Next: Source Rupture of the
Up: Ongoing Research Projects
Previous: Hayward Fault & Bridges
T. V. McEvilly, R. W. Clymer, R. Nadeau, A. Kirkpatrick, L.
Johnson, V. Korneev, E. Karageorgi
The HRSN network at Parkfield was installed in 1987 to provide a
direct test of two hypotheses critical to our understanding of the
physics of the earthquake process, with implications for
earthquake hazard reduction and the possibilities for short-term
earthquake prediction - major goals of the NEHRP:
1) That the earthquake nucleation process produces stress-driven
perturbations in physical properties of the rocks in the incipient
focal region that are measurable, and
2) That the nucleation process involves progressive and systematic
failure that should be observable in the ultralow-magnitude
-1 < M < 2) with high-resolution locations and
Analyses of the 13+ years of Parkfield monitoring data have
revealed significant and unambiguous departures from stationarity
both in the seismicity characteristics and in wave propagation
details. Within the presumed M6 nucleation zone we also have found
a high Vp/Vs anomaly at depth, where the three M4.7-5.0 sequences
occurred in 1992-94. Synchronous changes well above noise levels
have also been seen among several independent parameters,
including seismicity rates, average focal depth, S-wave coda
velocities, characteristic sequence recurrence intervals, fault
creep and water levels in monitoring wells. We have been able to
localize the S-coda travel-time changes to the shallow part of the
fault zone and demonstrate with numerical modeling the likely role
of fluids in the phenomenon. We can connect the changes in
seismicity to slip-rate variations evident in other (strain, water
level) monitored phenomena. Based on the ubiquitous clusters of
repeating microearthquakes, scaling laws have been developed that
can be projected to fit earthquakes up to M6, and they predict
unprecedented high stress drops and melting on the fault surface
for the smallest events. Exhumed fault-zone rocks provide
independent evidence for such source conditions. This hypothesis
is being debated vigorously in the current literature. Recurrence
interval variations in the characteristic event sequences (about one-half of
the microearthquake population) have been used to map fault slip
rate at depth on the fault surface, and this technique appears to
be applicable to other creeping fault segments. Along the way in
this exciting discovery process we have challenged the
conventional 'constant stress drop' source model, affirmed
characteristic earthquake occurrence and developed four-
dimensional maps of fault-zone microearthquake processes at the
unprecedented scale of a few meters. The significance of these
findings lies in their apparent coupling and inter-relationships,
from which models for fault-zone process can be fabricated and
tested with time. A more fundamental contribution of the project
is its production of a continuous baseline, at very high
resolution, of both the microearthquake pathology and the subtle
changes in wave propagation, providing to the seismological
community a dynamic earthquake laboratory available nowhere else.
This unique body of observations and analyses has also provided
much of the impetus for Parkfield as the preferred site for deep
drilling into an active seismogenic fault zone (the SAFOD
project), and we are expanding the network to improve its view of
the drilling target zone on the fault surface.
Scaling laws for earthquake source properties, statistical
description of earthquake occurrence, precise monitoring,
forecasting, estimating fault slip rates from repeating
microearthquakes, or virtually any careful analysis of
the earthquake process all demand an accurate estimation
of earthquake size. The difficulty is that these
kinds of investigations must operate at the M 0 level to
acquire sufficient data over the lifetime of a realistic
study For example, at M 6 on the Parkfield San Andreas
segment, definition of the recurrence statistics for the
repeating rupture of the fault requires data spanning
centuries, while at M 0, the same number if events can be
seen in 2-3 years, and there are hundreds of such repeating
sequences on the fault segment. The Parkfield event archive
has become the de-facto calibration data set for extending
new methods of microearthquake analysis to other segments
on active faults. There are few if any
complete catalogs of events with self-consistent
moment-magnitude relations at the microearthquake level
that can be integrated with conventional data sets at M>2
so that the above-mentioned studies can be projected into the vast
resource of well-established data bases of larger events.
For the Parkfield data we have compiled and carefully tested
a methodology for estimating seismic moment, calibrating the
HRSN moments with the regional NCSN preferred
magnitude catalog, producing a seamless relationship
from M<0 to M>6. Data from this calibration set
are being used in current research requiring accurately
merged catalogs (e.g., Burgmann et al.,
2000; Wiemer and Wyss, 2000).
Since last year several manuscripts in the community
have surfaced which attempt to explain the striking scaling
relation of recurrence intervals with earthquake size.
The original scaling for Parkfield microearthquakes was first published
in a peer reviewed journal
by Nadeau and McEvilly (1997), and later developed
in more detail by Nadeau and Johnson (1998).
