Earthquake research at Parkfield, California, 1993 and beyond; report of the NEPEC working group to evaluate the Parkfie

Report of the NEPEC Working Group to Evaluate the Parkfield Earthquake Prediction Experiment

By National Earthquake Prediction Evaluation Council Working Group

U.S. Geological Survey, Information Services

Any use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Cover: A low-angle aerial view of Parkfield, California, looking NNE. In the foreground is the San Andreas fault (red line), the source of the Parkfield earthquakes. Gordon P. Eaton, Director

CONTENTS

Overemphasis by the Public and Media of the Prediction Aspects of the Experiment

iii Fund Understanding of Societal Impacts of False Alarms . . .

1. Modified Mercalli intensity pattern for the 1966 Parkfield earthquake.

2. Significant earthquakes have occurred on the Parkfield section of the San Andreas fault at fairly regular intervals- in 1857, 1881, 1901, 1992, 1934, and 1966. The next significant earthquake is anticipated to take place within the timeframe 1983 to 1993. . . . . . . . . . . . . . . . . . 3
3. Conditional probability for the occurrence of major earthquakes along the San Andreas fault in the 30-year interval from 1988 to 2018.

4. Parkfield sign.

5. De Bilt, the Netherlands, east-west (DBN-EW) and north-south (DBN-NS) component seismograms for the 1922, 1934, and 1966 Parkfield events. Amplitude and time scales are constant.
6. Surface creep of the San Andreas fault at Middle Mountain (in mm) and the rate of seismicity under Middle Mountain (cumulative number of events with M ~ 1.5, depths~ 6.5 km since 1980) are shown. The Coalinga M 6. 7 earthquake caused the fault to slip left lateral and stopped the seismicity for slightly over a year. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
7. This cross section along the San Andreas fault cmnpares the amount of slip in centimeters during the 1966 mainshock (dark contours from Segall and Harris, 1986) to the seismic P wave velocities on the northeast side of the fault (shading and lighter contours in km/s from Michael and Eberhart-Phillips, 1991). The mainshock hypocenter (large star) was under a body of low velocity material but most of the slip occurred to the southeast where higher velocity material exists. White areas are small background earthquakes that tend to occur in the lower velocity material and around the edges of the mainshock. . . . . . .

8. Public Response Plans.

iv

10. The time of Parkfield status level, D through A, and their corresponding 72 hour probability of a M 6 earthquake are shown for period from the start of the experiment in 1985 through 1993. Although this report described a single A-level alert in late 1992, the experiment went to its second A-level in November 1993. The 72 hour probability for each status level is discussed in Parkfield Earthquake Prediction Response Plan [1988]. . . . . . . . . . . . . . . . . . . . . . . . . . . 12
11. Scientists from the USGS and officials from the California OES met with reporters on the lawn of the Parkfield Cafe for regular briefings during the level-A alerts in October 1992. . . . . . . . 13

v National Earthquake Prediction Evaluation Council Working Group

Ray Weldon

vi Massachusetts Institute of Technology

University of Oregon Summary

During the past century, earthquakes of M ,....., 6 have occurred with remarkable regularity on the San Andreas fault at Parkfield, California. Events occurred in 1857, 1881, 1901, 1922, 1934, and 1966. At least two of these events were preceded by large foreshocks and there is evidence for precursory creep of the shallow segment of the fault prior to the 1966 event. In 1984-1985, scientists developed and published a prediction, based on a model of "characteristic" earthquakes, that the next M ,....., 6 Parkfield event was expected in a time window centered on 1988, with 95% probability that the earthquake would occur by the end of 1992 [Bakun and Lindh, 1985).

Shortly after the publication of this prediction, with endorsement by the National Earthquake Prediction Evaluation Council (NEPEC), the United States Geological Survey (USGS) initiated the Parkfield Earthquake Prediction Experiment (the Experiment). With additional support from the state of California, the Experiment took on a public services aspect, as well as a geophysical aspect.

By late summer, 1992, the predicted event had not yet occurred. NEPEC chartered a Working Group to evaluate the Parkfield Earthquake Prediction Experiment. This group was asked a series of questions which are summarized below, along with the responses of this Working Group.

1) What is the current assessment of the prediction?

a) Is it still considered likely that an M ,....., 6 earthquake will occur in the short term?

Parkfield is still considered to be the most likely locality identified to "trap" a moderate sized earthquake. Empirically, no other location has demonstrated a sequence of earthquakes with as much regularity and as short an average recurrence time as the Parkfield sequence. The loading of the Parkfield segment of the San Andreas fault is unusually simple, with the creeping segment to the north leading to a continuous accumulation of strain on the locked segment near Parkfield. At this time, about as much strain has accumulated as was released in the previous event [Segall and Harris, 1986).

Estimates of the probability of the Parkfield earthquake occurring in the near future have been generated by a number of scientists. These estimates are based on a number of assumptions about the statistical behavior of faults, but all update the probability estimate to include the information that, as of the time of the estimate, the event has not yet occurred. Estimates of the probability of occurrence are clustered around a value of approximately 10% per year.

b) If the event does not occur by the end of 1992, what does that indicate about the original prediction? Was the basis for the prediction in error?

The original prediction was based on a rather specific set of assumptions. These include that the loading rate is constant, that failure of the same patch of the fault occurs at or below a threshold stress level, and that the stress drop is identical for each event [Bakun and Lindh, 1985). It was the adoption of this specific model that allowed the 1934 event to be ignored when evaluating the expected time of the next "characteristic earthquake," leading to a small uncertainty in the expected time of the event. This model is now known to be too simple to apply to Parkfield. In retrospect, the original prediction should have included an assessment of the probability that the model assumptions were correct, in addition to considering the uncertainties related to data noise in fitting the assumed model.

