The past year has seen a greater cooperative involvment with researchers from a variety of other institutions (USGS, USC, LLNL, Harvard, UCLA, SCEC, WG99) on a broad range of topics.
The high precision analyses we have been developing now take full advantage of the clustered and highly similar earthquakes recorded by the broadband borehole High Resolution Seismic Network (HRSN) at Parkfield which achieves a 2 to 3 order-of-magnitude improvement in spatial resolution over routine catalogues like the NCSN. This extremely high resolution has provided us with an unexpected and strikingly systematic picture of the space-time-size behaviour of earthquakes - including the existence of pervasive repeating earthquake behavior at all scales and a considerable interaction and mutual triggering of neighboring small earthquakes. The resolution and systematic behavior has since proven to be instrumental in identifying potential targets for the USGS sponsored proposal to NSF for deep scientific drilling into the San Andreas Fault (SAF) at Parkfield. As a result, the vital role Berkeley will play in the experiment has been firmly established.
In collaboration with Charlie Sammis at USC, Lane Johnson at Berkeley, and Bill Foxall at LLNL, our understanding of the physics of the earthquake sources using the repeating event phenomena continues to show ever more striking and potentially revolutionary results. Furthermore, it has laid the ground work for ongoing research focused on inverting recurrence intervals of repeating earthquakes to infer fault slip rate time histories and spatial slip rate variations both in depth and along strike.
Our catalogue of repeating event sequences are also being employed by the recurrence model working subgroup in their development of a recurrence model for the Bay Area Earthquake Probabilities Working Group report to be published sometime in 1999 (WG99).
In addition, we are using our ever expanding repeating event catalogue as a widely distributed set of repeating illumination sources for monitoring of temporal changes in various wave propagation attributes including a collaborative effort studying coda Q variations with Michael Antolik at Harvard. Our passive source monitoring efforts are being integrated and correlated with previous controlled source studies at Parkfield and with our ongoing monitoring of seismicity and our new approach at monitoring slip rate changes.
In a related study, we are investigating the phenomena of fault zone guided waves (FZGW), since they are expected to have a high sensitivity to changing conditions in the fault core. As part of this effort we are organizing a SCEC workshop on FZGW for later in 1998 with John Vidale of UCLA.
The primary source of data for these studies has been the HRSN, and while it has served well for the last 11+ years, it has recently been showing signs of failure due to age. As a result and in preparation for the Deep Scientific Drilling Experiment, plans have been made and some funds committed to the upgrade of the system with key features of the upgrade being 24 bit recording with exact time sampling, VCSN flow and seed format. This work is expected to be completed by next summer.
As mentioned above, Berkeley has been instrumental in helping identify targets for the proposed drilling effort. Furthermore, the space-time-size systematics we observe and the new view of fault heterogenity and earthquake physics they suggest have been critical to establishing feasibility of various drilling plans and in establishing alternative testable hypotheses in the experimental plan. These aspects of our research are described below and in the following section on the Physics of earthquakes.
In an article published last December, Nadeau and McEvilly (1997) find that studies at very high resolution of microearthquakes at Parkfield occurring since 1987 reveal a systematic organization in space and time, dominated by clustering of nearly identical, regularly occurring microearthquakes ('characteristic events') on 10-20 m wide patches within the fault zone. More than half of the 4000+ events in their 1987-1996 catalog exhibit these traits (See Figures 11.1 and 11.2). In general, recurrence intervals (0.5 to 2 yr.) scale with the magnitude of the repeating events for the on-scale range (Mw 0.2 to 1.3) in their study. The similar waveforms, superimposed locations, quasi- periodic recurrence and uniform size of these characteristic events permit relative hypocenter location accuracy of meters and predictable occurrence times within windows of a few months. Clustered characteristic events occur at depths as shallow as about 3 km, and these are feasible targets for deep scientific drilling and observation at the focus of a subsequent small earthquake within an active plate-boundary fault zone. At Parkfield the achievable location accuracy to which a hypocenter can be specified as well as the predictability of its occurrence time appear to be uniquely favorable for in-situ fault zone measurements.
