Landers/Big Bear Aftershock Data
- Network Code: XB
- Experiment duration: 1992-06 to 1992-08
- Number of events: —
- Prinicpal investigators: Adam Edelman, Frank Vernon
- Dataless SEED volume currently unavailable. Please visit the IRIS DMC
A collaborative study to record aftershocks was undertaken by several Southern California Earthquake Center (SCEC) institutions following the June 28, 1992 Landers and Big Bear earthquakes. The M7.6 Landers earthquake occurred at 04:58 PDT approximately 6 miles north of Yucca Valley along the southern extension of the Johnson Valley fault. The earthquake produced over 70 km of ground rupture, with cumulative right-lateral offsets of 3 to 6.5 m along segments of the Homestead Valley, Emerson Lake, and Camp Rock faults, all of which were involved in the sequence. The primary aftershock activity occurred in a narrow band from 33.6°N, -116.2°W north to 34.5°N, -116.6°W, with a separate cluster at 35.0°N, -117.0°W. The M6.5 Big Bear earthquake occurred at 08:04 PDT approximately 35 km west of the Landers epicenter in the San Bernadino Mountains, involving left-lateral slip on a fault that strikes N45OW and dips 70° SE. The Big Bear aftershocks occur over an area about 30km long, extending northeast from 34.4°N, -116.6°W
Personnel from the California Institute of Technology (Caltech), San Diego State University (SDSU), the University of California at Santa Barbara (UCSB), the University of California at San Diego (UCSD), and the University of Southern California (USC) participated in the installation and maintenance of the portable stations. Within 12 hours, four portable instruments were installed and started recording data, and an additional 18 were up and running within 48 hours. Approximately 10 Gb of raw data from thirty-one individual stations were acquired in the 2 months following the main shocks.
The task of processing these data was undertaken by UCSD and consisted of three main steps:
- Compilation of portable data. All timing corrections and event associations.
- Integration of non-PASSCAL data. AIl M4.0 and larger events from Lamont Doherty Earth Observatory's (LDEO) and SDSU's entire portable data sets have been associated with the PASSCAL set.
- Compilation of station information. All station coordinates, recording parameters, and response information have been compiled into database format.
In addition, all event header information and phase arrival data have been converted into the SCEC database format for this release of data, which contains all altershocks of magnitude 4.0 and greater or recorded at five or more of the portable stations and have an unambiguously associated Caltech CUSP ID.
The following participants collected or provided data for this product:
- James Batti, University of California at San Diego
- Michael Blackman, University of California at San Diego
- Rob Clayton, California Institute of Technology
- Steve Day, San Diego State University
- Adam Edelman, University of California at San Diego
- Ned Field, Lamont Doherty Earth Observatory
- Marina Glushko, University of California at San Diego
- Katrin Hafner, California Institute of Technology
- Egill Hauksson, California Institute of Technology
- Tom Henyey, University of Southern California
- Dave Johnson, California Institute of Technology
- Yong-Gang U, University of Southern California
- Harold Magistrale, San Diego State University
- Aaron Martin, University of California at Santa Barbara
- Bruce McManus, University of California at San Diego
- Jim Morl, United States Geological Survey
- Craig Nicholson, University of California at Santa Barbara
- Sanjay Phatac, San Diego State University
- Michelle Robertson, University of Southern California
- Jennifer Scott, California Institute of Technology
- Leon Teng, University of Southern California
- Frank Vernon, University of California at San Diego
- Hans Van De Vrugt, San Diego State University
- Joel Wedberg, San Diego State University
The initial response to the earthquake involved the mobilization and deployment of all available equipment to the main shock areas. The first day of deployment was accomplished with minimal coordination between the field teams due to the difficulty of communications and the delays inherent in processing and evaluating regional network data in such an extreme situation. The first portable stations were set up in the regions of the Landers and the Big Bear epicenters. By the end of the second day better communication procedures were in effect and all the SCEC seismic field teams were coordinating their efforts to maximize the coverage of the aftershock zones.
