Quantifying Historical Geographic Knowledge from Digital Maps.

Tenzing Shaw, Peter Bajcsy, Michael Simeone, and Robert Markley

National Center for Supercomputing Applications, University of Illinois at Urbana-Champaign

Microsoft Research eScience Workshop 2010
Berkeley, CA, October 11–13

An important question facing historians is how knowledge of different geographic regions varied between nations and over time. This type of question is often answered by examining historical maps created in different regions and at different times, and by evaluating the accuracy of these maps relative to modern geographic knowledge. Our research focuses on quantifying and automating the process of analyzing digitized historical maps in an effort to improve the precision and efficiency of this analysis.

In this paper, we describe an algorithmic workflow designed for this purpose. We discuss the application of this workflow to the problem of automatically segmenting Lake Ontario from French and British historical maps of the Great Lakes region created between the 16th and 19th centuries, and computing the surface area of the lake according to each map. Comparing these areas with the modern figure of 7,540 square miles provides a way of measuring the accuracy of French versus British knowledge of the geography of the Lake Ontario region at different points in time.

Specifically, we present the results following the application of our algorithms to 40 historical maps. The procedure we describe can be extended to geographic objects other than Lake Ontario and to accuracy measures other than surface area.