavas derived from Makushin and its flank vents contain distinctive crystal clots which are interpreted to be entrained cumulate material. Minerals that compose crystal clots are larger and more primitive than coexisting phenocryst phases. Chemical disequilibrium between the crystal clots and the host lava further suggests that the clots are of an origin external to their present host.

Clots found in the eruption products of flank vents are larger and more abundant than those from the central vent. Further, flank vent crystal clots are strikingly monomineralic while those in the central vent may be composed of any combination of clinopyroxene, plagioclase, olivine, orthopyroxene, and magnetite. The differences in mineralogical assemblage of these cumulates apparently reflects variations in the types of material available for entrainment. The low viscosity of the basalt in the flank magma chambers may enhance adcumulate growth. In contrast, the larger, evolved chamber of Makushin volcano is a rich source of diverse cumulate materials.

In extreme cases, crystal clots compose up to 40 percent of a rock and strongly influence whole rock geochemistry. Basaltic pyroclastic deposits derived from Pakushin volcano, a polygenetic flank cone on Makushin, have anomalously high Mg, Cr, and Ni contents attributable to entrained clinopyroxene and olivine. Extreme samples contain >10% MgO, >500 ppm Cr, and >80 ppm Ni. In contrast, effusive deposits from the same vent contain plagioclase crystal clots and have Al2O3 contents >19%. Thus a deposit from Pakushin may appear to be either a high-Mg or a high-Al basalt depending on its entrained cumulate content.

At least two major physical factors correlate with the abundance of entrained cumulate material: fluid density and eruptive flux. These parameters suggest that the retention of entrained material in a moving fluid may be approximated using Stoke's Law. Mafic magmas, with a high fluid density, are more likely to convey entrained material to the surface.

Deep Long-Period Events Associated with the 1992 Eruptions of Crater Peak Vent, Mount Spurr, Alaska
J.A. Power, A.D. Jolly, M.L. Harbin,and C.J. Nye

The 1992 eruption sequence of the Crater Peak vent of Mount Spurr initiated seismic activity at depths of 20 - 40 km , which consisted of a mix of Long-Period (LP) events, Volcano-Tectonic (VT) earthquakes, and Hybrid events. A few VT earthquakes were recorded in this depth range beginning 10 months prior to the first eruption on June 27, 1992. Activity increased following the June and August eruptions, and reached a peak about 30 days after the September eruption and preceded a strong shallow earthquake swarm in mid November attributed to a magmatic intrusion. Seismicity rates declined during late 1992 and early 1993 and continued at low levels through 1996. LP events in this depth range have poorly defined phases and extended codas, while VT's show well defined P and S phases and shorter durations. Stacked velocity spectra show a strongly peaked spectra between 1 and 3 HZ for LP events, while VTs have a broad spectra with significant energy from 1 to 8 HZ. Magnitudes of the largest located LP, VT, and Hybrid events at greater than 20 km depth are 1.9, 1.6, and 2.0 respectively.

1992 Crater Peak magmas are andesites containing about 57% SiO2, and are chemically homogenous throughout the eruption sequence. Compared with previous Mount Spurr magmas (and Aleutian andesites in general) the 1992 magma had high concentrations of Al, Na, and Sr, suggesting that the 1992 andesite was fractionated from basalt at greater-than-normal P and/or PH2O. The 1992 andesite is also different from that erupted in 1953, having lower concentrations of many incompatible elements yet higher silica contents. These results indicate that the chamber feeding the 1992 eruption was at mid- to lower- crustal depths and separate from that which fed the 1953 eruption.

Several independent lines of evidence suggest that seismic activity at 20 to 40 km depth defines the lower end of a magmatic conduit dipping to the southeast from the Crater Peak vent which fed the1992 eruptions. Possible source processes for deep seismicity include migration of magmatic fluids and brittle fracture resulting from stress perturbations caused by the movement or removal of magma from these depths. Factors contributing to deep seismicity at Spurr as opposed to other arc volcanoes include a relatively deep, more mafic magma source and the apparent short residence time in the upper crust.