Communications by fax and by phone with Southern California Seismic Network personnel provided necessary information to help make decisions about the locations where new stations should be deployed to complement existing network coverage and known portable installations. Within 12 hours of the Landers earthquake there were 4 operational portable stations and in the following 36 hours an additional1B stations were installed, including the LDEO portable stations that were installed concurrently in the eastern Coachella Valley. The station locations (Figure 1), and have been calculated to 1 meter accuracy with the use of differential GPS (Scott et af., 1994). Table 1 shows the station locations, number of triggers, and the range of epicentral distances, Table 2 shows the station equipment particulars, including recording parameters.
The first stations were deployed using equipment which had been initially located in southern California. By the second day additional equipment was received from the IRIS PASSCAL Instrument Center and from the USGS. There were five primary station configurations deployed. Three of these configurations (Figure 2) were high frequency stations using Mark Products L22-D seismometers, andlor Kinemetrics FBA-23 accelerometers recorded on either Reftek R172A-02, or Kinemetrics SSR-1 dataloggers. The remaining configurations (Figure 3) used either Streckeisen STS-2 or Guralp CMG-3T broadband seismometers with Reftek R172A-02 dataloggers. The broadband sensors were much more difficult to install and maintain due to their sensitivity to the environment and, in particular, to temperature gradients. But with proper siting and insulation it was possible to acquire high quality data from these sites. Each station was powered from 12 V batteries which were either exchanged during maintenance or charged in place. The charging was done by a solar panel or by a local 120 V AC source. Timing at all stations was based on the OMEGA time system.
Both the REF TEK R172A-02 and the Kinemetrics SSR-1 (SSR-1) dataloggers utilize 16-bit AID convertors. The R172A-02 was run in event trigger mode at either 250 or 100 s/s. The SSR-I was run in event trigger mode at 200 s/s. Both loggers employed an STA/LTA (short term to long term average) triggering algorithm.
The LDEO Coachella Valley stations recorded Kinemetrics SH-1 and SV-I velocity sensors, and Terra Tech FBA accelerometers with Kinemeterics PRS-4 dataloggers. Further information regarding the LDEO installations is available from Ned Field at the LDEO.
The total system response for each instrument/parameter configuration has been calculated and can be used to correct for instrument characteristics. Figures 4, 5, 6, 7, 9, 10, 11, 12, 13 show the normalized velocity and acceleration response. Amplifier gain and additional calibration information can be retreived from UCSD.
Twenty-one stations were operational by July 11, two of which were moved during the experiment to get better coverage of the northern aftershocks. After nearly three weeks of operation the network was reduced to 16 stations and by early August a skeleton array of 6 stations was left until early September. After the network was configured for 16 stations the field station visits were reduced from nearly daily down to a weekly or bi-weekly schedule. Figure 14 shows the status of each station throughout the experiment including the time periods that data were recovered along with the sensors that were deployed.
Sensor placement for each station was dictated by the local geology. The following stations had sensors installed directly on hard rock outcroppings of granitic composition: BN]I, EORD, ERWL, FORF, GRAV, HILL, IRON, LPMA, UCVF, UGGP, and YKNF. Most of these sensors were buried with loose gravel. The stations FIRE, GPOE, GRAC, LADY, RIMR, and VVST were located in very shallow sediments and were buried approximately 1 ft while BRDO and EDC2 were located in sedimentary rock. Station VlPC had its 515-2 mini-vault plastered directly to a hard-rock granite outcrop, and was covered with fiberglass insulation. The 515-2 mini-vault at JFRG was plastered to a slightly weathered granite outcrop, covered with fiberglass insulation and buried under a layer of soil. Station BRCC was installed inside a Joshua Tree Monument campground building with its 515-2 mini-vault plastered directly to the concrete foundation. The building was located on weathered granite. Station CG1S was located in a mine tunnel in hard rock.
The processing scheme required several steps: raw data retrieval followed by formatting, quality control, timing corrections, and event association. A Sun Sparc field computer was set up in a motel room several days after the main shocks. Much of the data reformatting, quality control and timing corrections were performed on this field computer. The computer was operated for about three weeks and data collected afterwards were processed at one of the participating institutions.
Initially, a REFTEK Exabyte drive was used to dump the data at each station, but it was found to be more reliable to exchange hard discs and copy data via the SCSI bus on a Sun Sparc station. After the raw data were downloaded to disk, TAR backup tapes were made. Next the data were converted to PASSCAL SEGY format and reviewed using PASSCAL's quick look program PQL to evaluate the station performance. Data recorded on SSR-ls were brought back to SDSU, then copied to a Sun Sparc station and converted to PASSCAL SEGY format.
