Elevation shows the shape of the landscape and may be represented with elevation colors, contours, terrain shading, curvature, or a combination of these layers.THE BASEMAP PROVIDES CONTEXT FOR THE TOPICS YOU map. Basemap features show relationships between the natural environment and the places people live, work, and explore. They show the structure of natural habitats and urban systems. Good map design offers the right amount of base information and positions it at the right level in the visual hierarchy. The basemap is usually background information—getting out of the way of other mapped information but still providing useful context.
Topographic mapping in the United States is led by the US Geological Survey (USGS), and they divide their effort into basic themes for The National Map: elevation, orthoimagery, transportation, land cover, hydrography, boundaries, structures, and geographic names. These themes are common for countrywide reference mapping from around the world. They are also the base information you select from as you design a map for a specific topic. The following sections describe design options and best practices for important aspects of raster and vector base data, as well as how this data is used through scale. Imagery, topographic layers, and individual themes are often available as services that load from the web as layers in a geographic information system (GIS) project. You can add them to your map with almost no effort.
Well-designed basemaps are built from the ground up, from landform to land use to line and point overlays:
Elevation shows the shape of the landscape and may be represented with elevation colors, contours, terrain shading, curvature, or a combination of these layers.
Images of the surface provide an overall understanding of the landscape, and specific layers for land cover, land use, zoning, and other property characteristics add detail.
Vector overlays of water features, boundaries, transportation, populated places, and points of interest add detail and structure.
All of this data needs to be selected at resolutions that suit the scale of the map, or series of maps, that you produce.
Figure 2.1 Smooth gradual elevation hypsometric tint at 1:120,000 scale, southwest of Ashland, Oregon, using a custom light beige to dark blue algorithmic color ramp. Lighter colors represent higher elevations and darker colors represent lower elevations. Data source: USGS. Map by P. Limpisathian, Department of Geography, The Pennsylvania State University (Penn State Geography).
The shape of the landscape sets the stage for both human and physical processes. Map readers will better understand a topic you map if they can see how it is distributed across the landscape, covering flat areas, running down valleys, and marching over ridges. Four common ways to represent landform from digital data are by shading terrain, setting colors for elevation, generating contour lines, and calculating curvature. Each of these is best derived from a digital elevation model (DEM), which is a raster data type that stores a regular grid of elevation values. A fine-resolution DEM works well for large-scale, detailed mapping, and a coarse-resolution DEM better suits the broad view of a landscape seen on smaller-scale maps.
The simplest view of elevation is a series of colors representing low to high. These colors may be continuous, from dark for low elevations grading through middle lightness to light for high elevations. Often, mapmakers vary color hues within this gradation. A classic treatment for small-scale maps uses greens for low elevations, yellows and oranges for middle elevations, brown for highlands, and then white for peaks. Appealing default color schemes of this sort are offered in GIS software, although they will be less suitable for larger-scale mapping that is not characterized by verdant lowlands and high mountain peaks—not your typical setting for a city map or state forest reserve. Green for lowlands in an arid region also may not offer a desirable view of the landscape. Color schemes are described in more detail in later chapters, but the example in figure 2.1 shows a simple approach, with a light-to-dark sequence that also transitions through related hues to provide more contrast between elevations.
The general term for elevation coloring is hypsometric tints (hypso). They may continuously change through elevation, as seen in figure 2.1, or they may be classified into ranges of elevation. For example, light yellow may represent 1,250 to 1,500 meters, light green 1,000 to 1,250, and middle green for elevations 750 to 1,000. Hypsometric tint bands may be approximated by classifying the DEM so that each pixel has a set color (figure 2.2). The overall visual effect will be a layering of colors in elevation order. Another way to create hypsometric tints is to close contour polygons, fill them, and order them with higher elevation polygons on top of the lower ones.
Contours provide specific metric information about the elevation of points on the landscape. An individual contour line connects points of equal elevation. In GIS-based mapping, they are generated from a DEM through interpolation. There are many choices for interpolation methods, such as inverse distance weighting and kriging, but here are some design tips for contour maps. Use the same contour interval throughout the map. For example, a 50-meter interval is used in figure 2.3. A constant interval allows readers to judge changes in the shape of the terrain. For example, as contours get closer together, the slope of the land increases. It is also common to set every fourth or fifth contour line darker or wider. These darker or wider contours are index contours, and the remaining lines are intermediate contours. The index contours provide a visual emphasis that shows a general shape and helps readers follow lines across the landscape (figure 2.3). Supplemental contours may be used to add some needed detail in flat areas. For example, 5- and 15-foot contours may sit between lines in a regular interval of 0, 10, 20, and 30. Supplemental contours are best given a different symbol, lighter or dashed, so the map reader does not think they show steepening slope.