Mafic Enclave Formation and Effusive Eruption at two Aleutian Arc Volcanoes
M L Coombs, D G Chertkoff, and J C Eichelberger

An apparent association between mafic enclave formation and effusive eruptive style is exemplified by two Aleutian arc volcanoes, Mount Dutton and New Trident volcano, where we have found evidence of introduction of mafic magma just prior to effusive eruption. Both the Holocene dacitic dome-building events at Mount Dutton (Miller et al., 1997) and the cone-building, andesitic-dacitic eruption of New Trident volcano in 1953-1963 produced enclave-bearing lava. Enclaves from both volcanoes were introduced into host silicic magma as mostly liquid mafic magma as evidenced by their diktytaxitic texture and the presence of chilled rinds on some large enclaves. However, whole rock chemistry of enclaves from the two volcanoes show that the origins of the liquids differ. Dutton dome lavas range from 62-65 wt.% SiO2 and enclaves range from 51-54 wt.% SiO2. The least silicic are high-alumina basalt. New Trident lava flows range from 58-65 wt.% SiO2 and enclaves range from 55-57 wt.% SiO2. Variation among Dutton enclaves result primarily from bulk assimilation of small amounts of dacite host. This is supported by the presence of host derived xenocrysts within enclaves and the fact that bulk compositions of more silicic enclaves lie on mixing lines between bulk compositions of primitive enclaves and host dacite. In contrast, New Trident enclaves are both more silicic and much lower in compatible elements such as Ni and Cr, indicating that if they do represent once primitive basalt, it underwent significant modification, probably through a crystal-melt segregation process at the magma reservoir's mafic-silicic interface.

While the process of enclave formation may differ in these two cases, evidence shows that at both volcanoes enclaves were introduced just prior to effusive eruption. The time between injection and eruption was not so short as to allow uncrystallized mafic melts to reach the surface, as represented by dramatically banded pumice in eruptions elsewhere. However, disequilibrium phenocrysts in Dutton dacites indicate insufficient time between mixing and eruption to destroy unstable mineral assemblages. Enclaves of both volcanoes contain abundant, microlite-free glass in contrast to the microlite-rich nature of the host groundmass. We attribute the microlite-free character of enclave glass to rapid cooling of a high-temperature, nuclei-poor melt within the crustal reservoir, followed by quenching at the surface by eruption shortly thereafter. The enclave melt experienced insufficient resident time to complete chemical equilibration within the melt phase or to form nucleation sites prior to eruption.

Enclave abundance in the New Trident flows increases from < 1% in the earliest flow of 1953 to 5-10% in the 1959 flow; this could be due to a continual influx of enclave material during the eruptive sequence. (Or, alternatively tapping of lower, enclave-richer magma during the later stages of eruption.) Continuing intrusion of basalt into the chamber during eruption has been suggested at Unzen volcano, Japan, based upon eruption of disequilibrium phenocryst assemblages over a duration an order of magnitude longer than expected chemical equilibration times (Nakamura, 1995). We speculate that slow influx of basalt into the chamber maintains the large DeltaT and DeltaC conditions required for enclave formation (Bacon, 1986) and for second-boiling (Eichelberger, 1980) that favors incorporation into the less dense silicic layer. This gradual replenishment is then compensated for by slow exit of freshly formed enclave-laden magma from the chamber with attendant degassing during ascent and resultant effusive eruption.

Monitoring and Analyses of Volcanic Activity Using Remote Sensing Data at the Alaska Volcano Observatory
D.J. Schneider, K.G. Dean, K. Engle, and S.L. Worley

There are about 100 potentially active volcanoes in the North Pacific Ocean Region (NPOR), which includes Alaska, the Kamchatka Peninsula, and the Kurile Islands, but fewer than 20 are monitored seismically. The NOPR averages about five eruptions per year, and more than 10,000 passengers and millions of dollars of cargo fly the air routes in this region each day. The Alaska Volcano Observatory (AVO) uses a variety of remote sensing data and techniques to monitor and analyze volcanic activity in the region. Remote sensing is a valuable tool which compliments other monitoring methods.

Both real-time and archival data are analyzed. Real-time data includes satellite imagery (AVHRR, HIRS, TOMS, GOES, GMS) and ancillary global gridded atmospheric data (UNIDATA). These data are routinely analyzed twice each weekday, and many times each day during crisis situations. Archival data are primarily Synthetic Aperture Radar (SAR), AVHRR, and an extensive collection of vertical aerial photography.

