Georeferencing and Digitizing a Bird’s Eye View Map
Historic Chicago Maps: Community Settlements
Georeferencing
Georeferencing a Bird’s Eyeview Map
On this page, I’ll present the process of georeferencing a bird’s eye view map into an electronic map, with all its components.
This map was originally created by Wesley Akin Hotchkiss, at the University of Chicago, and can be found at the Map Collection, Regenstein Library. The year of publication is not confirmed, but estimated to be 1948.
Following is a scan of the original map: a birds’ eye view map of Chicago, from 1940, showing the distribution of Catholic Churches relative to total population, as well as estimated location of the churches.
This map has two types of data: the number of persons per church in red, orange, and yellow; and the approximate location and number of the churches in each community area (inside Chicago), or municipality (suburbs of Chicago).
When looking closely, we notice a thicker black line within the map, which corresponds to the 1940s boundary of the city of Chicago. We base the georeferencing on this boundary, as its shape is easily identified. We are also able to identify the suburban municipalities by comparing the names on the current electronic map to the reference list on the left side of the map.
In a top view map, we usually use 5 to 7 control points to place a raster map to its geographic location. But due to the shape in this map, we used 24 control points. In the image below, you can see the control points on the map, and notice how the areas further away from the center distorted more than the central areas. In central areas, the polygons fit better on the original map, than the peripheral polygons. Some peripheral polygons overlap, but the shape and size might be different. In some cases in the West and South suburbs, the polygons intersect but don’t overlap completely.
The green crosses represent the control point on the raster image, while the red crosses represent the control point on the vector map. Ideally, both crosses should be perfectly aligned, however, they most likely will have a difference. The differences throughout all the control points are used to calculate the RMSE (root mean squared error), which serves as a parameter as to evaluate how well the raster map is aligned with the vector map, and therefore, the ground. The RMSE for this map is understandably high, at 1200km difference, also indicating that while we can connect the map to vector polygons, it would be hard to place the church locations exactly in the right place as intended by the original author.
As we add more points, ArcMap offers more warping options. After 3 points, we may use the “first order polynomial (affine)” option. After 6 points, we can use the “second order polynomial”. And finally, after 10 points, we can use the “third oder polynomial”. The differences in the three methods are shown in the image below. After testing several sets of control points and warping methods, deleting and replacing poorly performing ones, and checking for overlapping polygons, we’ve decided to keep the third order polynomial and these 24 control points.
After making the final decision in the georeferencing method, we can then proceed to digitize the features. But in order to do that, we need to build a map that combines the current boundaries of Chicago community areas and Illinois municipalities that are presented on the original map. The red polygons were kept as a final shapefile, or “final map” to be used with the data. This final map has underneath it a data table associated with each area.
Municipalities
Municipalities Identified from the Historic Map
Following the georeferencing, the municipalities identified from using a current shapefile and the bird’s eye view map are presented in the map below. The yellow polygons are the municipalities identified.
Churches
Number of Churches by Municipality
The map below shows the number of churches by municipality as counted manually from the original map.
Congregations
Persons per Church According to Colors on the Historic Map
The map below shows the range of number of persons per church as indicated by colors on the original map.