(left) Natural Earth II. (right) A National Geographic physical map painted by Tibor Tóth.
Compared to land cover that shows what exists on the land at a specific time, potential natural vegetation is a theoretical concept. Potential-vegetation maps assume that modern human-altered environments are a superficial and temporary phenomenon when viewed through the long lens of time. For example, representing Las Vegas with warm desert beige ignores today’s urban landscape veneer. New York City in the humid East would appear as forest green. Tóth’s palette for portraying potential vegetation uses colors that readerseven city dwellers with little day-to-day contact with the natural worldintuitively associate with the natural environment. Potential-vegetation maps depict the untrammeled natural environment not as it is, but as we think or imagine that it should look.
Making Natural Earth II
Creating Natural Earth II, a digital product, involved manual techniques similar to those developed by Tóth more than three decades ago. The starting point for production was the original Natural Earth land cover dataset created by the author in 2005. Modifications to the data in Adobe Photoshop included painting with the Brush tool, cloning, applying color adjustments, and masking.
Polar, desert, and highland environments in Natural Earth II are mostly the same as those in the original Natural Earth. These environments tend to have little human population and are still relatively intact, or affected in ways not discernable on a small-scale map. Modifications to arid environments involved replacing light yellow-green irrigated croplands with desert beige. By contrast, humid forest environments in the mid-latitudes and tropicsplaces most favored for human habitationreceived major modifications. In these areas, forest cover was restored to its estimated original extent before the significant growth of human population and activity in the last two or three thousand years.
Restoring forest cover to its original extent was an inexact exercise involving multiple sources of information and visual interpretation. The best indicators that forest once existed in an area are remnant patches of forest that exist on contemporary land cover data. For example, the original Natural Earth dataset shows scattered patches of forest throughout northwestern Europe on a landscape otherwise dominated by urban and agricultural land. This information, plus the prevailing moist climate and common knowledge that forests once covered the entire region, made the decision easy. A continuous green tone fills northwestern Europe on Natural Earth II.
Determining original forest extent was more difficult for drier areas where remnant forest cover is scarce today, such as central India and northeastern Brazil. Classic physical geography references helped with this task, including Küchler potential-vegetation maps, Köppen climate maps, and precipitation maps, all obtained from Goodes World Atlas, 1978.
Helpful online resources included the Global Distribution of Original and Remaining Forests, a map published by UNEP-WCMC. The Degree Confluence Project is a website where GPS users post photographs of the 11,076 places on Earth where lines of latitude and longitude intersect on land. Photographic evidence of forest at a coordinate intersection today is strong evidence that natural forest existed there in pre-modern times.
In drier forested regions, denser green tones occupy river valley bottoms and upland areas that are presumably moister. In some areas where small but significant forest stands exist, such as the Rocky Mountains of the western United States, appear with slightly more green. Conversely, in large regions with uniform forest cover, areas with relatively fewer trees received emphasis. The Black Belt prairie that traces a crescent through Mississippi, Alabama, and Georgia in the southeastern United States is one such example. These subtle modifications present a more complete picture of the natural environment and produce a more visually interesting map.
A moving target
In mapping potential-natural vegetation difficult questions arise about how far back or ahead in time one should go. Even without human influence, the natural environment is in constant flux. Southwestern Alaska, for example, is now covered with tundra vegetation even though the climate there is just warm enough to support forest coverit is hypothesized that not enough time has elapsed since the end of the last ice age for trees to colonize this remote area far from other forests and seed sources. In other regions, vegetation change can be rapid. The semi-arid Sahel bounding the southern edge of the Sahara is greener today than it was a few decades ago because of increased rainfall. The idea that plant succession will eventually lead to a climax community that will remain in stable equilibrium for a long time is refuted by scientists today because natural disturbances are not as rare as once thought.
When human influences are factored in, mapping potential-natural vegetation becomes even more complicated. That industrial societies have altered natural landscapes significantly is not debatablethe manicured suburb where I live is evidence of this. However, should the more benign activities of indigenous peoples in former times, such as intentionally lighted fires to clear forests or drive game, be a valid consideration for mapping potential-natural vegetation? For example, fires set by the pre-Columbian inhabitants of North America expanded the extent of the Great Plains grasslands into adjacent forestlands.
Alan Weisman in The World Without Us (2007) deconstructs how contemporary cities would revert to a natural state relatively quickly if the human species were to suddenly disappear from Earth (a scenario that is both troubling and fascinating). According to Weisman, in a matter of days weeds would start to grow in the cracked pavement of Manhattan, establishing a beachhead for other plants and the eventual return of forest. Mid-latitude forest environments, such as New York City, are resilient to change. More fragile environments elsewhere might never fully recover to what they once were because of human influences. Consider the drier parts of the Hawaiian Islands where introduced kiawe (mesquite) trees thrive on land that once was native grassland, a vegetation type unlikely to return ever again.