AVHRR and HIRS data are received in real-time at the Geophysical Institute, University of Alaska Fairbanks (UAFGI), and are available to AVO scientists within minutes after a satellite pass, both in full data format and as a graphical product on the world wide web. GOES and GMS imagery are acquired in full data format via the internet, and are routinely available within 90 minutes of collection. TOMS data from the Earth Probe satellite are provided to AVO by the NWS in Anchorage, via a NASA-sponsored receiving station. Global gridded wind and temperature forecasts are received twice per day via the UNIDATA program. This dataset is used to estimate the height of eruption clouds observed in satellite imagery, and as an input to PUFF, a volcanic ash trajectory and dispersion model developed at AVO.

The SAR data includes ERS, JERS and RadarSat data received at the Alaska SAR Facility, located at the GIUAF. Although these data are not available in real-time, they are a valuable source of historical data for mapping volcanic features and landforms in this frequently cloudy region. In addition, SAR is capable of detecting centimeter-scale crustal deformation. The application of this technique to volcano monitoring in Alaska is being investigated.

Near-Real-Time Volcano Monitoring via the Internet at the Alaska Volcano Observatory
L.K. Queen, C.J. Nye, J.P. Benoit, K.G. Dean, K. Engle, K.G. Lindquist, M.S. Servilla, and W.R. Hammond

The Alaska Volcano Observatory (AVO) uses an internal World Wide Web internet site as an essential part of its volcano monitoring program. The top page contains links to near-real-time seismic, satellite, and weather data as well as interactive log sheets and calendars. The page permits rapid distribution of diverse data between AVO's geographically distributed offices as well as off-site monitoring using home computers and modems. The latter feature is especially useful during protracted periods of around-the-clock monitoring.

Seismic data include plots of Real-time Seismic Amplitude Measurement (RSAM), spectrograms, reduced displacement, and synthetic helicorder traces. All these are automatically written every few minutes to graphics files viewable by common web browsers. Together, the plots permit detailed inspection of many facets of the seismicity, such as unfiltered and filtered amplitudes and frequency contents of the signals. The seismic displays make use of multiple hardware and software paths including PC Willie systems and the Unix-based automated seismic data processing package, ICEWORM, and Matlab. AVHRR satellite images are geometrically and radiometrically corrected, subsectioned, and the contrast is enhanced and a coastline embedded. GIF images of each file are posted. The subsectioning allows full-resolution image files small enough to be retrieved over modems. These processing techniques are automated and the images are available for viewing within 10 minutes of each satellite pass. Images of bands 2, 3, 4, and 4 minus 5 (useful for detecting volcanic ash) are produced for seven subsections of each of about 20 passes per day. The most recent images are posted, and a two-day archive preserved. NWS weather data, which aid in understanding seismic signals (e.g. ground-shaking by storms) and interpreting pilot and remote observer reports, are linked to the page. In addition the page is linked to automatically-generated output of the ash-plume dispersal model, PUFF.

Perl scripts are used to generate a log sheet of comments made by remote observers and an interactive calendar used for coordinating volcano watch duty. These scripts permit observations to be scheduled, reported, and discussed from diverse locations using standard web browsers.

Planned additions and improvements are legion.

The Alaska Volcano Observatory is a cooperative program of the US Geological Survey, UAF Geophysical Institute, and Alaska Division of Geological and Geophysical Surveys.

New Geologic Map of Aniakchak Caldera, Alaska
CA Neal, RG McGimsey, TP Miller, CJ Nye, CR Bacon, and TJ Felger

A new, 1:24,000 scale geologic map of Aniakchak Caldera portrays the recent eruptive history of the active 10-km-diameter, 0.5-km-deep caldera located 670 km southwest of Anchorage, Alaska. The map is compiled from several GIS databases and is one in a series of ongoing geologic investigations and volcano hazard assessments by the Alaska Volcano Observatory.

Following caldera formation ~3.5 ka ago, Aniakchak erupted andesitic to dacitic magma from more than a dozen intra-caldera vents producing lava domes, scoria cones, tuff cones, small composite cones, blocky lava flows, and maar craters. Vent distribution defines a crude arc that likely reflects a caldera-related ring fault. Two sites of repeated eruptions are Vent Mountain, a 430-m-high cone that is the source of most lava flows on the caldera floor, and Half Cone, the eviscerated remnant of a vent complex consisting of pumiceous deposits from explosive eruptions, as well as viscous lava flows and domes.