Figure 2.2 Classed elevation hypsometric tint in 250-meter class intervals. This tint uses a custom orange-to-green color ramp. Orange represents higher elevation classes, and green represents lower elevation classes.
Figure 2.3 50-meter-interval contour lines with bold index lines at 250-meter intervals.
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
Figure 2.4 50-meter-interval contour lines overlaying a classified elevation hypsometric tint, with bold index contour lines every 250 meters.
Figure 2.5 Simple black-to-white hillshade using default values.
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
A good way to add definition to an elevation representation is by overlaying classed hypso with contours generated from the same DEM. Use the same interval as the classes for a contour at each break, or use a finer interval with index contours at each hypsometric color break. This greater frequency of lines gives the reader specific information about detailed elevation variations that are harder to capture with more general hypsometry. Figure 2.4 shows index contours and hypsometric tints with an interval of 250 meters and intermediate contours with an interval of 50 meters within color bands.
A third method of terrain representation created from a DEM is hillshading. This method calculates artificial lighting of a surface with a specified direction (azimuth) and altitude (angle) of illumination. The default is from the northwest, and 45 degrees above the surface, so that the landscape looks illuminated from above and to the left by an imaginary sun (figure 2.5). That 45-degree sun angle can pick up shapes of a wide variety of features. There are many other factors to consider when creating a hillshade, such as whether high-elevation features will cast shadows on land nearby. Changing the azimuth and angle can give you a wide range of results, and they are worth experimenting with to better see landform shapes. (The terms “terrain shading,” “analytical shading,” “relief shading,” and “shaded relief” all have much the same meaning and are produced using hillshade geoprocessing in ArcGIS software.)
A hillshade from a good quality DEM gives you a view with no vegetation, no rock types, or other surface characteristics. It is not a picture of the landscape—it is calculated from a grid of elevation values. A well-done hillshade starts to look like an image, but do not be fooled. Carefully applying color ramps, combining multiple hillshades, and additional graphic-design processing can customize views. You might add different hues to slope aspects, such as peach on lit slopes and pale blues in shaded areas, or heightened contrast on rugged peaks and muddled smoothing in the lowlands to make room for cultural features. Expert terrain representation is fascinating in the detailed treatment of scree and rock faces, glaciers, and cliffs.
Often, the goal of more modest hillshade efforts is to create a background for detailed overlaid information. That role means they need to be fairly low contrast with light-colored shapes. Your care with illumination angles and color ramps may be hard to see below feature overlays but will improve the map appearance. Adjusting the hillshade color ramp from a white-to-black default (figure 2.5) to a white-to-gray ramp helps push the hillshade to the background so that it does not overwhelm your other content. Setting the layer so that it is partly transparent will produce a similar result. Another good ramp choice is from white to light yellow to brown (figure 2.6). With some adjustment, this ramp places white on the fully lit slopes, yellow on flat lands, and darker browns on shaded back slopes.
The default northwest illumination (315 degrees) will leave northwest-to-southeast trending features with less definition than features more perpendicular to the direction of illumination. For example, valley walls will look less steep when they run in this direction than those of smaller valleys that run across the lighting. A landscape with an overall trend in the direction of features can be better shown by adjusting the illumination direction. If you swing the azimuth around to south and east directions, however, the landforms will look like they are being lit from below and will invert for many map readers. Stream valleys will look like ridges, and ridges will look like valleys.
Using multiple lighting directions and then adding them together using map algebra can give a balanced view of all feature shapes. For example, adding together four hillshades with illumination from the north, northwest, west, and southeast (with perhaps extra weight on the northwest hillshade) can give you subtle detail on slopes that face all directions (figure 2.7). Strategies like this can be particularly useful for subtle terrains—we are not always mapping rugged mountain peaks, nor should every hillshade give that sense of an extreme landscape.
Unlike elevation colors, the scheme for hillshading always runs from light to dark regardless of the multiple images you add together or the hues that tinge your ramps. The overall form of the landscape will fall apart if you use a dark color at both ends or somewhere in the middle of a hillshade ramp because your reader will not see the illumination effect at the core of these representations.
Another way to emphasize subtle features that barely catch the light of a hillshade calculation is to add a curvature layer. Curvature has extreme negative values along the most concave features, such as V-shaped valley bottoms. It has extreme positive values at sharp ridges, the most convex areas of the image. Flat areas have zero values, and many intermediate features are usefully highlighted, such as cliff edges and tributaries draining into valleys. (You will need to fill pits before processing curvature.)
Figure 2.6 Colored hillshade using a custom white-yellow-brown color ramp.
Figure 2.7 Multidirectional hillshade using a black-to-white color ramp. Compare the northwest to southeast ridges and valleys (upper left corner to the middle of the figure) to their appearance in the northwest-only illumination seen in figures 2.5 and 2.6.
