Finish discussion of event classification

1
9 Feb 2009

• Finish discussion of event classification
• Tectonic earthquake swarms
• Driving mechanisms
• Examples
• Yellowstone
• Mount St. Helens
• Read about Benoit and McNutts global volcanic
  swarm database
• www.geo.mtu.edu/gpwaite/teaching/volcanoseismo/pa
  pers/BenoitMcNutt.Ann.Geoph.1996.GVESDB.pdf

2
Other seismic events/signals

• Rockfalls
• Emergent first arrival
• No clear secondary phases
• Cigar-shaped envelope
• Not harmonic
• Gradual increase and decrease in amplitude
• Can last less than a minute up to an hour
  (multiple rockfalls)
• Can dominate during dome growth
• Signals attenuate rapidly so they arent seen on
  distant stations
• Example observation from MBGA, vertical component
• LP phase is often accompanied by jets of steam
  and ash from the dome
• the source of LP energy is somehow linked to gas
  venting
• the LP event might trigger the rockfall
• Unzen rockfalls modeled as a sequence of forces
  resulting from the removal of a mass of lava from
  the dome

3
Other seismic events/signals

• Ice/glacier quakes
• Shallow, low-frequency
• Can look like LF events
• Mount Rainier gt

4
Lahars and pyroclastic flows

• Recent work by Zobin et al, JVGR, 2009 on Colima
  signals
• Lahars
• Duration of 10s of minutes to hours
• Peak frequencies from 6-8 Hz
• Pyroclastic flows
• Similar to rockfalls
• Durations of a few minutes
• Peak frequencies from 3-4 Hz
• Differences between PFs generated by dome
  collapse and eruption column collapse
• Differences btwn Lahars and pyroclastic flow
  spectra attributed to differences in mechanics of
  flow
• Recognition may be important for hazard analysis
  in real time

5
Source vs. Path Effects revisited

• Difficult to determine for shallow events
• One way to determine if LF signal is due to path
  or source is to examine different events (a VT
  and LP) that occur at about the same location
• Share the same path for most, so any differences
  attributed to source
• Hill and Pitt LV example
• Stacked spectra from 7 stations
• Two events 4 km apart
• Differences unlikely to be path-only

6
Source vs. Path Effects revisited

• Difficult to determine for shallow events
• Neuberg et al. show that large amplitude,
  low-frequency events are more likely to have the
  same spectral peak
• Tell me why?

7
1985 Yellowstone swarm
8
Earthquake Sequences

• Volcano-tectonic earthquake swarms

9
Rate Plot
Long Valley Caldera .05 m/s
Prejean, 2002
10
What causes earthquake swarms?

• What kinds of things could we investigate to
  determine the driver of swarm activity migration?

11
Other notable swarm characteristics

• Buildup to peak rate is different than decay from
  peak rate, suggesting different mechanisms
  (McNutt, 1996)

12
Other notable swarm characteristics

• Buildup to peak rate is different than decay from
  peak rate, suggesting different mechanisms
  (McNutt, 1996)
• Swarms occur in nonvolcanic areas
• Swarms may be modulated by tides
• Swarms may be triggered remotely
• Large, shallow earthquakes have triggered small
  earthquakes at distances of 1000s of km
• Triggering corresponds to arrival of surface wave
  train
• Response to very small stress changes (0.1 Mpa or
  0.01 bar)

13
What causes earthquake swarms?

• Stress transfer (Hill JGR, 1977 model)
• Series of en echelon dikes connected with cracks
• dikes oriented with their long dimensions
  parallel to the regional maximum principal
  stress, ?1
• Shear failure occurs along oblique fault planes
  connecting tips of adjacent dikes when a
  critical combination of fluid pressure in the
  dikes and difference between ?1 and ?3 is
  reached
• Slip on a particular fault will result in an
  incremental volume increase in the immediate
  vicinity, and subsequent fluid pressure drop in
  adjacent dikes. This will stabilize the immediate
  system of dikes and fractures, but will perturb
  the stress field in neighboring dikes triggering
  earthquakes in those systems.
• The stress perturbation in the adjacent dikes
  could induce the same type of earthquakes and the
  stress field perturbation would propagate away
  from the initial site of activity.
• In this model, earthquake activity migrates as a
  result of the stress perturbation and does not
  require transfer of fluids.

14
What causes earthquake swarms?
15
What causes earthquake swarms?
16
What causes earthquake swarms?
17
What causes earthquake swarms?
18
What causes earthquake swarms?
19
What causes earthquake swarms?
From S. Prejean PhD 2002
This model has been used to explain spasmodic
bursts of seismicity observed during many swarms
near Long Valley Caldera
20
What causes earthquake swarms?

• Model for magmatic dike intrusion Rubin, 1995
  Rubin and Gillard, 1998
• Four types of fracture are possible
• (1) slip on existing faults away from the tip
  cavity
• (2) slip on existing faults adjacent to the tip
  cavity
• (3) shear failure of intact rock adjacent to the
  tip cavity
• (4) mode I crack opening
• Magma pressure effectively reduces ?3 and allows
  slip along suitably oriented faults or shears
  intact rock
• Which of the 4 is most likely?

