Rectifying Maps for the NYPL

For this week’s “making things digital” class, I decided to do something a little different than digitizing text. When I saw Ben’s post of suggestions, I was immediately drawn to the last option: Rectify Maps for the New York Public Libraries. I had done a bit of basic GIS before and was interested that they had an in-site rectifying tool rather than requiring complex and expensive GIS software.

I went to the site, watched their video tutorial (not the best quality video, but it told me exactly what I needed to know), and decided to start giving it a try. Rectifying a map involves three main aspects: the historical map, the base map, and your control points. In order to rectify a map, the user places control points on similar locations on both the historical map and the base map. These control points are paired to each other. By carefully placing enough of these control points, the user can manipulate the historical map to match up with the modern base map.

The next step was to choose what kind of maps I wanted to rectify. I wanted to choose a place and scale I was familiar with so I started searching for historical maps of my home state, New Jersey. I found two maps that I found very interesting and began working on rectifying them. One is a 1795 engraving of New Jersey by Joseph Scott of The United States Gazetteer  (Philadelphia) and the other is a 1873 map of New Jersey from the Atlas of Monmouth co., New Jersey. Here are images of the historical maps before rectifying them:

NJ Map 1795

NJ Map 1795 (before)

NJ Map 1873 (some control points shown)

NJ Map 1873 (before)

I have decided to include some of my control points into the 1873 map so that you can see what they look like. In order to properly rectify a map, you must have more than one control point. The NYPL site requires that you have at least three control points in order to rectify the historical map with the base map. Also, the warper includes a mechanism that determines how off (margin of error) each of your control points are between your historical map and your base map. The tutorial video instructs you to make sure each of your control points have a margin of error of less than 10. Going into this, I assumed that  more control points linking my historical map to my base map would result in a the more accurate rectified map. However, this is only if you can get your control points under that margin of error of 10. Also, adding more control points can often distort the margin of error for your other control points. So it is not always best to have the greatest number of control points, but instead one should place control points in optimal positions yielding the least margin of error. Each map is also unique, so you need to find out what you think the best arrangement and number of control points are. I am not saying that my rectified maps are perfect (they are far from it), but I found that around six control points did the trick.

After placing these control points, I cropped the historical map a bit so that it would fit better on the base map, then I clicked “Warp Image!,” then played around with the transparency settings of the historical map in order to produce these new rectified maps:

NJ Map 1795 (after)

NJ Map 1795 (after)

NJ Map 1873 (after)

NJ Map 1873 (after)

Now I will offer a few final thoughts about the process (although I definitely expect to do this again). First, rectifying maps is a frustratingly precise process. Borders, state lines, and towns on the base map are often in very different locations (or non-existent) on the historical map. Also when you are placing control points you have to be constantly aware of not only whether or not your control points line up to the correct location on the historical map and the base map, but also of how each control point affects the margin of error of each other control point you placed. For example, I tried rectifying a map of the United States and was able to place three control points with very little margin of error for each. I placed them at the Northwestern part of Washington, the Southwestern corner of California, and the Southern-most point in Texas. However, no matter where I placed the next control point, the margin of error seemed to skyrocket for all four points as soon as I placed the fourth one. This might have been a problem with the first three points, but it did prompt me to scale down my efforts from maps of the entire United States to New Jersey maps.

There is one last thing that I wanted to comment on, and it deals with the base map. I was thinking about how the entire process of rectifying these maps concerned warping the historical map to fit the base map. This one-way process assumes that the base map is the accuracy standard and all other maps must conform to its scale and borders. I think that this assumption is something that is taken for granted. I understand the need to have a standard map, but could it not also be useful to have the program do the reverse? What if it generated  an overlay of the historical map on the base map AND an overlay of the base map on the historical map? What kind of value would an arrangement like that have? I am not sure, but I think it is something that at least needs to be considered. Also there are many historical maps that contain different information than the base map and are, therefore, incompatible with the rectifying process (although they are still listed on the site). I just wonder that if by placing such confidence in the base map, we are losing important information from the historical map.  I’ll finish this post by showing one of those maps that are listed on the NYPL site but could not possibly be rectified to our modern base map. There are many of them, but this one in particular stuck out as a very valuable and informative map that is completely incongruous with the base map.

1671 Depiction of Floridans

1671 Depiction of Floridans

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code class="" title="" data-url=""> <del datetime=""> <em> <i> <q cite=""> <strike> <strong> <pre class="" title="" data-url=""> <span class="" title="" data-url="">