The North Circumpolar Maps

NRCan, 1990 "North Circumpolar Region" MCR 198

An interesting pair of maps that recently came to my attention (through an article in Cartographica) are the 1990 and 2008 versions of the Canadian government's "North Circumpolar Region." High-resolution images for both are available online free from Natural Resource Canada (NRCan), so I'll just post some thumbnails here and details.

NRCan, 2008 "North Circumpolar Region" MCR 0001

The original map, MCR 198, was a tremendous success in 1990, and MCR 0001 (2008) is an attempt to revise, update, and of course improve it. It is at a slightly smaller scale (1:9,000,000 instead of 1:7,500,000) but many things have been added: international date line, additional arctic ocean floor detail, current pack ice limits, and so on.

But, which would I rather have hanging on my wall? Hands down it's the older version. It looks better. So what can we learn here?

What's striking about the more recent version is the difference in the shaded relief. The first map was done with conventional cartographic tools, and an artist rendered the shaded relief. The second map was done with current GIS technology, the shaded relief being computed.

According to the article by the authors (Mapping The North, R. Eric Kramers and Andrew Murray, Cartographica, vol 45, No. 3,  p.201), the new  shaded relief was generated from a DEM using GIS software. However, it didn't look right until it was "tempered  with the less-detailed, manually-drawn, 1990 relief." The two were blended in Photoshop to give the right balance of the high-resolution (computer generated) and low-resolution (hand-painted) components.

It's not the first time I've read this: computer-generated shaded relief is too accurate for the big view: although each peak and ridge is there, the overall impression of a mountain chain is lacking something. Tom Patterson (DEM Manipulation and 3-D Terrain Visualization, Catographica vol. 38, # 1&2 Spring/Summer 2001, p.92) wrote that high-resolution DEMs give the wrong effect at small scales.

Especially problematic are glaciated northern mountains comprising tightly packed ridges and valleys...which often appear as an irregular texture rather than a recognizable topography.

And here are Anna Leonowicz, Bernhard Jenny and Lorenz Hurni (Automated Reduction of Visual Complexity In Small-Scale Relief Shading, Cartographica vol. 45, No. 1, Spring 2010, p.73) pointing out that, at present, hand-painted relief is simply better:

Cartographers currently lack advanced methods for using [DEMs] to produce relief visualizations at a level of quality comparable to traditional, manually-executed relief representations.

Patterson recommended that you could at least get closer to a hand-painted  result by manufacturing a low-resolution DEM (i.e., down-sampling the original), and merging a shaded relief version of that with the  shaded relief from the high-resolution data. In a sense this is just what the authors of the 2008 map did.

Nonetheless the hand-painted relief, for my money, on the original 1990 map still looks better. A circumpolar map is a very grand overview. No one is going to refer to it for details, and it is designed to hang on a wall or adorn a floor. It gives an overall impression of planetary scale; so here I prefer the old style of shaded relief, where the mountain ranges stand out so strongly.

Western Canada, 2008-style

Western Canada, 1990-style

Compare Western Canada: the bold and well-muscled 1990 version, and the wrinkly (and pink!) 2008 version.

Yes, I'm not fond of the colour fill introduced on the more recent version to show national boundaries. It just seems distracting to me and I have a sense that they added it because the GIS software made it easy to do so. I want this to be a map of the physical landscape, not political landscape. The brightly coloured Scandinavian countries are like confetti over in the corner of the map, whereas in the former verison they were deliciously dark and complex entity.

Scandinavia, 1990-style

Scandinavia,  2008-style

I do appreciate the extension, below the 55 degrees north, of basemap features in the 2008 edition. On the 1990 edition nothing was shown south of that latitude but the water/land division. Now, the highway I live near, at 54º 40', is actually there, on the map.

However, something was thrown away! Kamchatka, the Sakhalin Islands and the Aleutian Islands, which the artist gave shaded relief to in 1990 throughout their full extents, now have it cut off! Russia needs the full extent of Kamchatka, to emphasize Siberia's size and the way it overhangs the Pacific Ocean. Rendered this way, it just looks silly.

Kamchatka, 1990-style

Kamchatka, 2008-style

And, as you can see in these details, the blues that represent ocean depth were deeper and darker in 1990, which also provided a better overall effect.

I don't want to go on ranting here, because I'm not that great a map designer myself. But for GIS people designing maps, the 2008 map is a good lesson. Sometimes it's harder to achieve the effect you want using the latest tools.