The systematics in space, time, size and waveform manifest in the HRSN data at Parkfield are in stark contrast to the seemingly disorganized nature of earthquake rupture and fault deformation observed with routine lower resolution datasets. This discrepency is attributable to the lack of definition in the surface recordings and routinely processed data. When high resolution data are processed using advanced techniques, the previous picture of diffuse seemingly uncorrelated seismicity largely resolves itself into a highly organized pattern of groups of repeating earthquakes having the same size, location, waveforms and with quasi-periodic recurrence. In addition, the hypocenters of the repeating earthquakes (and most earthquakes observed there) restrict themselves to a narrow planar core of seismic strain release within which the repeating earthquake sites cluster on several scales - a clusters of clusters organization also referred to as a discrete or hierarchical fractal distribution (See Figure 11.4). Fractal analyses of the distributions of these highly resolved hypocenters are currently underway in collaboration with Charlie Sammis at USC; Bill Foxall at LLNL; and Lane Johnson at Berkeley. Results of this work, discussed in more detail below, have led us to adopt a heterogeneous fault model which we are now developing and which has many implications for fault strength, lithologic vs. stress control of event occurrence and heat generation along the SAF.
In an article from the Bulletin of the Seismological Society of American published in June, Nadeau and Johnson (1998) used waveform data from a borehole network of broad band seismographic stations to study microearthquakes along the Parkfield segment of the San Andreas fault. Their Analysis of almost ten years of such data confirms that much of the seismicity in this region consists of repeating sequences, quasi-periodic sequences of earthquakes that are essentially identical in terms of waveform, size, and location. They estimated scalar seismic moments for 53 of these repeating sequences and combined them with equivalent estimates from 8 similar but larger event sequences from the Stone Canyon section of the fault (Ellsworth and Dietz, 1990) and the main Parkfield sequence. These estimates showed that seismic moment is being released as a function of time in a very regular manner. Measurements of the moment release rate, combined with an assumed tectonic loading rate, lead to estimates of the seismic parameters source area, slip, and recurrence interval (See Figure 11.3, left). Such parameters exhibit a systematic dependence upon source size over a range of 1010 in seismic moment which can be described by three simple scaling relationships (Figure 11.3, right). Several implications of these scaling relationships are explored, including the repeat time of earthquakes, average stress drop, strength of the fault, and heat generated by earthquakes. What emerges from this analysis of moment release rates is a quantitative description of an earthquake process that is controlled by small strong asperities that occupy less than 1% of the fault area. This means that the fault is highly heterogeneous with respect to stress, strength, and heat generation. Such heterogeneity helps to explain many of the apparent contradictions that are encountered in the study of earthquakes, such as why faults appear weak, why significant heat flow is not observed, how significant high frequencies can be generated by large earthquakes, and how various geologic features such as pseudotachylytes might form.
In an ongoing study, Nadeau and Foxall find that high resolution imaging of the frequency-space-time-size characteristics of a 15 Km segment of the SAF zone at Parkfield shows seismic slip there to be highly localized on, at most, a few sub parallel slip planes within a core of the fault zone less than 300m wide, and perhaps occurring on a single through-going fault plane. The distribution of seismically slipping fault patches is characterized by an extremely low fractal (3-D) capacity dimension (Do < 1.0) that is consistent with theoretical calculations which allow for characteristic repetition of events and assume coplanar slip and regional GR statistics. This relatively low Do and the repeating behavior are also consistent with the extreme clustering more broadly observed on this part of the SAF and with a model of deformation in which vast quiescent regions transfer slip aseismically between a comparatively few highly active seismic sites. Nadeau and Foxall show that the elevated estimates of the Do from previous studies at Parkfield (generally approaching 2) are artifacts resulting from relatively large uncertainties in hypocenters of routine catalogues, typically on the order of 1-2Km. Their physical model and observations conflict with previous spatial fractal models that require distributed deformation in the crust surrounding the main fault, with temporal fractal models that preclude quasi-periodic recurrence, and with the underlying concept of self-organized criticality. These high resolution observations are consistent with more broad scale observations of seismicity along the SAF from studies of lower resolution when hypocenter uncertainties are taken into account (Nadeau et. al., 1995; Nadeau and Johnson, 1998; Ellsworth and Dietz, 1990; Bakun and McEvilly, 1984). Furthermore, spatial scale invariance of fractal deformation models is generally assumed to hold so that the observed behavior of this short fault section may have important implications for large scale fault deformation models and seismic hazard analysis. Contrary to temporal fractal and self-organized criticality models, the strong quasi-periodic component in earthquake recurrence suggests that conditional probabilities of occurrence can be determined. The strong effect on Do of the 1-2 Km location uncertainties for even high resolution catalogues (such as the NCSN) also suggests that commonly employed smoothing kernels based on Do result in significantly over-smoothed seismicity and earthquake hazard maps.