But failure of this specific prediction does not negate the consensus of the Working Group that Parkfield is still the most likely place in the United States to trap an earthquake and that there is a relatively high probability that this event will occur on a time scale of a decade or so. In addition, Parkfield is unique because the location of the likely nucleation point can be estimated. And there is a long baseline of measurements already established there.

2) What have we learned during the experiment, both from the scientific and response community aspects? What have been the principal benefits that have come from both of these aspects of the experiment? While the major scientific impact of the experiment will

not occur until the next Parkfield earthquake is trapped in the dense web of instrumentation operating there, there have already been important benefits reaped from the real-time exercise created by the response community interacting with the scientific comunity. Because of its low population density, Parkfield has been an ideal location to begin this process. The Experiment has been a public policy success, with positive implications for response and mobilization to possible future earthquake alerts elsewhere. Cooperation at Parkfield has produced the California Short- Term Earthquake Prediction Response Plan. The "A"-level alert in October, 1992, provided a realistic "fire drill" to test the implementation of this plan. The press has learned how to portray alert levels and associated probabilities.

Several scientific results are also notable. For example, geodetic data have shown resolvable differences between the 1966 and 1934 events; Parkfield events are similar, yet still show differences large enough to violate the specific assumptions of the model on which the original prediction was based. In addition, the fault zone and asperities have been imaged at unprecedented resolution, with the identification of a low-velocity zone, perhaps related to high fluid pressure.

There has also been substantial technology transfer resulting from the Experiment, including increased expertise in siting of borehole instruments, experience with real-time seismic networks, and improved instrument design. These advances have made the installation of other instrumentation, such as that monitoring the Hayward fault, more cost effective.

3) Where should the experiment go in the future? What modifications should be made to scientific monitoring? What research efforts should receive highest priority? Should there be any modification in the agreements that govern the interaction between the USGS and the State of California with regard to hazard warnings?

The science of understanding the earthquake source is limited by the dearth of observations throughout the earthquake cycle. Parkfield is the most likely place yet identified to trap a moderate earthquake in a densely instrumented region and the best locale identified to answer a number of important scientific questions about the seismic source. Because the Parkfield segment of the San Andreas fault is loaded by the creeping section at one end, it provides a setting intermediate in scale between the simplicity of the laboratory and the complexity of most other faults. Substantial resources have been invested in setting up the Experiment and the marginal costs associated with continued operation of the experiment are minimal. Although the estimated annual probability of about 10% per year is the highest proposed for any specific location, and high enough to make the area scientifically and societally interesting, it is not high enough to ensure that an earthquake will occur on a time scale of a few years. Thus the USGS should view the Experiment as a long-term commitment.

In this context, the Working Group recommends that USGS continue the Parkfield Experiment and assign it a high priority. This includes having a committed, long-term Project Scientist with sufficient resources available to deal with the scientific, response planning, and public relations priorities of the Experiment. These aspects of the Experiment are likely to be in a state of high activity simultaneously when alerts are called.

A long-term plan is required for replacing failed or obsolete equipment, in particular, strain meters. This long-term plan should also include periodic peer review of the Experiment, with possible redirection and reordering of priorities. In recognition of the long-term aspects of monitoring the preparation zone throughout the earthquake cycle, and the problems due to transients and costs associated with new installations, the USGS needs to ensure access to monitoring sites. The USGS should attempt to acquire control of land where instruments are located to avoid disruption when landowners change.

There should also be a commitment to continue the public policy aspects of the Parkfield Experiment through the earthquake cycle. There is still much to be learned about public response to perceived false alarms, perception of risk assessment, and warning thresholds.

Introduction

Before the mid-1980's, the United States' earthquake prediction program was in a reconnaissance mode, with monitoring programs broadly distributed across regions of the nation that had high seismic activity. In 1984, scientists at the USGS began to plan spatially focused earthquake experiments, with a primary goal of evaluating possible precursory phenomena. The issue that had to be addressed at that time was where to use the limited resources available with the best chance for evaluating these phenomena. The immediate question that had to be answered was where to deploy the recently developed dual-frequency laser geodimeter. Sites under consideration included Parkfield and San Juan Bautista.

In 1985, scientists at the USGS published a prediction that the next Parkfield earthquake was expected in a time window centered on 1988, with a 95% probability that the earthquake would occur by the 'end of 1992 [Bakun and Lindh, 1985]. This prediction was based on a specific model of the "characteristic" Parkfield earthquake, discussed in more detail below. NEPEC reviewed the prediction favorably and the USGS decided to locate their focused experiment in Parkfield.

In April, 1985, the Director of the USGS sent a letter to the Governor of California informing him about the possibility of an earthquake at Parkfield. California and USGS each contributed $1M to begin instrumentation of the Parkfield area. Over 20 observational networks have been installed, including seismometers, creep meters, borehole strain meters, the two-color laser geodimeter, water wells, and magnetometers. Five of these networks are monitored in real time.

There were two scientific goals for the Parkfield Earthquake Prediction Experiment: 1) To record the geophysical details before and after the expected earthquake; and 2) To issue a short-term prediction. In addition, with the involvement of the State of California, the Experiment took on an important public policy aspect, serving as a test bed for communication between earthquake scientists and public officials. The rural nature of the community made it an ideal location to carry out the Experiment.

