Back to Fall 1999 AGU abstract list.
Back to the News
|
Statistical Methods for Trace Editing for Large 3D Reflection Seismic Surveys Barrie L Taylor 1 (303-779-8080; btaylor@lgc.com) Denise J Hills 2 (808-956-6055; djhills@soest.hawaii.edu) Gregory F Moore 2 (808-956-6854; gmoore@hawaii.edu) Nathan L Bangs 3 (512-232-3390; nathan@utig.utexas.edu) Tom H Shipley 3 (512-232-3230; tom@utig.ig.utexas.edu) 1Landmark Graphics, 7409 S. Alton St., Suite 100, Englewood, CO 80112, United States 2Dept. of Geology and Geophysics, University of Hawaii 2525 Correa Rd., Honolulu, HI 96822, United States 3Institute for Geophysics, University of Texas at Austin 4412 Spicewood Springs Rd., 600, Austin, TX 78759, United States The volume of data collected during a typical 3D seismic survey is now in excess of 20 million traces, for academic surveys, and on the order of hundreds of millions of traces for industry surveys. Mundane tasks, such as editing bad traces, can no longer reasonably be done manually and will increasingly be done through statistical methods. While statistical methods for trace editing have been used in industry for some time, it has only recently becoming necessary to use these methods for academic surveys. In this paper we demonstrate how statistical methods for trace editing were successfully employed on a recent large volume marine reflection seismic survey using ProMAX processing software. There are six basic steps for using trace statistics for editing a volume of data. 1) Calculation of the trace statistics. 2) Visualization of the resulting trace statistics in a graphical format. 3) Picking the statistical attributes that are meaningful for a particular dataset, and establishing the range of values to effectively kill undesirable data. 4) Testing these parameters on representative segments of the dataset. 5) Application of the statistical thresholds for trace editing to the entire dataset. 6) Post application quality control which may include a repetition of steps 1 and 2. Up to eight representative trace statistics may be calculated for a given time gate such as the average trace energy, the energy decay rate, the frequency deviation, and the peak frequency. These values are written to a database as well as to the trace headers and viewed in a graphical format using DBTools, a database display tool within ProMAX. Outlying values are graphically picked on one trace statistic, and the corresponding traces highlighted on the graphical representations of the other statistics. Next the ensembles (shot gathers in this case) which contain the traces with the chosen outlying statistical values are automatically brought up for viewing in a Trace Display as well as a corresponding graph of any number of the calculated trace statistics. Threshold values for editing bad traces are determined and applied to a select subset of the dataset for testing. Once the statistical values for editing the traces are determined, the whole dataset is run through trace editing. Trace statistics may be run a second time concurrently with the trace edits to allow for fast quality control of the output via DBTools. Using these tools the criteria for editing multimillion trace datasets may be established and applied with very few man hours required. |
Meeting: Meeting Section: Special Session: Index Terms: Theme: |
Last Modified: October 8, 1999
Comments: UTIG webmaster