Transcript Imaging the Magmatic System of Okmok Volcano, Alaska, with Ambient Noise and Tremor Matthew M.

Imaging the Magmatic System of Okmok Volcano,
Alaska, with Ambient Noise and Tremor
Matthew M. Haney
U.S. Geological Survey, Alaska Volcano Observatory
Seismic monitoring in Alaska
Method
Status
Needs
Quake location and mag
Operational
Multi-channel (network)
RSAM
Operational
Single channel
Real-time spectrograms
Operational
Single channel
Real-time helicorders
Operational
Single channel
Tremor detection
Beta
Multi-channel (network)
Ground-coupled airwaves
Beta
Multi-channel (network)
Infrasound array
Beta
Multi-channel (array)
Real-time quake/swarm
Beta
Multi-channel (network)
Beta/Research
Multi-channel (network)
Multiplets
Research
Single channel
Seismoacoustic coherence
Research
1 seismic + 1 infrasound
Detect small velocity changes
Research
Single channel
Tilt from seismic
Research
Single horizontal channel
Double-difference locations
Okmok
Volcano
Lu et al. (2005) JGR
Noise seismograms OKCEr-OKSP
0.3 Hz frequency
0.5 Hz frequency
Ambient noise tomography of Okmok
approximate
edge of caldera
Masterlark, Haney, et al.
(2010) JGR
E-W and S-N slices from 3D model
shallow low velocity zone:
weak caldera infill material
center of deep low velocity zone
(magma chamber) 4-4.5 km depth
below caldera floor
shallow low velocity zone
deep low velocity zone
During the same time period, Fournier et al. (2008) JGR
found a stable Mogi source 3 km below the caldera floor
from a network of campaign GPS instruments.
Why deeper than GPS or InSAR?
Mogi model assumes
homogeneous halfspace. Overlying weak
caldera material makes
the chamber appear
shallower than it is.
Masterlark (2007) JGR
Masterlark, Haney, et al.
(2010) JGR
Backprojection of Volcanic Tremor

Volcano seismicity often characterized by
emergent, low frequency, long duration
events lacking P- and S-wave arrivals

How to best analyze these events with
monitoring networks?

Backprojection of waveforms an attractive
option, but can do even better with small
networks
Backprojection of Earthquakes

Illuminating the source by summing over
stations
Ishii et al. (2005) Nature
Time reverse imaging:
Lokmer et al. (2009) GRL
O’Brien et al. (2011) GJI
Radial Semblance:
Dawson et al. (2004) GRL
Almendros and Chouet
(2003) BSSA
2008 eruption of Okmok
Final RED
of eruption
Larsen et al. (2009) EOS
Okmok seismic stations
2 broadbands:
OKSO, OKFG
5 short-periods:
OKAK, OKSP,
OKWE, OKWR,
OKRE
Several other
stations damaged
by eruption
Reduced Displacement 0.2-0.3 Hz
Izu Ooshima ~
1230-2380 cm2
Pinatubo ~
1070 cm2
MSH ~
260 cm2
McNutt and Nishimura
(2008)
Max DR ~ 300 cm2
Backprojection methodology

Spectral whitening, time shift, and compute
stack power for candidate source locations

At Okmok, virtually no path effects in the 0.20.3 Hz band (Haney, 2010)

Time shifting based on a homogeneous
surface wave velocity model of 2.7 km/s
(Masterlark et al., 2010)
Raw seismograms
Seismograms shifted at tremor location
Array Deconvolution

Problem: Impulse response of modest 7
station network lacks sharp resolution

Solution: Remove impulse response by
deconvolution

Two possible methods:
 Richardson-Lucy, Nishida et al. (2008) GRL
 Non-Negative Least Squares
Deconvolution of array response
Output or image point
i-th Receiver
k-th Source
j-th Receiver
1 hour of typical tremor at Okmok: July 23, 2008
Decon
shows
tremor
north of
Cone D
Backprojection
Predicted
backprojection
Theoretical
backprojection
for single source
Waveform inversion of tremor
OKFGz
OKFGe
OKSOz
OKSOe
OKFGn
OKSOn
8x1013 Nm


Finite difference
modeling/inversion code
by Chouet, Dawson, and
Ohminato
Moment-only solution
dominated by Mzz
E1 = 100 x Var(Misfit)/Var(Data) = 17%
August 2, 2008 tremor episode
Haney (2014) GRL
1-2 hours prior to tremor escalation at Okmok Volcano, 2008
Typical
tremor
location
0-1 hours prior to tremor escalation
Tremor movement
toward caldera wall
Haney (2014) GRL
A
B
C
D
Caldera Wall
Cone D
2008 Cone
Cone D
Lake
Apparent Tremor
Movement
NORTH
Haney (2014) GRL
OKNC-OKCE: late 2012
OKCD-OKCE: 2006
Haney et al. (2015)
OKNC-OKCE cross-correlation
Negative lag
Early
Positive lag
Late
Bennington et al. (2015) in revision
Agreement
between 4
estimates
Haney et al. (2015)
Annual variation of +/- 0.2%
between winter and summer
OKCE autocorrelation
not shown (noise)
Haney et al. (2015)
Annual variation of +/- 0.2%
between winter and summer
Snow
sites
Okmok seasonal variations
due to snow load in winter
Sites
without
snow
Hotovec-Ellis et al. (2014) JGR
High velocity
during eruption
Augustine 2005-2006
Expected
seasonal
variation
Low velocity
+/- 50 days
before and
after eruption
Explosive phase shown by
vertical dashed lines
Conclusions

Unique seismicity and structure at Okmok

Imaging of shallow (< 5km depth) magma
chamber with ambient noise tomography

First demonstration of backprojection for
tracking changing volcanic tremor

Time-lapse changes due to annual snow
cycle and possible magma intrusion at
Augustine
McGary et al. (2014) Nature
Haney and Tsai (2015) in revision
Talkeetna Mts
Alaska Range
Alaska Range
Average Moho
Depth ~32 km
WrangellSt. Elias Mts
Kenai Mts
Chugach Mts
Average Crustal
Vs ~3.4 km/s
Cook Inlet
Basin
Prince William
Sound
Average Upper Mantle
Vs ~4.5 km/s
Slab (?)
Interstation times during escalation
1-2 hours
prior to
escalation
OKFG
During tremor
escalation
OKSO
Location from waveform inversion
Error volume
slices: blue =
less error
2 km
Z
Tremor at
shallow
depth, < 1 km
Z