These analyses present strong evidence in
support of a highly heterogeneous fault zone with
scale-dependent stress drops where stress drops on small
scales are extremely high, while on the large scale,
averaged over large regions, they are small.
Subsequent research publications are taking issue with
this model, presenting alternative fault-zone processes
such as load-shielding (Anooshehpoor and Brune, 2000;
Sammis and Rice, 2000) or
creep-slip (Beeler, 2000) to avoid the
high stress drops we hypothesize for the small events.
Coincidentally however, ongoing work at
Berkeley correlating the energetics of formation of
fossil earthquakes (i.e. pseudotachylites) (Figure 13.1a)
with repeating earthquakes at Parkfield and elsewhere on the central
SAF system has evolved to the point where strong arguments can be made
based on direct field observation of features
of exhumed fault zones - the pseudotachylites - to support
the stress drop scaling (Figure 13.1b) and strong, scale-dependent,
heterogeneity of fault strength. (Wenk et al., 2000;
Nadeau et al., 2000).
Pseudotachylite (PT) structures on mesoscopic (a) and
microscopic (b) scales. (a) Hand sample of an isolated pseudotachylite
vein in a highly cataclastic gneiss. Note its planar geometry,
manifest by the 3-dimensional cut of the sample. This is typical of
PTs found in the study area. This vein corresponds to an earthquake of
-1 Mw, penny for scale. (b) Microstructures of microlites in plane
polarized light. Skeletal crystals of plagioclase nucleate
preferentially on fragments. They are surrounded by a fine groundmass,
mainly consisting of biotite. Large fragments are plagioclase and
quartz. These structures are indicative of very rapid energy release and
nearly instantaneous melting, providing additional evidence for the episodic
(earthquake) origins of these PTs. (c) Source area (A) verses seismic moment for repeating
earthquakes and PTs. Data points of 567 repeating earthquake sequences (over
-1 < Mw < 6 from the central San Andreas fault system (including parts of the
Hayward and Calaveras faults) (solid circles) and 290 pseudotachylite veins from the Peninsular Ranges (open squares) are shown. The fit to the combined data set is shown as a thick solid line. Homogeneous,
circular fault model representations are given by the 1 and 100 bar stress drop lines (dashed).
Typical earthquake stress drops appear to overestimate the source areas of these small events, by
up to 3 orders of magnitude for the smaller events.
As part of a more ambitious exploration of the extent to
which clustering of repeating and highly similar microearthquakes
characterizes active fault zones, we have devised a
Cluster Signature (CS) measure of such behavior for
any segment of a seismogenic fault. This characteristic helps
provide the underlying recurrence statistic with which
fault slip rate can be estimated, applying the
methodology developed at Parkfield. We have embarked upon a
Calibration/Extrapolation/Characterization of slip rate along
the San Andreas Fault from Parkfield to San Francisco
over the past 16 years, using the NCSN event catalog.
Initial results are fascinating, revealing
portions of the fault extending over many tens of kilometers
that exhibit coherent pulsing in slip rate.
In addition, the pulses appear closely related to the
occurrence of the Loma Prieta earthquake in 1989,
with quiescence prior to, and pulsing onset subsequent to Loma
Prieta. An expected exponential decay of slip rate is seen
in the LP aftershock zone following the main event.
Figure 13.2 illustrates the coherency in slip rate variations after
LP on 2 segments of the SAF SE of the LP rupture and separated
from each other by over 75 km. Also notable is the observed
quiescence prior to LP on its adjacent SJB segment.
This research is testing the utility of the
repeating earthquake statistic and the waveform similarity
characterizations in defining fault properties and fault segment
boundaries based more quantitatively on what appears to be a
robust measure of fault slip rate.
Slip rate histories for two segments of the San Andreas fault
obtained by applying to the NCSN catalog the method devised and
calibrated with the high-resolution HRSN data at Parkfield. (Top)
Occurrence timelines for 42 repeating sequences on a 12 km section of
the fault at San Juan Bautista. (2nd Panel) Cumulative slip for the
fault segment derived from the cumulative slip of all repeated events
occurring on the segment and scaled by the number of sequences.
(3rd Panel) Slip rate variations from the differences in
cumulative slip averaged over a 1.2 year time window. Note the
pulse-like slip rate variation at 3 year
intervals and the related clustering of event occurrences in the top
panel. The vertical line is the time of the Loma Prieta earthquake,
centered about 50 km northwest of the SJB segment. The three arrows are times
of the slow silent earthquakes at SJB reported by Linde et al. (1996).