A formal series of alert levels has been set up, triggered by phenomena such as fault creep or earthquake activity in

Lessons for the Scientific Community

regarding the response of the Parkfield region to smaller earthquakes nearby, and to larger earthquakes at some distance from the region (e.g., Kettleman Hills; Lorna Prieta; Coalinga; and Landers). These observations are extremely relevant to the design of monitoring installations and establishment of alert levels in other regions; they will also contribute to the revision of alert levels for Parkfield.

Scientific Results to Date The scientific experiments at Parkfield fall into three classes: those designed to monitor possible earthquake precursors, those designed to monitor the behavior of the region around the anticipated rupture nucleation point, and those designed to study the effects of earthquake-induced ground shaking on both natural and manmade structures. Many of the experiments will not yield their full scientific value until the earthquake has occurred, when the difference in baseline before and after the earthquake can be determined, and the effect of the earthquake on manmade structures can be analyzed. However, some of the instruments installed at Parkfield have shown temporal variations (particularly in levels of seismicity) that have led to important scientific advances even during the current pre-earthquake monitoring period.

One such example is the borehole seismic array that has been installed at Parkfield. Because of the extremely low noise present on the downhole seismometers, a complete seismic catalog down to magnitude near zero has been recorded for the region. These seismic observations have permitted the recognition of slow, microseismic slip events [Malin and Alvarez, 1992) and the identification of periods of higher seismicity that may be related to the locations of future slip events [Roeloffs and Langbein, 1994). Such events would have been impossible to identify with conventional surface seismic arrays, and have illustrated the importance of including downhole installations in other prediction arrays.

Because of the focus on Parkfield as a center of earthquake prediction efforts, the seismic history of the region has been scrutinized during the years since 1985. This has led to major advances in our understanding of the details of the previous Parkfield earthquakes. For example, differences in the extent of the rupture plane between the 1934 event and the 1966 event are now recognized [Segall and Du, 1993). The pre-1930 historical events in the region have been more closely studied, resulting in the recognition of 10 previously unrecognized events with magnitudes above 5.5 within 100 km of Parkfield [e.g., Toppozada et al., 1990). This new information has led to debate about just how similar the 1934 event and the 1966 event really are, and whether all of the earlier events were on the same fault patch, casting doubt on the validity of applying the "characteristic earthquake" hypothesis, in its simplest form, at Parkfield.

The focus on Parkfield has also resulted in reexamination of the predicted earthquake recurrence interval there based on the simplistic idea of uniform loading rate. For example, earthquakes on nearby faults, such as the 1983 Coalinga event, may have an effect [e.g., Simpson et al., 1988; Tullis et al., 1990). In addition, the results of viscoelastic relaxation following the great 1857 earthquake may lead to a decrease in recurrence time during the longer-term San Andreas seismic cycle [Ben-Zion et al., 1993). The possibilities that the segment boundaries at Parkfield are not geometrically well defined [Nishioka and Michael, 1990) nor visible as steps in the seismically defined fault zone [Eberhart- Phillips and Michael, 1992; Michael and Eberhart-Phillips, 1991) and that successive ruptures may overlap in spatial extent has been recognized, along with the tendency for earthquakes to cluster [Kagan and Jackson, 1991). The rate of strain accumulation since the 1966 event has also been examined, and used to constrain the moment deficit since 1966, which can be used for estimates of recurrence time [see summary by Roeloffs and Langbein, 1994).

Partially due to the focus on Parkfield, understanding of the statistics of "earthquake prediction" (in terms of estimation of recurrence times and their uncertainties) has advanced considerably. If we were to calculate the mean recurrence time and probability of the Parkfield earthquake as of today, we would do it by a more sophisticated technique, and using different uncertainties, than used in 1984. We would also recognize that various models of earthquake recurrence are possible. The question of whether to include the 1934 event in the probability calculation, and how to estimate the variability in the recurrence time ("shape factor") has led to further evaluations of the details of probabilistic calculations [Nishenko and Ruland, 1987; Savage, 1991; Roeloffs and Langbein, 1994). These advances in understanding of methods of estimation of earthquake probabilities have been applied to other regions along the San Andreas fault system [e.g., Working Group on California Earthquake Probabilities, 1988, 1990; NEPEC/CEPEC/SCEC Working Group, 1992; Jones et al., 1991] and elsewhere, including volcanic unrest at Long Valley Caldera [Hill et al., 1991].

Figure 6. Surface creep of the San Andreas fault at Middle Mountain (in mm) and the rate of seismicity under Middle Mountain (cumulative number of events with M ~ 1.5, depths ~ 6.5 km since 1980) are shown. The Coalinga M 6.7 earthquake caused the fault to slip left lateral and stopped the seismicity for slightly over a year.

Possible precursory phenomena have been scrutinized. The Parkfield project required a careful evaluation of various possible precursors, in order to decide what to measure and record at the start of the experiment. During the course of the experiment, it has become clear that some possible precursory phenomena were not being studied (e.g., the ULF electromagnetic signals) and that some other measurements related to precursory phenomena may be relatively problematic (e.g., rainfall-induced creep events). Theoretical modeling of expected strain accumulation prior to rupture [e.g., Left Lateral

6.5 Tullis and Stuart, 1992] has forced a careful look at the time scales and spatial scales over which precursory signals might be visible. This has shown that, for certain scenarios, we would expect to capture precursory signals on the currently designed and located instruments, but for other scenarios we would not be able to resolve these precursory signals. Thus, we have improved our knowledge of optimal experiment design for this location, and can now apply these techniques to future experiment design in other regions.