In a recently submitted paper, Sammis et al. (1998) investigates in more detail the implications of the findings of Nadeau and Johnson and Nadeau and Foxall with regard to the strength of asperity contacts and distribution of earthquake hypocenters. Nadeau and Johnson found that the smallest events occurred on patches having a linear dimension on the order of 0.5m., displacements of about 2 cm. and stress drops on the order of 2000 MPa, roughly ten times larger than rock strengths measured in the laboratory. The stress drop for larger events was observed to decrease as a power law of the seismic moment reaching the commonly observed value of 10 MPa at about magnitude 6. Sammis et al. show that these large strengths are consistent with laboratory data if the preexisting microcracks are all healed. A hierarchical fractal asperity model is presented which is based on recent laboratory observations of contact distributions in sliding friction experiments. For this model, a fractal dimension of D=1.0 is shown to be consistent with the observed poser law decrease in stress drop and increase in displacement with increasing event size. The spatial distribution of hypocenters in the Parkfield area is shown to be consistent with this model, and with a hierarchical clustering of asperities having a discrete rescaling factor of about 20.
Our catalogue of repeating event sequences is being employed by the recurrence model working subgroup to develop a statistical recurrence model for the Bay Area Earthquake Probabilities Working Group (WG99). Specifically the recurrence model subgroup has chosen the 53 repeating (characteristic) earthquake sequences (221 earthquakes) from the Nadeau and Johnson (1998) to be part of a data set used to assess various proposed recurrence models and from which final model parameters will be determined. Berkeley's 221 characteristic events represent over 2/3rds of the natural events being employed in this effort.
We are continuing to expand our repeating event catalogue for use as a set of repeating illumination sources for monitoring of temporal change in various wave propagation attributes. These passive source monitoring efforts are being integrated and correlated with previous controlled source studies at Parkfield (Karagerogi et al., 1992 and 1997) and with our ongoing monitoring of seismicity and our new approach of monitoring slip rate changes with repeating earthquake recurrence intervals. Summaries of these efforts follow.
Using the differential coda Q method of Aster et al. (1996), Nadeau and Antolik (1997) have extended the search for temporal change in coda Q at Parkfield, CA (Antolik et. al. 1996) by significantly expanding the number, distribution and time base of the earthquake multiplets used. Additionally, they constrain the similarity of events in the multiplet sources by requiring they contain only characteristically repeating earthquakes.
Nadeau and Antolik process 184 characteristic multiplets ( earthquakes) yielding about 2000 event pairs for analysis. This large number of pairs in conjunction with the characteristic nature of the sources should provide the quality and redundancy of data necessary to constrain estimates of temporal change in coda Q to less than 2% for 3-30Hz frequencies at .
They also expand the spatial coverage of the source-receiver travel paths and extended the time base (1987-96, inclusive) to better characterize the temporal stability structure of the region and to investigate the possibility of a connection between the subtle variations in coda Q reported in the previous study - of the order of 5% or less - and recent changes in overall seismicity, M4+ activity, and recurrence intervals of repeating earthquakes - all of which may be related to processes taking place in preparation of the expected M6 earthquake at Parkfield.
