Global Road-Networks in the Mid-20th Century: Digitizing the AMS Map Series 1301
ISBN 978-85-88783-11-9
Authors
1Fuhrer, R.; 2Hunziker, P.
1ETH ZURICH Email: raphael.fuhrer@ivt.baug.ethz.ch
2ETH ZURICH Email: hunziker@icr.gess.ethz.ch
Abstract
Advances in GIS technology over the last decades have yielded numerous global data sets on transport infrastructure. Beyond their usage for navigational purposes, these data have allowed novel analyses of transport systems and the role of accessibility for economic development and policy. However, comparative analyses of accessibility over long time-spans and at a global level have so far not been possible due to a lack of comprehensive historical data. This paper attempts to fill this gap by introducing a new historical road-network data set in vector format, representing the state of road infrastructure across the globe in the mid-20th century. The information underlying the newly introduced data set originates from a collection of standardized topographical map sheets at a resolution of 1:1mio, created and compiled by the US Army Map Service (AMS) in the Cold War era as part of an effort to chart the entire globe's landmass. More precisely, this paper pursues four goals. First, we discuss the merits and potential shortcomings of using the AMS 1301 map series as a source for compiling consistent, global road-network data for the mid-20th century. In particular, we review the available alternatives and elucidate the numerous advantages associated with employing the AMS 1301 series. Second, we describe the newly developed digitization process which was used to extract the road-network data from the approximately 950 physical map sheets containing the original information. The employed digitization procedure is almost fully automated, and relies on machine learning techniques to extract and combine road-network information from scanned map images in a consistent and fast manner. Moreover, the proposed digitization procedure is implemented exclusively in open-source software, making it a potentially useful resource for further cartographic digitization projects. Third, we discuss and evaluate the newly digitized data. In particular, we attempt to assess the quality of the new data set by comparing it to multiple contemporary road-network data sources. We find that in many areas, the road-network data extracted from AMS maps feature substantially more detail than contemporary and openly available road-network data collections, especially in rural areas of developing countries. In addition, we make use of the network structure of the newly introduced data to perform basic descriptive analyses of the state of accessibility across the globe in (approximately) 1950. Finally, we outline future avenues of research made possible by the availability of the newly digitized data. In particular, we propose combining the new road-network data with contemporary data sources in order to analyze the interplay between accessibility, economic development, and state-building processes over long-time spans and in a cross-country setting.
Keywords
Road-networks; Historical data; Automated digitization