In a series of ongoing experiments, we are applying a recently developed object recognition algorithm to the quantification of 2-D pore networks in thin-section. Developed at the Distributed Applications and Systems Lab (DASL) at University of Montana (http://www.umt.edu/ MERGE), this algorithm has been already been successfully applied to the analysis of large volume satellite imagery. Key to our petrographic image analysis is the enhanced resolution provided by the MERGE algorithm. Whereas most petrographic image applications process images on the order of 1000x1000 pixels (i.e. 1,000,000 pixels), the MERGE algorithm has been successfully applied to images as large as 625,000,000 pixels using only well equipped local workstations. We are currently modifying and applying this rule-based algorithm to the spatial analysis of pore networks, framework grains, and diagenetic constituents in a suite of thin-sections cut from core samples of the middle Jurassic upper Bowes Member (Piper Formation) from the Rabbit Hills oil field in north-central Montana. The upper Bowes is a bioclastic carbonate that comprises reservoir facies in the Rabbit Hills field. Samples for thin-section were derived from the Flynn Trust 7-15 well. Preliminary petrographic analysis of the reservoir facies suggests that it should be ideal for this sort of petrographic image application. Reservoir facies consist mainly of quartz-bearing ooid and bioclastic grainstones. Shelter porosity and moldic porosity after molluscan detritus are volumetrically the most important porosity types. A fringing calcite cement has partially occluded both primary and secondary porosity. Direct porosity measurements of the reservoir facies average 16.8% (n=19). Kmax measurements average 66 md (n=19). Both pores and pore throats are easily resolved petrographically. We are currently conducting spatial analysis of pore volumes, diagenetic constituents, and framework components from this well, and we anticipate a reasonable comparison between porosity and permeability values computed from our petrographic image analysis and those obtained from direct measurement.

AAPG Search and Discovery Article #90928©1999 AAPG Annual Convention, San Antonio, Texas