Terrain in Photoshop: Layer by Layer

Last year, I successfully used this space to prepare a version of my NACIS 2015 talk prior to giving it, and if you’ll indulge me, I will do so again this year. Turns out it’s helpful to write your thoughts down before delivering them to a roomful of 150+ people.

EDIT: If you’d like to see the final talk that came from this, you can view it here: https://youtu.be/npTWWbuZ3AA — it’s an abbreviated version of what you’ll find below, but also contains a sort-of-nice peroration that isn’t found in the text.

For the past couple of years, I have been mapping the terrain of Michigan. Though the phrase “past couple of years” may be misleading: I worked on a first draft for a couple of months in 2014, and a second draft for a month in 2015, and now it’s still sitting there, almost-but-not-quite done. When it’s properly done, I’ll post more about my rationale for making it & some design decisions, but in short: it’s a love song to my home, a chance for me to explore and understand its landforms more intimately, and an opportunity for me to build some terrain mapping skills—putting into practice all of the lessons I’ve absorbed over the years from colleagues at NACIS.


There are some odds and ends yet to take care of (some of which I’m only noticing now as I write this post), but this is mostly how it’ll look.

I’m often intimidated by the complex maps of others, feeling like I could never achieve the same thing. But seeing how a piece was made—besides being educational—often helps dispel that feeling. So, let’s break this down in Photoshop, layer by layer. This is more of an overview than a step-by-step of what buttons to click; I think it’s more valuable to explain why I beveled something, and leave the Internet to explain what buttons to click to actually apply the bevel effect. Oftentimes I feel like good mapmaking is just a matter of fumbling around, trying to copy others, so that’s what I’ll try and put all of you in a position to do. It worked for Bob Ross’s viewers, after all.

Intertwined in all of this is a story of the people who made this map possible through all of the wisdom they have shared over the years with me, in personal conversations and through presentations I’ve attended. So, you can try and copy me while I tell you about all the people I’ve copied.

Let’s start at the bottom, with land cover. We open with a simple light green fill. This will stand in for most of the non-woody vegetation in the state: grasses, crops, etc.

01 - Base Color.png

To that, we’ll add some trees in a darker green. To do this, I make use of a tree cover layer from the National Land Cover Database. It tells me, per pixel, what percentage of that area is covered by tree canopy.


(Aside: NLCD only covers the United States. This map covers part of Canada; for the most part, I’m not going to talk about dealing with the Canadian “Circa 2000” land cover data because it was a frustrating distraction from what I want to convey here. I’d probably look at using a Landsat-derived source if I were doing it over; maybe for Draft 3.)

To add this to my map, I use it as a opacity mask on a darker green fill layer. So, places where our data show more tree canopy get more of this darker green blended in.

If we zoom in, you can see that I’ve used the Dissolve blending mode. What this basically means is that it’s using the tree mask to control the density of the green pixels, instead of their transparency (as we might expect an opacity mask to do). So the green fill gets sparser or denser based on where the data say there are more trees.


I like this because it gives a little texture. If you want the texture rougher, you could merge these two layers and run a median filter.


Now that we have some basic vegetation, let’s add some other land cover types. Again, all from NLCD data (for the US, at least). Let’s start with wetlands: plenty of those in Michigan. I give them a blue-green color, and put them on top of the vegetation using the Multiply blending mode. This treats the layer sort of like stained glass: it darkens & modifies the vegetation colors underneath, so now we have some waterlogged vegetation.


Setting layers to Multiply in Photoshop (and Illustrator) is a simple but invaluable trick, and it also introduces to us the first character in our story: Tanya Buckingham. Tanya runs the UW Cartography Lab, and has been my mentor since my student days. And she loves setting things to Multiply. I’ve forgotten the origin of so many of the little things I know how do to, but I remember picking up a love for Multiply from her; I use it all the time in mixing colors. In the bigger picture, Tanya is the reason I started going to NACIS, and therefore is indirectly responsible for most of the rest of the stuff I’ve learned that she didn’t teach me personally.

Next up is impervious surfaces: cities & roads. NLCD has some handy data that show not only if an area is mostly concrete, but what percentage impervious surface it is. I blend that in to the landscape again using the Multiply blending mode, so it darkens what’s underneath without totally greying it. I don’t want these areas to stand out too much; just enough to see that they’re there. Some people make these red or purple or otherwise pop them out of the landscape, which makes sense for some applications, but in my case I’d like to keep the cities from distracting us from the vegetation.


Finally I take the Bare Land classification from NLCD and add that in. This land cover type represents areas without vegetation, which could mean it’s bare rock, but in Michigan usually means it’s sand: the state has some of the world’s finest sand dunes, and they’re a big tourist attraction. I give this layer a sort of sandy color using the Color Overlay effect, which just replaces all your pixels with a selected color (which makes it easy to change it layer). Then I drop it on top of everything.


There’s only last piece to the land cover puzzle here, and that is void data. In Canada, there were some spots where the land cover data I had simply listed areas as “unknown.” Since they were small, and they were outside of the map’s subject area, I did what any lazy good mapmaker would do: I filled them in with made-up data.

To do this I used Photoshop’s Content-Aware Fill ability. And herein we introduce another character: Alex Tait. Alex is the Geographer at the National Geographic Society, and among other things he co-founded Practical Cartography Day at NACIS, which is where I picked up a lot of my most useful cartographic ideas. So, like Tanya, he’s indirectly responsible for a lot of what I know. At NACIS 2010 he gave a presentation on “Photoshop Tips for Practical Cartographers,” in which he showed off using Content-Aware Fill to remove pesky clouds from satellite imagery, and so when I saw these void areas in my land cover, I thought back to his talk.

In short, CAF looks around your image and finds a similar-looking area and uses it to fill in your selection. So, I select my voids and tell it to look at the rest of the map and use those patterns to fill them in. After that, there’s also some spot checking and manual adjustments to these made-up areas.


(Aside: I filled these voids in 2014; since then I’ve become aware of other land cover products that I’d probably use instead to avoid this problem, or at least patch in the gaps. But I keep this section here because I think it’s useful to show off Content-Aware Fill.)

