Diapositive 1 - Earthquakes: nucleation, triggering, and relationship

advertisement
The long precursory phase of many
large interplate earthquakes
Michel Bouchon, Hayrullah Karabulut,
Virginie Durand, David Marsan
With: Mustafa Aktar, Jean Schmittbuhl, Serdar Ozalaybey
The reason why foreshocks occur before some
earthquakes and not others is unknown.
Two mechanisms are proposed to explain their
presence:
- A cascade model where a first shock randomly
triggers other shocks and one of them triggers the
earthquake. The fact that this leads to a large
earthquake is simply the result of a random throw.
- An aseismic slip model where a patch of the fault
begins to slip slowly and the foreshocks are the
breaking of the fault asperities resisting the slow slip
(Jones and Molnar 1979 ; Ohnaka 1992; Dodge Beroza and
Ellsworth 1995,1996 ; Abercrombie and Mori 1996; Zanzerkia
Beroza and Vidale 2003; McGuire et al. 2005).
If the first model is correct, there is no hope to
ever predict earthquakes.
If slow slip is what generates foreshocks, one
may hope that in the future this slow slip will be
directly measurable and that the preparation
phase of some earthquakes will be recognizable
before rupture occurs.
Let us first look at the 1999 Izmit
earthquake.
It is the largest well-recorded
strike-slip earthquake to date and
broke the North Anatolian Fault
(the interface between two
horizontally-moving plates).
Photo A. Barka
Spot image interferometry of Izmit rupture, Michel and
Avouac, 2002 : « The surface ruptures associated with
the Izmit earthquake seem remarkably simple »
As the rupture begins on a
segment with simple geometry,
one may hope to clearly see the
nucleation phase (if it exists !)
Spot image interferometry of Izmit rupture, Michel and
Avouac, 2002 : « The surface ruptures associated with
the Izmit earthquake seem remarkably simple »
The 150km-long Izmit rupture and the closest stations to the epicenter
45 minutes before the earthquake, first triggered window:
20 minutes before the earthquake:
18 shocks are visible in the 6 triggerred windows:
As the largest shocks are recorded at several
stations and the time difference between their
P-arrival and the one of the main shock are
the same, they come from what will be the
earthquake hypocenter
Spectra of 9 of the shocks:
before the first
foreshock
the 40s prior to
the earthquake
Cross-correlating the record with the template
Let us look again at the first triggered window :
filtered
record
Izmit nucleation
Observation: During the 44 minutes that precede the
earthquake, the hypocentral area emits a seismic
signal which repeats itself over and over sometimes
only a few seconds apart. These bursts become more
frequent as the time of the earthquake approaches.
They seem (maybe) accompanied by a continuous
low-frequency seismic noise.
Interpretation: A patch of the fault located at the
bottom of the brittle crust has begun to slip slowly 44
minutes before the earthquake. An asperity which will
become the hypocenter of the earthquake is resisting
this slow slip and is continuously reloaded by the slow
slip around it. This phase of slow slip accelerates in
time...
Is Izmit just a special case ?
or could it be a common case ?
Are many large earthquakes
preceded by a phase of slow slip
accelerating (irregularly) in time ?
All the M≥6.5 (interplate and intraplate) earthquakes which occurred in
these two densely instrumented zones between 01/01/1999 and 01/01/2011
Cumulative
seismic moment as
a function of time
in a zone of 50km
radius around the
epicenter of the
interplate
earthquakes.
The traces end just
before the
earthquake.
Cumulative
seismic moment as
a function of time
in a zone of 50km
radius around the
epicenter of the
interplate
earthquakes.
The traces end just
before the
earthquake.
Although different
time windows are
considered here,
an increase in
seismic activity
seems to precede
most interplate
earthquakes.
zoom
Cumulative number of
events in the 4 days
prior to 70% of the
interplate earthquakes.
The largest event is
circled in red.
Cumulative number of
events in the 4 days
prior to 70% of the
interplate earthquakes.
The largest event is
circled in red.
zoom
70% of interplate
earthquakes are
preceded by an
increase of seismic
activity in the day or
hours before.
Why the difference ?
Why the difference ?
The most logical (?) explanation:
Slow slip of the plate interface generally
occurs prior to its rupture
Characteristics of the Preparation Phase
Characteristics of the Preparation Phase
Its timing :
Stacks of cumulative
numbers of events
prior to the interplate
earthquakes
(each sequence is
given the same
weight)
Stacks of cumulative
seismic moments
Characteristics of the Preparation Phase
Its spatial extent :
Plate interface
Transform
Plate interface
Subduction
Plate interface
Subduction
Transform
Interpretation
The large interplate earthquakes are generally
preceded by a period during which the two plates in
contact begin to move slowly.
The foreshocks are produced by the breaking of the
frictional asperities resisting the slow slip.
This slow slip accelerates irregularly in the days or in
the hours before the earthquake...
Large subduction earthquakes seem often preceded
by the slow slip of a large part of the subducting plate.
USGS catalog for
the 100km-radius
zone surrounding
the Tohoku
epicenter in the 6
months prior to the
earthquake
For details on
the Tohoku
precursory
phase see Kato
et al. (2012)
Some events which occur close in time have almost identical
spectral shape:
The 150km-long Izmit rupture and the closest stations to the epicenter
Download