Visualizing 36 Hours on the Alcan Highway

A man’s move to Alaska becomes his sister’s data viz pet project

When my brother Robby asked me to join him on his drive to Alaska last summer — he was moving from Del Rio, Texas, to Palmer, Alaska, for his next U.S. Air Force assignment — I’ll admit I wasn’t pumped about the idea of using five vacation days to sit in a car. I was, however, delighted by the possibility of having extended one-on-one bonding time with him. (As a father of three youngsters, he isn’t exactly swimming in free time, and we’ve lived about 2,000 miles apart for over five years.)

I was also thrilled because it seemed like an experience ripe for data viz — we’d surely have ample time to collect data on the drive, and there would be a lot of changing variables: geographical position, temperature, and country, to name a few. I’d been meaning to do an analog, collect-your-own-data visualization project for a while, and finally I had the perfect opportunity.

My brother Robby drove from Texas to Alaska in August 2019

The Data Collection Process

Robby began his drive with his family in Texas. After dropping his wife and daughters off at the Denver airport, he picked me up at Calgary International Airport in Alberta. We spent a day hiking in Banff and officially began our drive from there.

We collected data hourly. Each hour, starting at 12:45 p.m. on August 21 after an indulgent brunch at Toulooloo’s in Banff, we recorded both the temperature and the number of cars we could see on the road in either direction. We also took a picture of the environs and then flipped the camera and took a selfie. We did this every hour, even at 1:45 a.m. when it was pitch black outside and I had to wake up for the data collection.

My brother’s high-tech truck automatically switched between the metric system and United States customary units when we crossed over from Canada into Alaska; we recorded the temperature as such (i.e. Celsius then Fahrenheit). When recording the hour, we started using the 12-hour format but switched to the 24-hour format (a.k.a military time) on the second day of the drive.

In addition to the hourly data collection, we also decided to record any wildlife we saw along the drive.

We did a decent job collecting our data “on time,” though there were a few hours that we were a bit late.

Inspiration Strikes

When we began the drive, I had no idea what sort of visualization I wanted to build. I’ll collect the data now, I thought, and figure out how to visualize it later. But at some point in the drive (must’ve been somewhere between finishing Born a Crime and listening to our third episode of Stuff You Should Know), a visual form started taking shape in my mind. I was describing Charles Minard’s famous “Figurative Map of the Successive Losses in men of the French Army in the Russian campaign 1812–1813” to my brother when I thought Why not emulate Charles Minard’s famous piece? Mimicking the greats is usually a good way to learn, after all.

The data I wanted to visualize was much less grave than Minard’s, but it presented a similar problem: how to encode multivariate data with a predominantly geographical component.

Multivariate data

In the visualization, there are nine variables encoded:

• The number of other cars on the road at the hour of data collection: indicated by the thin line in the bottom half of the viz
• Temperature at the hour of data collection: indicated by the thick “fuzzy” line on the bottom half of the viz
• Driver at the hour of data collection: indicated by the color (green for me, blue for Rob) of the lines
• Time: day is labelled along the lines (photos also shift downward / upward on each day change), and hour is along the x-axis of the bottom half (I decided to label the time using military format to fit the theme)
• Time zone: indicated by vertical dotted lines in the bottom half of the viz
• Geographical location: the route of the drive is mapped along the top half of the viz. I rotated the traditional ‘north is up’ orientation such that the generally northwestward bearing of the route would proceed left to right on the page.
• What it looked like outside at the hour at the hour of data collection: bottom photo
• What we looked like at the hour at the hour of data collection: top photo
• Country: indicated by flags at the border

The Data We Didn’t Collect

Because I did not have a visual format in mind when we began collecting data, there are some things I wish we had done differently. In retrospect, I wish we had recorded:

• Elevation: given that the Alcan highway took us in and out of the Rockies, our drive was topographically interesting.
• Exact time of stops: we did not record exactly when we stopped for food and sleep so I had to estimate that.
• Exact time of driver switches: we also did not record exactly when we switched drivers, so I decided to change the color of the line halfway between the hour marks.

Making the Map

I constructed the map over several months, working on it here and there when I had free time. I’m not sure how many hours I spent in total, but I’d like to think it was fewer than the number of hours visualized.

The materials I used to create my visualization were: pencils, pens, markers, colored pencils, ruler and yard stick, glue, scissors, large paper, and 1"x1" photos

Throughout the process, I frequently used the internet to do research on various topics, such as map projections and Canadian time zones. I also referred to Minard’s double-page visualization in Tufte’s Beautiful Evidence for design inspiration.

And this was my process:

As always, I started with a few sketches. Then I created a list of encodings I wanted to use in the visualization (e.g. driver: color, temperature: height of line, etc.).

Next up, drawing the route itself:

I researched different types of map projections. After much deliberation over what type of projection to use (see the “What I learned” section below for more details), I pulled the directions from Banff, AB to Palmer, AK from Google Maps and printed out the route.

