K a t e H u i h s u a n C h e n , R o l a n d B ür g m a n n , R o b e r t M . N a d e a u
Do the repeating earthquakes at Parkfield talk to each other?
Berkeley Seismological Laboratory, University of California, Berkeley, CA 947204760, USA
Contact: [email protected]
Abstract
The large population of characteristically repeating earthquakes at Parkfield provides a unique opportunity to study how these
asperity ruptures interact with each other and nearby non-repeating earthquakes. Here we analyze M -0.4 ~ 3.0 repeating
earthquake sequences to examine the variation of recurrence properties in space and time. We find that 67% of quasi-periodic
repeating sequences (i.e., coefficient of variation in recurrence interval less than 0.3) correspond to zones of low seismicity,
suggesting that these more regular repeating events are more isolated in space and from perturbing stress changes. We find that
closely spaced repeating sequences show evidence of strong interaction in time, reflected in temporally clustered event
recurrences. The temporal correspondence appears to be a function of separation distance from nearby earthquakes rather than the
relative size of the events. The response of the earthquakes to the occurrence of large earthquakes provides the clearest
documentation of the interaction process. A ccelerations of repeating sequences are associated with the significant seismic moment
release in the vicinity, and following rhe 2004 Parkfield earthquake, a large number of sequences exhibit accelerated recurrence
behavior. The characteristically decaying afterslip pattern is not obvious for some of the repeating sequences located close to the
largest co-seismic slip area, suggesting either that the stress changes are very heterogeneous, or that the rupture erased or shut off
some of the sequence source areas. Building on the above observations, we will be able to develop mechanical models that test the
extent to which fault interaction in the form of static stress changes and transient postseismic fault creep produces the observed
aperiodicity in the occurrence of these events, and furthermore, attempt to improve predictions of the times of future event repeats.
1.
RES: Repeating earthquake sequence
COV: Coefficient of variation
What controls the aperiodicity?
For overall neighboring events
(a)
Fig. 2. COV in recurrence interval of each repeating
sequence versus number of its neighboring
earthquakes in the distance of 1 km (red circles), 2 km
(blue circles), and 3 km (black triangles). (a) Overall
neighboring earthquakes are presented. (b) Only the
earthquakes with larger magnitude than the target
repeater are presented.
For larger neighboring events
(b)
Simply considering the total number of nearby
events does not produce a notable trend.
Cluster C (M1.88 - 2.09)
Separation: 0.392 km
Role of surrounding earthquakes ?
SF
Separation: 0.408 km
within 1 km
within 2 km
within 3 km
LA
SF
3k
Separation: 0.545 km
m
More isolated RESs tend to recur in a regular
manner.
Repeater
LA
RESs with a COV < 0.3 have a smaller number of
larger events in their immediate vicinity.
Neighboring
events
2004 Sep.
Parkfield earthquake
COV<0.2 Repeating earthquake sequences (RESs)
Scientific Question
What determines the timing of earthquake recurrences and their regularity?
2.
Historical M6 Parkfield earthquake
2004
Separation distance <3 km
What controls recurrence acceleration?
all repeating events
repeating events
without post-PK effect
(a)
Data without post-PK effect repeaters
1966
1934
1922
12
0.36 km
Repeaters associated
with M3+ events
10
0.42 km
T2
T3
T4
1950
2000
2050
Extremely shortened Tr that
gradually increase with time
Tr av=(Tr 1+Tr 2+.....Tr n)/n
Do the nearby M>3 events
correlate with unusually short Tr?
Separation distance <2 km
23
distance
from ref. seq
11
0.40 km
14
No repeater
in 1 km
1993.9 M5 event
(but the difference > 0.2 yr)
distance
from ref. seq
9
m
3k
[Waldhauser et al., 2004; Thurber et al., 2005]
NCSN repeating earthquake sequences (RESs)
from all repeaters
from data without post-PK effect repeaters
Repeater
(b)
Neighboring
events
Cluster A
50% of the NCSN REQSs are
characterized by a COV<0.3
Coefficient of variation =
SAFOD
target
NW
SE
C
A
standard deviation divided by the mean of
recurrence interval (Tr)
D
3.
Do nearby RESs correlate in time?
