Scale-dependent heterogeneity in fracture data sets and grayscale images

Lacunarity is a technique developed for multiscale analysis of spatial data and can quantify scale-dependent heterogeneity in a dataset. The present research is based on characterizing fracture data of various types by invoking lacunarity as a concept that can not only be applied to both fractal and non-fractal binary data but can also be extended to analyzing non-binary data sets comprising a spectrum of values between 0 and 1. Lacunarity has been variously modified in characterizing fracture data from maps and scanlines in tackling five different problems. In Chapter 2, it is shown that normalized lacunarity curves can differentiate between maps (2-dimensional binary data) belonging to the same fractal-fracture system and that clustering increases with decreasing spatial scale. Chapter 4 analyzes spacing data from scanlines (1-dimensional binary data) and employs log-transformed lacunarity curves along with their 1^st derivatives in identifying the presence of fracture clusters and their spatial organization. This technique is extended to 1-dimensional non-binary data in chapter 5 where spacing is integrated with aperture values and a lacunarity ratio is invoked in addressing the question of whether large fractures occur within clusters. Finally, it is investigated in chapter 6 if lacunarity can find differences in clustering along various directions of a fracture netowork thus identifying differentially-clustered fracture sets. In addition to fracture data, chapter 3 employs lacunarity in identifying clustering and multifractal behavior in synthetic and natural 2-dimensional non-binary patterns in the form of soil thin sections. Future avenues for research include estimation of 2-dimensional clustering from 1-dimensional samples (e.g., scanlines and well-data), forward modeling of fracture networks using lacunarity, and the possible application of lacunarity in delineating shapes of other geologic patterns such as channel beds.