[en] Image processing methods have been used very effectively to estimate physical properties of isotropic porous earth materials such as sandstones. Anisotropic materials can also be analyzed in order to estimate their physical properties, but additional care and a larger number of well-chosen images of cross sections are required to obtain correct results. Although low-symmetry anisotropic media present difficulties for two-dimensional image processing methods, geologic materials are often transversely isotropic. Scalar properties of porous materials such as porosity and specific surface area can be determined with only minor changes in the analysis when the medium is transversely isotropic rather than isotropic. For example, in a rock that is transitively isotropic due to thin layers or beds, the overall porosity may be obtained by analyzing images of cross sections taken orthogonal to the bedding planes, whereas cross sections lying within the bedding planes will determine only the local porosity of the bed itself. It is known for translationally invariant anisotropic media that the overall specific surface area can be obtained from radial averages of the two-point correlation function in the full three-dimensional volume. Layered materials are not translationally invariant in the direction of the layering, but we show nevertheless how averages of cross sections may be used to obtain the specific surface area for a transversely isotropic rock. We report values of specific surface area obtained for thin sections of Topopah Spring Tuff from Yucca Mountain, Nevada. This formation is being evaluated as a potential host rock for geologic disposal of nuclear waste. Although the present work has made use of thin sections of tuff for the images, the same methods of analysis could also be used to simplify quantitative analysis of three-dimensional volumes of pore structure data obtained by means of x-ray microtomography or other methods, using only a few representative cross sections chosen from the full three-dimensional data set. This approach provides a quick and easy way to get initial estimates of physical properties