Pyroclastic Flow Induced Tsunami Captured by Borehole Strainmeter during Massive Dome Collapse at Soufriere Hills Photo: Voight Photo: Voight (1) Department of Geosciences, University of Arkansas Volcano, Montserrat, West Indies: Observations and Theory (2) Department of Geosciences, Pennsylvania State University (3) Applied Fluids Engineering, Inc. (4) Department of Terrestrial Magnetism, Carnegie Institution Washington Elizabeth Van Boskirk (1), Barry Voight (2), Phillip Watts (3), Christina Widiwijayanti (2), Glen Mattioli (1), Derek Elsworth (2), Dannie Hidayat (5) Division of Earth and Ocean Sciences, Duke University Pyroclastic flows entering the ocean at the Tar River (2), Alan Linde (4), Peter Malin (5), Jurgen Neuberg (6), Selwyn Sacks (4), Eylon Shalev (5), R. Steven J Sparks (7), Simon R Young (2) (6) School of Earth & Environment Earth Sciences, Leeds University Valley delta (7) University of Bristol, Department of Earth Sciences 0.8 HARR seismic envelope III Analysis Abstract 0.6 I II a b c IV A 0.4 Figure 9. Spectra for Strainmeters at three Caribbean Andesite Lava Island Precision-Seismo-geodetic Observatory Normalized Amplitude 30 15 Figure 10. Spectra of 0.2 Figure 5. Eruption Seismicity and Dilatation. GEOWAVE simulation 25 1 5 10 35 (CALIPSO) borehole sites recorded dilitation offset by 7-12 minutes from the seismic signals leading 38 20 TRNT strainmeter. ) A: Normalized seismic amplitude for HARR output. 3 200 m up to 12/13 July 2003 dome collapse eruption of Soufriere Hills Volcano, Montserrat, West Indies. 6 Unfiltered 12/13 July 10 surface broadband seismometer. All records Spectrum1d of GMT 100 Pyroclastic flows were observed entering the ocean at the Tar River Valley (TRV) delta at 18:00 local B start at 23:00 (UTC) 12 July 2003. Individual was used to compare 2003 eruption data. Each Cumulative Cumulative 0 Vo l um e ( HARR Z-cmp seismogram 1 sample of the strain is a time (LT) 12 July. The eruption occurred 23:30 LT 12 July, with dome collapse consisting of 210 x 5 pyroclastic flows are annotated on seismic GEOWAVE wave C 0 x 10 sampling of the 106 m3 (DRE). Dilatometers at three sites recorded complex, correlated long period oscillatory wave -1 envelope and major divisions are shown as amplitude outputs with 8 Time [hour UT on 13 July 2003] unobserved signal. packets with periods ~250-500 s. The strongest oscillatory signal was at Trants, the site nearest to HARR spectrogram Roman numerals. B: Cumulative collapse the TRNT dilatation. 6 volume from HARR-normalized amplitude in A. a) Spanish Point b) the TRV delta and the coastline. GEOWAVE was used to model wave initiation and propagation to C: HARR vertical component seismogram. D: Airport c) Offshore test that pyroclastic flows and dome collapse into the ocean at the TRV delta stimulated tsunami. D 4 Spectrogram for seismogram in C. Note that Trants, 250 m from Frequency (Hz) These tsunami are hypothesized to be signal the strainmeter is observing. The spectral power 2 most of power is in 1–3 Hz band and that higher TRNT d) Average of simulated ocean waves were compared to observed strain at Trants. Simulated wave heights Figure 1. 12/13 July 2003 Eruption frequency energy is observed during significant Trants 250 m radii using a) Soufriere Hills Volcano (SHV) taken May 31, 2003. View Table 1. Strainmeter TRNT dilatation collapse events. E: TRNT dilatation highpass 5 gauge files e) Average correlate with observations and demonstrate strainmeter sensitivity to geophysical processes not 400 SSE. E 0 filtered at >0.002 Hz. Peak amplitudes in of Trants 500 m radii data was sampled at envisioned in the design of the installation at SHV. b) Taken August 12, 2003 from the same location. Nanostrain -400 different wave packets TRNT spectrogram dilatation that occur 7–10 min after same events using 6 gauge files. The 3 3000 Approximately 210 M m of the dome collapsed during the 0.008 are recorded as seismic energy at HARR. F: lines signify the throughout the eruption. 2000 July 12/13 eruption/dome collapse. 0.006 Spectrogram for TRNT dilatation in E. Most of correlation of increased 1 b) sample 2 c) sample 3 1000 CALIPSO Borehole Sites c) July 12-13 2003 dilatometer and vertical seismometer F d) sample 4 e) sample 5. 0.004 power Figureis between 3. 0.004 and 0.006 Hz, and spectral power at records from TRNT site, which is 5.8 km from SHV. Frequency (Hz) during Mattiolipeak etcollapse al., 2006 event III, energy is spread associated with a broad Each CALIPSO site is equipped with a three component seismometer (Duke CIW Hz-1kHz), Pinnacle 0.002 Timescale is in seconds with 0 at July 12 00:44 (Mattioli et to higherGeology frequencies (Mattioli et al., 2006). peak at 0.01 Hz in all Technologies series-5000 tiltmeter, Ashtech U-Z CGPS receiver with choke ring antenna with SCIGN 0.000 al., 2004). 