So now we have a land cover layer (except water) that will be the basis of the map’s colors. Some people do this with a lot more classes: they might use a separate color to show crops, or show deciduous forests differently than coniferous ones, etc. NLCD has all that data (they have 20 land cover types, in fact). But for my purposes, this was the right balance. It gives the basics, without being too busy. And I like building a color layer from land cover data because it has some realism without being as noisy and cluttered as satellite imagery. It’s a simplified, but recognizable, version of the truth. Good generalization is a caricature, and that’s what cartography comes down to.


Next up comes a little color adjustment: making the land cover more green. I do this here because I sent an earlier draft of this map to Tom Patterson, and he suggested making it more vibrant. As a general rule, I do what Tom Patterson says, because his reputation as one of the foremost cartographers of our era is well-earned. But also I agree with his critique in this instance.

I met Tom at my first NACIS conference, in 2009. I’d admired his maps as a student at UW–Madison, and had already started to idolize him before seeing him in person. I remember walking up to him and basically interrupting his conversation to introduce myself. Which, given that I’m a very shy and introverted person, indicates how badly I wanted to meet him. Since then he has been a great mentor, colleague, and collaborator. I have learned a great deal from his presentations, his articles in Cartographic Perspectives, and our personal conversations. He’s a very friendly and helpful guy and you should definitely all get to know him.

So, he was nice enough to give some feedback on an early draft, and because of that I made things a little greener. The difference is really subtle, but it’s there.

There's a difference, I swear. It's more noticeable when you turn the layer on & off. My career mostly involves tiny tweaks that make things better without anyone noticing.

There’s a difference, I swear. It’s more noticeable when you turn the layer on & off.

(Aside: you’ll see that all the land cover work done up to this point is flattened into a single layer. I did this to save computing power because I was working with giant, 2–4 GB files.)

Atop this, I add a hue/saturation adjustment layer. It’s got a mask on it, so that it’s only affecting the areas outside of Michigan. I use this layer to make everything a little greyer, a little lighter, and a little bluer. This helps create a distinction between subject and non-subject area.

Next up is the one major piece of land cover we’ve been missing: water. It’s actually built out of 3 pieces.


First off is the water layer from NLCD, combined with vector water bodies from the National Atlas, shown above in white. That covers the lakes pretty well.

Next up is streams, which are often too narrow to appear on NLCD, so—like the water bodies—we need to supplement with vector data. The National Atlas has some great 1:1 million-scale stream data, and that’s where the black lines above come from. I don’t want the map to be dominated by streams, though, so I break them into two classes: major & minor. The minor ones are thin enough that they’re more in the background: on the final map, you can see them if you look close, but they’re not prominent. On the other hand, I want the major rivers of the state to stand out. The problem is the the National Atlas data doesn’t have a way of sorting major from minor. So, here I turn to another data source: Natural Earth.

Natural Earth’s vector data is spearheaded by our 4th character, Nathaniel Vaughn Kelso, a longtime NACIS member with a passion for developing tools and resources to help others make better maps. Most of you are likely familiar with Natural Earth, a data library that underpins a great deal of present-day cartography by providing free, clean, consistent vector & raster layers that align with each other. NVK & Tom Patterson worked together with an army of volunteers to make possible so much of what mapmakers have done since then.

Natural Earth is actually too coarse for what I want (their finest data are at a 1:10 million scale), but it is perfect for telling me which parts of the National Atlas data to use. If a river is significant enough to appear in Natural Earth’s 10 million data, I take the 1 million National Atlas data and mark it as “major” and give it a thicker stroke. I also taper the strokes a bit in Illustrator before bringing it into Photoshop.

Finally, I make all of those layers look more water-like by appling two layer effects in Photoshop: a color overlay to simply turn them all light blue, and a bevel effect. The bevel effect is something else I got from Tanya Buckingham. Adding it gives a little depth to the water by putting a thin shadow on one side and a highlight on the other, embedding the water into the land. Like most good effects, you can’t really tell it’s there on the final map, but you can tell when it’s not.


Next up is the shaded relief. This is a map about landforms, after all. I got my DEM from the National Elevation Dataset, which fortunately covers Canada as well.

Tom Patterson is the shaded relief guru, and my admiration for his work is one of the major reasons I got into doing terrain mapping. He’s got a ton of great advice on his website, so I won’t repeat most of it here. But among other things, I took his advice to keep the relief from being too detailed by downsampling my DEM so that it was about half my intended print resolution.

The relief I’m using was generated with Blender, and I’ve got a discussion & tutorial about it here, so I won’t go into detail. In short: it’s a program I started working with because I think it produces much more natural-looking shadows.


Mixing relief with a color layer can be tricky. I used to have fairly amateurish ways of doing it through simple transparency overlays that muddied my colors and/or darkened everything too much. In recent years I have developed better methods, which owe much to advice I received from both Tanya Buckingham and Tom Patterson (you’ll note I come back to them a lot). Each of them has their own slightly different way of doing it, and mine is a sort of modified hybrid of their approaches. I have a full tutorial here, but I’ll give you the short version.

Basically, I take two copies of my relief and use Levels adjustments in Photoshop to leave just the shadows on one copy and just the highlights on the other.


I then set the shadows to the Linear Burn blending mode, which causes them to darken the land cover underneath (in a way that preserves the color vibrancy better than using the Multiply mode), and set the highlights to the Screen blending mode, which lets them lighten the land cover. So I’m left with a land cover layer that’s lighter and darker as the relief dictates, but still has the same color palette.


Finally, atop that, I use Levels to lighten up the relief in the areas outside of Michigan. Here I’m using that same state-shaped mask that I used to lighten the landcover.


There’s one more piece of landform puzzle, and that’s to add a texture shade. Texture shading is a technique developed by Leland Brown, a mathematician and hiking enthusiast in Los Angeles. It basically helps emphasize edges and small textural details in the landscape.


A raw texture shade is cool to look at, but it’s most valuable when paired with shaded relief. Here, I use the same Linear Burn/Screen method as described above to mix it in with the relief & land cover.


Leland’s work has been a boon to mine, and everyone else who does terrain mapping. He first presented texture shading at NACIS 2010, where he came from outside the world of practicing cartographers. His presence at that meeting was partially due to Tanya Buckingham, who organized the conference that year. She wrote the Call for Participation that found its way to Leland, and her invitation made him feel like his participation, as a hobbyist rather than a professional, would be welcome. NACIS is Nicest, as we say.

At that conference, he met Tom Patterson, with whom he started working on promoting texture shading and making it more easily available via a standalone application. Through Tom’s connections to Brett Casebolt of Natural Graphics, texture shading is now a part of Natural Scene Designer.