Errors in timing were also corrected at this point. The common timing errors that were present in the raw data resulted from improper leap second settings and unlocked Omega time clocks. Unfortunately, not all the RT72A-02 had the same firmware revisions. Firmware revisions 2.44, 2.45, and 2.46 were all used and each has a different leap second setting. To further complicate matters, an additional leap second was inserted on June 30, two days into the experiment. The timing shifts were accomplished by applying a shift to the trace start time via the TotalStatic field in the SEGY header using the program SEGYSHIFT.
The next step involved event association between the portable stations. This was accomplished by running the UCSB_TABLE and REAP programs. UCSB_TABLE creates a listing of event times, while REAP associates the data according to specified time and array parameters. Both the LDEO and SDSU datasets were then associated with the P ASSCAL dataset based on event time windows. At this stage all data meeting certain time criteria were organized into a sequence of events, and both P and 5 phases were picked. In order to determine event locations and magnitudes, two methods were used to associate the portable station records with the SCSN catalogue. An initial event-time association was performed, utilizing an ORACLE relational database, which correlated SCSN origin times and the events generated by REAP. This did not create a unique association so we utilized the program dbassoc_arrivals (written by Dan Quinlan at the University of Colorado) which associates predicted arrivals from an origin table using the IASPEI91 travel time tables and the actual P and S phase picks. Due to the high level of activity during the first few days of the deployment, approximately 10% of the records contain more than one event which resulted in ambiguous associations. Some of these problems also occured when the SCSN catalogue contained multiple locations within a small time window leading to more than one location for a single portable event. These records will be reviewed and re-associated at a later date. Finally, all event header information and phase arrival data were converted from CSS 3.0 format to the SCEC database format (Appendix 1). The data will also be converted to SEED format with all response information included, and be available from the IRIS Data Management Center at the University of Washington.
This release of data comprises 2345 events recorded at 5 or more stations, each having an associated SCSN CUSPID. Figure 15 shows the distribution of these aftershocks. Figure 16 shows magnitude distribution, indicating that the majority of events fall between MI1.5 and MI4.5, and Figure 17 shows a histogram of number of events recorded versus number of stations. Figure 18 shows the range of magnitude versus epicentral distance, and Table 3 shows initial SCSN locations of all Ml4.0 and larger events.
For the majority of the data, the timing corrections that were applied were necessary to correct for known offsets due to leap second discrepancies, and resulted in accurately timed data. There were some stations whose time clocks were either faulty, or were not reliably synchronized to the Omega transmitting stations. Station UCVF never locked on to an Omega transmitter, and it is apparent that the timing drifts an unknown amount (+/-10 seconds) throughout the data set. Station UGGP has unreliable timing for days 194-198 due to frequent lock and unlock cycles of the Omega timeclock. These data were left uncorrected, instead of applying estimated shifts based on phase arrivals. Table 4 shows the time shifts that were applied to the data to correct for the known discrepencies. To further examine any systematic timing errors, travel time residuals have been calculated with the dbassoc_arrival program, Figure 19 (a, b, c, d) shows these residuals for each station for all arrivals within 4 seconds of a predicted arrival. Figure 20 shows the residuals for stations UCVF and UGGP during periods of unreliable timing.
The data set consists of a wide range of earthquakes, from records with S-P times of less than 0.5 seconds, to events with anomalously low frequency waveform characteristics throughout the record, as well as many co-located events with similar magnitudes. Figure 21 shows the vertical traces of an event in the Big Bear region with an epicentral distance of approximately 150 meters from station ERWL. Figure 22 shows the vertical traces of an event recorded by 14 portable stations as well as 4 TERRAscope stations displaying low frequency characteristics.
In addition to the data contained in this product, both the USGS and UCSB have data sets containing aftershocks from the Landers area. UCSB deployed two 5-element arrays of Kinemetrics SSAZ accelerographs, and has data available via ftp (contact Jamie Steidl at UCSB for further information). The USGS deployed 34 GEOS instruments recording both L-22 and FBA-23 sensors including 5 that were in operation at the time of the main shock. The resulting data set consists of 484 events that were recorded between May 18 and August 23, 1992 (Hough, S. E. et al., 1993).
