Abstract
Carbonate reservoirs are often comprised of a heterogeneous pore system within a matrix of variably distributed minerals including anhydrite, dolomite, and calcite. When describing carbonate thin sections, it is routine to assign relative abundance levels to each of these components, which are qualitative to semi-quantitative (e.g. point-counting) and vary greatly depending on the petrographer. Over the past few decades, image analysis has gained wide use among petrographers, however, thin section characterization using this technique has been primarily limited to the pore space due to the difficulty associated with optical recognition beyond the blue-dyed epoxy associated with the pores. Here, we present a new method of computerized object-based image segmentation (Quantitative Digital Petrography: QDP) that relies on a predefined rule set to enable rapid, automated thin section quantification with only minor human interaction. We have developed a novel work flow that automatically isolates the sample on a high-resolution (i.e. <1μm/pixel) scanned thin section, segments the image, and assigns those segments to predefined categories – e.g., pores, cement, grains, etc. Using this technique, statistically relevant numbers of thin sections can be rapidly processed and quality controlled, thereby allowing quantitative data such as MICP, wettability, and surveillance data to be integrated with the petrographic observations for a more complete description of the carbonate rock. Our technique can also incorporate multiple layers, such as cross-polarization, Back Scatter Electron (BSE) imaging, and elemental maps, which allow additional information to be easily integrated with results from QDP. The QDP approach is a significant improvement over previous digital image analysis methods because it 1) does not require binarization, 2) eliminates the subjectivity in assessing abundance levels, 3) requires less hands-on time for the petrographer, and 4) provides a much fuller dataset that can be incorporated across an entire well or field to better address common challenges associated with carbonate reservoir characterization, such as understanding pore type and cement abundance, pore connectivity, grain distribution, and reservoir flow characteristics.