So, Tanya helped encourage Leland to come to NACIS, where he met Tom (& me), and because of all that I can now use texture shades on my maps.

As with the relief, I tone down the texture shade outside of Michigan using levels. Notice also that I also have a mask on the texture shade shadows—this mask contains the shaded relief. The basic idea is: let’s not darken the same area twice. If it’s already darkened by the relief, the texture shade is made much more subtle, to keep things from getting way too dark.


Finally, we have the bathymetry, to give some texture to the Great Lakes.  I downloaded data from NOAA that had lake depths (as well as incomplete land elevation).

I use a mask to confine the data to just the Great Lakes, and then adjust the levels so that I have a black-to-white gradient.


Then I use a gradient map to change the color scheme. This is a simple tool in Photoshop that lets you turn a greyscale layer into a color one, by simple mapping each level of grey onto a color gradient. So, here I turn the white areas to light blue and the black to a darker blue. It’s all editable on the fly, so if I want to change colors, it’s quick and easy.


And that’s the basemap! If you’re playing along with our home game, the total layer stack is:


It’s a lot of pieces, and the final result is complex, but by breaking it down piece-by-piece, hopefully I’ve shown that it’s often composed of conceptually simple parts. A little green here, a little bevel there, etc. I did not conjure this fully-formed from nothing. It’s just built up from a lot of little operations, and so hopefully I’ve demystified some of this if you’re intimidated like I probably would be.

The final piece of the finished product is the labeling. I went with Minion Pro, because I liked the swash capitals. I won’t get into details about the typographic choices here. Instead, I’ll just highlight a couple of things.


First off, I add a white outer glow to my type, which is based upon a concept I picked up from Tom Patterson a few years ago at a presentation he made to the UW Cartography Lab. I like glows because they’re less heavy than adding a vector halo around the type. Again, I’m trying to do things that people don’t notice are there, but would notice if they weren’t. The glow is modulated a bit by a mask: they show up stronger when the map base gets darker. They hardly show up at all when the type is against light areas and it’s not needed.


The other thing I want to point out is a challenge that I don’t normally face in mapping: what to name things.

There are plenty of interesting landforms in Michigan, but they often have poorly-documented names, or no names at all. The US Board on Geographic Names had some things pretty well covered, but there were plenty of hill ranges, uplands, and more that lacked official toponyms. And even when a feature had a toponym (official or otherwise), I sometimes found only vague or conflicting information about the actual extent of the feature. I spent a lot of time digging through various websites and old maps, and sometimes making gut decisions about what to label things. Hopefully when this map is released it will prompt people to send me improved information so I can prepare a more authoritative draft.


I looked at a lot of maps, including these, and documented my sources and choices in a spreadsheet.

One particularly valuable source was some work by Randall Schaetzl, a geographer at Michigan State University. Among other things, he published a paper dividing the state into a number of physical regions, from which I drew a lot of label inspiration. Co-author on this paper is another NACIS colleague, and co-organizer of this year’s Practical Cartography Day, Carolyn Fish. I’d known her for years through NACIS before stumbling upon her name in this unanticipated context.

Since I was making a lot of semi-arbitrary (but research-backed) decisions, I asked for feedback from yet another colleague I met through NACIS: Leo Dillon. He works at the State Department, and is also a member of the US Board on Geographic Names. He primarily works with foreign names, but he was kind enough to review my domestic naming decisions and run them by some Michigan natives in his office. I also asked for feedback from Tom Patterson, as well, but you’ve already heard enough about how awesome he is.

And there you have the current map. It’s in a second draft; will need one more draft someday when I get around to it, but it’s mostly there.


There are some practical mapmaking takeaways here, and hopefully I’ve inspired some folks to dig into how some of these tools work, or give new techniques a try. But the bigger takeaway is this: the connections I have made to all of these people are what enabled this work far more than any particular knowledge of programs & tools. This is a profession driven by who you know, but not in the traditional sense of “who can get me a job,” but instead “who can keep me growing and learning?” I have been grateful to find that cartography is absolutely brimming with people who want to share, who want to help, and who are willing to dedicate the time to help their colleagues do better work. I owe so much of the quality of my work to these people, and I know that I am not nearly unique for being in that situation. We all depend on each other to keep growing.

I doubtless owe unacknowledged debts on this project, and I apologize to those of you who taught me things that I left off. As the years go on, the sources of my knowledge sometimes fade. Feel free to chime in if you’re responsible for something I’ve done =).

A Career Built on Side Projects

I am asked, from time to time, how I have managed to make it as a freelancer (so far). For those who are unwise enough to rely upon me for advice, I generally offer two major comments: first off, the network of comrades I have developed via my participation in NACIS has been a big piece of the puzzle—they send work my way and help me out when I’m stuck on projects.

But the second big piece of advice I give to students and other inquirers, besides getting actively engaged in the carto-community, is to make a lot of stuff that no one is paying you for.

At present count, there are 29 projects in my portfolio. Of these, only 4 are paid client work. The other 25 are a mixture of pro-bono projects and, mostly, random things that I wanted to try making. I have unfurled lakes, I have diagrammed rivers, and I’ve messed around with Penrose tiles (that one’s not in my portfolio, though maybe it should be). Some of these are afternoon projects, but several of them have been significant undertakings, eating up dozens of hours spread out over months. Usually when I’m done with one of these projects I have no idea what to do with it, since no one asked for it.

In sum, my portfolio represents a large amount of unpaid and unasked-for effort. But these various “side projects” (a potentially inaccurate label, given how much of my time they have taken up) have probably been invaluable to my career, for a number of reasons:

Artistic Freedom

The problem with client work is that, while it pays, it doesn’t always feed the soul. The particular needs of clients don’t always permit me the creative freedom to do something that I find personally pleasing: they want me to use this set of colors, wedge this logo in there, add that probably-unnecessary scale bar, etc. And, of course, sometimes I have an idea that I think will look cool, and the client wants me to go in another direction.

But with unpaid projects, I have no one to answer to. I choose the area, the data sets, and the techniques, and I can make whatever I can envision. Sometimes (or, maybe, often) it doesn’t work out, but when it does, it’s usually very satisfying in a way that client work does not often match.

The constraints which clients place on me can offer fun (or frustrating) challenges, but in either case, sometimes I want to make something that fits my taste and wishes, rather than someone else’s.


