I live with a couple of cyclists, remedy who spend many of their summer days out on the trails west and south of Madison. A few months ago, buy one of them asked me to make a bike map for him, pointing out what he felt was a shortcoming of the ones available to him: it’s hard to figure out where the hills are. This is particularly critical if you ride in places like the Driftless Area, as my roomates do. A map showing you where to turn and which roads have wide shoulders and low auto traffic is very useful, but it doesn’t tell you how rough the next hill is going to be.
Figure 1: The above is a draft of one of my first attempts, in this case depicting a particular ride that one of my roommates hopes to participate in this summer. click to see fullsize
So I set to work considering how best to show elevation changes along a route, and I came up with a relatively simple concept: encode the elevation of points along the route using line width.
The symbology here is, I think, fairly efficient. By varying the width I am encoding three pieces of information: the elevation of the path, the slope, and the aspect. The first is not particularly useful, but the other two are the critical pieces of information for the cyclist. Importantly, both need to be on the map together – knowing the slope of a hill is great, but you also have to know whether, as you head along the road, you’ll be climbing up that steep grade or coasting down. Getting all that information into one symbol is not necessarily that hard. Both slope and aspect are derived from elevation, so it’s really just a matter of producing a map which shows elevation in a way that makes it easy to see the pattern of how elevation changes. Show the one variable, and your brain fills in the other two. But, it works a lot better if the symbology makes it easier for your brain to figure out how elevation changes. Compare the two maps below:
Figure 2: Using the visual variables of lightness or size to encode data
One encodes elevation along the path by width, and the other by color value. In my opinion, slope is much easier to figure out when line width changes than when the color value does. The color at A is darker than the color at B – but can you quantify how much darker? And can you do it as easily as you can tell how much wider the line is it at A vs B? Speed and ease of understanding are, I think, particularly important given how the maps are to be used. I am told that these will be read by people who don’t even stop their bikes while reading the map (I don’t really know anything about biking – I’m not usually permitted outside the confines of the UW Cartography Lab). So, the map has to work when they’re not looking closely or long at it. The second advantage of line widths over something like color variations is that line widths are more robust – they won’t vary according to lighting conditions, as the users bike in and out of the shade of trees and in varying levels of cloud cover.
The map on the left (using lightness) does have a couple advantages of its own. A small one: by not changing line width, we don’t have to worry about lines getting too wide (causing crowding) or too narrow (and thus being hard to see). The other advantage is really more of a lack of a disadvantage – the highest elevations are not dominant. Look back at Figure One for a moment – notice how the south-center part of the map stands out the most. It’s at the highest elevation, so it has the widest lines. But it’s also mostly flat stretches, which means that it’s not a big deal to our cyclist – they want to know about the hills, about the changes. Encoding elevation by colors keeps the reader from focusing attention as much on the high elevations, which won’t stand out quite so badly.
Instead of encoding something the cyclist doesn’t care about (elevation) and letting them figure out the things they do want to know (slope and aspect), we could just encode the latter directly. Again, though, we need both for it to be useful, and so here’s where it gets tricky. Slope isn’t so bad – it’s just however many degrees the angle is, so that’s something we can pretty easily put into a color ramp, for example. But aspect is the hard part, since it depends on which way you’re going down the road. It’s uphill one way, and downhill the other. You could put little arrows or some other indicator next to the road to indicate which way is uphill. Or perhaps encode the aspect in color hue (red for north, blue for east, etc.) while changing the lightness of the color to indicate the grade. Or, you might try this:
Figure 3: The arrow points downhill, and larger arrows or darker ones indicate steeper slopes.
There are more possibilities, obviously. But I am of the opinion that these solutions are somewhat weaker than simply showing elevation directly – the reader has to process two different symbols (or two properties of the same symbol) and extract two pieces of information. Maybe that’s still easier than processing the symbols to extract elevation, and then calculating slope and aspect internally. But I do not think so. If you present someone with a map they intend to use to figure out the lay of the land, they’re expecting to see the terrain – hills, valleys, etc. Figure 3 above is getting too abstract. It doesn’t feel like land anymore, and so it’s harder to interpret. This is why people like hillshading – mountains look like mountains, and that’s something we can understand without a lot of processing.
I imagine a reader could train themselves to interpret something like Figure 3 faster and easier, since it does show what they want to know with about the least amount of ink possible, and without showing anything extraneous. But that will take effort and learning. Right out of the gate, I think a map showing elevation is easier to understand, because it’s a lot easier to figure out what the landscape is going to look like.
Thinking about the landscape was, in fact, what led me to the initial technique of encoding elevation by line weight. I had simply thought of it in terms of looking down on the world from high up. Roads which are at a higher elevation would be closer to your eye, and so appear larger than those far down in valleys. Whether or not this particular concept is working in the back of people’s minds when they see these maps is another matter, but it at least provided the inspiration. The more academic analysis came later, much of it while I was writing this up.
While I appreciate any general feedback readers would be so kind as to provide, I’m particularly interested to know if anyone’s seen anything like this before. It’s not a terribly complicated symbology idea, so I imagine someone somewhere must have thought of this.