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
Figure 2.8 Curvature generated from a filled DEM using default values.
Figure 2.9 A combination landform representation using three layers: (1) an overlay of curvature with a custom white-to-black color ramp at 70 percent transparency, (2) hillshade with a black-to-white color ramp at 40 percent transparency, and (3) hypsometric tint gradation with a custom dark-teal-to-white color ramp.
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
Usually, curvature is used in a landscape representation by using a black-to-white sequence (figure 2.8), with valleys represented as dark and ridges represented as light. Hue can be integrated into this sequence as well. This layer may lie below a transparent hillshade to add emphasis, add glints to peaks and ridges as a very transparent overlay, or be combined with the terrain layers using raster processing.
There are many DEM-based calculations that can be made, such as slope, aspect, viewsheds, and flow accumulation. But the four DEM-based layers that are most useful for creating a basemap to show the shape of the land are the following:
hypsometric tints to show relative elevation
contours to show specific elevation values
hillshading to show the shape of features
curvature to deepen valleys and highlight ridges
A combination of three methods is seen in figure 2.9.
Map layers that show the sorts of activities that characterize a landscape are orthoimagery, land cover, and land use from parcel data or zoning data. All of these may be used as basemap layers to support other map topics, or they may be the main focus of a map.
The detailed textures and colors of a landscape and the types of human activity on it are seen in aerial photographs and satellite imagery. These images are products of high-resolution remote sensing. When images showing the landscape from above are corrected to align with accurate mapping, they are referred to as orthophotos and orthoimages (“ortho” is a short form for “orthorectified”). Orthophotos and orthoimages are used for compiling GIS content and as seamless backgrounds for many types of mapping. Orthoimages collected when leaves are off deciduous trees are especially useful for compilation because roads and small features will be less obscured by trees. Leaf-on images may offer more attractive views of a location.
Orthoimage services at many map scales are available online, and many communities use fine-resolution orthophotos as key information in managing their infrastructure. Mapping over imagery (figure 2.10) is challenging because it includes many surface colors and a full range of dark and light areas. That means that any color you choose for a line or label will not be easy to see over the basemap. Later chapters discuss details about using halos and other effects with labels to make them more readable over varied backgrounds.
Keep in mind that an orthoimage does not need to be seen at full contrast to be a useful context for map reading. Before you mask the landscape with contrasting label halos and line casings, consider pushing the image to the background by making it lighter. That may mean increasing the image transparency to lighten it or using a partly transparent white layer between the image and the overlaying labels and lines. Overlaying the orthoimage with a partly transparent hillshade can also both push it to the background as well as add form to the image (compare figures 2.11 and 2.12). This overlay can be helpful because northern hemisphere images often have illumination from below (southeast sun angles are common with leaf-off images) and may look inverted to map readers. An orthoimage can also be set to a white-to-black or another monochrome ramp to keep urban or forest textures but remove hue contrast that interferes with reading overlaid symbols.
Figure 2.10 Orthoimage of Glens Falls, New York, at 1:200,000 scale.
Figure 2.11 Orthoimage southwest of Ashland, Oregon, at 1:60,000.
Figure 2.12 Multidirectional hillshade at 55 percent transparency overlaying an orthoimage, southwest of Ashland, at 1:60,000.
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
Another view of landscapes generated from remotely sensed data is land cover. In the United States, National Land Cover Data (NLCD) from the MRLC (Multi-Resolution Land Characteristics Consortium) is an example of a readily available product for basemap layers. It provides sixteen surface categories calculated from Landsat satellite data and is available through time (for example, 2001, 2006, and 2011). The data is fairly coarse, at 30-meter resolution, and better suited for mapping at scales smaller than 1:120,000. NLCD also offers US data for percent tree canopy (How dense is tree cover?—figure 2.13) and for a percent imperviousness layer for developed surfaces (What has been built up or paved?—figure 2.14).
Figure 2.13 NLCD percent tree canopy cover raster of Glens Falls at 1:120,000, using a modified color ramp from light to dark green to represent tree cover percentage values of 1 to 100, with white for zero tree cover.
Land cover describes both natural and human landscapes (figure 2.15). Example categories from NLCD include the following:
open water or ice and snow (coded with pixel values of 11 and 12—“10s”)
developed at low, medium, or high density; or developed open space (20s)
Figure 2.14 NLCD raster of percent developed imperviousness for Glens Falls. Dark purple represents the least permeable surfaces.
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
deciduous, evergreen, or mixed forest (40s)
shrubland (50s)
herbaceous grassland or Alaskan covers of sedge, lichen, or moss (70s)
planted pasture or hay and cultivated crops (80s)
wetlands that are woody or herbaceous (90s)
These categories are based on remotely sensed reflectance from the land surface. Although they suggest land uses, they are not based on property information such as residential zoning.