The Parkfield experiment required a working collection of monitoring instrumentation designed to observe long-term and short-term changes in the fault near the inferred point of rupture initiation. As the experiment progressed, equipment failure and changes required modification of both the hardware (e.g., the cable connections on some downhole equipment) and the science plan (e.g., the alert levels triggered by creep events during heavy rainfall). Many of the bugs related to real-time operation of this monitoring system, and related to real-time operation of this monitoring system, and to design and installation improvements, are now worked out. Similar monitoring systems are now being installed along the Hayward fault system, near San Francisco, and along parts of the southern California fault systems. There has been a great savings in cost for these installations because of the experience gained at Parkfield. The real-time aspect of such data collection has been greatly advanced by the expertise gained in the Parkfield experiment, so that Parkfield serves as a starting model for the design of similar, but younger and more sophisticated, systems.

Lessons for the Response Community

The Parkfield Earthquake Prediction Experiment has been a success from the response community perspective as demonstrated by the response of emergency management services, in conjunction with the scientific and media communities, to the October, 1992, "A"-level alert. During the Experiment, the emergency response community has been able to prepare and exercise its response plan for the "A"- and "B" -level alerts, including the notification of local governments. Cooperation at Parkfield between the USGS and the California Office of Emergency Services (OES) has resulted in the State of California Governor's Office of Emergency Services Parkfield Earthquake Prediction Response Plan [OES, 1988], which was the basis for the response actions during the "A" -level alert. Based on the Parkfield Earthquake Prediction Response Plan, California OES developed and published the California Short- Term Earthquake Prediction Response Plan [OES,1990]. This second document describes state agency and county government actions to be taken in response to any scientifically-driven earthquake alert or prediction at any other location in the state.

Figure 9. Flow of information resulting from an A-level alert at Parkfield. Signals from USGS instruments at Parkfield (1) are telemetered to the USGS Western Region Headquarters in Menlo Park (2) for analysis by USGS scientists. The California Office of Emergency Services (OES) in Sacramento {3) is notified by the USGS for all A-, B-, and C-level alerts. A-level alerts are immediately transmitted by OES to county governments (4) using established OES emergency communication channels. County officials also have established procedures to relay the warning to such groups as law enforcement agencies, fire departments, utility companies, and school districts.

Problems Identified

ment was fully operational, there has been little provision for ongoing review. The annual expenditure and the importance of this project to the prediction community requires periodic external review (suggested annually) to assure its viability and credibility. An assessment of the progress and quality of the project by a panel of experts would ensure that appropriate scientific and response objectives are kept at the forefront of the project and that the Experiment changes in response to new developments, as well as doing much to dispel criticism of the experiment. The review might be done by a subset of the NEHRP Review Panel, which could oversee the general worth of the project in the context of the NEHRP program and give advice on the direction and disposition of financial support.

Scientific Data Not Readily Accessible

There is a perception that some investigators funded under the Parkfield project have been too busy, inadequately supported, or perhaps reluctant to release their data in a timely manner. These delays have reduced the usefulness of the data to the Parkfield team. An implicit philosophy of the Experiment is that all investigators must work in a cooperative effort in order to provide the data in a timely fashion for time-dependent decisions regarding designations of alerts and to assure timely review of data. If data gathered within the Parkfield experiment are important enough to be funded, they are important enough to be made available on a short enough time frame to contribute to the decision making process.

The question of data accessibility is also related to the philosophy of data management. It is problematic that there is not now a modern data base management scheme implemented for the Parkfield data. Individuals who wish to examine the data must be familiar with the location(s) of particular data files in USGS computers (which require secure passwords) in order to retrieve the data. The user then must plot the data with his or her own software and has little information on attributes or an understanding of how the data may relate to the earthquake process. This is a serious limitation and does not allow ready access to the user.

While we realize that data acquired in sophisticated experiments must be carefully processed and scrutinized by individual investigators, we suggest that because of the specific NEHRP financial commitment and national stature of the Parkfield Experiment that a modern Data Base Management System be implemented for archiving and managing the Parkfield data. This requires that all on-line recorded data be available immediately after processing and that offline and low-rate data, requiring editing and processing, should be archived within a sufficiently short time to be useful to the decision makers. This aspect is also important for providing on-line access to designated users. A user-friendly Geographical Information System available with sub-licenses to the participating Principal Investigators would also be useful to manage, correlate, and display the archived data.

We also note that some additional data, now collected at Parkfield, should be considered for transmission and recording at Menlo Park. These include the electrical resistivity array and the ULF EM measurements.

A possible restriction in the Parkfield data distribution scheme is that the downhole and acoustic data, which are recorded on tape at Parkfield, are then sent to Duke University and UC Berkeley for analysis. Because of these time-consuming steps they are not available to the USGS team in a timely or efficient manner. A plan should be made to make these data more readily available to the USGS in Menlo Park quickly enough to be useful in a short-term alert and prediction mode.

Overemphasis by the Public and Media on the Prediction Aspects of the Experiment

The original objectives as stated by Bakun and Lindh [1985) were that the Parkfield experiment was to monitor the details of the final stages of the earthquake preparation process. The instrumental aspect of the Parkfield project was designed primarily as a surveillance project. However, with the involvement of California funding, the Experiment took on a short-term earthquake prediction objective that is perceived by the scientific and emergency response community as the "national" prediction experiment.

The general public now perceives the Experiment primarily as a short-term earthquake forecasting project with an inherent expectation to accurately predict an earthquake, while the scientific community views it not only as a short-term prediction experiment, but also as an effort to trap a moderate earthquake within a densely instrumented network. It is important to educate the public that there is great value to this monitoring effort even if the prediction effort is unsuccessful.