From time to time, one of my side projects spreads around a bit on social media or news sites. The result is that more people find out about me, including potential clients. Added bonus: more people, in general, become aware that “cartographer” is a real job and maybe think slightly more about how their maps are made.

Client work can, of course, turn into free advertising, as well, and I’ve certainly seen this happen to colleagues. But, unpaid work is perhaps more likely to, simply because I have the flexibility to make more broadly appealing maps. Clients often ask me to make straightforward maps of mundane subjects, like real estate properties, or episcopal sees in Poland (this one comes up much more often than I would have expected). These maps are often in greyscale, or otherwise simple in style. They’re valuable in their own context, but also less likely to attract random passersby on the Internet. With unpaid projects, though, I have the option to stretch my wings and produce something grander, bolder, and on a subject that may draw more eyes.

A note of caution here: all this makes sense if you’re doing the work for yourself. You may run into unscrupulous types who want to instead get you to do work for them for free in exchange for the benefit of “exposure.” Sometimes this even takes the insidious form of a contest—”hey designers, create a new transit map for our agency and we’ll use the winning map for free and tell everyone your name!” Don’t give other people free work. But, if you’re not doing anything else with your time, and no one is going to pay you to do the cool thing you want to do, then getting some exposure out of it is a positive.

Trying New Things

Sometimes, I see someone demonstrate something cool at NACIS—a new piece of software, a trick in Illustrator, etc., and I want to try that out. Or maybe I’ve got a weird technique idea of my own that I’d like to experiment with. Unpaid projects give me the opportunity to learn new skills. It’s nice if I can get paid to do it, but that’s not always possible; I often don’t feel like waiting for someone to happen to need this idea I’ve been wanting to test.

Side projects give me a chance to mess around with ideas without the pressure of having to succeed. For example, I spent many hours exploring shaded relief in Blender, trying to figure out how to make it work and what the right parameters were. It is unlikely I would have found a client who wanted to pay me for all the time needed to figure out how to do that, just so I could stick a subdued relief background on their map. And, if I decided to learn the software without charging them the extra hours, I still would have been under time pressure to learn new software and solve unique problems while getting their map made by the deadline. By doing it on my own, I now have the knowledge and experience to make use of it whenever it’s needed without having to worry about whether or not I can learn it fast enough.

My side projects are investments in my skill set. When clients come calling, I have a bigger toolkit I can draw on, and that makes me feel more confident in telling them, “yes, I can do this for you.” Instead of pacing around trying to come up with The Idea that solves their problem, I’ve got a lot of existing options to try out, some of which I’ve only used before on side projects. Sometimes the clients even make it easy by pointing out such a project in my portfolio and saying, “make it look like that one!” The unpaid work can make the paid work easier and faster (note to self: stop charging by the hour).

Filling the Shop Window

Having a good portfolio online is important to getting work, since a lot of people who want to hire me would understandably like to see what I’ve done before. This is as true for freelance work as it is for finding regular employment.

Side projects are an important part of my portfolio. As I mentioned, they’re presently 86% of what I’m showing off there. First off, not all of my client work can be shared—sometimes the information contained therein is not for public consumption. But also much of it is simply not that eye-catching, as I mentioned above. I’m not trying to demean those projects. I appreciate solidly-executed, no-frills mapping. I make a point of including some of those items in my portfolio (and, as I reflect in this post, I should really put some more of them in there). But I’m also trying to impress people in order to convince them to give me their money, and so I want a lot cool stuff in there; my unpaid work is often my most interesting (“how about I make this  thing that no one would ever need, and therefore never pay for, but which is awesome?”). And, because I’m often experimenting with new ideas and techniques, these unpaid projects also give me a chance to show a wide range of cartographic styles, so that a potential client is more likely to see something in there that looks like what they want. This also ties back into the creative freedom aspect, above: I want to show clients the sort of work that I want to do.

Print Sales

Despite these being side projects, some of them do end up earning me money. I’ve put a number of projects on Zazzle over the years, which is a print-on-demand service. I upload a file, and then they send me money sometimes. It’s pretty easy. I never produce these maps with the expectation that they will make money (and often they do not), but sometimes they do and that’s a nice way to get a little something back for the investment of time.

Over the last several years as a freelancer, I’ve had a lot of time periods in which I had no clients. But these intervals have not been downtime. I have used these spaces in order to try new techniques and satisfy creative urges, the results of which sometimes earn me client work or, rarely, a little money in print sales. So, my unasked-for advice to you, if you’re underemployed in cartography, is to fill the spaces: find something you’re passionate about and make maps for yourself. It may well pay off in the end.

(Of course, this sort of approach means I have a zillion unfinished projects, some of which have been lingering for years—but we won’t talk about that)


Advertising the Physicality of Old Maps

This morning, thanks to the magic of Twitter, I was alerted to an article on Slate about the maps at the Osher Map Library, at the University of Southern Maine. One of the things that the article points out, and which was the focus of Gretchen Peterson’s tweet about it (and, as I later realized, was the article’s alternate title and suggested Twitter text), was that not everything important about a map survives the digitization process.

“As soon as you turn a primary source into an image, you start to lose something,” Edney suggested.

Second (and more difficult to reconstitute on a computer screen) are the physical details of an object—its size, its smell, the grain of the paper. These are the features that can help us situate an object within its vanished lifeworld, showing us what it meant to those who made it, along with the ways it helped them make meaning from the world more generally.

If you know me at all, you won’t be surprised that I would agree with these sentiments. I’m always rambling about artsy aesthetic things, and I love print materials. But the article also made me wonder if there’s a way to recapture at least some of what is lost. The Osher Map Library, as the article points out, tries to photograph things like ragged paper edges or book bindings, which is a great idea.

I think it would make sense if a digitization project also included multiple photographs of each object, including wide and detail shots. Here, I’m thinking about the model used for selling prints online. Have a look at the way Axis Maps includes detail shots of their typographic maps, for example:


And here’s how National Geographic shows off its wall maps:


Potted plant sold separately.

Marketers want people to get a sense of the object they’re buying. Not just its informational content, but how big it is, what it’s made of, and what it’s going to feel like when it is in your hands or on your wall. They’re trying, insofar as a photograph will let them, to convey the physical aspects of the object. The focus on popular map sales is often aesthetic, rather than informational.