Tephra correlation suggests a hiatus of significant explosive activity for ~ 1 ka after caldera formation. The oldest documented pumice-fall deposit (~2320 14C yr BP) is overlain by a thick sequence of fine-grained tephras which may correlate with a cluster of andesitic tuff cones in the southeast part of the caldera. Above this sequence are at least three dacitic tephras, all < 800 14C yr in age. One of the youngest (~ 400 14C yr) and the most voluminous post-caldera explosive produced ~1km3 of chemically zoned, pyroclastic flow, surge, and fall deposits from Half Cone. Flows and surges from this eruption overtopped the north wall of the caldera and fallout is traceable at least 50 km to the north where it is a 15-cm-thick ash and lapilli horizon. Aniakchak last erupted phreatomagmatic tephra and lava from three vents along the west margin of the caldera in May-June 1931; fallout of fine ash was reported 600 km to the north.

The caldera contains a 2.75 km2, 19.5-m-deep lake which formerly stood as much as 170 m higher than at present and covered more than half the caldera floor. This higher stand - previously deduced from subaerially exposed fine-grained lacustrine deposits, remnant wave-cut terraces on the caldera wall, and the modified morphology of tuff cones - is further supported by subaqueous textures of lava domes and indurated tuffaceous sediments identified in the western part of the caldera. The lake drained catastrophically sometime before ~500 14C yr BP (Waythomas et al., 1996, GSAB v. 108, n. 7); pending 14C data should further constrain the age of the flood.

Real-Time Monitoring and Eruption Warning Responses by the Alaska Volcano Observatory
TEC Keith

The Alaska Volcano Observatory (AVO), a cooperative program of the U.S. Geological Survey, the University of Alaska Fairbanks Geophysical Institute, and the Alaska Division of Geological and Geophysical Surveys, is responsible for monitoring the >40 active volcanoes of the Aleutian volcanic arc and giving timely warnings to mitigate volcanic hazards, particularly to the international air carrier community.

In a typical year as many as five volcanoes in the Aleutian arc - some are eruptions are small and short-lived, others continue for months to years. Several large silicic caldera-forming eruptions have occurred during the past 4 ka and the 1912 plinian eruption of Novarupta remains the largest eruption of the 20th century. AVO uses a combination of monitoring techniques dictated by logistics, severe weather conditions, and remote locations. Seismic networks consisting of at least 6 stations have been installed on 16 Alaska volcanoes during the past few years. Data is telemetered in real-time to Fairbanks and Anchorage where it is analyzed and archived. Remote sensing using satellite sensor data processing is used for monitoring volcanic activity and detecting volcanic ash clouds throughout the Aleutian Arc. Data from several satellites are scanned routinely twice daily and more frequently when activity is suspected. Close working relations have been established with Russian volcanologists in Kamchatka where several eruptions occur each year.

The real-time hazards warning system for Alaska has evolved as residents have experienced major explosive volcanic eruptions, most recently Augustine 1986, Redoubt 1989-90, and Spurr 1992. The AVO emphasis is on interactive, 24-hour response and availability to the public during crisis times. Rapid and timely communication from AVO to the public includes a telephone call-down list to critical agencies and facilities located near the volcano and a faxed and emailed Update or Information Release to over 200 locations (federal and state agencies, air carriers, local response contacts, media) giving pertinent information on the actual or anticipated eruption. A Level of Concern Color Code was developed to help relay the status of volcanic activity quickly and simply to the public. AVO, National Weather Service, Federal Aviation Administration, U.S. Air Force, and Alaska Division of Emergency Services have developed a Volcanic Ash Warning Plan for Alaska and the Russian Far East. This plan is mainly for the large number of jet aircraft transiting the North Pacific air routes in one of the world's busiest air cargo areas and exposed to hazardous volcanic ash several times a year. This document gives procedures for complex interagency communications regarding the identification of volcanic eruptions and drifting volcanic ash clouds and the distribution of various products internally and externally to give public warnings. Because of frequent volcanic events in Alaska, this plan is used several times a month and modified every two years.