How Should the Experiment be Modified in the Future?

The Working Group recommends that the Experiment continue as a specific coordinated scientific effort in monitoring through the earthquake cycle, as well as in earthquake prediction. Although the annual probability of the expected characteristic event occurring, about 10% per year, is sufficiently high that the Experiment should continue, it is not so high that we can expect the Experiment to be over on a time scale of a few years. Thus, it is prudent to take a long-term perspective in contemplating the future of the Experiment. Issues to be addressed include costs, relocating the Experiment, mechanisms for ongoing evaluation of the Experiment, and future USGS response efforts.

What are the Incremental Costs Associated with the Experiment?

In addressing possible modifications of the Experiment, it is important to place the budget in context. The internal budget for the Parkfield Experiment is $1.4 million/year, but this amount is much more than the true incremental cost of the Experiment. The great majority of these expenditures are for salaries of scientists (prorated based on estimates of the fraction of time spent on the Experiment). Since these salaries would continue to be paid if the Experiment were ended, termination of the Experiment would not allow salary costs to be cut, although the effort could be deployed elsewhere. Real-time monitoring is highly automated, making use of computer systems that already exist. In the judgment of the Working Group, turning off the Experiment completely would save less than $200,000 per year in the internal program. The external program has a budget of rv$400,000 per year. Thus the total incremental cost of the Experiment is approximately $600,000 per year -just over 1% of the USGS NEHRP budget.

Should the Experiment be Moved?

USGS is under tremendous pressure to "do something" in more heavily populated areas, such as near the Hayward fault. The argument is made "How can we justify spending money trying to predict an earthquake in an area with as few taxpayers as Parkfield, when so many more people would be affected by an earthquake on the Hayward fault?" In answering this question it is important to remember that we do not at this time know which, if any, of the instruments monitoring pre-earthquake activity will measure premonitory signals. Thus,· we do not know the requirements for instrumentation to install to be useful for earthquake prediction in heavily populated areas. It is extremely important to answer this question as quickly as possible. Impatient as we may be with the lack of the expected event, Parkfield remains the most likely place identified to trap a moderate earthquake. It is probable that precursors would only be detected by instruments located quite close to the earthquake preparation area. Parkfield represents one of only a few places where this preparation area has been specifically identified. In addition, the long baseline of measurements already made at Parkfield represents an investment that should be used, not walked away from.

While there are benefits to having the monitoring effort visible to the public, evaluating the possibility of precursors should have highest priority. The best place to find out if they exist is at Parkfield, where the target area is well defined. It is important that adequate resources be made available to keep the Experiment alive and evolving, not placed in mothballs.

Long-Term Aspects of the Project

Commitment for Long-Term Management

There has been an apparent lack of commitment by the USGS for long-term management and identification of importance for the Parkfield project. The project has included several Project Chiefs whose terms have been less than that of their corresponding USGS Branch Chiefs, with some in that position for as little as one year. The Experiment is sufficiently important that it deserves a commitment by the USGS for a long-term management team with the requirement that those individuals become totally familiar with the objectives and results, and have a good working relationship with participants. A Parkfield Project Chief should remain in that position for sufficient time to be able to make critical decisions based on a broad experience of observations and predictive models. Further she/he should have the status of major contributor of the NEHRP program and be involved in its planning and implementation.

Acquisition of Land Rights at Parkfield

There is apparently a problem with long-term rights of access to instrumentation sites at Parkfield. We believe that the importance of this experiment merits a cooperative commitment from federal, state, and county officials to assure land accessibility and that an effort should be made with private land owners to gain long-term commitments and accessibility. This aspect of the project would be enhanced by a local coordinating committee with members from the Parkfield community. This may require consideration of the instrumentation sites as designated land within a state or federal easement program, as a national earthquake study area, or other designated scientific establishment similar to the Stanford Linear Accelerator, the Superconducting Super-Collider site, etc.

Reassess Project Periodically to Modify, Upgrade and Acquire New Equipment

The scientific objectives of the Parkfield experiment are now reasonably well supported with modern instrumentation. It is imperative that these instruments, particularly the strain meters, be maintained, upgraded, and replaced when they fail. In addition, there are some methodologies that deserve additional consideration at Parkfield:

Broad Band Seismological Studies - An important issue is that there is only one broad-band, continuously recording system at Parkfield, installed in late 1991; it is not apparent to the Working Group how data from this instrument are used by those involved in the Parkfield Experiment. In the past few years, broad-band seismic data, 50 Hz to 0.03 Hz, recorded on wide-dynamic range digital recorders (either on site or via digital transmission) have been shown to be very useful for retrieving source and transmission properties of moderate to large earthquakes. In the recent earthquake for which the "A"-level alert was issued, broad-band techniques could have been used to rapidly obtain accurate source properies (e.g., rupture geometry, time and space properties, relationships to geometrical features of the proposed nucleation site at Middle Mountain, etc.). The Committee was unaware of to what extent, and how rapidly, broad-band data had been used in evaluating this event.

With the availability of new wide dynamic range broad band seismometers (such as Guralps, Streckheisens or equivalent) and 24-bit recording for example with ReiTeks, there is a wealth of new information that can be gained by large dynamic range-wide band technology. For example, attenuation (weak vs. strong ground motion) over broad ranges of magnitudes and distances, source properties of small to moderate events, and accurate foreshock and aftershock assessment may yield important information on earthquake precursors. We suggest that a component of wide-dynamic range broad band coverage of the Parkfield area be implemented.