The small map scale, vegetation boundaries with fuzzy edges, and generalized classification of Natural Earth II partially circumvents these problems.
Using a generalized classification brought other benefits to Natural Earth II. Because it is a base for mapmaking, a generalized classification with fewer environmental colors improved legibility. A base map that is too colorful and varied can distract from thematic information overlaid on it. A practical concern was limiting the shades of green representing forest types to a number that a reader could easily differentiate, a task made more difficult by colors that merge together on the map just as forest types do in nature.
Understandable terminology was another reason for using a generalized classification. Many land cover and vegetation maps use scientific terms that are unfamiliar to average readers. In the Natural Earth II classification, simpler terms were preferred. For example, consider the broad forest belt that encircles the earth in the northern hemisphere immediately south of the Arctic Circle. Natural Earth II calls this environment northern forest and depicts it with cool green. For the intended audience, the author rejected the terms boreal and sub-arctic as too arcane. Other terms for common tree types found here were not inclusive enough. Needle leaf and coniferous do not take into account the many birch and aspen trees that comprise northern forests. In addition, the presence of these broad-leaf deciduous trees and larch, a needle-leaf deciduous species, ruled out using the term evergreen.
Terrestrial environment classification used for Natural Earth II.
The Natural Earth II classification contains three forest types (tropical, temperate, and northern) and two types of partial forest, (Mediterranean vegetation and open forest/savannah). Coloring these forests involved visually comparing Köppen climate maps and the restored forest extent on Natural Earth II and applying an assigned color in Photoshop where they coincided. For example, brownish green Mediterranean vegetation found in scattered pockets around the globe correlates with Mediterranean (Cs) climates in Köppen’s system. Following this approach, northern forest correlates to Köppen’s Db, Dc and Dd climates, tropical forest to Af and Am climates, and temperate forest to several intermediate climates in his system.
A generalized environment classification also must come with omissions. Natural Earth II does not classify shrub land or wetlands as distinct environments. Shrub landa vegetation type that gets little respectis subsumed by Mediterranean vegetation, open forest, grassland, and desert in the Natural Earth II classification. Wetlands are also part of other classifications. The Everglades of Florida appear as grassland and the bogs of the Western Siberian Lowlands, the largest wetlands on earth, appear as lightly tinted northern forest.
Delineating the boundary between grassland (steppe) and forest environments was difficult. Natural Earth II reclassifies herbaceous land cover (a category that includes croplands and pasture) in the original Natural Earth dataset as grassland. Factoring in the absence of intentional burning by humans, forest cover encroaches on and blends softly into the Hungarian Alföld, Great Plains of North America, Llanos of Venezuela, and other grasslands of the world.
Natural Earth II is an artistic interpretation of our planet in a more natural state than which exists today, based on physical geography references. It is not scientific data.
Environment colors:..From the original Natural Earth dataset with manual enhancements in Photoshop (downsampled from 32,400 x 16,200 to 16,200 x 8,100 pixels).
Shaded relief:..Rendered from CleanTOPO2 (upsampled from 10,800 x 5,400 to 16,200 x 8,100 pixels).
Rivers:..Rasterized World Data Bank 2 vector data adjusted manually for better registration with shaded relief. Rasterization size: 16,200 x 8,100 pixels.
Water bodies, version 1:..Rasterized World Data Bank2 vector data. Rasterization size: 16,200 x 8,100 pixels.
Water bodies, version 2:..MODIS VCF satellite data (downsampled from 32,400 x 16,200 to 16,200 x 8,100 pixels).
Note: Aral Sea and Lake Chad shorelines updated from May 2007 Google Maps imagery.
Printing and resolution
Offset printing:..At a resolution of 270 dpi, the 16,200 pixels in Natural Earth II translates to a world wall map 1.5 meters (5 feet) wide. Printed maps of the continents are possible at about tabloid size, depending on the size of the continent.
Inkjet prints:..These are possible at much larger sizes, especially for maps viewed by people standing a meter or two away. The question is: how low of a resolution are you willing to accept? At 100 dpi, a world map would measure 4.1 meters (13.5 feet) wide.
Web:..The upsampled shaded relief in natural Earth II, created to accommodate offset printing, looks too generalized when viewed at 100 percent size in a web browser. The author recommends downsampling Natural Earth II to about 75 percent of original size or less for online viewing.