This sounds to me like the exact antidote to the Slate article’s comment about losing “the physical details of an object.” Nothing will substitute for seeing the real thing, to be sure, but it would certainly help to see a detailed scan of an old map alongside beautiful shots that highlight the paper grain, the impression made by the press, and how large it is — and here I am thinking about the part of the article that says,“researchers are still sometimes shocked when they request an item only to find that ‘you have to put eight tables together to unroll it.’” It’s tough to get a sense of the size of objects when they’re all on screen, and sometimes being able to imagine details like that are actually important in a research project.

Photography like this will help to stimulate viewers’ imaginations, helping them fill in the blanks imposed by not being able to hold the object in person.

Maybe some map digitization projects already do something this; I’m certainly not an expert on the subject, but at least so far all of the digitized maps I’ve seen have been single shots, capturing the object flatly. That’s certainly the most critical kind of imaging to do. And, of course, these libraries are usually on very tight budgets and are often lucky to get money to digitize maps at all. But, hopefully we’ll someday reach a point where some more effort can be applied toward imaging the map as a physical object, in addition to an information container, perhaps by borrowing a few pages from the advertising playbook.

On an unrelated note: this blog used to be pretty heavy on tips, tricks, and showing off map projects. It has, over time, begun also involving more idle and sometimes uninformed musings like the above. But I have no intention of abandoning the old type of content; merely supplementing it.

Naming the Golden Minutes

There is a special span of time that is known to freelancers of all stripes: that magical period after you’ve delivered a draft to a client, but before they have had a chance to reply to you with comments & revisions (or, rarely, immediate acceptance).

It’s one of my favorite times on the job, and as I reflect on it, I realize there are three threads of feeling associated with it.

There is pride in accomplishment: in finishing a draft, I’ve produced something, which gives me a sense of satisfaction (and, for me, that continuing feeling of creative productivity has proven to be critical in maintaining my mental health).

It is also a time of calm and unwinding after the effort and stresses of production. Responsibilities are lifted for a short while, at least for this project. I can take a walk, chat with friends, or maybe just work on something else less pressing, unburdened by the need to feel “on the clock.” Now the ball is in the client’s court. All other things being equal, I like being the one waiting on other people, rather than the one who is holding everything up.

Though the pressure is off for a short spell, there’s also a tendril of nervousness that comes with it. My work is being scrutinized. Will they like it? Will I have to make major changes or start over? Did I end up spelling something wrong? What awaits me in that next email that comes from the client?

Overall, though, the feeling of calm accomplishment is usually a high point on the emotional roller coaster of production work.

The reason I wrote this post: I feel like this period needs a name. Maybe other sorts of freelancers (or even other mapmakers, or probably other people who just have jobs in general) have a name for it. But I don’t. Lacking one leads to long-winded explanations:

“Hey Daniel, how are you doing?”
“Great. I’m in the middle of a time-period-in-which-I-have-finished-a-draft-but-the-client-hasn’t-commented-yet!

So what do you think? What should this magical span be called, or what do people already call it?

I was going to offer a couple of my own suggestions, but I can’t think of anything that isn’t terrible, except for maybe “the Cease-fire,” but that’s a little adversarial, and I only feel like that about clients sometimes. So hopefully you folks come up with something. Or just use the comments to talk about your own feelings about delivering drafts.

Update: Marty Elmer’s suggestion is currently my favorite:

A Matter of Perspective

Today I wanted to share with you a little project of mine from a few months ago, which may best be described by the question: What happens if you take the shoreline of a lake, cut it, and unfurl it?Unfurling Slide-01

The once-closed shoreline of the lake now becomes linear, providing a new perspective on a familiar feature. Warm up your scrolling finger, because here’s what happened when I linearized Lake Michigan:

Draft 5

Click for an even larger version. Take a while to browse around.

A drive around the lake becomes a reasonably straight line. Not only that, but the map is actually continuous — the roads running off the bottom of the map are the same as those coming in at the top. It provides a unique perspective on the way people arrange themselves around the lake.

I’ll get to the how of all this in a little bit, but first the big question: after seeing my map, most people ask me, “Why did you do this?” Actually, “What’s the point?” is usually how they put it.

Why Make the Map

Like many of my projects, this can be filed under “stuff I spent a huge amount of time making, and now don’t know what to do with.” Partially I did it to see if I could; I’m pleased with the technical achievement of having figured out how to construct the map. But the end result is not just an idle novelty, like the Penrose binning was.

I made this map because I wanted to show space referenced against a natural feature, rather than figuring locations based on the cardinal directions of north/south/etc. I think it’s a very human perspective, grounded in how we relate to the lake, rather than how it looks from space. Rob Roth just wandered by while I was writing this and said that this depicted “configural knowledge,” so there’s your search term if you want to read the academic side of this sort of thing.

As the idea took shape in my mind, it reminded me of a couple of other things that I’d heard about in the past. The first is the 1849 Petition of the Ojibwe Chiefs, sometimes attributed to Kechewaishke.


You can follow the link to read more about the interpretation, but the short version is that the map depicts the relationships of various natural features in northern Wisconsin. There is spatial information here, but it’s not presented against the grid to which we are accustomed. The vast expanse of Lake Superior is compressed into the thick horizontal blue line; its true size and shape is not relevant here. Instead, the map depicts understandings of relationships, rather than physical measurements.

While working on my map, I was also reminded of this Slate article on geocentric directional systems (thanks Alasdair Rae for reminding me where to find the story!). On Bali, you might find directions described not in terms of north/south/etc., but instead as to whether one is moving clockwise or counterclockwise around the island, or moving toward or away from the nearest major mountain. It’s orientation with reference to surroundings, rather than a superimposed grid.

I’m not sure if either of these were direct inspirations, but they were certainly in the back of my mind as I got to work, as examples of other uncommon ways of thinking about and depicting space.

Besides an opportunity to play around with a fun perspective, this is also a sentimental and personal project. I’ve lived almost the whole of my life on this map, in one part or another. Lake Michigan has been a dominant feature in my personal geography, and it anchors my understanding of my homeland. If I’d been raised in another context, out in the plains or along the ocean coasts, I don’t think I would ever have thought to do something like this. But, whether or not my upbringing figured in, this map makes perfect sense to me; it feels right. It also emphasizes just how far apart my current home (Madison, WI) seems from my former home (Kalamazoo, MI) when the only path between them involves driving around a vast expanse of water.

