Transcript Reorientation of small faults as a possible mechanism of LURR

Exploring the underlying
mechanism of LURR theory
Can Yin
Supervisor: Prof. Peter Mora
QUAKES, Department of Earth Sciences
University of Queensland
Load/Unload Response Ratio –
promising but controversial
• As a phenomenon, high LURR prior to major earthquake
has been found in many earthquakes by retrospective
examination. It is also observed in several real time
predictions.
• Since accelerating seismic moment release is also observed
in many earthquakes, it is attractive to find a common
underlying mechanism for these two phenomena. If they
share a common root, does the critical region size in AMR
apply to LURR?
Definition of LURR
LURR is originally defined in differential form, making it a state variable
depending only on the state of the system at that moment.
Y
X
R
; X  lim
P  0 P
X
In current implementation, we define the loading duration instead of loading
increment itself as P and introduce the integral form of response rate as
follow:
1
X 
T
T

0
dR
1
dt 
dP
T
Accordingly, LURR is defined as
 E(t )
tT
N
m


E (t )
E

i 

i 1
T
tT

   
T  N

E
(
t
)
  E im 

 i 1

tT



1
T
X
Y   
1
X
T
If the averaging time is long enough, T+ is almost equal to T-.
However, the sacrifice of compromise to the integral
definition of LURR is it now involves many other
factors, whose relations with LURR are yet to be found.
The new LURR is related to the history of system
evolution and thus dependent of many extrinsic
parameters, such as magnitude range of selected
earthquakes, the spatial range of earthquakes used,
fault parameters, etc. For example, when defining
effective Coulomb Failure Stress for criteria of
load/unload, we simply define it either as that on the
fault plane of ensuing main shock or that on an optimal
fault plane determined by the tectonic stress field. In
the former case, not all the earthquakes occurred on
the main fault plane; while in the latter case, it is proved
that the tectonic stress field changes direction prior and
after a major earthquake. In either case, the definition
of Coulomb Effective Stress is kind of ambiguous.
Re-examine critical region sizes
by LURR
• We examine our study on last 5 major earthquakes in California
with magnitudes greater than 6.5 between 32° N and 40°N
latitude since 1980. The catalog used in this study is the Council
of the National Seismic System (CNSS) Worldwide Earthquake
Catalog, which is accessible via the World Wide Web at the
Northern California Earthquake Data Center
(http://quake.geo.berkeley.edu/cnss). According to the relation
between critical region radius and ensuing mainshock
magnitude in AMR, radii of the critical region of all 5
earthquakes are larger than 100km.
• As in Australia, due to limited data, we only examined the last
two well-recorded earthquakes with magnitude greater than 5.0.
1983.05.02 Coalinga Earthquake (M6.7)
1987.11.24 Superstition Hills (M6.6)
1989.10.18 Loma Prieta (M7.0)
1992.06.28 Landers Earthquake (M7.3)
1994.01.17 Northridge Earthquake (M6.7)
1997.03.05 Burra, South Australia. M5.0
Burakin, West Australia
(a series of M>5.0 earthquakes since late 2001 till early 2002 )
Discussion
Note: All plots here are plotted with incorporation of disturbance of random process, which is
expressed by Y90, i.e. LURR= Y/Y90. LURR>1.0 means it can be considered as abnormality at
90% confidence level. LURR=0 indicates that data is insufficient to make significant calculation
(the minimum sample size is 20 events in a group).And upper limit of magnitude of earthquake
taken into calculation is 4.0.
From the figures above, it is obvious that smaller the region, the
more prominent the LURR abnormality will be. How to explain
it? Two possible answers:
(1) Either the method used in AMR to correspond the region size
where LURR has peak value to critical region size does not
apply to LURR; or
(2) there is simply no critical region. More close to the future
epicenter, more critical. This is a very intuitive explanation.
• Look at the epicenters of
these earthquakes. If the
seismogenic time of an
earthquake is decades
long, given the proximity
of time and space
between two earthquakes,
say Landers(7) and
Northridge(8), how could
you distinguish which
precursory events belongs
to which major
earthquake? If they are
not distinguishable, the
critical regions of each
earthquake will be
intermingled. Then how
large is each region?
Re-orientation of small faults as a possible
mechanism of current LURR implementation
• Some researchers found that the tectonic stress field rotate
before and after the major earthquakes in California.
• In order to find a clue for the relation between stress
rotation and LURR abnormality, we calculate LURR under
different fault planes, which is uniquely determined by
tectonic pressure axis (P-axis). Below are some examples
of LURR change with optimal P axis direction. P axis
change from –90 deg(West) clockwise to 0 (North) until
+90 deg(East).
Loma Prieta, CA
Northridge, CA
Burakin, West AUS
Discussion
• LURR abnormality pattern in California is obviously different from
that in Australia. In Northridge, for example, LURR has peaks in many
directions at different time. In Burakin, however, LURR has peaks
only in direction along North-South. It is likely that the tectonic stress
field fixes in this direction for a long time. This may be the result of
the relatively stable tectonic stress field in Australian plate.
• When the crust is far away from criticality, small faults tend to be
randomly orientated, LURR calculated for any fixed direction will
surely fluctuate around unity. But when the crust is close to failure, the
number of fractures along a particular direction determined by the
built-up tectonic stress will get dominant, causing calculated LURR
along this direction high up. Such rotation mechanism partly solves the
dilemma brought by the ambiguous definition of Coulomb Effective
Stress in current implementation of LURR.
Summary
• The method to identify critical region size in AMR doesn’t
apply to LURR. Failure to find better power-law curve
fitting at smaller region in AMR may be due to
observational limit rather than physical meaning, because
there is not enough seismic events to obtain a statistically
significant curve-fit within smaller region.
• LURR may not share a common underlying mechanism
with AMR, but may be caused by critical sensitivity which
is reached by the “phase up” of previously randomly
oriented small fractures re-orientating towards a certain
direction determined by tectonic stress prior to main shock,
resulting in higher LURR in this direction. Such reorientation may be a kind of underlying mechanisms of
LURR phenomenon for current implementation.