A Statistical Approach to Assessing Depth Conversion Uncertainty on a Regional Dataset: Cooper-Eromanga Basin, Australia. Issue 1 (1st December 2016)

• Record Type: Journal Article
• Authors: Kulikowski, David; Hochwald, Catherine; Cooke, Dennis; Amrouch, Khalid
• Is Part Of: ASEG Extended Abstracts (Online). Volume 2016:Issue 1(2016)
• Page Start: 1
• Page End: 7
• Publication Date: 2016-12-01
• Subjects: Cooper Basin -- Eromanga Basin -- 3D seismic data -- depth conversion -- velocity modelling; Prospecting -- Geophysical methods -- Periodicals; Prospecting -- Geophysical methods; Periodicals
• DOI: 10.1071/ASEG2016ab200
• Languages: English
• ISSNs: 2202-0586

Abstract

Abstract : Deciding on the most accurate grid based depth conversion method can often be an arbitrary choice made by geophysicists, particularly if previous research is limited. The importance of accurate depth conversion is particularly crucial in the Cooper-Eromanga Basin, where the presence of oil rich, low relief structural traps are questionable depending on the method used. Previous depth conversion studies are limited to local scales, limited well control and a focus on select horizons. To investigate the depth conversion uncertainty on a regional scale, this research performs a comprehensive and regionally extensive depth conversion analysis utilising 13 3D seismic surveys with 73 interpreted TWT grids and 657 wells. Depth conversions were performed using the 4 most commonly used methods; average (pseudo) velocity, time-depth trend, kriging with external drift using TWT, and kriging with external drift to tie stacking velocities to average well velocities. To manage the large volume of data, a looping script was written to automate the depth conversion process and utilise the cross-validation, or blind-well method (use n wells to predict the n th +1 well). Statistics on several variables were captured after each loop, with cluster analysis performed on the final data set to test variable significance on depth conversion accuracy. A database of approximately 10000 error calculations found that although the average velocity method is the most accurate at a high level (average absolute error ~24.9 feet), the best method and the expected error changes significantly (tens of feet) depending on the combination and value of the most significant variables. The variables which impacted uncertainty the most were; location (3d survey), formation, distance to the nearest well, and the spatial location of the predicted well relative to the existing well data envelope.