Fault and fracture networks channel or impede fluid flow in the subsurface. They become major drivers of the flow dynamic in low-porosity rocks, where their geometry and topology govern the groundwater flow on a regional scale. As topology is related to the percolation threshold, it represents one of the crucial metrics to characterize fluid flow properties. The objective of this study is to design an efficient and robust method to characterize networks in terms of fluid-flow properties and link the components of the network to raster data for further analysis.
We present an automated framework for data extraction and analysis based on graph representation of 2D fault and fracture networks. Initially, the geometric parameters and their distributions are extracted, and the subsequent analysis is based on a georeferenced graph that is linked to raster data, such as elevation models, magnetic intensities, or gravimetric data. This approach allows for characterizing geometric and topologic properties of the entire network, for determining potential sub-networks, and for applying standard graph algorithms.
Application of the framework to a synthetic data set and a real-world case study from the Musgrave Province in South Australia demonstrates the efficiency of our method in handling natural fault networks, in analysing these networks statistically, in assessing connectivity, and in linking the networks to geophysical data. The algorithm can also be applied to surface water drainage networks.
The framework produces a georeferenced graph with edge and vertex weights derived from raster data. Minor flaws in lineament digitisation such as gaps in fault traces are automatically corrected. The completeness of mapped network can be evaluated based on the statistic and topological analysis. The graph representation can be used to assess dominant groundwater flow direction, or it can be used to derive equivalent permeability fields for regional groundwater models.