Figure 2.15 NLCD land cover raster of Glens Falls with default color values. Open water and wetlands are in blues, developed lands in reds, barren land in gray, forested areas in greens, and agricultural lands in yellow and brown.
Categories may be collapsed to suit simpler land-cover basemap needs. For example, using only raster forest and agricultural categories (combining three and two categories, respectively) will provide a basic land-cover background that fits around city information from other sources (such as a road network) and water from vector hydrography data instead of the raster (figure 2.16). Another alternative to grouping the forest categories is to use the percent canopy for a forest layer with more variation in lightness in the background (figure 2.13).
Figure 2.16 NLCD land cover raster of Glens Falls with aggregated classes for forest and agriculture at a smaller scale (1:300,000).
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
Parcel data inventories the details on property lots such as precise boundaries, owner names, assessed values, and identifying numbers. Parcel maps are also called cadastral maps. Parcel data may also have general land-use attributes, such as single-family residential, or these attributes may be extremely detailed, such as parking lot, high school, manufacturer, unimproved beach, and mini mart. Although they are intricate, parcel types may allow you to build a colorful portrait of the details of a town that can support other mapping projects (figure 2.17).
Parcel data can also provide a strange but useful version of roads for base information. Most gaps between blocks of urban parcels are roads. Dissolving parcels together to create city block outlines gives you roads that are limited (they are the negative space without names or attributes) but can act as a detailed road background. These sorts of road shapes can suffice when vector road lines are too generalized to register well with your other data for mapping at large scales.
Figure 2.17 Classified parcels of Glens Falls, New York, at 1:22,000. The parcels are based on generalized activity using a modified standardized zoning classification color scheme with residential parcels in yellows; commercial in reds; recreation and entertainment in blue-green; community services in blues; industrial-manufacture in purple; public services in gray; and wild, forested lands, and public parks in greens.
Data source: Warren County, New York State. Maps by P. Limpisathian, Penn State Geography.
A town’s zoning rules restrict land parcels to general functions such as residential, sales and services, manufacturing, transportation and utilities, educational, and agricultural. Zoning is often the main topic of a map. In the United States, the American Planning Association (APA) recommends particular colors for common zoning classes, and zoning maps often combine these with varied textures. A zoning map may be well supported by using an underlay of orthoimagery so map readers can tell which buildings fall in and outside of particular zones.
Figure 2.18 Classified zoning based on generalized activity using a modified standardized zoning-classification color scheme.
Zoning can also be used as a basemap layer that shows the general character of a town (figure 2.18). The residential areas will be colored differently than the commercial and industrial areas, and those can be distinguished from parks and other recreation sites. When you use zoning data this way, you may aggregate land-use classes and adjust colors as needed to support the overlaid main-map topic (figure 2.19). Parcel data may also have enough zoning information in the attributes to create a zoning map that can then be generalized to a basemap.
Figure 2.19 Generalized land use based on reclassified zoning, excluding commercial, overlaid on an orthoimage.
Data sources: USGS, Warren County, New York State. Maps by P. Limpisathian, Penn State Geography.
Parcels and zoning let you develop a land-use map with more accuracy and detail than a land-cover map. Figure 2.20 shows transparent land-use colors overlaid on an orthoimage with roads produced using gaps between blocks of parcels and labeled using names from more generalized road lines. This combination of layers provides a detailed basemap that can support other urban mapping topics.
Figure 2.20 Combination overlay of road name labels derived from invisible road lines, dissolved parcels with only the block outlines shown, classified zoning data at 30 percent transparency, and an orthoimage.
Lines and polygons are the classic content of basemaps. Standard basemap layers are hydrography, transportation, boundaries, and cultural point features. These layers are generally on top of the raster and polygon layers for landforms and land cover or land use in the basemap. An overlay of vector data may need to be edited to register with other layers, even after differences in coordinate systems are resolved. Mismatches are particularly a problem if layers come from different sources or different scales. For example, rivers may not run down valleys in the hillshade, or roads may not run through their corresponding gaps in the forest canopy raster. The general rule is to move the vectors to match the physical characteristics that underpin them if those layers are orthorectified.
Hydrographic data is quite detailed, but typical surface water features used on a basemap are rivers, lakes, wetlands, and coastlines. In the National Hydrography Dataset (NHD) for the United States, rivers, streams, and canals are represented as lines, and sometimes as polygons when river banks are far enough apart to be drawn separately. Bodies of water, such as lakes, ponds, and reservoirs, are polygons. Mapmakers often use the short form “hydro” to refer to a full set of water features (not to be confused with “hypso” for hypsometry—elevation).
