Illumination
Figure 1 Up or down? Bright Angel Canyon, Arizona appears as a proper canyon when illuminated from the northwest (left) and like a mountain range when illuminated from the southeast (right). Source: Both images were digitally generated from 10 meter DEMs
Upper left illumination: Of cartography's many conventions, few are as rigorously followed as the use of upper left illumination (northwest on a north-oriented map) to portray shaded relief. When an illumination source other than the upper left is used the resulting shaded relief tends not to look right. In extreme cases relief inversion occurs, a perceptual problem whereby mountains appear as valleys and vice versa. Relief inversion is most problematic when lower right (southeast on a north-oriented map) illumination is used, when the light source is perpendicular to linear landforms, and when linear landforms are situated amidst flat surroundings. Canyons are especially susceptible to relief inversion, perhaps because most readers are accustomed to visualizing positive landforms (i.e. mountains) on map shaded relief rather than the inverted world of canyon country (figure 1).

The use of upper left illumination for cartographic shaded relief is a curious standard for depicting topography in a supposedly realistic manner. In the natural world the Sun almost never shines from northwest except during midsummer in the very highest latitudes, where few people live to witness the event. Instead, our preference for upper left illumination is largely due to ergonomical considerations and cultural preference. Because most people are right handed, when we write or draw we tend to place desk lighting above and to the left so as not to cast shadows over our work. Moreover, our work generally progresses from upper left to lower right, minimizing the possibility for smearing.

When depicting shaded relief and other 3D phenomena on a flat sheet of paper, we naturally take a cue from the lighting within our work place environment and apply shadows to the lower right sides of slopes. Centuries of depicting topography in this manner has ingrained this cartographic convention. This preference for upper left illumination/lower right shadowing pervades all of western graphical culture and can even be seen in the graphic user interface (GUI) of computers—note the upper left highlights and lower right shadows that are used to depict 3D icons and palettes. The Roman alphabet also shows this preference. It is not coincidental that we write from the upper left to lower right and that many of the lower-case letters contain their identifying characteristics on the lower right side (for example: g, h, q, b, and y).

Rotating illumination: Conventional upper left illumination works well for depicting most shaded relief but not in all cases. For example, a northwest trending ridge facing into the illumination source would appear flat because of inadequate highlight and shadow definition compared to similar ridges with different orientations. The accepted solution to this problem is to rotate the global illumination source from the northwest—30 degrees in either direction is the rule of thumb first espoused by Eduard Imhof and subsequently adopted by all others—depending on which way the major landforms trend. Upper left illumination is a resilient convention that can withstand moderate deviations from a strict upper left azimuth (315 degrees) without jeopardizing the overall legibility of a shaded relief.

Regardless of the care given to selecting a global illumination direction, shaded relief of large complex landscapes often contain areas of subordinate topography that remain poorly defined. The problem can be alleviated by varying the illumination direction for localized problem areas within the shaded relief—the 30 degrees of rotation rule is also applicable to internal adjustments to the light source. Although making localized adjustments to the illumination direction introduces an element of geographic inconsistency to shaded relief, the sacrifice is made for the greater good of improving overall graphic clarity. When skillfully applied, localized adjustments to the illumination direction go unnoticed by all but the most discerning map readers, because the inherent complexity of topography makes small variations in shadow patterns difficult to detect. Modifying the illumination direction is not necessarily easy. For instance, when a ridge trends directly toward the global illumination source, the choice must be made as to which aspect of the slope will receive illuminated and shadow tones. It is generally advisable to maintain consistency with the illumination and shadow patterns found on nearby slopes unaffected by the localized illumination modifications. All factors being equal, my preference is to apply shading to right-facing aspects, because I am right handed. You should avoid the confusion of placing shadows or illumination on facing adjacent slopes (figure 2, left).

In cases where the illumination shines perpendicular to linear landforms, changing the light source to a more oblique direction enhances the shadow and illumination detail on subordinate slopes, better explaining the landscape.

