Web map projections: How to reduce distortions

All maps that are displayed using a map projection will have some distortion of area, shape or distance. However, web maps have exacerbated this issue because web maps can be zoomed in and out, which changes the map scale but not the projection. So, when a web map user is viewing a map of a city and then zooms out to display the continent using the same map projection, distortions can become evident. Small scale maps of Canada are particularly vulnerable to map distortion due to our geographic location on the globe. Read this blog post to get a better understanding of map projection issues and how to reduce their effect.

I’ve been working on map distortion issues for a long time, and because the world is round and maps are flat, there is still no magic representation that will efficiently fix this problem. However, I was intrigued by a recent blog post that gave some background on the history of the issue and provided some suggestions on how to reduce map distortion effects. The blog is entitled “Mercator, it’s not hip to be square”, by Dr. Kenneth Field.

Canadian web mappers have struggled with this area distortion problem for some time, and the blog characterizes the issue as a cartographer’s nightmare. Dr. Field states: “… I’ll explain … using the recent 2019 Canadian general election data. Canada is a cartographer’s nightmare. It has a huge landmass that extends across a large north-south extent which sits squarely in the Web Mercator zone of highest distortion. It also has vastly differing population densities across its area from a more densely populated south to an increasingly sparse population in the north.” So, as we have seen many times in the past, Web Mercator projections (and most map projections) are particularly problematic for representing the entire Canadian landmass.

Here are examples from an article about the 2019 Canadian election results displayed by riding. The map on the left is displayed using the Web Mercator projection, and the map on the right uses the Canada Albers Equal Area Conic projection.

Dr. Field then goes on in his post to give three alternative ways of publishing web maps using anything other than Web Mercator. Here are his suggestions:

1. Create vector tiles

For those who are not yet familiar with vector tiles, they are much more adaptable than raster tiles because the data is divided into small areas or squares (tiles) of data that can be stylized by the viewer software; be displayed at any map scale or resolution; and can also be re-projected to any suitable projection by the map viewer. So, if you wish to allow the user to scale and project the map as the user wishes, then vector tiles are the way to go.

2. Create raster tiles

If you want to fix your map into a specific projection and include all the annotation and symbology to make the map more consistent and understandable, then the solution is to create a raster cache (aka image tiles) of your map. A raster tile cache is a database of tiny images of the map that are individually produced at each of the various map scales and are requested and displayed automatically as required by the user’s map viewer. This has been the tried and true method of producing web maps since the early days of web mapping.

3. Using an ArcGIS Online workaround

Since ArcGIS Online does not yet support user-defined reprojection, Dr. Field suggests finding and loading a hosted tile layer that is available in the projection that you want. Then, load your thematic layer onto the projected layer, which will automatically re-project the thematic layer to the same projection as the hosted tile layer.

While the Web Mercator projection does create a lot of distortion, especially in Canada’s North (above 60°N), it is quite useable in more southern latitudes in Canada. Also, when the web map is zoomed into city scales, the issues with Web Mercator are much less of a concern. So, it’s quite alright to continue to use the Web Mercator projection for larger-scale maps, but for national or northern maps of Canada, alternative projections may be considered more useful.

Here are examples of the 2019 Canada Federal Election Results Map from top to bottom, left to right: 1) Canada Albers Equal Area Conic; 2) Canada Lambert Conformal Conic; 3) WGS 1984 Web Mercator Auxiliary Sphere; 4) North American Equidistant Conic.

So, the Canadian landmass is challenging to display in a web map for many reasons, most of which are just basic physics and mathematics complications. Web Mercator still works relatively well at most southern locations in Canada and larger scales; still, as one displays larger areas or more northern areas, distortion issues related to area or distance may crop up in your map. However, several sensible workarounds can be used to display thematic data of Canada at various map scales from local (large scale) to global (small scale). Try such solutions if you think the distortions within your web map need to be reduced.