Network attributes of hydrographic data that allow it to be used for hydrologic modeling are perhaps excessive for basemap construction, although they can be useful in many ways. For example, attributes that list calculated upstream drainage area or cumulative flow allow the mapmaker to taper the stream lines. Tapering means lines get progressively thicker as flow increases—you only need about six line widths to make this work. Variation in line widths gives a visual sense of the stream network (figure 2.21). Tapering keeps numerous small tributaries from dominating the look of the map and visually competing with the major rivers into which they drain. Upstream drainage area, or cumulative flow, also provides criteria for setting some river labels larger than others, just as highway names are set larger than local road names. Rivers with a higher upstream drainage area or cumulative flow are large rivers downstream of small feeder tributaries and can be accorded larger labels.
Networked hydrographic data also has artificial paths and connectors running underneath bodies of water (figure 2.22). You may not need these if you show all the water features on a map, but they are useful for labeling oddly shaped river areas cleanly down the middle (label the line rather than the area). You may also use artificial paths and connectors instead of a river polygon at smaller scales for a simpler linear feature. These paths also let you remove a clutter of small lakes and ponds from the map without leaving gaps in stream lines.
Figure 2.21 Waterbodies overlay tapered flowlines drawn using total upstream drainage area and medium resolution NHD south of Ashland, Oregon, at 1:200,000. Data sources: USGS, NHDPlus. Map by P. Limpisathian, Penn State Geography.
Boundaries come from numerous sources. Administrative boundaries that are commonly used on basemaps are state, county, township (minor civil division, MCD), incorporated place, and reservation boundaries. Others that may be useful are boundaries of recreation or conservation areas such as forest reserves and parks at local, state, and national levels. Boundary options are wide ranging and often have an implicit hierarchy. They may also be nested with overlaying lines that create visual clutter as they fall one on top of the other. This can be a messy problem if you would like to use dashed lines, because the dashes overlay and fill in each other’s gaps. Another problem with boundaries is they may not perfectly overlay if they are from more than one source. There are two basic options for cleaning up the appearance of coincident boundaries.
Figure 2.22 Hydro at high resolution with artificial paths and connectors shown in red, near Glens Falls, New York, at 1:40,000. Data source: USGS NHD.
If boundaries overlay well, then use a boundary line made up of two lines, one dashed and one solid sitting right below it (figure 2.23). Set the solid line at the same width, or slightly wider, to cover up any lines below it. It may be the same color as the background or have a color that is part of the boundary design. This strategy also lets the line remain visible on a background that varies in lightness, such as hillshading or orthoimagery. The line will appear as light dashes over dark background areas and dark dashes over light background areas.
A second strategy is to convert polygons to lines and use a GIS operator that removes coincident lines (or delete the extra lines by hand). You will likely want to use two versions of area features in the map. The original polygons may carry color fills and also area labels, but have no outlines. The boundary lines will be separate with redundancies removed. This strategy is also useful because you have the flexibility to position the fill below most other features (and set it transparent), while the lines may be placed above features such as hydro and roads.
Figure 2.23 An overlay of county, minor civil division, and incorporated and unincorporated place boundaries in Glens Falls, New York, with appropriate hierarchy of symbols and solid lines to mask dashes below. Data source: USGS. Map by P. Limpisathian, Penn State Geography.
Figure 2.24 Classified hierarchy of road and rail lines for Glens Falls at 1:200,000.
Figure 2.25 GNIS POI in Glens Falls, New York, at 1:100,000.
Data source: USGS. Maps by P. Limpisathian, Penn State Geography.
Roads are key reference information for other mapped data and can be overwhelmingly busy on a map. A good road dataset will have many levels of importance as attributes for the lines (figure 2.24), so you may use sparse or detailed roads that maintain a sense of networked structure. Typical US road hierarchy includes categories such as interstate, US, and state highways; primary, secondary, and connector roads; and local roads. You will likely want more detailed importance attributes so you can control which roads are labeled and remove minor roads that are less connected within the network. But you cannot always get what you want.
Transportation data also includes other features such as railways, unimproved or four-wheel-drive roads, trails, service roads, private roads, cul-de-sacs, alleys, parking lots, bike routes, ferry crossings, and bridges. Some of these may enliven a basemap, but many will be best removed using their respective attributes—if you map all transportation in a complete dataset, prepare for a mess. Roads also may have multiple names and numbers. A stretch of a single highway may be best labeled with road shields for its state and county road numbers as well as its local name. The level of detail you attend to with these problems depends on whether a map is used specifically for navigation or the roads function as a base for other operational layers.
Populated places may be represented with point symbols instead of administrative areas as described in the previous section. These points may be used as town spots, or they may be used only to control positioning of place labels and not shown with a symbol.