(Bottom Panel) Results of the same analysis applied to a segment 75 km
to the southeast of SJB. Three-year pulsing is again visible,
perhaps advanced slightly from the SJB pattern, and the depressed slip
rate seen at SJB prior to the LP earthquake is not present on this more
There has been a lot made of so-called fault-zone guided waves
(FZGW). Much of it has been directed toward modeling wave
propagation in relatively simple, vertical low-velocity structures
in order to match discrete observations of the late, low frequency
arrivals sometimes recorded near the fault trace. We are
approaching this problem from a somewhat different direction,
using the extensive observations of these waves in the Parkfield
network, the 3-D P- and S-velocity model for the fault zone, and
our Vibroseis results that place an apparent strain-related zone
of changing wave propagation parameters within the shallow (the
upper 3 km) part of the fault zone (Karageorgi et al., 1992, 1997;
Korneev et al., 2000). To investigate FZGW more quantitatively,
we have begun to characterize the distribution throughout the
fault zone of source-receiver paths that produce strong FZGW
signals from earthquakes. The goal of this research is to be able
to first determine the patterns of generation and propagation of
FZGW, to characterize the wavefield in terms of velocity and
particle motion relative to the fault zone, and to model
the phenomenon numerically using new 3-D guided-wave algorithms
under development. We are also using the numerous and widely distributed
sites of repeating earthquakes as illumination sources for imaging
temporal changes in FZGW propagation in search of evidence
relating to processes of fault healing or large event nucleation.
Figure 13.3 shows the type of data set that can be constructed from the
HRSN waveforms. In the Parkfield archive there are thousands of
microearthquakes available with which to build a receiver gather of a
sources for any component of motion throughout the ten stations.
Stacking of the traces is very effective because of the uniform
source mechanism common to neighboring events on the fault. Our
initial work is suggesting that the strong generation as well as
the propagation of typical low-frequency FZGW in the coda of S is
controlled by a well-defined feature within the fault zone that
appears to be the plunging NW edge of the M6 asperity - the green
region under station MMN (bottom panel Figure 13.3).
(Top) A receiver gather for station MMN, horizontal component, of 544
earthquake sources at depths 3.3 to 3.8 km (see swath shown in bottom
panel) along a 25 km stretch of the SAF at Parkfield. Traces are
stacked into 100 m bins, a legitimate procedure due to the uniform
focal mechanisms. Note the spatial variation in the relative
generation of the fault-zone guided wave (FZGW) along strike, and
the corresponding drop in S phase amplitudes. (Bottom) Ratio of the
S-coda to S-phase energy as recorded at station MMN for
earthquakes through-out the fault zone. The position of the dots
represents the earthquakes hypocenters projected along the fault surface.
The dark grey dots indicates earthquakes whose coda energy is less than 20
S-phase energy as recorded at MMN, Medium grey indicates a ratio of greater
the color change with the onset of the strong FZGW illustrates the
significance of the color change. This type of display provides
an easy assessment of travel paths to MMN from events through-out
the fault zone which generate strong guided wave energy. A similar plot
for earthquakes recorded at station EAD (not shown) provides a reverse
profile image of wave propagation through the zone which shows a converse
relationship (light grey to the NW and dark grey to the SE). This argues for a
source of the FZGW which is located, in depth, along the NW edge of the
M6 Parkfield asperity at the transition between creeping and locked
behavior (roughly in in the light grey region under MMN).
The Vibroseis monitoring investigation reported significant
travel-time changes in the coda of S for paths crossing the fault
zone southeast from the epicenter of the 1966 M6 earthquake.
Progressively decreasing travel times in the anomalous region
reached 50 msec or more by the end of the study. Changes in
frequency content and polarization were also found and those
effects, too, could be localized to the zone of common nucleation
and rupture onset for the previous M6 earthquakes, and, possibly,
the region of slip initiation for the great earthquake of 1857.
The temporal pattern in these variations appears to be synchronous
with changes in deformation and seismicity measured independently
(Nadeau and McEvilly, 1999). Because similar variations are not
seen in the waveforms recorded from microearthquakes in the same
part of the fault, Karageorgi et al. (1997) conclude that changing
fluid conditions in the uppermost section of the fault zone in
response to deeper, tectonic stress perturbations are the likely
cause of the temporal variations. Korneev and McEvilly (2000)
modeled the variations numerically, and
successfully explain the observations as interaction (reflection
and transmission) of the shallow wavefield with a 200-meter-wide
low-velocity fault zone in which the velocity increases by 6
we hypothesize, to hydrological changes accompanying a significant
pulse in slip rate and seismicity that was evident in independent
data. We will pursue the modeling of wave propagation influenced
by the fault zone in a more complex and 3-D medium.