The 10-station borehole-installed high-resolution seismographic network at Parkfield has been in operation since 1987 and provides data for a large number and types of investigations of fault-zone processes at very fine scales. The data base of over ten years extent includes more than 4000 earthquakes and 52 controlled- source data sets in which the network was illuminated by our S-wave Vibroseis at up to eight source points each with three polarizations. Our purpose in the Parkfield work has been to observe at maximum possible resolution the microearthquake process and the properties of propagating elastic waves in a localized part of a fault zone presumed to be late in the earthquake cycle at the M6 level (or possible M8 if Parkfield is the initiation point of the 1857 event). Monitoring and modeling of controlled-source data show clear evidence of systematic changes in the elastic wave properties in the upper kilometer or so of the section along the fault zone and above the presumed nucleation zone (Karagerogi et al., 1992 and 1997). However, early results from similar investigations using deep characteristically repeating microearthquake sources fail to show corresponding changes even for late arriving fault guided phases - generally presumed to sample properties in the fault most directly. On the other hand, significant changes in the repeat times of the characteristic microearthquakes and the ratio of repeating to non-repeating events are widely observed and correlate well in time and space with both the Vibroseis results and with changes documented in other measurements, e.g., fault creep, water-well levels, seismicity, etc. This raises the tantalizing suggestion that we may all be seeing different parts of the same time-varying fault zone process, and that there may indeed be reason to expect unique changes in the patterns associated with nucleation of a large earthquake at Parkfield.
In a paper to be submitted this Fall, Nadeau and McEvilly take advantage of unique attributes of the characteristic microearthquakes at Parkfield California (constant size, highly repeatable waveforms, collocation, short recurrence times, and the distribution of large numbers of sequences in depth and along the fault segment) and theory developed in Nadeau and Johnson (1998) to map inferred slip rate history at depth and along strike from variations in recurrence intervals between repeating small earthquakes. At present Nadeau and McEvilly have identified 151 sequences suitable for the analysis which comprise a data set of 853 earthquakes and 702 recurrence interval times spanning over 11 years.
The study focuses on the variations in recurrence intervals for individual characteristic sequences from their mean values, assuming that the instantaneous recurrence interval is a measure of the local fault slip rate. It is thus possible to infer the spatial-temporal distribution of slip-rate on the fault surface from the changes in recurrence rates of the repeating sequences. A test of the method using the available seismicity record shows systematic spatial and temporal changes in the slip rate that correlated well with observed deformation at the surface and that were synchronous with earthquake activity and other indicators of fault-zone deformation (Figure 11.5). It is found that localized changes in estimated slip rates, at times exceed 40-50 percent, and have occurred in apparent response to the recent M4+ activity at Parkfield. Changes in slip occurred both at close range and at considerable distance (>5 km) from the recent M4+ events, although the inferred changes were more pronounced at shorter distances and show a delayed response time at greater distance. The computed changes were significant to depths below 10km.
If this phenomenon is found through other high-resolution studies to be generally common behaviour in active faults, it forms the basis for a new method to monitor the changing strain field throughout the seismogenic zone.
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 have
begun to explore this problem from a somewhat different approach, 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 the zone of changing wave propagation parameters
within the shallow (the upper 3 km) part of the fault zone (Karageorgi
et al., 1997). In this investigation we have begun to characterize the
distribution throughout the fault zone of source-receiver paths that
produce strong FZGW signals. The goal of this part of our 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 finally, to model
the phenomenon numerically using new 3-D guided-wave algorithms under
development at LBNL (V. Korneev, personal communication). Figure 11.6 is
a display of the type of data set that can be constructed from the HRSN
waveforms. In the 1km-wide strip at depth there are more than 500
sources with which to build a receiver gather 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 on this is suggesting that the
strong generation as well as the wave propagation is a shallow feature
of the fault zone. We plan to take this study farther as part of
our ongoing Parkfield research and
as part of this effort we are
currently in the process of organizing a SCEC workshop on the FZGW for
later in 1998 with John Vidale of UCLA.