Closeness of individual events
in RES pairs?
Cluster A
Quantifying the “interaction”
Number <5 NCSN repeater
17
0.27 km
6
0.28 km
5
25
10
15
25
14 9
22
23
19 19
PK
5
distance
from ref. seq
3
0.10 km
1
0.45 km
19
0.50 km
Fig. 3. (a) Normalized recurrence interval as a function of year. Recurrence data associated
with nearby M3+ events are highlighted in red.(b) Close-up of short normalized Tr of less than
0.5. Number indicates the RES’s identity. (c) Event chronologies of the target RES (highlighted
in pink) and the RESs within 1 km from the target RES.
(a)
(M1.34 - 2.41)
A large magnitude difference within
a cluster does not correspond to a
low interaction coefficient.
Event pairs with dM < 0.3 (M w )
Overall event pairs
Separation: 0.303 km
pair
d
do
RA
time
Fig. 4. (a) Interaction coefficient between NCSN RESs pairs vs. separation distance. Filled
The breaks may suggest the temporal correlation circles indicate overall RES pairs in Clusters A to E. Crosses mark the pairs with similar
becomes insignificant when the separation
distance > 3 km, for the M<3 earthquakes we magnitude (magnitude difference (dM) of less than 0.3 Mw unit). (b) Interaction coefficient
consider.
between NCSN RESs pairs vs. magnitude difference.
0
RA
time
d1
do
d
SA
time
0
SA
time
(d)
hA
NCSN + HRSN repeating earthquake sequences (RESs)
Magnitude different (M w unit)
pair
(c)
tc
(b)
3000
τs
breaks
Separation: 1.407 km
Less than 0.5 yr : 1
Overall separations: 7
Interaction coefficient = 1/7 = 0.143
A’
(b)
pair
pair
A
(b)
Closely adjacent RESs with
comparable size show
strong correlation in time.
Separation: 0.499 km
Stress-free asperity (SA) : ttt=0
Creeping area (CA): bbb=0.03
Repeating aspertity (RA): bbb=0.0
Distance (m)
Separation: 0.169 km
Event-time separation:
(a)
-3000
-3000
Separation: 0.141 km
M6 event
Future work - mechanical model
500
pair
1. Interaction Coefficient
Number >=5 NCSN repeater
COV = 0.0 (pair)
COV = 0.0 - 0.3
COV = 0.3 - 0.5
COV = 0.5 - 0.7
COV = 0.7 - 1.0
1966 M6
17
15
The trend of interaction coefficient decreasing
with increasing separation becomes clearer for
the pairs with small size difference (dM<0.3).
1984-2005 Relocated seismicity
E
0.21 km
Normalized Tr
COV=0 : Perfect periodicity
COV~1 : Poissonian recurrence
COV>1 : Temporal clustering
B
Normalized Tr (Tr / Tr av )
extremely short Tr are mostly from post-PK repeating events
Separation distance <1 km
Close-up
15
Pa
SAFOD
target
4.
Do nearby RESs have similar recurrence properties?
(a)
Pa
tc
When the separation distance is small,
the recurrence histories appear to be
similar.
hA
’
(a)
N u m be r
1984-2005 Double difference earthquake catalog
No rma l iz e d Tr
HRSN RESs recurrence histogram
Cumulati v e Mo /1E+ 2 1 d yne - cm
Background seismicity:
Short recurrence intervals are
associated with nearby moment
release of M>3 events.
Sh e ar st r ess D is p la c e men t
NCSN (1984-2004) 30 M1.3~3.0 sequences (178 events) [Nadeau and McEvilly, 2004]
HRSN (1987-1998) 187 M-0.4 ~ +1.7 sequences (1123 events) [Nadeau and McEvilly, 2004]
HRSN updated (1987-Sep. 15 2006) 25 M-0.51~2.16 sequences (462 events) [Nadeau et al., 2005]
All repeaters
ref. seq
1857
1900
distance
from ref. seq
S hear s tr es s Dis pl ac em ent
T1
.......
1881
1850
DA TA GA P
[ separation in time < 0.2 yr
separation in space < 4 km]
1901
Repeating earthquake data:
ref. seq
25
No r malize d Tr
Repeating Earthquake Data
(c)
Depth (m)
This question cannot be answered without a statistically sufficient
observations of recurrence properties in natural earthquake populations.