Recent examples of AVO's successful volcanic mitigation efforts involving populated areas and $billions in property include volcanic activity at Redoubt 1989-90, Spurr 1992, Akutan 1996, Pavlof 1996, Okmok 1997.

Seismicity in the vicinity of the Katmai Group of volcanoes, Katmai National Park, Alaska; July 1995-March 1997
A.D. Jolly, S.R. McNutt, M.L. Coombs, S.D. Sthiler, and J. Paskievitch

The Alaska Volcano Observatory (AVO) re-established a short-period seismic network in July 1995 near the Katmai group of volcanoes: Martin, Mageik, Trident, Novarupta, Griggs, and Katmai caldera. The initial network consisted of 5 resurrected USGS short-period instruments (circa 1987) located in the Valley of Ten Thousand Smokes (here named the VTTS subnet). The VTTS subnet recorded nearly 500 locateable earthquakes and operated until April 1996 when a telemetry failure interrupted data collection. In July 1996 the VTTS subnet was reserviced and 6 new short-period stations were installed in the vicinity of Martin and Mageik (here named the MM subnet). The combined 11 station network recorded over 1500 locateable earthquakes between July 1996 and March 1997, including an intense swarm on October 16-30, 1996 at Martin-Mageik volcano. A second telemetry failure at the VTTS subnet in April 1997 degraded the ability to locate local seismicity. This failure prompted AVO to replace the ageing VTTS subnet with 5 new short-period instruments in July 1997.

Analysis of the October 16-30, 1996 Martin-Mageik swarm is complicated by changes in the recent station history. The addition of the MM subnet lowered the effective detection threshold from ML 1.0 to ML 0.5 at Martin-Mageik and from ML 0.7 to ML 0.5 at Trident, but was unchanged at Katmai caldera. The changes increased the rates of locateable seismicity from about 3 events per week to about 9 events per week at Martin-Mageik and from about 4 events per week to about 7 events per week at Trident. Rates of seismicity remained nearly constant at Katmai caldera (about 3 events per week). During the October 16-30, 1996 swarm, rates of seismicity increased by almost two orders of magnitude at Martin-Mageik ( 225/week) and the b-value increased from 0.92 to 1.54. Martin-Mageik returned to background rates (9/week) but retained an anomalous b-value (1.56). Error estimates for weighted least-squares b-values are about +/- 0.06.

Seismic Monitoring at Aniakchak Volcano, Alaska
WR Hammond, AB Lockhart, CA Neal, RG McGimsey, and JF Paskievitch

In 1997, a new network of 6 short-period seismometers was installed at Aniakchak Volcano, a 10-km-wide caldera ~ 3.5 ka in age, located 670 km southwest of Anchorage, Alaska. This effort is part of the continuing expansion of seismic monitoring in the central Aleutian Arc by the Alaska Volcano Observatory (AVO). A well-established record of numerous, explosive, post-caldera eruptions, including a significant historical eruption in 1931, and its position beneath the heavily traveled jet air routes across the North Pacific made Aniakchak a priority target for new, real-time seismic monitoring by AVO. The network is designed to detect seismic unrest related to volcanic activity, verify and monitor eruptions, and contribute to detection and location of regional earthquakes. This network replaces a single seismometer that was installed in the summer of 1994. Data from this single site had been intermittent, however, several small swarms of volcano-tectonic earthquakes were noted in 1995.

The seismic network at Aniakchak was engineered to withstand extreme environmental conditions including high winds, abrasion, darkness, cold, snow load, and animals typical of the Alaska Peninsula. Electronic components are housed in custom aluminum enclosures designed to provide protection from the severe weather conditions. Robust solar-battery and primary-cell power systems ensure uninterrupted operation during long periods of winter darkness, cloudy weather or volcanic eruption. All stations use short-period sensors and standard analog FM telemetry components with proven long-term reliability. One station inside the caldera is a three-component sensor. Analog signals are telemetered to a receive site and telephone drop at an FAA communications building in Port Heiden (25 km west) for transmission to AVO in Fairbanks.