Operation, maintenance and archiving of the HRSN data has become much more laborious in recent years due to the failing acquisition system (Figure 11.7). Hence an acquisition system upgrade has become a critical part of our research plan. Continued research with data from the Parkfield borehole network appears very promising so that serious consideration must be given to the obsolescence of the recording system and the need to upgrade. We have worked hard at finding the most economical approach to an upgrade, and the technological developments for 24-bit digitization, telemetry and workstation- based acquisition platforms has reached an attractive situation in that sense. The network structure is in place so that we can use the power systems, enclosures, possibly radios and antennas, the central site tower, power and instrument space, and we already have links to Berkeley for the EM, GPS and BDSN broadband instrumentation already operating there. Some funds have been committed to the upgrade of the system and key features will include 24 bit recording with exact time sampling, VCSN flow and seed format. This work is expected to be completed by next summer.
Antolik, M., R. M. Nadeau, R. C. Aster and T. V. McEvilly (1996). Differential analysis of coda Q using similar microearthquakes in seismic gaps. Part 2: Application to seismograms recorded by the Parkfield High Resolution Seismic Network, Bull. Seism. Soc. Am. 86, 890-910.
Bakun, W. H. and T. V. McEvilly (1984). Recurrence models and Parkfield, California earthquakes, J. Geophys. Res. 89, 3051-3058.
Ellsworth, W. L. and L. D. Dietz (1990). Repeating earthquakes: characteristics and implications, Proc. of Wrkshp. XLVI, the 7th U.S. - Japan Seminar on Earthquake Pred., U.S. Geol. Surv. Open-File Rept. 90-98, 226-245.
Harris, R.A. and P. Segall (1987). Detection of a locked zone at depth on the Parkfield, California, segment of the San Andreas Fault, J. Geophys. Res. 92, 7945-7962.
Karageorgi, E., R. Clymer and T. V. McEvilly (1992). Seismological studies at Parkfield. II. Search for temporal variations in wave propagation using Vibroseis, Bull. Seism. Soc. Am. 82, 1388-1415.
Karageorgi, E., T. V. McEvilly and R. Clymer (1997). Seismological studies at Parkfield IV: Variations in controlled-source waveform parameters and their correlation with seismicity, 1987 to 1995, Bull. Seism. Soc. Am. 87, 39-49.
Langbein, J., R.L. Gwyther and M.T. Gladwin (1998). Possible increase in fault slip rate at Parkfield in 1993 as inferred from deformation measurements from 1986 to 1997, Seism. Res. Lett. 69, 151, 1998.
Nadeau, R. M. and M. Antolik (1997). Expanded Analysis of Differential Coda Q Using Characteristic Earthquake Multiplets at Parkfield CA, EOS Trans., AGU 78, F482.
Nadeau, R. M., W. Foxall and T. V. McEvilly (1995). Clustering and periodic recurrence of microearthquakes on the San Andreas Fault at Parkfield, California, Science 267, 503-507.
Nadeau, R. M. and W. Foxall (1996). Implications of characteristic microearthquakes at Parkfield, CA, for fractal crustal deformation models (abstract), EOS Trans., AGU 77, F503-F504.
Nadeau, R. M. and T. V. McEvilly (1998). The Parkfield Experiment: A New View of Fault-Zone Process, Seism. Res. Lett. 69, 151.
Nadeau, R. M. and T. V. McEvilly (1997a). Characteristic microearthquake response to the December 20, 1994 M5 event at Parkfield, CA: changes in recurrence interval and inferred slip rate (abstract), Seismo. Res. Lett. 68, 325.
Nadeau, R. M. and T. V. McEvilly (1997b). Seismological studies at Parkfield V: characteristic microearthquake sequences as fault-zone drilling targets, Bull. Seism. Soc. Am. (in press).
Nadeau, R. M. and L. R. Johnson (1998). Seismological Studies at Parkfield VI: Moment Release Rates and Estimates of Source Parameters for Small Repeating Earthquakes, Bull. Seism. Soc. Am. 88, 790-814.
Sammis, C. G., R. M. Nadeau and L. R. Johnson (1998). How Strong is an Asperity?, submitted to J. Geophys. Res.