After writing an article on handcrafted data visualization a few months ago, I kept my eyes open for new examples. As they accumulated, I began to notice a few commonalities that displayed the power of graphics created by hand: distortion, precision, physicality, and art.
I’ll start with a painting (above) of Yosemite National Park. Here’s the Government Publishing Office’s description: “Looking down Yosemite Valley from west to east, an alpine panoramist depicts the Yosemite National Park’s wondrous rock forms, hanging valleys, waterfalls, lakes, and streams with El Capitain and Half Dome forming the central spectacle with Bridalveil and Yosemite Falls.”
Commissioned by the National Park Service and painted by Heinrich Berann, the map manipulates perspective and distorts elevation to provide an exaggerated sense of the landscape. Famous features stand out from the background. Highly visible roads wind through Yosemite’s steep terrain.
Despite being unrealistic, it’s an eminently understandable view of the park. Berran intentionally manipulated the view to make the geography readable, and guides a viewers eyes to the important features.
(If you like this panorama, Betsy Mason wrote about the whole series for National Geographic: Gorgeous Panoramic Paintings of National Parks Now Online.)
In addition to being a skilled landscape painter, Berran was one of the most important cartographers of the 20th Century. His Ocean Floor Map was the culmination of a series of maps that showed the structure of the sea floor for the first time. Working with Bruce Heezen and Marie Tharpe in the 1960s and 70s, Berran illustrated the mid-ocean ridges, oceanic trenches, and island chains that provided compelling evidence for the (then new & controversial) theory of plate tectonics.
The map skillfully distorts elevation and perspective to show bathymetric details that would otherwise be invisible due to differences in scale—even if they weren’t hidden under miles of ocean water.
Check out the Hawaiian Islands—each island appears perched on a pillar due to the extreme vertical exaggeration. Seamounts (underwater mountains not tall enough to become islands) cast shadows on the permanently dark abyssal plain (the flat bits of the ocean bottom). The perspective is (like Yosemite) impossible, but in this case it’s uniform over the entire map. Tom Patterson (recently retired from the National Park Service) coined the term “plan oblique relief” for this technique. He writes:
Compared to shaded relief, which assumes a theoretical viewpoint directly above the earth, 3D relief shows landforms “standing up” in partial profile. Because 3D relief closely resembles how mountains appear to humans on the surface of earth, it is easier for people to understand at a glance. High solitary mountains…actually look like the spectacular mountains that they are when viewed obliquely in 3D.
The map clearly shows the form of features on the ocean floor by rendering them in an unrealistic manner.
Another cartographic classic is this relief-shaded painting of Switzerland’s Walensee by Eduard Imhof. Here the distortion isn’t primarily in perspective (although every attempt to re-create a three-dimensional surface in two dimensions is inherently distorted), but in representation of the landscape. Imhof’s intent was not to create a perfect replica, or even a map, but “the visual experiences of the landscape painter and his artistic conception.”
The shape of the painting is precise, but the color is designed to give the sense of “silvery light on beautiful summer days”. Alpine summits are rendered in warm yellows and reds, while pastel greens and blues dominate the forested lower slopes and valley farms. (A color sequence echoed in many of the hypsometric scales developed by Imhof to show elevation.) Ridgelines are highlighted with a sharp contrast between sunlit and shadowed mountain faces (lit from the upper left—impossible at such a northern latitude).
Compare the painting with a satellite mosaic of the same area. The summer landscape—with the exception of urban areas and a few stray clouds—is entirely shades of green. Mountainous topography is flattened out by the vertical perspective, and the Walensee itself is hard to distinguish from its surroundings.