I’m certainly not the first to do something like this. Shortly after this post went up, I learned that Nick Martinelli sketched out a hand-drawn example of this linearizing idea earlier this year, in which he straightened out Oregon. I imagine there are other cool examples out there that I don’t know about.
Martinelli Oregon

Nick Martinelli’s linear Oregon sketch. Hand-drawn maps are just the best.

There are plenty of other linear maps out there that people have brought to my attention, though the comparison might sometimes be inexact. Works like Ogilby’s maps of British roads start with features that are already linear, rather than straightening out a closed shape.

Design Thoughts

Before I get into how the map was made, I have some random observations/thoughts on the design.

  • The whole concept behind the map is weird, and the end product looks odd. Because of that, I wanted to, as far as possible, “naturalize” this distorted perspective — to make it feel like it’s perfectly normal to see the world this way. The warm colors are meant to feel more organic, and also to feel like something you’d see in any run-of-the-mill map. It’s not drawing attention to itself or shouting “hey look I made a weird distorted perspective!” Attractive, but understated.
  • The brownish tones are a population density raster. I included it partly as a matter of visual interest, so things wouldn’t be so flat. I considered putting city boundaries instead, but showing the noisy distribution of humans in this way, rather than with hard political boundaries, has a more natural feeling to it. Again, trying to make the map feel more organic.
  • I didn’t include a legend for the density. I feel like readers can figure out quickly that “browner = more people,” especially with the city labels being on there. I didn’t like the distraction of adding a legend, especially since the actual density numbers are irrelevant. I have a (possibly annoying) fondness for not really putting much explanation on my maps.
  • I decided to go with Gill Sans for the city and highway labels, which I’ve not used much before. It feels a little older (because it is), and it’s a classic. Again, I think it fits the idea of an attractive, understated aesthetic. Using a typeface like this sells the idea that this is a normal and natural way to see the world.
  • I used Sorts Mill Goudy for the lakes, because a website told me to. Here’s a real typography geek secret: we often look online to see what other people think, and then if their opinion makes sense we might adopt it. I did a search for “what serif to pair with Gill Sans,” and went with a suggestion that made sense to me.
  • The map is formatted to print at the fabulously inconvenient size of 10″ × 60″
  • One of the most difficult decisions was how to rotate the map. I actually started with a horizontal orientation, with the land at the bottom and the water at the top. To me, that seemed to embody the feeling of sitting on land and looking out toward the water. However, then Evan Applegate pointed out to me that if I ever put it on a website, people would dislike having to scroll horizontally. So, I rotated it 90º and relabeled it. This is me surrendering to the tyranny of modern digital devices.
  • Where to center it was another concern. The map above starts with Chicago at the top, a familiar location. In print, I’d probably put Chicago near the center or just above. If someone’s looking over the whole object at once in person, I don’t think they’ll start at the top the way they’re forced to when they see the digital version in this post. And I want Chicago to be that familiar anchor that people see early on to begin to understand what’s happening.
  • All this boils down to: digital displays really limit your ability to appreciate this map, and it would be great if we could all go back to printed maps.

How I Made the Map

So, let’s get into how I made this thing. I began by compiling data in ArcMap, pretty much as though I were making an ordinary map: roads, population density, state boundaries, and hydrography.



US National Map Small Scale, and the Atlas of Canada 1 million National Frameworks. I selected just the roads that were Interstates, US Routes, or those that were part of the Trans-Canada Highway or Canadian National Highway System.


Same sources. There are a lot of rivers in these data sets, and I so I thinned the network heavily. I actually took just the rivers that were in the Natural Earth 10 million scale data. The Natural Earth linework itself was too coarse, but it was a guide as to which of the more detailed 1 million vectors to use from my other sources.


As above, plus two more detailed sets — TIGER and Canadian Census Boundary Files. The more detailed sets had three purposes:

  1. I wanted more detail on the Lake Michigan coastline, since a lot of attention would be drawn there.
  2. The more generalized Lake Michigan polygon didn’t always line up well with the population data, so using the more detailed coastline helped fix gaps. As I went along, I manually generalized the linework here and there where it was too detailed to look good.
  3. The coarser, US 1 million scale data were fine for most smaller lakes, but sometimes the lakes were weirdly in the wrong place, so I occasionally had to use the TIGER data to manually correct the position of the more generalized data. I also eliminated all lakes under 0.5 sqmi, just to keep things clean and simple.


US National Map Small Scale.

Population Density

For the US, I used Census data. Rather than join the Census data to shapefiles myself, I found some handy ones that already had selected demographics already in the attribute table. For Canada I grabbed some data from Statistics Canada.

As said, I wanted the population density layer to add some visual interest. I wanted it to be fairly finely textured, so I initially used Census blocks, the smallest geographic unit available. But there were way too many of them, and the files made Illustrator unhappy. But if I stepped up to larger units, like block groups or tracts, there was too little detail in rural areas. So, I used a mix: blocks in rural areas, and tracts in urban areas. I used the Census shapefiles for urbanized areas to make the distinction. Doing this drastically reduced my polygon count while keeping all the detail I wanted in rural areas. In urban areas, the downgrade in resolution was unnoticeable at my map scale.


I used my mixed census shapefile to make a sort-of-unclassed choropleth of population density. In reality, it’s got 50 classes, so it’s close enough to unclassed that you can’t tell (since I’m not aware of how to do a properly unclassed one in ArcMap). However, the classing is not linear. Instead, it shows the square root of the population density. I’m a big fan of giving unclassed maps a non-linear treatment (and drafted an unfinished blog post about it years ago). In this case, using the square root reveals small changes in low-density areas, details that would otherwise be washed out. In a linear choropleth, there would be major cities visible and not much else; the density of Chicago or Milwaukee would skew the color scheme so much that small villages and hamlets in the country would be unseen. Taking the square root of the data de-emphasizes the densest areas, helping fix this. Actually, I’ll tell you a secret: I showed the 2.25th root of density, not the square root, because I’m way too detail obsessive. That extra 0.25th of a root made a tiny difference that I liked and which you will never notice.
Nonlinear Density

Notice all the towns that appear once freed from the tyranny of linear color schemes.

Finally, before applying the classing, I clipped the top 1% of the data. So, the very densest places all look the same, and the colors only change when you’re outside of that, in someplace that isn’t in the top 1%. This was, once again, to keep the very densest outliers from skewing everything.