Eroded volcanoes and other conical landforms that bow towards the light source pose difficulties when applying illumination and shadows. At some point when applying alternating illumination and shadows to the radial ridges, a "breaking point" is reached, where shadows or illumination face one another. My workaround solution is to select a minor ridge as the breaking point to receive the full force of illumination, disguising the problem and sacrificing the ridge within a blown-out highlight.

Figure 2 Poor (left) and good (right) examples of locally adjusted illumination for Grand Teton National Park, Idaho/Wyoming. The highlight box on the left image shows illumination on facing adjacent slopes; the right image properly alternates illuminated and shadowed slopes.
Sources: The right image was created by Bill VonAllmen (retired) U.S. National Park Service. The left image, also owned by the NPS, was created by an unknown artist 32 years ago.
Illumination elevation: Most computer programs used for shaded relief production assume a default illumination source 45 degrees above the horizon, which generally works well. For areas of extremely high local relief, such as high alpine mountains, a slightly higher elevation is advantageous for illuminating detail in the shadowed slopes. However, take care not to raise the illumination elevation too high or the landscape will become flattened, yielding shaded relief reminiscent of the hachured slope shading technique used during the nineteenth century and no longer considered to be a viable relief presenation method (figure 3, right). By contrast, a shallower illumination elevation works better for landscapes with low relative relief and at small scales where shadowing needs to be emphasized at the expense of detail.

In the process of manual relief shading the elevation of illumination is rarely considered, at least on a conscious level. The tones applied to slopes are largely governed by aerial perspective, a purely graphical consideration discussed in next section.

Figure 3 Illumination elevation governs the amount of shadowing or detail that appears on shaded relief. The computer-generated scenes show the upper Yosemite Valley, California. Source: 30 meter USGS DEMs.
Aerial perspective
Figure 4 Aerial perspective: A pyramid shown in side view (left) and from above (middle and right). Exaggerated aerial perspective has been applied to the middle pyramid. The right pyramid uses constant shading.
The aerial perspective effect is an elegant graphical technique that is used for making two-dimensional shaded relief appear more three-dimensional. The concept was developed by Eduard Imhof, the late Swiss authority on traditional relief presentation, and is based on his real world observations of landscapes. The idea is very simple. When we look toward the horizon from the top of a high mountain, landscape features further away appear fainter than those in the foreground because of atmospheric haze. This creates a sense of depth. Imhof applied this natural haze, which he called the aerial perspective effect, to the vertical (z) dimension on shaded relief maps viewed from directly above. With the aerial perspective effect applied, highlands, which are theoretically closest to the reader, are depicted with more contrast than similar landforms in the lowlands further away (figure 4, middle).

Applied to shaded relief, the aerial perspective effect conveys to the map reader subconscious cues about relative elevation, imparts a sense of three three dimensionality, and deters relief inversion. For instance, the topsy-turvy shaded relief of Bright Angel Canyon (figure 1) could be remedied with a strong application of the aerial perspective. Although the aerial perspective effect has been a cornerstone of manual relief shading for decades, none of today's software applications used for shaded relief production takes it into account (figure 5, right). For example, on a digital shaded relief a 20-degree slope would receive the same amount of shadow or illumination regardless of whether it was at high or low elevation. By contrast, with the aerial perspective effect applied, a 20-degree slope would receive either darker or lighter tones depending on its relative elevation.

Figure 5 The middle image shows manually drawn shaded relief that uses the aerial perspective effect. It was interpreted from the contour map show at left. The right image is a digitally generated shaded relief of the same area without aerial perspective. The images in figure 5 correspond to those in figure 4.
Shades of gray
Figure 6 Ta‘u, American Samoa with a balanced range of tonal values (left) and without midtones (right).
Source: Relief by Dr. Michael Wood, University of Aberdeen, Scotland. Used with permission from the University of Hawaii Department of Geography.
Shaded relief is the representation of topography on maps using modulated light and shadows to simulate a three-dimensional landscape viewed from above. Being comprised solely of continuous-tone gray smudges, shaded relief is art in its most basic form. For reasons of clarity, other graphical elements (color, textures, lines, etc.) are usually, but not always, absent on shaded relief so as not to interfere later with surprinting map information.