Other point data for US national mapping includes structures, such as energy facilities, medical centers, transportation terminals, emergency response and police stations, government buildings, water treatment plants, and other points of interest. Public attractions data include obvious locations such as landmark buildings, but also places such as cemeteries, ski areas, sports arenas, golf courses, and historic sites. Point locations may also be physical features such as summits. Point locations for symbolizing and labeling these types of features for US locations (figure 2.25) can be downloaded from GNIS (Geographic Names Information System). They may also be collected from volunteered geographic information sources such as Open Street Map or, of course, your local knowledge.
These sorts of resources offer the perpetual challenge of having many features of differing importance all classed together. A major medical center and a local podiatrist may be in the same health-related data. Feature codes may let you separate and cull minor locations, or they may not. Selected points of interest (POI) on a map provide key locations which help people understand the importance and character of a mapped distribution.
Figure 2.26 shows the usual vector basemap layers together—hydrography, transportation, boundaries, and populated places.
Combinations of basemap layers are shown previously in figures 2.9 (hillshade, curvature, and elevation), 2.19 (generalized zoning and orthoimage), 2.20 (parcel-based roads, zoning, and orthoimage), and 2.26 (places, administrative boundaries, transportation, and hydrography). The following maps show additional combinations from among the themes described in this chapter. Figure 2.27 shows a basemap with incorporated place and MCD boundaries, roads and railway lines, hydrography, and contours. This is a plain vector basemap with a range of feature types that would support many operational overlays.
Figure 2.26 Combination overlay of classed administrative boundaries, classed road lines and rail features, waterbodies, hydro lines, and GNIS populated place points in Glens Falls, New York, at 1:200,000. Data sources: USGS, NHDPlus. Map by P. Limpisathian, Penn State Geography.
Figure 2.27 A plain basemap combination shows the overlay of MCDs and incorporated place boundaries, roads and railways, tapered hydro, and contour lines at 50-meter intervals with index contours at 250-meter intervals, for southern Ashland, Oregon, at 1:40,000. Data source: USGS. Map by P. Limpisathian, Penn State Geography.
Figure 2.28 is dominated by raster land cover and a hillshade (75 percent transparent) with a supporting overlay of hydro and major roads. This gives a sense of the landscape and landforms with less human detail.
Figure 2.28 The landscape view of Ashland, Oregon, at 1:120,000 overlays generalized roads, waterbodies and tapered flowlines, hillshade at 75 percent transparency, curvature with a modified white-to-black color ramp at 75 percent transparency, and land cover.
The last combination, shown in figure 2.29, is larger scale and includes roads and hydro over a more detailed landscape created from an orthoimage with a 60 percent transparent hillshade. It could support mapping of natural resources information.
Figure 2.29 This natural resources basemap of southern Ashland, Oregon, at 1:40,000 combines road lines, hydro with tapered flowlines, hillshade at 60 percent transparency, and an orthoimage.
Data sources: USGS, NHDPlus. Maps by P. Limpisathian, Penn State Geography.
Some of the examples are fairly high contrast, verging on reference map designs rather than basemaps ready for additional map information to be overlaid. Basemaps are background, so they need to be light (or dark) enough to allow the main themes to stand out against them. They are important content but not always the main content. They support reader understanding with physical and cultural contexts.
Base data you select should be well matched to the scale of the map display. Using raster data that are too coarse produces a blurry, grid-like, or angular look. For example, land-cover data at 30-meter resolution is suitable for map scales starting at 1:120,000. If you use these data with 1:25,000 topographic mapping, the land cover areas will look like collections of squares, and their edges will not align with other data, such as overlaid roads or forest edges in orthoimages. The scale is too large for the data resolution. Using vector data at larger scales than intended means lines will look angular or overly smoothed. As you download base information, pay attention to the scale or scale range for which it is intended. Does that resolution match your map’s scale?
The cell size or resolution of a DEM determines the range of map scales for which you will use it. You want to choose a DEM that will give you the best looking terrain when raster data download services offer a variety of resolutions. For example, The National Map Viewer for US data has many elevation products available for download for each location, from 1/9 arc second to 200-meter cell-size resolutions. Kimerling’s recommendations on matching resolution and appropriate mapping scales for desktop screen resolutions (100 pixels per inch) are useful—a simplified version is seen in table 2.1.
Figure 2.30 The wrong resolution DEM for the map scale produces a blurry result. The hillshade from a 1-arc-second DEM is shown at 1:15,000 but is suited to mapping at 1:120,000. The mapped location is centered on Ragged Mountain in the Mount Greylock State Reservation near Adams, Massachusetts (exported at 400 dots per inch [dpi]). Data source: USGS.