Time difference < 0. 2 yr
Recurrence interval
dτ=0.0125 MPa
Repeating earthquake sequences
D e pth ( k m)
[Nadeau and McEvilly, 2004]
HRSN updated repeater
NCSN and HRSN repeater
1966 M6
Quantifying the “interaction”
2 . Similarity in recurrence property
(b)
(a)
DATA GAP
Closely adjacent RESs show similar
recurrence history.
Events rapidly recurred after the 2004 Parkfield earthquake, which
suggests a characteristically decaying afterslip pattern, whereas
in the lower panel, the afterslip pattern is not clear.
dτ=0.1MPa
seq.
24
18
Seismicity
[Waldhauser et al., 2004; Thurber et al., 2006]
Slip model (Kim and Dreger, 2007)
Pre- PK : 1984 - 27Sep.2004
Post-PK : 28Sep.2004 - 2005
4
20
7
(b)
The relationship between recurrence
history and magnitude difference is
unlikely to follow a linear pattern
8
References
dτ=0.8 MPa
K im, A., and D. S. Dreger (2008), Rupture process of the 2004 Parkfield
earthquake from near-fault seismic waveform and geodetic records, Journal of
Geophysical Research, doi:10.1029/2007JB005115, in press.
N adeau, R. M., and T. V. McEvilly (2004), Periodic pulsing of characteristic
Along-strike distance (km)
10
microearthquakes on the San Andreas fault, Science, 303, 220-222.
14
N adeau, R. M., M. Traer, and A. Guilhem (2005), Repeating earthquake and
nonvolcanic tremor observations of aseismic deep fault transients in Central
California,Eos Trans. AGU, 86(52), Fall Meet. Suppl., Abstrac G44A-01.
25
Fig. 1. (a) Fault-parallel section showing background seismicity and NCSN-derived repeating sequences. Fill color/shades are
keyed to the COV in recurrence interval. Dashed red circles indicate the clusters of interest (with inter-repeater distances of
less than 1 km). Number indicate the ID of NCSN repeating sequences. (b) Fault-parallel section section showing the
distribution of HRSN and NCSN repeating sequences (filled circles), updated HRSN repeating sequences (white crosses),
background seismicity (open circles color coded for pre- and post-2004 earthquake), 1966 M6 hypocenter, and their
relationship to the slip distribution of the 2004 Parkfield mainshock [Kim and Dreger, 2007]. Fill color/shades are keyed to the
COV in recurrence interval. Number indicate the ID of the updated 25 HRSN repeating sequences.
13
T hurber, C., H. Zhang, F. Waldhauser, J. Hardebeck, A. J. Michael, and D.
2
Fig. 5. (a) Event chronology for two HRSN REQS clusters (see white crosses with number for
locations in Fig. 1b). (b) Inter-event time spans (recurrence interval) as a function of time for each
RES in the two clusters. The similarities between recurrence history curves are illustrated by crosscorrelation coefficient.
Fig. 6. (a) Similarity in recurrence history (Figure 4b)
between HRSN RES pair as a function of separation
distance. (b) Similarity in recurrence history curve as
a function of magnitude difference.
Eberhart-Phillips (2007), Three-Dimensional Compressional Wavespeed Model,
Earthquake relocations, and Focal Mechanisms for the Parkfield, California, Bull.
Seism. Soc. Am., 96, S38-S49.
W aldhauser, F., W. L. Ellsworth, D. P. Schaff, and A. Cole (2004), Streaks,
multiplets, and holes: High-resolution spatio-temporal behavior
seismicity, Geophys. Res. Lett., 31, doi:10.1029/2004GL020649.
of Parkfield
Fig. 7. (a) Schematic planar representation of the fault
model in cross-section. (b-c) Schematic diagrams of stress
versus time and displacement versus time at isolated
repeatable asperity (RA) and freely slipping zones (SA). (d)
Evolution of shear stress of two RA (solid circles indicate
earthquakes). Stress evolution as a function of stress
increment induced by the other rupture (dτ), for two patches
have the same size and same stressing rate.