In a project carried out by Ann Kirkpatrick, we explored the degree
of improvement possible over the Michelini and McEvilly (1991) 3-D P-
and S-wave velocity models estimated early in the Parkfield project,
when there were only 169 events used the inversion. Now, with another
decade of data, it is possible to build a much more extensive data
set. About 4800 and 2100 P and S arrival times, respectively, were
selected for uniform raypath illumination throughout the study
volume. We have a suite of new models using various permutations of
grid spacings and data set combinations. The gross features of
these models are similar with each other and with the 1991 model,
however, the new models include a larger
geographic scope and more earthquakes and additional auxilliary
data sets. These additional data primarily help to fix
the edges of the model and to extend the model in the along
fault direction (both NW and SE). As a result, the event
locations on the ends of the network have straightened out
significantly (including the those in the vicinity of the
SAFOD drilling site). The apparent dip of the events is reduced
somewhat, but the hypocenters are still biased to SW of the fault
trace and the USGS locations. The salient features in
velocity models and the Vp/Vs ratio are not significantly
different from the 1991 inversion results. This study is
being prepared for publication.
Anooshehpoor, A. and J.N. Brune, Quasi-Static Slip-Rate Shielding by Locked and Creeping Zones as an Explanation for Small Repeating Earthquakes at Parkfield, submitted to Bull. Seism. Soc. Am., 2000.
Beeler, N.M., A simple stick-slip and creep-slip model for repeating earthquakes and its implication for micro-earthquakes at Parkfield, submitted to Bull. Seism. Soc. Am., 2000.
Burgmann, R., D. Schmidt, R.M. Nadeau, M. d'Alessio, E. Fielding, D. Manaker, T.V. McEvilly, and M.H. Murray, Earthquake Potential along the Northern Hayward Fault, California, Science, 289, 1178-1182, 2000.
Karageorgi, E., R. Clymer and T.V. McEvilly, Seismological studies at Parkfield. II. Search for temporal variations in wave propagation using Vibroseis, Bull. Seism. Soc. Am., 82, 82, 1388-1415, 1992.
Karageorgi, E.D., T.V. McEvilly and R.W. Clymer, Seismological Studies at Parkfield IV: Variations in controlled-source waveform parameters and their correlation with seismic activity, 1987-1994, Bull. Seism. Soc. Am., 87, 39-49, 1997.
Korneev, V.A., T.V. McEvilly and E.D. Karageorgi, Seismological Studies at Parkfield VIII: Modeling the Observed Controlled-Source Waveform Changes, Bull. Seism. Soc. Am., 90, 702-708, 2000.
Michelini, A. and T.V. McEvilly, Seismological studies at Parkfield: I. Simultaneous inversion for velocity structure and hypocenters using B-splines parameterization, Bull. Seism. Soc. Am., 81, 524-552, 1991.
Nadeau, R.M. and L. R. Johnson, Seismological Studies at Parkfield VI: Moment Release Rates and Estimates of Source Parameters for Small Repeating Earthquakes, Bull. Seism. Soc. Am., 88, 790-814, 1998.
Nadeau, R.M., L.R. Johnson and H.-R. Wenk, Are Pseudotachylites Fossil Earthquakes?, submitted to Seism. Res. Lett., 2000.
Nadeau, R.M. and T. V. McEvilly, Seismological Studies at Parkfield V: Characteristic microearthquake sequences as fault-zone drilling targets, Bull. Seism. Soc. Am., 87, 1463-1472, 1997.
Nadeau, R.M. and T.V. McEvilly , Fault slip rates at depth from recurrence intervals of repeating microearthquakes, Science, 285, 718-721, 1999.
Sammis, C.G. and J.R. Rice, Repeating Earthquakes as Low-Stress-Drop Events at a Border Between Locked and Creeping Fault Patches, submitted to Seism. Res. Lett., 2000.
Wenk, H.-R. L.R. Johnson, and L. Ratschbacher, Pseudotachylites in the Eastern Peninsular Ranges of California, Tectonophysics, 321, 253-277, 2000.
Wiemer, S. and M. Wyss, Combined Mapping of the Earthquake Size Distribution and Stress Tensor Orientation Offers New Insight into Properties of Faults, submitted to Science, 2000.
Next: Source Rupture of the
Up: Ongoing Research Projects
Previous: Hayward Fault & Bridges
The Berkeley Seismological Laboratory, 202 McCone Hall, UC Berkeley, Berkeley CA 94720
Questions and comments to firstname.lastname@example.org
Copyright 2000, The Regents of the University of California.