While real-time recording is perceived as costly to some, modern seismological instrumentation has been designed for on-line recording and is standard throughout the world's seismological community. These data would cost the same, if not more, if they were recorded on site or at Menlo Park in a time-delay mode. Moreover, if they were recorded onsite and transmitted to Menlo Park, then the timely nature of the data would be negated for warnings and alert. We suggest that real-time recording of seismological and related high sample-rate data be continued.

Parkfield Network Calibration- A general problem common to most short period seismic networks, including CAL-NET and hence Parkfield, is the lack of a systematic absolute calibration of the complete seismometer-recording systems for true ground motion. The USGS indicates that calibration pulses are recorded daily on the short period seismometers, but in reality little is done with the calibration data. The short period seismometers are calibrated only when they are brought to Menlo Park for bench tests where they can be absolutely calibrated (presumably very infrequently, i.e., every few years). If there is a large earthquake at Parkfield and the short period instruments have not been calibrated accurately, a major source of information, namely accurate ground motions (for the on-scale events for even small to moderate events), will be lost. This information is especially useful for assessing large ground motions and attenuation.

Network and other instrument calibrations should become part of the Parkfield archive accessible to all users. Additional information in a data base should include a history of each station, model numbers of components, maps of station locations, instrument modifications, calibration constants, etc. The public must be assured that the data acquired from this and any similar project are validated.

Paleoseismicity- The basic premise of the Parkfield experiment is the recurrence of "characteristic" M rv 6 events. However, because there is little trenching information in the Parkfield area (in part due to geography and other logistical problems) the long-term Holocene record has not been evaluated. Thus, we do not know if the typical event size and recurrence interval assumed from the six historic events also dominates the longer-term seismic patterns. For example, if paleoseismological studies revealed the presence of larger events, this would change our view of the typical Parkfield event. This information is crucial for making statistical assessments of earthquake probability and for evaluating the characteristic earthquake model. Some studies have been carried out addressing the feasibility of trenching near Parkfield. These studies need to be documented and critically assessed.

Assessments of Costs and Productivity of the Experiment

The Experiment should continue to evolve, and external review should be part of this process. We suggest that the project be reviewed periodically by a panel of independent scientists. An annual report, discussing productivity and cost, should be prepared and submitted to this panel. The report should contain a summary of the scientific and emergency response aspects of the project, as well as a listing of products (reports, catalogs, data archives, papers, presentations, etc.) supported in part or totally by the Parkfield experiment. This part of the report should be presented at a national meeting such as the AGU or SSA, with consideration of publication in a journal such as Eos, to inform the community of the status of the project.

An accurate assessment of the cost of the Parkfield experiments should be included in the report to the panel, assessing individual project costs and related salaries. A review of funds for USGS employees and contract employees should be scrutinized and the USGS management should ensure that funds for the Parkfield project are expended according to their intended use. Equipment acquisitions should be listed and a comparison of operating costs ofthe network and other systems should be made and reported periodically to the review panel. Further, the report should provide a list of instruments or projects that were considered each year for implementation, continuation, or deletion.

Recommendations for Future USGS Response Efforts

Strengthen the USGS Response Role

The USGS should recognize and provide support for the Experiment as a scientific experiment in the broader integrated context of an actual public policy activity. In general, the USGS must recognize the importance of developing written plans to provide the scientific and public information support that is needed during all stages of public alerts. The USGS should also consider institutional recognition for scientists who are committed to devoting portions of their careers to public policy and education without jeopardizing potential career advancement. The USGS should consider formal and institutional recognition of the important public policy role that scientists can play for all phases of natural hazards prediction and response.

The USGS should establish a formal protocol describing agency functions and personnel functions required to adequately support the media, the state and local governments, and the emergency management community, after issuing earthquake predictions. This protocol should be written in the form of a Standard Operating Procedure which describes the tasks to be performed by dedicated USGS personnel during earthquake alerts. The USGS should develop these Standard Operating Procedures by querying representatives of the media, state and local governments, and emergency management who depend on earth science information during alert periods. In the case of Parkfield, the USGS should be committed, as an institution, to preparing for an "A"- level alert. To date, the USGS has relied on individual scientists to provide a presence and continuity for interaction with state and local government officials. Even though individual Parkfield scientists have had a continuous commitment to providing earth science information to the media the USGS was unable to provide adequate representation t~ address media questions regarding the prediction during the "A" -level alert.

Figure 10. The time of Parkfield status level, D through A, and their corresponding 72 hour probability of a M 6 earthquake are s~own for period from th~ start of the experiment in 1985 through 1993. Although this report described a single A-level alert 1n late 1992, the expenment went to its second A-level in November 1993. The 72 hour probability for each status level is discussed in Parkfield Earthquake Prediction Response Plan (1988].

The USGS has strengthened its response role since the October, 1992 "A"-level alert by recognizing that the Parkfield Chief Scientist, Project Chief, and the Public Information Officer roles must be adequately covered by three scientists rather than one scientist, as has been the case for most of the duration of the Experiment. The Parkfield scientists at the USGS, Menlo Park, have already implemented this additional coverage of duties at least until the closure of the fiscal year, September 30, 1993. For additional "A"- level alerts, one scientist will run the field laser experiments, one scientist will coordinate the Menlo Park operations, and a three-scientist team will mobilize to Parkfield to provide earth science information to the media.

