There is an increasing interest both academically, commercially and at a governmental level to spatially understand and model the performance of infrastructure networks at local, regional and national scales. Particular interest lies in understanding the vulnerability and probabilistic risk faced by large scale physical infrastructure networks to spatially localised extreme events such as natural hazards. In order to perform such analysis and modelling in a holistic manner one requires the ability to represent and model the dependencies and interdependencies that exist between different infrastructure networks, which in turn requires the ability to derive a ‘network-of-networks’ spatial representation and expression of physical infrastructure. Traditionally, spatial networks and their subsequent analysis have been performed within a GIS environment. However, the ability of GIS packages to explicitly represent large interdependent networks comprising of potentially many hundreds of thousands of nodes and edges has been limited. Moreover, the analytical network functionality provided by GIS packages often falls some way short of the complex analysis required to understand how failures propagate through individual networks, between dependent networks and feedback when interdependent relationships exist.

In order to address these issues we have developed a fully relationally compliant open source spatial database schema and a python coupled-interface to the NetworkX graph analysis package for the representation, encoding and analysis of networks and their dependencies/interdependencies. In particular we have used collaborative rapid-prototyping to develop our database schema for network representation and dependence/interdependence encoding using PostgreSQL and the spatial extension PostGIS. The schema is exposed through plpgsql functions, which are provided to the user wrapped in python class methods, forming the software API.