21
What causes earthquake swarms?

• Magma intrusion
• Most dike intrusions do not make it to the
  surface Gudmundsson et al., 1999 so we never
  see most dike intrusions
• Rate and distance of dike propagation depends on
  magma viscosity (?), dike width (w), thermal
  diffusivity of the host rock (?), dimensionless
  parameter (?) that depends on temperature
  difference between the host rock and magma,
  latent heat of crystallization, and the heat
  capacity (Rubin, Ann. Rev. 1995)
• The cooler the magma, the wider the dike must be
  to avoid freezing in the same amount of time
• More difficult to propagate granitic dikes, than
  basaltic dikes
• ? for basalt is 0.5
• ? for granite is 0.9
• Example t10 days, ?1.5 X 10-6 m2/s
• Basalt dike of width lt 1.5 m would freeze
  completely
• Granite dike of width lt2.7 m would freeze
  completely

22
What causes earthquake swarms?

• Magmatic dike propagation distance and rate
  (after Rubin 1995 and Rubin and Gillard 1998)
• Assume w1 m, ?25000 MPa, ?0.2, l1 km
• ?P30 MPa
• Now plug in viscosities for basalt (102 Pa s) and
  granite (107 Pa s)
• vbasalt 100 m/s (8.5 km/day)
• vgranite 1 mm/s (8.5 cm/day)
• How far will the dikes get in 10 days?
• Basalt dikes in Hawaii and Iceland 1 - 10s of
  km/day
• What conditions would make it easier for more
  silicic magma to propagate?

23
What causes earthquake swarms?

• Water/gas migration
• Seismicity induced by reservoir impoundment 0.1
  - 1 km/day
• Activity migration used to estimate hydraulic
  diffusivity from the seismic diffusivity (e.g.,
  Sharpiro et al., 1997)
• DSL2/t where L is the distance activity migrated
• Values for DS range from 0.5-50 m2/s
• In injections, DS depends on well pressure
• What are the likely earthquake types associated
  with these kinds of swarms?

24
Diffusivity
25
Diffusivity
26
Yellowstone earthquake swarms
27
What caused this swarm?
28
What caused this swarm?
29
What caused this swarm?
30
1980 Mount St. Helens Earthquakes

• Quiet prior to March 1980
• Swarm began and peaked - explosion
• Settled into relatively constant rate

31
1980 Mount St. Helens Earthquakes
32
1980 Mount St. Helens Earthquakes

• Quiet prior to March 1980
• Swarm began and peaked - explosion
• Settled into relatively constant rate
• Early earthquakes spanned range of depths and
  distances from volcano
• Later events directly beneath (within) cone

33
1980 Mount St. Helens Earthquakes
May 17
Feb 15
34
1980 Mount St. Helens Earthquakes
May 17
May 17
Feb 15
Feb 15
35
1980 Mount St. Helens Earthquakes

• Quiet prior to March 1980
• Swarm began and peaked - explosion
• Settled into relatively constant rate
• Early earthquakes spanned range of depths and
  distances from volcano
• Later events directly beneath (within) cone
• Syneruption and posteruption activity is much
  more intense!

36
1980 Mount St. Helens Earthquakes
37
1980 Mount St. Helens Earthquakes
Jun 17
Feb 15
38
1980 Mount St. Helens Earthquakes
Jun 17
May 17
Apr 17
Mar 17
Feb 15
39
1980 Mount St. Helens Earthquakes
May 17
Feb 15
40
Volcano Tectonic Earthquake Swarms

• Types of swarms defined by temporal distribution
• Most common is the case where activity increases
  to a peak just prior to eruption (Zobin)
• Sometimes have multiple peaks, but with largest
  peak near the beginning of the swarm (MSH 1980)
• Otherwise there is a single peak in the middle of
  the swarm (like 1985 Yellowstone example) and
  decrease in activity before eruption

41
Volcano Tectonic Earthquake Swarms

• Types of swarms defined by temporal distribution
• Most common is the case where activity increases
  to a peak just prior to eruption
• Sometimes have multiple peaks, but with largest
  peak near the beginning of the swarm (MSH 1980)
• Otherwise there is a single peak in the middle of
  the swarm (like 1985 Yellowstone example) and
  decrease in activity before eruption
• Pre-eruption swarms vary in length from just a
  few hours to thousands of hours (hundreds of
  days) (remember swarm duration is subjective)
• Short swarms more common for basaltic systems.
  Why?
• For more on earthquake swarms http//kiska.giseis
  .alaska.edu/dbases/swarmcat/GVESD.HTML

42
Benoit McNutts swarm database

• Swarm data from volcanoes all over the Earth
• Not comprehensive or uniform, but instructive to
  study
• Swarms not associated with eruptions less likely
  to be reported
• Volcanic monitoring networks vary in minimum
  magnitude of completeness
• Duration of swarm defined differently
• Not all the same data area available for all
  swarms (max magnitude, event types, b-values,
  etc.)

43
Benoit McNutts swarm database

• Swarm type defined based on temporal relationship
  to eruption
• Type IV are eruptions w/o swarms

44
Benoit McNutts swarm database

• Swarm duration is reported for most swarms

45
Benoit McNutts swarm database

• For swarms that precede eruptions, 9 days is
  common duration

46
Benoit McNutts swarm database

• Database useful for examining trends
• Type 1b swarms
• Short duration swarms (lt100 hours) are not
  correlated with eruption duration
• Long duration swarms of 100 days or longer are
  positively correlated with eruption repose time
• What does this mean?

A new tool, WOVOdat, may enable much more
detailed, comprehensive investigations of
similarities, trends, etc., in volcanic eruption
data
47
Benoit McNutts swarm database
48
McNutts Generic Volcanic Earthquake Swarm Model