To adequately depict subtle topography the tonal range on a shaded relief original should be large, even if the piece is to be ghosted for final printing. Tones ranging from 5 to 95% black are the often stated ideal for grayscale images in the graphics industry. This is also an appropriate target for shaded relief presentation, especially if the relief is destined to be printed with light colored inks. However, utmost care must always be taken to show subtle detail within the darkest shadowed slopes—these areas have a tendency to plug up in printing. A wide tonal range does no harm because it is relatively easy and safe to subtract value from dark images. Conversely, increasing the value of light images is not advisable because data must be interpolated where little or nothing exists.

Every shaded relief should contain a harmonious balance of light, midtone, and shadow values. When one of these classes of value dominates or is diminished, the effectiveness of the entire shaded relief suffers. Depending on the topographic characteristics of any given landscape the values comprising a shaded relief must be adjusted accordingly. For example, midtones are essential for graphically bridging areas between illuminated and shadowed slopes and provide the lowland base tone from which all other topography projects upward. But when midtones are inadequately portrayed or absent, the entire shaded relief lacks form and appears spindly (figure 6, right). In another common example, shaded relief that is dominated by large areas of shadowed slopes, such as those typically found on large-scale mountain maps, should employ less dense tones in the shadows so as to avoid a heavy ponderous appearance. Small-scale shaded relief is generally more tolerant of darker shadows and brighter highlights than those at large scale. When portraying illuminated slopes highlights (areas without any value—i.e white) should be used sparingly only for the most exposed slopes at the highest elevations. Overuse of highlights blows out the topography, denying the reader vital information. So do overly dense shadows that obscure detail.

Shaded relief is often displayed too heavily relative to other map information, especially with manual shaded relief. After devoting huge amounts of time and effort devoted to drawing manual shaded relief, the cartographer invariably feels compelled to present his or her magnum opus prominently. From hard learned experience I advise holding back. No map has ever been ruined by a relief that prints too light. The same claim cannot be made about relief that prints too dark.

Details, details...
Figure 7 Small-scale shaded reliefs centered on Lander, Wyoming portrayed with more (left) and less (right) generalization (and very different artistic styles).
Sources: Left relief by Herwig Schutzler, MapQuest.com (retired). Right relief by Richard Edes Harrison for the "National Atlas of the United States of America."
Please forgive the diatribe that follows, but I must address the most egregious sin of digital shaded relief: excess detail. It is not uncommon to see published maps, created by otherwise reputable organizations and individuals, containing digitally-generated shaded relief so dense with detail that it appears as an illegible texture or, maybe, Einstein's brain. While many of these offenders go to great lengths to carefully thin vector linework to achieve the appropriate level of generalization for the map scale being used, the same principle is not extended to shaded relief. To be fair to my lapsed colleagues, there are practical reasons for this dichotomy. Most cartographers have less familiarity manipulating raster images than vector objects—after all, our profession is linework centric (virtually every new map, regardless of its final appearance, begins life as lines). Nevertheless, the procedure for generalizing digital shaded relief is relatively simple: 1) Downsample the DEM before rendering a shaded relief, 2) Apply a modest amount of median filter in Photoshop to the shaded relief after it has been rendered. The median filter is superior to the various blur filters in Photoshop for generalizing shaded relief, preserving areas of maximum contrast, such as ridges and canyon bottoms, while at the same time removing small irregularities and noise on intervening slopes. It works like magic—so go forth and generalize.

The excessive detail found on digital shaded relief may be partially explained as an over reaction to the high level of generalization used on most manual relief. Every microform depicted on a manual shaded relief exacts an expensive toll in time, eye strain, and tedium (conditions that still, unfortunately, plague automated cartography, as anyone who has built a DEM from scratch can attest). Considering the trying circumstances under which it is produced, traditional relief tends to portray only as much detail as is needed to satisfy a map's intended purpose. For example, a shaded relief intended for use as a backdrop on a thematic map would be more generalized than a relief showing physiographic regions (figure 7). Because maps generally contain more surprinting information in the inhabited lowlands, the shaded relief shown in these areas can be more generalized than in the highlands, and this has the added benefit of bolstering the aerial perspective effect.