This aftershock experiment collected the largest amount of data of any RAMP to date, approximately 1O Gbytes from more than 13,000 events. Considering that the mainshocks occurred on a Sunday morning, and that field teams each had a minimum of 3 hours travel time, it is quite remarkable that the first event recorded on multiple portable stations occurred less than 12 hours after the Landers mainshock. The equipment support by the IRIS PASSCAL Instrument Center, the USGS Branch of Earthquake and Landslide Hazards, and the Rocky Mountain Front Experiment was essential for the maximum deployment of 24 simultaneously recording stations.
After reviewing the field operations for this experiment, it is clear that there are some essential procedures which should be followed in the future to maximize data recovery and minimize redundant effort. The most important component for a successful experiment is a communications plan. The field personnel need a dedicated contact point at the nearest regional seismic network center which acts as a message center. This will allow field teams to coordinate their efforts when they are working in different geographical areas. The contact point should also be responsible for providing the most recent updates on the aftershock activity. This information is critical for field teams to decide where to deploy the next stations. Due to the difficulties of getting to a working telephone, particularly in remote areas like the Landers aftershock zone, the contact point should be constantly manned for the first several days. During this experiment several members of the SCSN staff fulfilled this role. From the field side, it is important that all groups use this contact point until a local field headquarters can be established. Ideally, all field groups would have mobile radios, but the reality is that the pay phone will be the primary method of communications. Once the field headquarters is established then one person should be tasked with manning this post and that person should have the responsibility to coordinate field groups and maintain contact with the regional seismic center. In this manner the co-location of equipment with identical sensors can be avoided and stations can be distributed efficiently and adaptively in a dynamic situation.
The other important component is the establishment of a data processing computer at the field headquarters with an individual dedicated to managing the data as it returns. This is one of the most important elements of the field exercise since this allows quality control capability at the outset and allows initial data processing to proceed without delay. The quality control issues which must be addressed immediately are whether ail components are working and gains set correctly along with an evaluation of the time quality and accuracy. The field computer should have enough disk space to be able to load 2 days' data from the array from ail stations so that full event associations can be made. In the case of Landers at least 4 Gigabytes would have been needed. In this experiment a computer was not installed for several days and, in retrospect, having one would have improved the efficiency of the field operations as well as the overall data recovery. The initial volume of data was so large that the first order processing and quality control was not completed until months after the experiment was over. In fact, new data sets have surfaced as late as one year after the start of the experiment requiring association with the final data product.
There are several software areas which need development to improve the data processing scheme. The first is to incorporate an automatic phase picking algorithm which is appropriate for both broadband and short period data. This should be followed by a more sophisticated algorithm to associate arrival phases into events. The current algorithms are based purely on a time-based association which works well in a single event per trigger environment, but during Landers there were several thousand events associated by this algorithm which were actually multiple events. At the present time these are sorted out manually, yet many of these multiple event triggers could have been identified automatically. Once a proper set of arrivals are chosen then a standard location program could be used. The fundamental routines to solve these data processing problems exist now and are being prepared as a software package for future aftershock studies.
The final lesson learned from this experiment is the benefit of using short period seismometers, strong motion accelerometers, and broadband seismometers. During the first days it was extremely important to get stations operating quickly and this meant using short period sensors and accelerometers. Once the basic network was functioning, then the extra hour or two time investment for installing broadband sensors proved to be worthwhile and yielded recordings of previously unidentified seismic signals from local earthquakes, as seen in the low frequency signals present in Figure 21.
In the final preparation of this release, we encountered problems with ambiguously associated events affecting approximately 10% of the data. We plan to examine the phase data and re-associate these events with the SCSN, as well as location calculations for future release.
Several institutions contributed resources which helped us to complete this quality product. Art Lerner-Lam and Gene Humphrey promptly removed equipment from their Rocky Mountain Front experiment for loan to SCEC which enabled us to deploy the broadband stations. Tom Henyey and Leon Teng from USC loaned instruments from their Los Angeles Basin experiment which helped speed the deployment in the early stages. Personnel at the IRIS P ASSCAL instrument center and the USGS Branch of Earthquake and Landslide Hazards provided several stations worth of equipment. Danny Harvey and Dan Quinlin at the University of Colorado provided software and helpful discussions during the processing procedure. Personal communications with Sue Hough at the Pasadena branch of the USGS were helpful in obtaining information regarding their deployment and data set. Funding from SCEC fund #40140A helped support the task of maintaining the stations and processing the data.
