Thursday 9 January 2014

Old atlas, new projection

Charles O. Paullin and John K. Wright's Atlas of the Historical Geography of the United States, first published in 1932, is an award winning work of art. A magnificent historical atlas published ahead of its time and including innovative thematic representations. Each one of the nearly 700 maps makes it a truly fantastic cartographic work deserving of a new audience.

The Digital Scholarship lab of University of Richmond have gone to great lengths (and no doubt great expense) to bring this atlas into the 21st century by creating a digital version. It's described as an enhanced edition, a composite work of the combined efforts of students and staff with support from the Andrew W. Mellon Foundation. Please do go take a look because if you've never seen the original the new version opens it up to new eyes and a significantly wider audience. That's a huge positive.

The digital version effectively takes the scanned content from the atlas and places it atop a digital map so the atlas is reincarnated in an online form. Text panels are included, legends are thoughtfully added as insets on the page and there are some really sensible, intuitive and simple controls to help navigate the work, include textual context panes and provide interactivity.

In addition to general enhancements, there are three main improvements over the original that the new online edition claims in terms of functionality. Firstly, the maps are now clickable. Great! Now we can access the data to go with the maps and mine the information for specific places. This is one of the main benefits of any online map of course. Second, many of the maps in the original formed a series which were represented using side-by-side small(ish) multiples but now they are animated (effectively flicking between the overlayed small multiples). Again, web mapping supports animation very well and it's been used to good effect here. Rollovers are also used well, for instance to hover over the isochrones on the rates of travel maps and reveal location specific details. 

Third...the new edition contains "Georectified maps...warped so they can be placed consistently on top of a digital map".

<rubs eyes in disbelief> Come again?

While placing the maps into a consistent projection to support some of the other objectives (like animation) is cited as a reason for the georectification I'm completely bemused.

Paullin and Wright chose their projection perfectly. Albers is an equal area projection. It is absolutely the right choice to represent thematic information because it maintains area across the map meaning that no one part has its visual importance in our perception exaggerated.  This is crucial if we're to accurately compare the values on the thematic maps and not have those impressions modified by a projection before they even hit our eyes. Yet here, they've taken the original plates and manipulated them into Web Mercator. In so doing, not only have they changed the originals in terms of their dimensions, they've warped the message the maps communicate because they've distorted area. The unnecessarily curved text is really just an inconvenience of the process...there are bigger implications though it does show clearly the impact of changing projections on a map. Northern latitudes suffer most from the warping so they now appear larger in relation to southern latitudes.

Take the following as comparison. There is more yellow on the georectified top map than the original shown beneath. Your first impression is slightly different depending on which map you look at. The way your brain assimilates the image is based entirely on the area that each colour fills. This impact is subtle on some maps, profound on others.





And what of the impact on those nice round proportional symbols?



Not round any more so our ability to assess relative size is impeded.

At least you can toggle between the georectified maps and the original plates but here's my very simple beef with this work...why not use an Albers digital basemap? Why go to all that effort of making the old maps fit a completely inappropriate projection when the original authors had already chosen the right projection? Why let the tail of the default web map projection dictate the needs of the new digital version? There was nothing about the digital version that demanded Web Mercator. Just reconfigure your map app to use Albers and you not only save a considerable amount of work processing all those maps to nicely fit Web Mercator, you actually preserve the use of the correct projection.

What a shame...a terrific project with so many good elements but one utterly incomprehensible cartographic howler. They're not the first to fall into this trap though. Plenty of other large, highly lauded projects have also taken the Web Mercator route rather than tackle the issue of projections properly.

No comments:

Post a Comment