Re: Suggested Generic Standard for SegY files of stacked data. CSEG Recorder, October 1999 "240 Byte Trace Headers"

The comment that bytes 115-118 may be set, but should not be required, etc... will create problems if this information is not provided. Following are several reasons:

  1. Read errors during a copy or archive can (and often) result in shorter or longer traces, depending on the type of errors and the recording density of the tape. When the data no longer resides on magnetic tape, there are no "interblock gaps" to rely on for block (trace) sizes.
  2. We have seen data traces become longer within very long marine data lines. e.g. going from shallow water (4 second data) to deep water (6 second data). The binary header does not provide for multiple "number of samples per trace" within a line. The only way this data could be loaded into a workstation was to ignore the "number of samples per trace" in the binary and use the value from each trace header.
  3. With increasing on-line delivery of seismic data (mostly stacks), where tape interblock gaps no longer separate data traces, trace rotation often occurs when totally dependent on the binary header.
  4. The SEG Standard "strongly recommends" they be set.

Suggestion: The "number of samples per trace" should always be specified in the Trace Headers and should always be a multiple of the "Data Sample Code (bytes per sample) for both stack and SegY field data.

Have seen instances where both binary and trace header information was incorrect, with respect to the "samples per trace" field. This creates havoc during on-line downloads.

Best Regards,
Norm Pascal,
NRI On-Line Inc

Editors’ Response

Thank you, Norm, for your comments. These byte locations should be set for variable length traces, and in most versions of SEGY they are always set. One motivation for the Generic SEGY article is to simplify and speed up data loading over a wide variety of workstations and since not all workstations can read variable-length traces, the fixed length version is preferred.

Thanks
– Editors

End

References

Share This Column