The hand-crafted image is much more readable than the “realistic” view from a satellite. Writing about the differences between aerial photographs and maps, Imhof himself said:
Within the limits of photographic resolution capabilities and under suitable conditions of good lighting, etc., aerial photographs (especially in color) are realistic instantaneous pictures of the earth’s surface, albeit only its superficial aspects. However, such pictures are often full of deceptive features and obscuring conditions. Similar things may frequently appear to be different. Important objects may not be visible, while incidental and unimportant things may stand out clearly. The topographic map, on the other hand, is a generalized image, conditioned by its scale, its purpose, conventions, and the artifice of its maker, and portraying significant and, more or less, permanent conditions and objects. Similar things appear the same. Important things are emphasized; unimportant things are suppressed. Even if the map image is made to resemble nature as closely as possible, it remains, in essence, abstract and more or less subjective, i.e. dependent on its maker.
(Cartographic Relief Presentation, pages 50–52.)
Sometimes, distortion is a matter of necessity. In Minard’s 1864 map of French wine exports, shipping volume is encoded by line width—which causes problems when the quantities involved meet local geography.
For example, France shipped a lot of wine out of Marseilles and nearby port cities (where the Rhône River provided easy access to vineyards in the French interior) through the narrow Strait of Gibraltar—Minard simply widened the strait to accommodate the data. Similarly, Minard pried open the Bay of Biscay (squashing Iberia almost beyond recognition) to allow for the tremendous volumes of wine exported from Bordeaux.
The severe distortion of coastlines works because the focus of the graphic is on the quantity of exports, not the geography. Knowing the purpose of a graphic (along with the intended audience) is a key component of creating an effective visualization.
Henry Beck’s 1933 Tube map is justifiably regarded as a landmark in information visualization, but it followed a decades-long evolution in which the geography of London became separated from the topology of the Underground.
The final stops of that evolution were a series of maps by F H Stingemore. The example above, from 1927, more-or-less accurately retains the complicated shape of the Underground’s many lines, but expands and collapse distances between them (compare to Macdonald Gill’s 1920 map). The distortion of distance allows far-flung stations to be included on a compact map, and (perhaps more importantly) provides the elbow room to print readable station labels. Surface features are entirely absent, with the exception of the sinuous Thames and its many bridges (brilliantly denoted with negative space).
The simplicity—a network of nodes linked by gentle curves—allows the map to be readable despite the relative complexity of the system.
Like any tool, distortion in visualization can be used for political or self-serving ends. One exceptional example is Juan de la Cosa’s Mappa Mundi, published in 1500.
De la Rosa—a Spaniard and member of several of Columbus’s expeditions—used the first known map the Americas to maximize Spanish territiorial claims. Drafted shortly after the Treaty of Tordesillas divided the non-Christian world into Spanish and Portuguese territories along a line “three hundred and seventy leagues west of the Cape Verde Islands” (shown in green), the map pushed Brazil far to the west of its actual location. Despite this obfuscation, Portugal ultimately colonized Brazil, which remains the only Portuguese-speaking country in South America.
On the surface, my final example is anything but hand-crafted—it’s a rendering of image data collected by a space probe, modified with the help of a computer. But it’s also the result of careful handiwork (tracing the surface featues of the comet to break the landscape into segments at different distances) and, perhaps more importantly, a purposeful manipulation of data.
This view of Comet 67P/Churyumov-Gerasimenko was collected by the Philae Lander in 2014. On the left is the view broadcast back by the lander. On the right is a version with one key addition—a simulated atmosphere. This subtle fuzziness, increasing with distance, mimics the cues we’ve learned to interpret throughout our lives. Here’s how Mattias Malmer described his creation:
I wanted to help my eyes perceive the landscape in a more familiar way, so I added some human-readable depth cues. I carefully segmented the landscape, following every edge that I could find, and based on the overlaps I created a very crude depth estimation. To simulate atmospheric scattering, I added increasing amounts of bluish tint and a loss of contrast as the landscape moves farther away from the camera.
For me it brings the landscape alive a little, and helps me make sense of it.
The common thread running between these examples is the intentionality of the distortions—when you’re building something by hand the creation of every element requires thoughtfulness, time, and effort. Each modification was done with a specific purpose in mind, which enhances the effectiveness of the visualization to communicate. Which is why we make graphics in the first place, right?
Next up: Precision.