Next: Conclusion Previous: The experiment
Next: Conclusion
Previous: The experiment
Data sorting
Cultural noise in urban areas can cause false triggering
of the recorders; we used triggering thresholds to identify earthquake
events.
To make it easy for others to use the data set, we have sorted the
approximately 200,000 files into events according to the
event catalog provided by the SCEC data center.
The theoretical P-wave arrival-time for an event to a given station,
T, and the triggering time of all the files
recorded by the station ( 
), where i is the file index, were
calculated first.
The difference between the two, 
, was
then calculated and the file with the minimum 
was
copied into the event directory if 
.
Apparent timing errors were taken into account in the sorting process.
In all 1550 events were recorded by at least
one of the stations, and 37,107 files out of the over 200,000
files were sorted into the 1550 events. Therefore about
80% of the files were false triggers.
If two events occurred close to each other in time (less than 60 s),
the above automatic sorting scheme may put a file
in both event directories. In fact, the file may contain waveforms
from both events.
Data format and file names
Data dumped from the station disks were converted to SEGY format by use of the REF2SEGY (version 93.285) routine originally written by Early et al. at the PASSCAL Instrument center. The resulting SEGY format data can easily be converted to other formats such as SAC, AH, SEED, and SIERRA by using routines provided by the IRIS data center.
The original files were identified by the starting hour, minute, and second of the file, the DAS number, and the channel number. The frequent movement of the DAS's (Table 3(a);(b);(c)) made it difficult to identify the station at which a particular file was recorded. Information about DAS and sensor types were not included in the file names. The new file names contain the information about station and channel number, starting hour and minute of the event, and DAS and sensor types. For example, a file named sssccyydddhhmmDxSy indicates that it is an event with origin time at year yy, day ddd, hour hh, and minute mm; it was recorded by channel cc of station sss, where cc=01 is the vertical component, cc=02 is the NS component, and cc=03 is the EW component; the Reftek DAS is 0x, where x=6 is for Reftek 72A-06, and x=7 is for Reftek 72A-07; and the sensor type is y, where y=1 represents L4C, y=2 represents L22, and y=3 represents L28.
Earthquake Information
Figure 4 indicates that more than 1000 of the 1550 events were recorded by less than 10 stations. Source information on the fifty-three events recorded by at least 40 stations is listed in Table 4(a);(b), and their locations are shown in Figure 5. Thirty-two of the 53 events were used in a previous report by Gao et al. [1996]. The majority of the events were located in the northern part of the aftershock zone. It must be pointed out that the files were sorted according to the original USGS-Caltech earthquake catalog, while in Gao et al. [1996] a modified catalog provided by Jim Mori at USGS was used. Only strong events are included in the modified catalog. The number of events in the modified catalog is about 1/3 of that in the original catalog. There is almost no difference between the two for data sorting of the strong events.
Log and error files
Log and error files provide important information about the data, such as recording parameters, clock state, volts per bit, and timing errors. The log files are named as sss.ddd.rrrr.log, and the error files are named as sss.ddd.rrrr.err, where sss is the station name, ddd is the day when the data was dumped from the disk to computer, rrrr is the DAS number. The total size of the log and error files is 26 Megabytes.