I overlaid a paper onto the printed map and rotated it such that the origin and destination would be at the same height (i.e. so the route would run generally left to right) and then traced. I knew that my final visualization had to be at least 36” wide if I wanted to include 1"x1" photos — any smaller and they wouldn’t really be visible — so next I needed to scale up the map. I drew a 1cm x 1cm grid over over the traced route, and also a larger (2in x 2in) corresponding grid onto over-sized paper. With the assistance of the grid system, I completed a freehand transfer of the route to the larger version. I then labelled major* cities along the way.
*generous use of the word major, given that we were in the middle of nowhere for most of the drive.

Mapped route with overlaid grid

Next up, plotting the data:

I marked the hours along the bottom of the paper; ultimately I’d decided that I wanted the line graph portion of the viz to be the same width as the route. I also drew axes and lines for both temperature and number of other cars (see “What I learned” section below for more details).

Next, I colored the lines by who was driving. By this point, to my delight, a color theme was beginning to emerge — Robby and I had on blue and green shirts for much of the drive, and a lot of the photos involved blues and greens and the aurora borealis we saw was green — so I decide to go with blue for Robby and green for me.

Late in the evening on Aug. 21, Robby and I saw Aurora Borealis.

I glued the photos to the map at the hour marks (using the timestamp of the photo to ensure accuracy) and added additional photos along the route (e.g. pictures of Robby taking a dip in a mountain-fed stream). These photos were taken when we saw or did something cool, rather than on the hour.

I also added in milestones, like where we saw the aurora borealis and where we crossed into the US. Finally, and possibly the most time consuming part of it all, I erased all my pencil marks.

Robby and I collaborated to create a title for the visualization: “Illustrated Depiction of the hourly progress of Captain Robert J. Mello and his navigator K. A. Mello on their journey northwestward to joint base Elemendorf-Richardson”

What I learned in the process:

• The dual axes controversy — Going into this project, I knew that there was controversy around using dual axes. After doing some research, I found that there are certainly cases to be made against using dual axes. In the end, however, I decided to use dual axes because (1) I did not intend the reader to draw conclusions about the relationship between temperature and number of cars of the road (the components of the line chart with dual axes) and (2) I wanted to use the space underneath the mapped route as efficiently as possible. I tried to make the difference between the temperature and cars line very obvious by directly underlining the labels in the thin or thick line accordingly.

Dual axes (temperature on left and # cars on right)

• Map projections — Before this project, the only thing I knew about map projections was that I didn’t know enough not to research them before embarking on this viz. So research I did. In the end, I ended up sticking with the Mercator projection used by Google Maps because 1) I was mostly concerned with preserving the shape of the route and 2) it wouldn’t require doing my own custom route mapping using the lat-long of the photos we took. So…laziness I guess?

Unique Challenges

Visualizing a 3,000-plus mile drive came with a set of interesting challenges, such as:

• We crossed time zones twice! I decided to indicate this with vertical dotted lines & corresponding timezone labels, on the line chart.
• Two different temperature scales The car switched from the metric system (degrees Celsius) to United States customary units (degrees Fahrenheit) automatically when we entered the US (technology these days…). I decided to account for this by labeling the left Y-axis with both Celsius and Fahrenheit.
• To left-to-right or not to left-to-right The advantage of rotating the compass so the route proceeded left to right was that I was able to mimic Minard’s design more closely. The drawback is that it is a little deceptive. Having the map directly above the line chart makes it seem as if the map shares the x-axis with the line chart. In other words, it makes it seem as if the point on the route that intersects with an hour marking is where we were at that hour in the drive. For some hours, this is correct. For others, it is not. I played with the idea of using vertical lines to connect hours with points on a map (for the hours where I can pinpoint) as Minard did, but ultimately decided they may be too distracting.

Compass rose

Conclusions

The driving force behind this visualization was artistic curiosity rather than practical necessity. I wanted to create something cool and learn something along the way. Though I wasn’t hoping to draw major conclusions about the road trip or make any data-informed decisions, the visualization and the creation thereof did help me realize a few things:

One big takeaway from the visualization is that Robby drove way more than I did. (I hope I was a good copilot!) Another big takeaway is I am not photogenic at 1:45 a.m. in the morning. Next time I’m having my photo taken for a major award, I’ll insist it be during the day.

Finally, I learned that anything can be a data viz project. It all starts with a pencil and somewhere to log your data!

I certainly did learn a lot by immersing myself in this analog data viz project; if this project seemed at all fun or interesting to you, I recommend you find your own excuse to start collecting data!

Shoutout to my helper kitty, Bella Luna, who provided moral support throughout the artistic endeavor; to Robby for providing the gas money and being willing to record hourly data; and to the DVS community for weighing in on questions that came up along the way.

Thanks to Alyssa Bell for help editing this article!