00:00 02:00 04:00 06:00 the synthetic gauge time mount and radome, and single component, very-broad-band Sacks-Evertson dilatometer. Time (HH:MM UTC 13 July 2003) series. GEOWAVE GEOWAVE uses a Boussinesq model for wave propagation combined with TOPICS, which is input into the program FUNWAVE, to Sensitivity Test Results simulate the wave initiation and development of a tsunami (Watts and Waythomas, 2003; Watts et al., 2003; Wei and Kirby, 1995). Detailed The GEOWAVE simulation entry bathometry for the region of Montserrat is smoothed to reenact the underwater pyroclastic flow. The equations to generate a pyroclastic flow Figure 11. Wave amplitude parameters that significantly affect the height in time for are exactly the same used as to generate a wave from a subarial debris flow. GEOWAVE handles frequency dispersion, simulating deep- shoreline wave amplitudes are: water waves. The model treats the flow as a solid block that decelerates from an initial velocity. Underwater deformation of the flow is not GEOWAVE simulation. a) Entry at Tar River •final depth and runout length included in the model (Watts and Waythomas, 2003). GEOWAVE simulated runnup and allows enough entry sources to recreate the b) White’s Bottom Ghaut •runout time geometry of the TRV delta. c) Spanish Point •landslide volume d) Airport 50 sec 100 sec •runout time Figure 7. Montserrat Bathymetry. Figure 8. GEOWAVE e) Offshore Trants, 250 m Red stars are CALIPSO borehole simulations. from TRNT In all cases, the values for the sources sites and TRNT, GERS, and AIRS 16.8°N Colors (red for wave f) Average of Trants 250 m of the edge waves (S1 and S2) cause had operating strainmeters during heights above and blue for radii using 5 gauge files g) changes in the gauge maximum wave the eruption. Green circles wave heights below sea Average of Trants 500 m amplitudes. nasa represent entry sources used in level) are scaled to model radii using 6 gauge files. 16.7°N 6.0 4.0 GEOWAVE. The simulated 4.5 3.0 extremes. Simulations are 3.0 2.0 3 pyroclastic flow has five entry 1.5 1.0 for 7 Mm pyroclastic flow 0.0 0.0 meters -1.5 meters -1.0 sources to recreate the geometry of -3.0 -2.0 entering sea at ~70 ms–1 -4.5 -3.0 the entry at the TRV delta. Blue -6.0 -4.0 with volume to width ratio Figure 3. AIRS CGPS site. Figure 4. SHV Monitoring Sites and orange circles represent of 16700 m2. Z max is Conclusions July 14, 2003. Note ~11 cm of ash Left: Map of CALIPSO borehole sites (red 300 sec Z max gauges used in GEOWAVE. maximum wave height GEOWAVE, along with the methods applied by Voight and Watts provide parameter limitations to simulate a accumulation from the July 12/13 squares) and other monitoring stations around Several gauges (orange) are over entire temporal 16.8°N pyroclastic flow entering the TRV delta in which to create tsunami similar to the 12/13 July 2003 dome collapse. Schematics of the very-broad- SHV. placed as a radii around Trants and domain. Eastern band Sacks-Evertson dilatometer Right: Color Infrared image taken by ASTER. the results are averaged in space Montserrat, from TRF to collapse and pyroclastic flow events leading up to the peak collapse. Detailed simulations confirm that the (Mattioli et al., 2002) The TRV delta is the grey pyroclastic flow for a fixed time. The next step is to Trant’s Bay, experienced unique strainmeter observations may be explained by ocean loading from pyroclastic flow generated deposits on the eastern side of SHV (Mattioli et calculate the Green’s Functions for maximum waves of few 16.7°N tsunami (Mattioli et al., 2006). To quantitatively compare the signals of the strainmeter to the simulated 2.0 al., 2004). a load on a uniform elastic half 1.5 4.0 meters, with heights as 1.0 3.0 tsunami, however, one needs to solve Green’s function, which determines the energy transfer for an elastic 0.5 space. We will examine the load 0.0 2.0 high as 14 m localized meters meters -0.5 1.0 half-plane. Also, to further recreate the actual collapse events of 12/13 July 2003, multiple pyroclastic flows over a several km area offshore -1.0 proximal to TRF. (Mattioli -1.5 0.0 Trants since the volumetric strain is -2.0 et al., 2006). causing ocean excitation should be simulated. 62.2°W 62.1°W 62.2°W 62.1°W July 12/13 2003 Eruption Timing directly proportional to the dVof the The largest dome collapse of Soufriere Hills Volcano removed ~210x106 m3 DRE of material.