The Distortion

Now comes the fun part: linearizing the map. First, I drew a simple polygon around the lakeshore in Illustrator. Then I bisected each angle in that polygon to create skewed quadrilaterals going both inland and into the lake.


I drew the lines such that each one going inland was the same length. This created a sort of rough buffer around the lake, so that my final map would end up including all places within a certain distance of the lakeshore. For the lines going into the lake, I made them only long enough to cover peninsulae, nearby islands, and any other bits of land that fell on the other side of my original simple polygon.

To actually make the linear map, I needed to take each of these skewed quadrilaterals, un-skew them into rectangles, and then stack them up. But, how does one straighten a quadrilateral in this way?


This is where I drew upon the expertise of Christopher Alfeld, a mathematician/programmer friend, because the answer is: lots of math. I once briefly understood why we were doing what we were doing, but that knowledge has since vanished into the ether. What I am left with are the equations which he determined that I needed to use:

x = x0 t s + x1 (1 − t)s + x2 t(1 − s) + x3 (1 − t)(1 − s)
y = y0 t s + y1 (1 − t)s + y2 t(1 − s) + y3 (1 − t)(1 − s)

Here, s and t are the new coordinates of our transformed quadrilateral. After asking Wolfram Alpha to solve for those variables, we get:

Fortunately, Wolfram Alpha outputs equations as images, so I didn’t have to re-type all this. Unfortunately, Wolfram Alpha outputs equations as images, so I did have to re-type all of this when I actually wrote the code to make the map.

To implement this, I wrote a Python script. It loads in an SVG, applies the transformation, and kicks out a new SVG. I didn’t find an existing Python SVG parser that I could immediately figure out, so the biggest part of my script was writing a simple one.

Obligatory code screenshot. I should share the real thing, but I think it needs some tweaks first.

Once all that was settled, I could make the map. All I needed to do was:

  1. Go to the untransformed map in Illustrator and select one of the skewed quadrilaterals;
  2. eliminate all the artwork outside the quad;
  3. save as an SVG;
  4. run the script;
  5. load the newly generated SVG into Illustrator;
  6. stack the unskewed quads together in a new Illustrator document; and
  7. style everything.


There were a few wrinkles along the way. I had to increase the point density of the linework (e.g., a straight line between two points becomes a straight line between 4 points), so that the shapes could warp a little more smoothly. Also the equation for whatever reason didn’t work right until I first rotated the skewed quad such that one line was horizontal.

What would have been smart is if I’d done this in JavaScript and run it as a script within Illustrator itself, rather than sending it out to an SVG and back. So, that might be the next thing.

City Labels

Finally there was the labeling of cities. I decided not to do every one of them, but only the significant ones. The problem was how to decide what is “significant.”

Labeling every city above a certain population threshold wouldn’t work, as it would miss locally-important small towns while giving attention to every single suburb of Milwaukee or Chicago. Instead, I needed a way to measure regional significance.

I ended up with a process fairly similar to the one I developed for locally-enhanced hypsometric tints. Short version: I used my population density shapefiles to make a raster. For each pixel in the raster, I checked to see how many standard deviations above or below the regional mean its density was (the region being a 25km radius circle). Now I knew whether each pixel’s density was regionally significant, or whether it was merely about as dense as its surroundings. Then, for each city, I summed its pixels to get a score for how regionally relevant it was. Then I labeled the highest-scoring cities.

It worked out pretty nicely. A Chicago suburb of 10,000 people doesn’t make the cut, but a small town of 1,000 in the middle of nowhere gets labeled, as it’s a landmark for the area. It’s very much like the scale ranks that occur in Natural Earth’s populated place shapefile, but I have no idea how they did theirs.

Next Steps

As said, I’m not sure what to do with this right now, other than stick it here. A couple of people have asked about getting prints, and so I’ve also put it up on Zazzle, where you can find a vertical or a horizontal version.

I’ll be presenting on this work, and showing off a print version, at NACIS2015. I’d like to do all the Great Lakes, and put them onto one impractically sized poster that maybe someone would want. Such maps might also make nice wallpaper borders, I suppose, if I went back to a horizontal orientation. Mostly, though, I just want to get them out there and in front of people’s eyes, and provide them with a new perspective on a familiar feature.

If you made all the way to the end of this post, I have a bonus for you: I also did Lake Superior!

Draft 1

Click for an even larger version. Take a while to browse around.

On Faith

My work is driven by small miracles.

People pay me to come up with ideas, but I am not certain where they come from. I meet with a client, and they tell me what sort of information they want to convey, and why. They hand me some data, and maybe a sketch or two. I promise that I will meet their needs by a certain deadline, and then I leave the room, having no clue what the end product will look like.

This is the life of a creative professional. I know how to make perfunctory maps, but to make something good and unique? That requires magic. I cannot will it into existence, but must wait patiently for it to occur. First you’re working on something that looks terrible, and then the Idea appears that fixes it and makes it good. Or maybe it doesn’t, and things keep looking terrible for a little longer. Each of us has our own rituals and practices to make the Idea more likely to come around — meditating, looking at other people’s art, talking to friends, or, if you’re Don Draper: napping, drinking, and movies.

It can be frustrating, and it can be worrying, to sit there staring at the blank canvas and hoping that you will not fail to deliver on your promises. To be without a set of clear, well-defined steps to reach excellence. To keep moving forward requires faith. To understand that you have the tools to succeed, and that the correct combination of synapses will eventually fire in your brain sooner rather than later.

Not all of mapmaking is so dependent on magic, to be sure. There are many tasks which are rote and uncreative, and there are clients who simply want you to produce something you’ve already done before (or to follow their specific instructions). In truth, that’s most of my work. But occasionally someone says, “it should look scholarly and modern,” or “these data sets should be shown in an entirely new, cutting-edge way,” and those aren’t really instructions. Now you must have faith that the Idea will come to you.

Having faith is about more than just the Idea, too. Even if someone comes to you with a specific style example to copy, it may be something you’ve never done before. I’ve never gone from A to B before, but now I need to decide if I want to commit to taking an unfamiliar path, by a certain deadline, and do a good job along the way.

This is a business, like many others, that requires faith in oneself. That you will figure it out; that you will eventually stumble upon that Idea; that you will figure out software or the data or whatever is necessary to deliver on what someone asks of you. That you have it in yourself to succeed at ill-defined tasks.

