Researcher Clair Bond (University of Glasgow) in association with Midland Valley and GX Technology has been inviting tradeshow attendees to take part in a test of their seismic interpretation acumen.
Uncertainty
ODIN picks up on the current vogue for uncertainty and risk management by trying to quantify the risk of an interpreter getting it wrong! As Bond notes, ‘If you put geoscientists in front of the same data, you get as many interpretations as interpreters! These derive from different assumptions, bias and experience. The question is, does this ‘concept uncertainty’ have a greater impact on prediction than models built with a Monte Carlo-type approach?’
Guinea pigs
The ODIN methodology started with the creation of a 2D geological model built with Midland Valley’s 2D Move package. A synthetic seismic section was then created by GX-Technology. The ODIN Guinea pigs—seismic interpreters who were bushwhacked at various recent tradeshows—were invited to interpret the ODIN ‘data’ and fill out a questionnaire.
200 geoscientists
The results from some 200 interpretations showed that interpreters were frequently biased by their ‘prior knowledge’. Interpreters tended to shoehorn their interpretations into the tectonic regime with which they were most familiar. Those working in extensional settings saw extension. Thrust belt workers saw thrusting. In fact the section was an inversion problem—a fact only spotted by 43% of the sample.
Future
One encouraging result from the survey was that experienced interpreters fared better than newbies. Especially if their experience came from an extensional setting! To remove such bias, the ODIN project is to expand into other tectonic settings. Future results will be evaluated for statistical bias and will target specific groups of interpreters.
© Oil IT Journal - all rights reserved.