- Anderson, J., Farrell, W., Garcia, K., Given, J., Swanger, H., CSS Version 3 Database: Schema Reference Manual, September, 1990.
- Hough, S. E., Mori, J., Sembera, E., Glassmoyer, G., Mueller, c., Lydeen, S.,Southern Surface Rupture Associated With the 1992 M7.4 Landers Earthquake: Did It All Happen During the Mainshock?, Geophysical Research Letters, 1993.
- Scott, J., Hauksson, E., Kanamori, H., Mori, J., Global Positioning System Re-survey of Southern California Seismic Stations, Bulletin for the Seismological Society of America, submitted 1994.
|BNJI||1992181||1992207||34.1625||-116.321||0.827||Benji Road, Yucca Valley, Calif.||ss||–||–||–|
|BRCC||1992182||1992246||34.0721||-116.3915||1.207||Black Rock Campground, Joshua Tree National Monume||ss||–||–||–|
|BRDO||1992184||1992190||33.8296||-116.149||0.483||Berdoo Canyon, Joshua Tree National Monument, Cali||ss||–||–||–|
|CGTS||1992190||1992224||34.9514||-116.8639||0.694||Callico Ghost Town, Barstow, Calif.||ss||–||–||–|
|EDC2||1992182||1992206||33.9166||-116.3266||0.48||East Deception Canyon 2, Joshua Tree National Monu||ss||–||–||–|
|EORD||1992182||1992246||34.6561||-116.718||1.16||East Ord Mountain, San Bernadino Co., Calif.||ss||–||–||–|
|FIRE||1992181||1992189||34.0486||-116.578||0.749||Fire Station, Morongo Valley, Calif.||ss||–||–||–|
|FORF||1992181||1992200||34.0883||-116.9205||1.596||Forrest Falls, San Bernadino Co., Calif.||ss||–||–||–|
|GPOE||1992182||1992206||34.4227||-116.7169||0.894||George Poe's Place, San Bernadino Co., Calif.||ss||–||–||–|
|GRAC||1992181||1992206||34.2668||-116.3882||0.913||Green Acres, Landers, Calif.||ss||–||–||–|
|HILL||1992181||1992220||34.3614||-116.4541||0.891||Hill, San Bernadino Co., Calif.||ss||–||–||–|
|IRON||1992185||1992246||34.6167||-116.5671||1.152||Iron Ridge, San Bernadino Co., Calif.||ss||–||–||–|
|JFRG||1992185||1992246||34.0457||-116.6215||0.862||Jumping Frog Lane, Morongo Valley, Calif.||ss||–||–||–|
|LADY||1992189||1992220||34.1306||-116.3139||0.811||Lady's Place, Yucca Valley, Calif.||ss||–||–||–|
|LPMA||1992182||1992206||34.3603||-116.4192||0.733||Los Padres Mine Access, San Bernadino Co., Calif.||ss||–||–||–|
|RIMR||1992181||1992220||34.1865||-116.4634||1.099||Rimrock, San Bernadino Co., Calif.||ss||–||–||–|
|UCVF||1992182||1992207||34.0108||-116.306||1.441||Upper Covington Flats, Joshua Tree National Monume||ss||–||–||–|
|UGGP||1992185||1992246||34.7499||-116.6606||0.857||Under Ground Gas Pipeline Road, San Bernadino Co.,||ss||–||–||–|
|VIPC||1992184||1992246||34.0177||-116.1903||1.246||Very Important Person Campground, Joshua Tree Nati||ss||–||–||–|
|VVST||1992181||1992206||34.3596||-116.5789||0.912||Valley Vista Road, San Bernadino Co., Calif.||ss||–||–||–|
|YKNF||1992181||1992206||34.2075||-116.4457||1.094||Yellowknife Road, San Bernadino Co., Calif.||ss||–||–||–|
URL: http://eqinfo.ucsd.edu/deployments/portable_aftershock_studies/landers_big_bear.php [Last updated: 2015-10-22 (295) 22:24:32 UTC]