Sensitivity of Input Parameters in the "Puff" Ash-Cloud Tracking Model Using Satellite Image Observations
Jonathan Dehn, Ken Dean, and Craig Searcy

Advanced Very High Resolution Radiometer (AVHRR) satellite images are used to confirm and improve model simulations that track airborne volcanic ash clouds in the North Pacific Ocean Region. Puff, the model used in the simulations, requires real-time, four-dimensional, windfield data to predict the location of airborne ash. Using observations or estimates of plume height and assumptions of reasonable particle distributions for the eruption, the plumes behavior is modeled through advection, fallout and turbulent diffusion. Model results accurately show the location of ash particles when the simulations are compared to satellite observations, however, the particle distribution may be greater than that shown on satellite images. By modifying the input parameters the simulations can be "tuned" to more precisely match the satellite image observations. The comparison between simulations and satellite imagery helped identify the sensitivity of the model to initial input parameters such as; diffusion coefficient and initial plume geometry. The results also suggest that the optimum input parameters may vary between eruptions.

Tsunami generation during the 3500 yr B.P. caldera-forming eruption of Aniakchak Volcano, Alaska
C.F. Waythomas and C.A. Neal

A discontinuous, pumiceous sand, a few centimeters to tens of centimeters thick, is located up to 15 m above mean high tide within Holocene peat along the northern Bristol Bay coastline of Alaska. The bed consists of medium-to-fine, poorly sorted, pumiceous sand in a 2-m-thick peat sequence and contains rip-up clasts of peat and tephra. The bed is unique in the peat sequence. Major-element compositions of juvenile glass from the deposit and radiocarbon dating of enclosing peat support correlation of the deposit with the Aniakchak caldera-forming event. The distribution of the deposit and its sedimentary characteristics are consistent with emplacement by tsunami. The deposit most likely represents redeposition by tsunami of climactic fallout tephra and beach sand during the ca. 3.5 ka Aniakchak caldera-forming eruption on the Alaska Peninsula. We propose that tsunami were generated by the sudden entrance of rapidly moving, voluminous pyroclastic flows from Aniakchak into Bristol Bay. A seismic trigger for the tsunami is unlikely because tectonic structures suitable for tsunami generation are present only south of the Alaska Peninsula. The pumiceous sand in coastal peat of northern Bristol Bay is the first documented geologic evidence of tsunami related to a volcanic eruption in Alaska.

Volcanic tremor and ground-coupled airwaves observations during the 1996 eruptions of Pavlof Volcano, Alaska, and their implications for source locations
J P Benoit, S R McNutt, M A Garces, and N Husen

Volcanic tremor and ground-coupled airwave amplitudes were measured throughout the Sept. 15 - Dec. 29, 1996 eruptions of Pavlof volcano. Systematic variations in tremor amplitude betw een different stations and components suggest the existence of at least two spatially distinct tremor sources. Volcanic tremor was continuously recorded using a new network of 5 vertical and 13-comp. short-period seismometers. Reduced displacement (DR), a normalized measure of tremor amplitude, was calculated in near-real time. Continuous tremor with a DR>=3cm^2 was recorded from early Oct. to Dec. 4. Three 15-45 min. bursts of stronger activity (>=10cm^2) occurred on Oct. 19, Nov. 4, and Nov. 23. On Nov. 4, a ash plume was observed to ~8km; no observations are available for Oct. 19. or Nov. 23. On Dec. 4 tremor amplitude abruptly declined to pre-eruption levels, then on Dec. 10, ampl itudes rose exponentially for 24hrs peaking at 12cm^2. Pilot reports confirmed a plume reaching 9.5km 18hr s after the strongest activity. Tremor amplitudes decreased over 3 days to below detection levels. On Dec. 26, t he last episode began with tremor amplitudes increasing exponentially, peaking at 20cm^2 24hrs later. D ue to bad weather no visual observations were made. The tremor amplitude remained at ~7cm^2 for 36h rs and then abruptly stopped. The first ground-coupled airwaves from explosions were detected on Sept . 25. The amplitudes and numbers of airwaves varied throughout the eruption with peaks occurring on Sept. 29 (100/day), Oct. 11 (70/day), and late Nov. (150/day). Significant variations in the tremor amplitud e ratio between the radial and vertical components at the closest station corresponded well with the numbers of airwaves detected. When large numbers of airwaves were recorded more seismic power was directed horizontally, suggesting a shoaling of the tremor source. Variations in the amplitude ratios between stations during the De c. 10 and Dec. 26 episodes also suggest the existence of more than one source location for the tremor.