My last word about generalization echoes the advice given about printing density: less is often more.

A matter of style
Figure 8 Tahaa, French Polynesia (left), my first attempt at relief shading in 1980. It depicts drainages and ridges with too much linearity—a typical characteristic of student projects. I drew Pohnpei, Federated States of Micronesia (right) several years later. It more realistically depicts the topography as faceted slopes. Both islands have similar topography.
Source: Pohnpei relief used with the permission of Manoa Mapworks, Honolulu, Hawaii.
With the advent of automated production, shaded relief has undergone a stylistic transformation. One digitally-generated shaded relief looks much like another, as if ordered from an "ACME" catalog, regardless of the software application used for its creation. Shaded relief has become one of the most homogeneous design elements on maps. This is a dramatic departure from the halcyon days of traditional production, when every shaded relief looked distinctively different depending on the skill, experience, and drawing style of the artist who produced it.

In many cases the style of a particular relief artist has defined the look of entire mapping organizations: Eduard Imhof cast his long shadow over the entire Swiss topographic map series—and well beyond, Tibor Toth's open comfortable style is synonymous with National Geographic Maps, and, Hal Shelton created richly textured pieces for the Jeppeson Map Company (now defunct). Moreover, these talented pioneers of twentieth-century relief shading also created fine art pieces, natural science illustrations, and panoramas. Behind the Map Library reference desk at the Library of Congress is a stunning large-format landscape of the Grand Canyon painted by Hal Shelton, and Imhof's pieces are displayed at Alpine Museum in Berne, Switzerland. In some cases protégés were cultivated to carry on the shading styles of their organizations after the master retired. For example, the work of John Bonner for National Geographic Maps is virtually indistinguishable from that of Tibor Toth (now retired, working as an NG contractor), resulting from of years of side-by-side work. And at the Swiss Bureau of Topography, Paul Ehrlich faithfully produces his pieces to match Eduard Imhof's famous style.

For students undertaking relief shading for the first time, developing a style is a secondary or tertiary concern after the basic mechanics have been mastered. Developing a style comes naturally with practice. It took me a couple of years. My greatest hurdle was visualizing topography not as a skein of linear ridges and drainages, but as an areal phenomenon comprised of a quilt-like mosaic of faceted slopes (figure 8). I have a preference for relief with a clean soft appearance, like the creases in freshly laundered sheets (perhaps as a result of my upbringing in the weather-worn Adirondack Mountains), while in Europe the stylistic preference is for sharper contrast, perhaps motivated by the angular peaks of the Alps. We seem predisposed to visualize topography through an environmentally deterministic lens. Such preconceptions are then filtered through our personal drawing style before the shaded relief is finally put to paper. This is analogous to the illustrations by some early European explorers who portrayed native peoples similarly to themselves. The challenge for cartographers is to present shaded relief accurately, with natural realism, and a dash of artistic style. For me, given my preference for subtlety, it is a constant struggle to include appropriate amounts of sharp contrast to the eroded canyons of the American southwest or the glaciated peaks of Alaska.

Although all digitally-generated shaded relief presently looks stylistically interchangeable, I foresee the day when digital technology will offer much greater choices in presentation style—perhaps in ways that have yet to be imagined. A hint at that possible future can be seen in Painter, the natural art software application sold by Corel. With the click of the "auto Van Gogh" button a photograph can be transformed into work closely approximating the style of the famous Dutch master. Can "instant Imhof" be far off?

Additional reading

Cartographic Relief Presentation by Eduard Imhof (de Gruyter, 1982) is the definitive text about shaded relief and other forms of relief presentation on maps. Although published before the advent of the digital revolution, this geography classic thoroughly covers the concepts and presentation issues that remain relevant to relief shading today, whether by manual or digital techniques.


Home
Continue: Manual enhancement