To this point, this review has focused largely on processes related to the precipitation of diagenetic cements. However, the increased focus on detailed reservoir characterization described in the introduction has also led to greater efforts to develop techniques for scaling up basic petrographic and core analysis data to predict interwell porosity and permeability variations. The basic problem is that reservoir heterogeneities on the order of tens to hundreds of meters typically control the recovery efficiency of a reservoir. Unfortunately, permeability data are seldom available at these scales. As such, statistical and probabilistic techniques for preserving the variability observed at the scale of thin sections and core plugs, coupled with sophisticated new 3-dimensional seismic imaging techniques, and improved borehole logging are becoming an important component of reservoir quality prediction. One only needs to go to recent editions of journals published by the Society of Petroleum Engineers to see this new emphasis (e.g. Gómez-Hernández and Journel, 1994; Tyler et al., 1994).
Systematic classification of porosity and permeability, unbiased by variations from one petrographer to another, is one of the fundamental components of integrated reservoir quality analysis. With the increased availability of smaller, faster computers over the past few years, we have also begun to see a wide variety of sophisticated algorithms for computerized image analysis of petrographic data. Most of these programs combine transmitted light, backscattered electron and/or cathodoluminescence imaging of petrographic thin sections, with image digitizing, processing and analysis to assess such geometric variables as pore size, surface area, shape, and connectivity, all major controls on permeability (eg. Ehrlich et al., 1991, and McCreesh et al., 1991). In recent applications of these techniques, researchers have been able to collect quantitative mineralogic and pore structure data from a variety of different environments and rock types in an automated fashion, with the prospect for increased use of these techniques in the future (e.g. Evans et al., 1994).