Transcript presentation

Statsei4
Analysis of complex seismicity pattern generated
by fluid diffusion and aftershock triggering
Sebastian Hainzl
System
Toni Kraft
Introduction
A
Closed System = “plate boundary scenario”
Assumption:
tectonic loading + earthquake induced effects
Statistical Earthquake Models:
- long-term mainshock occurrence: Stress-Release model
(Vere-Jones, 1978)
 talk: Bebbington
 poster: Kuehn & Hainzl
- short-term clustering:
ETAS model
(Ogata, 1988)
Epidemic Type Aftershock Sequences
Introduction
B
Open System
= “intraplate scenario”
Assumption:
Examples:
tectonic loading + earthquake induced effects
+ external forcing
- volcano related seismicity
- postglacial rebound
- fluid intrusion
Introduction
In the latter case, statistical modeling has to take care
of the spatiotemporally varying external forcing.
Two examples are shown:
1) Unknown external force:
(Hainzl & Ogata, JGR 2005)
“Vogtland Swarm Activity”
2) Known hypothetical source:
“Seismicity at Mt. Hochstaufen”
1) Vogtland swarm activity
(Hainzl & Ogata 2005)
magnitude
swarm 2000
episodic occurrence
of earthquake swarms:
1896/97, 1903, 1908/09,
1985/86, 2000
time / date
Possible mechanism:
“...fluid overpressure
in the brittle crust”
(Braeuer et al., JGR 2003)
1) Vogtland swarm activity
(Hainzl & Ogata 2005)
Statistical modeling by means of the ETAS model
Each earthquake has a magnitudedependent ability to trigger aftershocks:
f(M) = K exp( a M )
The aftershock rate decays according to
the modified Omori law:
h(t) = (c+t)-p
external triggering
tectonic loading +
pore pressure increase
aftershock triggering
induced stress + pressure changes
1) Vogtland swarm activity
(Hainzl & Ogata 2005)
Method to extract the forcing signal:
fit of the ETAS model by maximum likelihood method
estimation of the ETAS parameter in a moving time window
forcing rate [#/day]
Results:
1.
external triggering accounts only
for a few percent of all events
2.
temporal variation of the forcing
time [days]
signal is correlated with phases of
(i) diffusion-like spatiotemporal migration (Parotidis et al. 2003)
(ii) enhanced tensile components (Roessler et al. 2005)
3.
method is successfully tested for model simulations:
Fluid signal can be reconstructed!
1) Vogtland swarm activity
Unknown driving force:
reconstruction of the spatiotemporal pattern of the
external force is possible
revealed pattern can be compared with competing
source models
Indirect test of seismicity models
2) Seismicity at Mt. Hochstaufen
- spatially isolated activity
- earthquakes are felt since more than 700 years
- seasonally variations
 hypothesis: rainfall induced (Kraft et al., 2006)
2) Seismicity at Mt. Hochstaufen
Analysis of the high-quality data from year 2002
INPUT:
daily measured rainfall
OUTPUT:
earthquake catalog
> 1100 events
> 500 locations
2) Seismicity at Mt. Hochstaufen
2) Seismicity at Mt. Hochstaufen
 lambda=0.3, c=4600 day/bar, D= 0.32 m2/s
 80% rain-triggered & 20% background events
2) Seismicity at Mt. Hochstaufen: RESULTS
rain
pressure
comparison:
pressure increase
&
earthquake rate
2) Seismicity at Mt. Hochstaufen: RESULTS
Coefficient of Correlation as a function of the delay time between
daily seismic rate &
daily rain
2) Seismicity at Mt. Hochstaufen: RESULTS
Coefficient of Correlation as a function of the delay time between
daily seismic rate &
daily rain
daily seismic rate &
pore pressure increase
 high correlation with the pore pressure diffusion model
2) Seismicity at Mt. Hochstaufen:
Summary:
- direct test of the hypothesis of rain-triggered activity
- model yields high correlation with observation
- this suggests that very tiny stress changes are able
to trigger earthquakes