Our library is undertaking a project to map historical data regarding 18th and 19th century British and American printers and publishers. One of the problems that we have encountered is the different granularity of location data. For example, one publisher may list a street number, street name, and city whereas another publisher may only mention a street name or even only the city. In such cases, we are wondering the best way to georeference the location and present this approximate data.

Does the spatial humanities community have rules or conventions for dealing with these cases?

Also, what are the most effective ways for visualizing “approximate vs precise” geospatial data on the same map?

  • this is normally accomplished with multiple locators. It can still be run as one geocode but would "hold" exceptions to run with the next level locator, and so on until all locators have been run. – Brad Nesom Apr 9 '12 at 17:52
  • Hi, when only city is mentioned I would use coordinates of the historical center. There should be some historical building in the center of the city, which can be used as navigation point - for example a church. – Juhele Apr 10 '12 at 5:50
  • At what scale, size, and medium is the map (or maps if more than one) to be presented? What is the focus of the map and who is the target audience? If the focus is to simply plot the approximate locations of publishers, then the best approach may simply be to state that locations of points are approximate, rather than try to create a symbology that may only serve to confuse casual viewers. – blah238 May 9 '12 at 0:46

This is an interesting challenge. Here's my suggestion.

Define each uncertain location as a polygon defining an area that you are "reasonably certain" contains the actual location. Such polygons could cover streets, cities, city blocks, a single address, etc.. For visualization the symbology could scale the polygons' fill transparency inversely with the amount of area they cover, i.e. large (imprecise) polygons would be rather feint, whereas smaller (more precise) polygons would appear more distinct. I think this would keep the large polygons from visually dominating.

I'm not sure how this would work out in a case where multiple polygons cover the same area, e.g. an entire city. One solution to this would be to offset such polygons from each other a bit, as points in a box plot can be "jiggled" apart. Precisely known locations could be identified with opaque points on top of the polygon layer(s) and displayed at a size that fits the location or map scale.

| improve this answer | |
  • 1
    On a single map whose extent encompasses both the US and England, I'm not sure creating polygons would be worthwhile since even entire cities would only appear as specks. Perhaps simply categorizing and symbolizing points by relative accuracy would be sufficient. But is displaying the accuracy the focus of the map? If not, simply stating that the locations of the points is approximate may be sufficient. – blah238 May 9 '12 at 0:43
  • That makes sense; I was picturing a scalable map. I completely agree that a disclaimer of location accuracy is sufficient if the relative precision with which locations are known is not the purpose of the map. – Gregory May 9 '12 at 1:22

Not the answer you're looking for? Browse other questions tagged or ask your own question.