Johannes H. UHL, University of Colorado Boulder, United States
Stefan LEYK, University of Colorado Boulder, United States
Keith BURGHARDT, University of Southern California, United States
Yao-Yi CHIANG, University of Minnesota, United States
Kristina LERMAN, University of Southern California, United States
Craig A. KNOBLOCK, University of Southern California, United States
Recently released geospatial datasets offer the opportunity to develop novel methods to reconstruct and analyze historical urban and rural road networks. Such datasets include industry-generated property data [1,2], public building stock data [3] containing building age information, and collections of scanned and georeferenced historical topographic maps (e.g. [4]) depicting historical snapshots of settlement and transportation infrastructure. Moreover, detailed data on contemporary road networks are available at increasing levels of completeness and geographic coverage [5]. We are developing innovative data integration strategies that make use of the historical information contained in cadastral property and building databases, as well as historical map data to reconstruct historical road networks based on the contemporary road network. These methods make use of the spatial (i.e., proximity) and semantic (i.e., address-based) relationships between historical records and contemporary road network data to infer the age of each street segment in the road network. Our methods involve geospatial analysis and image processing, and these methods have been applied to model urban street networks in the United States over a time period of more than 100 years. For example, we implemented a probabilistic framework to infer the likelihood of contemporary road existence in the past for each segment of the road network based on color information collected from underlying historical maps. Currently, we are testing this and other methods on data sources in Europe and elsewhere.
Mots clés : historical road networks|spatio-temporal networks|geographic data integration|topographic map processing|long-term urbanization dynamics
A103919JU