Geoscientists are obliged to deal with complex, highly interconnected, spatially extensive Earth systems. The long tradition of addressing these by reduction and simplification has increasingly been complemented by approaches that attempt to deal with these complex systems more holistically, and to understand the synergistic interconnections in addition to the individual mechanistic and historical relationships. Many complex Earth systems can be represented as networks. Graph theory, a branch of mathematics well suited to network analysis, is thus emerging as a powerful tool in the Earth and environmental sciences. Network approaches in many disciplines (including Earth and environmental sciences) have highlighted a linkage between system properties and dynamics or behavior that can be addressed using graph theory. The purpose of this paper is to review graph theory applications, introduce geoscientists to some of the more promising techniques, and develop a synthesis of graph and network-based approaches in the geosciences.
An Earth system can be characterized as a set of interconnected components. These may be locations, sources, sinks, or nodes in flux networks; objects (e.g., landforms, mass or energy storage compartments); processes or process bundles or regimes (e.g. weathering, moisture advection; isostatic adjustments); or phenomena or events (e.g. overbank flows, tropical cyclones, earthquakes). These components are connected by fluxes of matter and energy, feedbacks, spatial or temporal sequencing or adjacency, statistical correlations, and process-response relationships. Table 1 summarizes some general examples. Thus Earth systems can often be represented as networks, whether in a literal sense (e.g., fluvial channel networks; rock fracture patterns) or as a conceptual and analytical tool. The components represent the nodes (vertices) of the network, and the relationships between them the edges (links). The box-and-arrow diagrams commonly used in geosciences may be treated as graphs, with the boxes as nodes and the arrows as edges or links.