Review the Threshold Criteria Determining Alert Levels

The USGS should review the probability percentages, which act as threshold criteria for the Experiment alert levels, with respect to the following three concerns:

1) the level of accuracy and the statistical uncertainty associated with these probability percentages;

.001 A

2) the effect of "false alarms" on the credibility of the earthquake prediction process; and
3) the appropriate actions for the public and response community as reflected by these probability percentages and alert levels. An "A"-level alert for the Experiment has been assigned

a 37% probability for the occurrence of an earthquake of M rv 6 within 72 hours. The public and the media infer that the accuracy of this percentage is very high, reasoning that it has not been stated as, e.g., 36% or 38%. Since the uncertainty in this estimate is much larger than the reliability associated with a probability expressed to two significant digits, the scientists are implying to the public unrealistic accuracy. This leads to the misperception that scientists believe their probabilities to be more accurate than they are.

There is also a problem in the public perception of the Experiment created by the highest level alert - the "A" -level alert. The alert scale is saturated at a low probability level. It is easy (albeit incorrect) to associate the highest available alert level with a high absolute level of probability and the need for media attention. When the expected event does not follow the posting of the highest available alert level (as should usually be the case for the alert levels as defined), the alert is easy to perceive as a false alarm. Will the frequency of perceived false alarms associated with future alerts diminish the intended response and preparedness actions of the public, as well as begin to discredit the scientific credibility of the earthquake prediction process? The false alarm rate needs to be more fully understood by local governments and the public. Perhaps the most straightforward remedy would be to define a new alert scale, e.g., I-V, with the current "A" -level assigned level II. (Level I might be triggered, for example, by the "early warning" system under study.) Although this is more of a public relations issue than a science issue, public relations are an important aspect of the Experiment.

References

Jones, L. M., K. E. Sieh, D. Agnew, C. Allen, R. Bilham, M. Ghilarducci, B. Hager, E. Hauksson, K. Hudnut, D. Jackson, A. Sylvester, K. Aki, and F. Wyatt, Short-term earthquake hazard assessment for the San Andreas fault in southern California, U.S. Geological Survey Open-File Report 91-92, 1991.

Kagan, Y. Y., and D. D. Jackson, Seismic gap hypothesis: Ten years after, J. Geophys. Res., 96, 21,419-21,431, 1991.

Karageorgi, E., R. Clymer, and T.V. McEvilly, Seismological studies at Parkfield. II. Search for temporal variations in wave propagation using Vibroseis, Bull. Seismol. Soc. Am., 82, 1388-1415, 1992.

Malin, P. E., and M. G. Alvarez, Stress diffusion along the San Andreas fault at Parkfield, California, Science, 256, 1005-1007, 1992.

Michael, J., and D. Eberhart-Phillips, Relations among fault behavior, subsurface geology, and three-dimensional velocity models, Science, 259, 651-654, 1991.

Michael, A. J., and J. Langbein, Earthquake prediction lessons from Parkfield Experiment, Eos Trans. AGU, 74, 145-55, 1993.

Michelini, A., and T. V. McEvilly, Seismological studies at Parkfield: I. Simultaneous inversion for velocity structure and hypocenters using cubic B-splines parameterization, Bull. Seismol. Soc. Am., 81, 524-552, 1991.

NEPEC/CEPEC/SCEC Ad Hoc Working Group on the Probabilities of FUture Large Earthquakes in Southern California, Future Seismic Hazards in Southern California Phase I: Implications of the 1992 Landers Earthquake Sequences, California Division of Mines and Geology, 1992.

Nishenko, S. P., and R. Buland, A generic recurrence interval distribution for earthquake forecasting, Bull. Seismol. Soc. Am., 77, 1382-1399, 1987.

Nishioka, G. K., and A. J. Michael, A detailed seismicity study of the Middle Mountain zone at Parkfield, California,Bull. Seismol. Soc. Am., 80, 577-588, 1990.

OES, California Short-term Earthquake Prediction Response Plan, State of California, 1990.

OES, Parkfield Earthquake Prediction Response Plan, State of California, 100 pp., 1990.

Parkfield, California, Earthquake Prediction Scenarios and Response Plan, U.S. Geol. Surv. Open File Rep. 87-192, 1987.

Parkfield Working Group Report, Parkfield: First short-term earthquake warning, Eos Trans. AGU, 74, 152-153, 1993.

Roeloffs, E. and J. Langbein, The Parkfield,California,Earthquake Prediction Experiment, Rev. of Geophys., 92, 315, 1994.

Savage, J. C., Criticism of some forecasts of the National Earthquake Prediction Evaluation Council, Bull. Seismol. Soc. Am., 81, 862-881, 1991.

Segall, P., andY. Du, How similar were the 1934 and 1966 Parkfield earthquakes?, J. Geophys. Res., 98, 4527-4537, 1993.

Segall, P., and R. Harris, Slip deficit on the San Andreas fault at Parkfield, California, as revealed by inversion of geodetic data, Science, 299, 1409-1413, 1986.

Simpson, R. W., S. S. Schulz, L. D. Dietz, and R. 0. Burford, The response of creeping parts of the San Andreas fault to earthquakes on nearby faults: Two examples, Pure Appl. Geophys., 126, 665-685, 1988.

Toppozada, T. R., C. Hallstrom, and D. Rausam, M7, 5.5 earthquakes within 100 km of Parkfield, California, Seismological Research Letters, 61, 42, 1990.

Tullis, T. E., W. D. Stuart and R. W. Simpson, Instability model for Parkfield earthquakes, including the effect of New Idria, Coalinga, and Kettleman Hills events, Eos, Trans. AGU, 71, 1632, 1990.

