- journal_title：Bulletin of Canadian Petroleum Geology
- Contributor：Zhuoheng Chen ; Yexin Liu ; Kirk Osadetz
- Publisher：Canadian Society of Petroleum Geologists
- Date：2012-09-01
- Format：text/html
- Language：en
- Identifier：10.2113/gscpgbull.60.3.142
- journal_abbrev：Bulletin of Canadian Petroleum Geology
- issn：0007-4802
- volume：60
- issue：3
- firstpage：142
A meaningful exploration decision depends not only on a reliable estimate of resource potential, but also on a robust geological risk evaluation of the exploration targets. We propose a Support Vector Machine approach for geological risk evaluation of petroleum occurrence in a petroleum play. The resulting risk map can be used to highlight the geographic locations of potential undiscovered petroleum resources and help visualize regional exploration risk. Treating risk evaluation as a two-category classification with uncertainty, this method integrates all available data and geoscience information from regional study and petroleum system analysis to interrogate and evaluate exploration drilling results. From this it makes inference on the likelihood of petroleum occurrence for untested exploration targets in a petroleum exploration play. Together with estimated sizes of remaining untested prospects, it provides a risked resource map for ranking exploration opportunities and helping to make business decisions affecting exploration strategy. This approach can also be used in evaluation of unconventional resource plays to highlight ‘sweet spots’ and estimate ‘dry well’ risk. The proposed method and workflow are illustrated using an example from the late Triassic–early Jurassic structural play of the Heiberg Group in the western Sverdrup Basin, Canadian Arctic Archipelago.