Figure 2.30 shows a result of downloading the wrong DEM to make a hillshade. The 1-arc-second DEM is too coarse to produce a useful hillshade at the large scale of 1:15,000. The pixels are about 30 meters square, the same as NLCD land cover data, and can be seen as square shapes in the image at this large scale. A hillshade created with smaller pixels (1/9 arc second, or about 3 meters) at the same scale is shown in figure 2.31. These high-resolution DEMs from US portals are often lidar-based and provide details such as rock textures, road cuts, parking lot edges, and sometimes building footprints.
*For an output map pixel density of 100 pixels per inch (40 pixels per cm).
Source: A. Jon Kimerling at http://blogs.esri.com/esri/arcgis/2011/02/28/dem-resolution-output-map-pixel-density-and-largest-appropriate-map-scale/.
Figure 2.31 The correct resolution for the scale produces a crisp result. The 1:15,000 terrain map is created by hillshading a 1/9-arc-second DEM for the same extent as figure 2.30 (exported at 400 dpi).
Figure 2.32 Larger pixels work well for the 1:40,000 terrain map, with a hillshade from a 1/3-arc-second DEM. The extent of figure 2.31 at Ragged Mountain is shown by the yellow rectangle (exported at 400 dpi).
Figure 2.33 One-arc-second pixels (shown at too large a scale in figure 2.30) are seen here at the suitable scale of 1:120,000. The extent of figure 2.31 at Ragged Mountain is shown by the yellow rectangle (exported at 400 dpi).
Figure 2.34 Pixels 100 meters in size are seen here at the suitable scale of 1:360,000. The extent of figure 2.31 at Ragged Mountain is shown by the yellow rectangle (exported at 400 dpi).
Data source: USGS.
The next three figures are produced from standard downloaded DEM products and default hillshade settings (figures 2.32 to 2.34). Each is shown at a suitable scale, with the fine yellow box showing the size of the largest scale (figure 2.31).
Figures 2.30 to 2.34 were all produced at high resolutions (400 dpi) suited to book publishing. The next two figures repeat figure 2.33 at 100 dpi export, akin to a desktop computer display. You can see how the first at 1/9 arc second (too fine) has become strangely granulated during rendering (figure 2.35), and the second at 1 arc second still looks good (figure 2.36 appears much the same as figure 2.33, which was exported as a much finer resolution image). The fine lines of the scale bars certainly look worse at 100 dpi. Not only does excess detail affect file size, but it can also degrade image quality for the reader. Chapter 4 goes into more detail on resolution and file formats for exporting maps, but the main message here is to choose (or create) data with a resolution that matches the intended map scale. You can also smooth a DEM or resample it to larger pixel sizes to produce data at desirable scales for your terrain mapping.
As you move from detailed contour mapping to smaller scales, you can get away with selecting a subset of contours from a more detailed interval, such as keeping the index contours only. You will likely become quickly dissatisfied with this strategy because you will have sparse lines that are each overly detailed. Generalizing contour lines with simplification and smoothing operations to try to get them to work at smaller scales is a poor choice. The way they nest and follow each other in shape will become corrupted as the shape of each line is independently changed. The landscape form will no longer be clear, and the lines will be inaccurately positioned, making for a messy map. The best way to prepare contours for a smaller scale is to go back to the DEM, smooth it, and then regenerate the contour lines. You may smooth the DEM, resample it to coarser pixels (such as 10 meter pixels resampled to 100 meter pixels—100 pixels are averaged to get one value for the larger pixel), or do both. Resampling, smoothing, and other DEM processing suited to a smaller-scale representation sets you up for creating better contours as well.
Figure 2.35 Pixels 1/9 arc second in size are too fine to image well at the small scale of 1:120,000 when exported at 100 dpi.
Figure 2.36 One-arc-second pixels image well at 1:120,000 and 100 dpi (coarser than figure 2.33).
Data source: USGS.
NHD data includes medium resolution for 1:100,000 mapping—NHDPlus provides upstream drainage area attribute tables at this scale—and high resolution for 1:24,000 mapping. Figure 2.37 shows a sample of the high and medium resolutions. You can see that the high-resolution data (figure 2.37A) is quite difficult to work with at smaller scales and would not provide a useful context for most map topics that are not intensively related to hydrography. There are too many lines for other content to compete with. This high-resolution view is shown with artificial paths and connectors turned on through waterbodies and river areas to see channels that could be used for cartography if areas are removed. The medium-resolution NHD data (figure 2.37B) is shown with the river areas replaced by centerlines and some small waterbodies removed to create hydro better suited to the scale. All of the river flowlines in the data are shown in figure 2.37B—they are much sparser and simpler than the high-resolution data. It matters which dataset you choose for a hydro basemap.
Figure 2.37 Hydro data for the area around Glens Falls, New York. High-resolution data are mapped at a scale smaller than is suitable (A), compared to medium-resolution data with some small polygons removed to suit this scale, 1:200,000 (B). Data source: USGS NHD. Maps by P. Limpisathian, Penn State Geography.