This sort of faith and confidence has long been a weak point for me, but becoming an accidental freelancer has been very good for teaching me that I can figure things out. My work is a constant source of self-surprise. Almost nothing I’ve made is something I would have imagined myself as making just a short time before. But as time goes on I have learned to have faith in my ability to get through each new task, and to be open to the right moment of inspiration.

It helps, as well, to be surrounded by a network of helpful and friendly colleagues, without whose aid I would much worse off. They show me the way from A to B when I am not certain how. And sometimes, when I am waiting for the Idea to come, it visits them instead, and they share it with me. Having such support helps give me the confidence to look clients in the eye and say, “Yes, I can do that for you.” Even if I don’t know how just yet.

As Bradbury said: Go to the edge of the cliff and jump off. Build your wings on the way down.

The Most Important Thing

Longtime readers will know that I have had the privilege of serving as one of the two Editors of the Atlas of Design for the last few years. Now, however, my term has come to an end, and I am stepping away from the project. It’s been an amazing experience, and I hope you will indulge me in a little reflection.

My involvement is entirely due to the patience and generosity of Tim Wallace. In 2010, he was serving as the student board member of NACIS, and was tasked that October with coming up with a plan for NACIS to publish an anthology of great maps. Nick Springer had previously created two editions of Cartography Design Annual, and NACIS was hoping to pick up where Nick had left off.

Tim and I were both students in the University of Wisconsin Cartography Lab, and as he began to work on the book idea, he would share his ideas and updates with me. I was very interested in the project, being in love with printed materials and also having previously had work featured in Cartography Design Annual #2. After a short while, I started giving Tim a lot of unsolicited suggestions, and not-so-subtly hinting that I would like to be a part of the action. He handled this with graceful patience, and in February of 2011 he offered to let me help.

Now officially aboard, I kept spamming him with ideas and involving myself in the decision-making. When I get excited about something, I can turn into a wall of mental force. After the first month, Tim, quite rightly, reminded me that this project was his charge, as I was starting to get a bit carried away. Soon, though, we “hit a rhythm” (as Tim puts it) and formed a partnership as co-Editors.

My excitement barrage notwithstanding, we had a great working relationship. When we joined forces, on this project and others, Tim and I were more than the sum of our parts. We inspired each other to new heights as we passed ideas back and forth. Being able to work on something you love with people you love is basically the best experience life has to offer.

By April 2011 we’d put together a vision for a new publication, a beautiful anthology of contemporary cartography intended to inspire professionals and educate laypersons. A global honor in cartography. Something important. We had big dreams and were excited to see them through. A high percentage of our meetings consisted of statements like “This is going to be amazing!” “So excited!” “Wow!” I’m sure we used a lot of exclamation points.

We plugged away on and off for the next 12 months, figuring out what it took to actually make a book. We ran a competition, received 150+ entries, recruited judges, made final selections, got quotes from presses, worked out page sizes, bindings, and materials, and many other things. It was an amazing experience and a lot of fun to learn about the publication process.

AoD Workday

Hard at work.

The NACIS Board of Directors had a rough idea of how much this book would cost, which was based on the idea that it would be printed and bound inexpensively. But Tim and I felt like the materials of the book needed to be high-quality, to reflect (and respect) the quality of the contents. So, we aimed higher, and in April 2012 we asked the Board for about three times as much money as they had been contemplating spending. I remember being very nervous prior to the meeting, and Tim and I brainstorming in the days beforehand how to convince them to fund our dream. It was a sort of shift. Initially, this was something the Board had asked Tim to do. Now it was something we were asking the Board to do.

I called in to the meeting via Skype. I laid out our vision, and then told them how much we wanted to spend. I remember hearing at least one person gasp. Tim and I had figured some scenarios in which we could do things cheaper, though at lower quality, and we figured on maybe getting funding to do one of those, but we had decided to start with a go-for-broke request for everything we wanted. But, to my surprise, everyone completely supported our higher-cost vision. The treasurer started shuffling money around between parts of the budget to accommodate our unexpectedly-high request. What I remember most is that people started offering money out of their own project budgets, and I started getting teary-eyed as people kept stepping up to help make our dream happen. I’m not sure, but that may be the only instance of tears of joy/relief at a NACIS Board meeting. I remember reporting back to Tim that we had won the lottery. The Board took a big risk on us, layout out a very large fraction of its money to print a book that no one was certain would sell.

Then it was back to work, to make the book in time for NACIS 2012, that October. We designed the layout, got copyright releases from contributors, did copyediting, figured out fulfillment channels, wrote an opening essay, and plenty more. I got to visit the press while the book was being printed, which was really cool. Finally, the book was finished, bound, and ready to send to us. Actual quotes from emails between me and Tim:

“Yessssssssss!!!!! Aaaah! I love paper!!!!”
“OH MY GOSH, FELLAS!” [“Fellas” being a nickname, of obscure derivation, we had for each other]
“Aaaaaaaaaasaaaaaaaaaaaaaaah! Goooooooooooooooooooooooooo! Yesssssssssssssssssssssssssss!”
“I’m freaking out!”

Editing a proof

Editing a proof

The book was done on a Friday, but the driver wouldn’t bring it to us until Monday. I was so excited that I asked the printer for a sneak preview — could they take some photos of the book for us? They did one better: someone came to my house and hand delivered a copy to me Friday night. As I told her at the time, it’s rare to have a vision for how something will turn out, and for it to in fact come out exactly that way. But that’s what it was: everything I’d hoped for. All that work and dreaming had turned into a solid reality, and to some degree it was anti-climactic now that it was in my hands, but it was also very satisfying.

The book debuted in October 2012 at the NACIS Annual Meeting and was well-received. When I got back to Madison I went about the business of taking orders, packing books up, and shipping them around the world (which was also kind of fun to learn about). Some of our customers were libraries, and eventually it got put on WorldCat.

Seeing my name, alongside my friend Tim’s, in a library catalog entry for a book we made together is one of the highlights of my life.

Editing the first two volumes of the Atlas of Design is probably the most important thing I’ve done in my career thus far. It’s been an immensely satisfying and, at times, deeply moving experience. It’s a bit difficult to move on, and to let the project live on and change under the direction of others. But I know it is left in the good hands of the current editorial team of Sam Matthews, Marty Elmer, and Ginny Mason. I look forward to seeing what their dreams are.