Tullis, T. E., and W. D. Stuart, Premonitory changes prior to a model Parkfield earthquake, Eos Trans. AGU, 79, 397, 1992.

Working Group on California Earthquake Probabilities, Probabilities of large earthquakes occurring in California on the San Andreas Fault System: U.S. Geological Survey Open File Report, 88-998, 62 pp., 1988.

Working Group on California Earthquake Probabilities, Probabilities of Large Earthquakes in the San Francisco Bay Region, California, U.S. Geological Survey Circular 1059, 1990.

Periodicals

Earthquakes & Volcanoes (issued bimonthly).

Preliminary Determination of Epicenters (issued monthly) .

Technical Books and Reports

Professional Papers are mainly comprehensive scientific reports of wide and lasting interest and importance to professional scientists and engineers. Included are reports on the results of resource studies and of topographic, hydrologic, and geologic investigations. They also include collections of related papers addressing different aspects of a single scientific topic.

Bulletins contain significant data and interpretations that are of lasting scientific interest but are generally more limited in scope or geographic coverage than Professional Papers. They include the results of resource studies and of geologic and topographic investigations, as well as collections of short papers related to a specific topic.

Water-Supply Papers are comprehensive reports that present significant interpretive results of hydrologic investigations of wide interest to professional geologists, hydrologists, and engineers. The series covers investigations in all phases of hydrology, including hydrogeology, availability of water, quality of water, and use of water.

Circulars present administrative information or important scientific information of wide popular interest in a format designed for distribution at no cost to the public. Information is usually of short-term interest.

Water-Resource Investigations Reports are papers of an interpretive nature made available to the public outside the formal USGS publications series. Copies are reproduced on request unlike formal USGS publications, and they are also available for public inspection at depositories indicated in USGS catalogs.

Open-File Reports include unpublished manuscript reports, maps, and other material that are made available for public consultation at depositories . They are a nonpermanent form of publication that may be cited in other publications as sources of information.

Maps

Geologic Quadrangle Maps are multicolor geologic maps on topographic bases in 7 112- or 15-minute quadrangle formats (scales mainly 1:24,000 or I :62,500) showing bedrock, surficial, or engineering geology. Maps generally include brief texts; some maps include structure and columnar sections only .

Geophysical Investigations Maps are on topographic or planimetric bases at various scales; they show results of surveys using geophysical techniques, such as gravity, magnetic, seismic, or radioactivity, which reflect subsurface structures that are of economic or geologic significance. Many maps include correlations with the geology.

Miscellaneous Investigations Series Maps are on planimetric or topographic bases of regular and irregular areas at various scales; they present a wide variety of format and subject matter. The series also includes 7 1/2-minute quadrangle photogeologic maps on planimetric bases that show geology as interpreted from aerial photographs. Series also includes maps of Mars and the Moon .

Coal Investigations Maps are geologic maps on topographic or planimetric bases at various scales showing bedrock or surficial geology, stratigraphy, and structural relations in certain coal-resource areas.

Oil and Gas Investigations Charts show stratigraphic information for certain oil and gas fields and other areas having petroleum potential.

Miscellaneous Field Studies Maps are multicolor or black-andwhite maps on topographic or planimetric bases on quadrangle or irregular areas at various scales. Pre-1971 maps show bedrock geology in relation to specific mining or mineral-deposit problems; post-1971 maps are primarily black-and-white maps on various subjects, such as environmental studies or wilderness mineral investigations .

Hydrologic Investigations Atlases are multicolor or black-andwhite maps on topographic or planimetric bases presenting a wide range of geohydrologic data of both regular and irregular areas; principal scale is 1:24,000, and regional studies are at 1:250,000 scale or smaller.

Catalogs

Permanent catalogs, as well as some others, giving comprehensive listings of U.S. Geological Survey publications are available under the conditions indicated below from the U.S . Geological Survey, Books and Open-File Reports Sales, Federal Center, Box 25286, Denver, CO 80225 . (See latest Price and Availability List.)

"Publications of the Geological Survey, 1879-1961" may be purchased by mail and over the counter in paperback book form and as a set of microfiche.

"Publications of the Geological Survey, 1962-1970" may be purchased by mail and over the counter in paperback book form and as a set of microfiche.

"Publications of the Geological Survey, 1971-1981" may be purchased by mail and over the counter in paperback book form (two volumes, publications listing and index) and as a set of microfiche.

Supplements for 1982, 1983, 1984, 1985, 1986, and for subsequent years since the last permanent catalog may be purchased by mail and over the counter in paperback book form .

State catalogs, "List of U.S. Geological Survey Geologic and Water-Supply Reports and Maps For (State) ," may be purchased by mail and over the counter in paperback booklet form only.

"Price and Availability List of U.S. Geological Survey Publications," issued annually, is available free of charge in paperback booklet form only.

Selected copies of a monthly catalog "New Publications ofthe U.S. Geological Survey" are available free of charge by mail or may be obtained over the counter in paperback booklet form only. Those wishing a free subscription to the monthly catalog "New Publications of the U.S . Geological Survey" should write to the U.S . Geological Survey, 582 National Center, Reston, VA 22092.

Note.--Prices of Government publications listed in older catalogs, announcements, and publications may be incorrect. Therefore, the prices charged may differ from the prices in catalogs, announcements, and publications.

Where this page came from

This page was imported from U.S. Geological Survey. Published by the U.S. Geological Survey and, as a work of the United States government, in the public domain.

Nobody has written it yet — it is the source material at a new address, which is why search engines are asked to skip it and why no one earns from it. It is up for grabs: take it on, and it is yours to rewrite and to earn from.