Datasets for populated places are also chosen or altered to suit the map scale. The three datasets shown in figure 2.38 are from Natural Earth, an open source for high-quality data for small-scale mapping. Natural Earth offers data for themes such as boundaries, populated places, roads, rivers, and terrain generalized to suit 1:110M, 1:50M, and 1:10M. These are small-scale maps, as you can see from the large extent shown around Shanghai. All three are shown together at a too-large scale of 1:5M in figure 2.39. The cartographers generalize the urban areas for each smaller scale. Compared to the blue areas, the green outlines for maps five times smaller are simpler and smoother. Urban areas are merged when they are close together and many smaller urban areas (which are still big cities) are eliminated. When the map scale becomes very small and the urban areas become very small, they are better collapsed to points. This is the same type of operation as with river area polygons, collapsed from an area to a line.
Figure 2.38 Urban areas centered on Shanghai, China, and coastlines in the region are shown at their intended scales of 1:10M, 1:50M, and 1:110M.
Data source: Made with Natural Earth. Maps by P. Limpisathian, Penn State Geography.
Figure 2.39 Urban area data are shown superimposed and enlarged to 1:5M for comparison: 1:10M (blue), 1:50M (green), and 1:110M (purple point). Polygons are simplified, smoothed, merged, and finally collapsed to points to generalize them to suit the map scales.
Another example of vector data through scale is choosing from among census geographic entities. The hierarchy and nesting of enumeration units for aggregating US census data are shown in figures 2.40 and 2.41. For example, blocks fall within block groups, and those nest within tracts and then within counties. Counties nest within states, but school districts, congressional districts, and places such as towns and villages (figure 2.41A and F) may overlap statistical boundaries. Comparing figures 2.41C and F shows block groups do not nest in incorporated places, for example. This seems like trivia, but as you plan to cache a series of maps of demographic data, you plan which enumeration units to use through scale ranges. Block groups may be great for large-scale levels. Changing to tract, then county, then state data for smaller-scale levels will provide a useful sequence for an online mapping tool.
Figure 2.40 Standard hierarchy of census geographic entities with lines connecting nested enumeration units. This diagram does not include hierarchy detail for American Indian, Alaska Native, and native Hawaiian census geography, such as tribal areas. Data source: US Census Bureau (http://www.census.gov/geo/reference/hierarchy.html).
Figure 2.41 Five US census geographies shown together (A) for Glens Falls, New York, at 1:240,000. Hierarchy of census geography shows nesting of blocks (B), block groups (C), tracts (D), and counties (E). Census place polygons (F) are an example geographic entity that falls outside the nesting hierarchy. Data source: US Census Bureau. Maps by P. Limpisathian, Penn State Geography.
Census geographies are offered as generalized files to suit small-scale mapping at a variety of scales. As you download US census enumeration units, you will choose the level to suit your map purpose. For example, in addition to fully detailed TIGER (Topologically Integrated Geographic Encoding and Referencing) county files, cartographic boundary files for counties are served by the US Census Bureau at three levels:
500K = 1:500,000 (the largest scale)
5M = 1:5,000,000
20M = 1:20,000,000 (small scale—whole United States on a page)
None of these boundary files are detailed enough to function as good boundaries on a large-scale map, but they are much better choices for a map showing demographic data across a region. The simplified lines do not form coalescing and overly complicated shapes that interfere with reading the map. Figure 2.42 shows the full TIGER data, 1:500,000, 1:5M and 1:20M at the same scale for comparison. The cartographic boundary files also have the advantage of including generalized coastlines.
There are many changes made to geographic data to prepare it for mapping at particular scales. You cannot keep all the detail of a real place as you make it tiny on the page. The sections in this chapter approach this topic from the perspective of selecting data at the right resolution for a mapping problem. There are many geoprocessing tools you can use to change the geometry of your data to improve the look of your map. The following are the operations mentioned in this chapter:
eliminate
reclassify
simplify
aggregate
collapse
merge
smooth
Other ways to improve data for smaller scales are to displace and exaggerate the data.
Having attributes in your GIS data that let you systematically eliminate features and labels within categories is key for multiscale cartography. These attributes include population or economic importance for places, class or traffic volume for roads, numbers of flights for airports, upstream drainage area or cumulative flow for rivers, and readily calculated areas for polygons.
Figure 2.42 US Census Bureau cartographic boundary files mapped at 1:5M (green in A). These 1:5M-resolution data are overlaid on detailed TIGER counties in light gray to show their generalization (C). Finer resolution 1:500K data are shown in blue (B), and coarser resolution 1:20M data are shown in purple (D). The small black rectangle shows the enlarged location of B to D on the 1:5M map (A). Data source: US Census Bureau.
