LARSE '93 DATA PROCESSING


The initial uncompressed data set was recorded on approximately 150 Exabyte and Data tapes, and comprised 87 Gbytes total with some station data recorded in Reftek format and others in SEGY format. Using the PASSCAL software program ref2segy the Reftek DAS format was converted to IRIS SEGY format.

The software processing comprised several phases:

FINAL DATA PRODUCT

The final format of the dataset is SEGY format with DC mean removed and clock drift timing corrections applied. This final dataset has been archived on the Metrum mass storage system or other media for use by other institutions in scientific analysis. At UCLA the final dataset was merged, and then cut into segments containing the teleseism and local events that were collected during the experiment. This process was applied to the SEGY format data and transferred to UCLA via ftp. Logfiles were recorded at almost every station for each week. These are included in the final data product in storage.

TIMING CORRECTIONS

In order to make the appropriate timing offset and drift corrections, the log files had to be edited in order to be used properly by the PASCALL program refrate. Because of the older Reftek CPU versions, the required timing corrections were written incorrectly to the log files, which refrate uses to make a record of the time corrections it applies to the data. Refrate reads the log files and pairs certain statements regarding timing, such as "unlocked-phase error', 'phase error-DSP CLOCK SET', etc. These pairs create intervals for which timing corrections are calculated and used by clockcorr. Clockcorr reads the output files of refrate and searches for the intervals which span the time at which events begin. Most corrections are simple and straightforward. However, there are some problematic cases, as outlined below.

All of the log files are in the form I.** where ** is the DAS number of the Reftek at the sites. The log file lines containing phase error or DSP set reports contain the essential timing correction. Normally, the corrections would list a seconds and milliseconds portion of time offset. However, for almost all of these cases, the milliseconds portion was not recorded in the log file. Moreover, if the offset was positive, a whole second was added to the correct milliseconds portion of the offset.

In summary, in all cases of a positive phase error, a DSP line of (phase error -1) was added. The only case where this was not needed was when there were no DSP lines after the time jerk. This meant that the total correction was the same as the phase error and the time jerk completed the correction. Refrate uses the value of the phase error to calculate drift.

There still remain some problems with the absolute timing in the seismograms. Most of the known one-second (leap second) errors were removed during the data processing from the stations with omega timing receivers. However, a few stations may still contain these errors if they occurred in the middle of the week, or if station timing was pulsed with an omega signal only once. These errors should be relatively obvious to spot in the data set. In addition, 10-second errors were also removed from some stations with omega timing receivers. However, as in the one-second case, stations for which the error occurred in the middle of the week were not corrected. These timing errors are also easy to spot in the data set.

REFERENCE


LARSE HOME PAGE LARSE '93 LARSE '93 INSTRUMENTATION LARSE '93 EVENTS