GEOG 486
Cartography and Visualization

Color Spaces


Color spaces are essentially scientific systems for specifying and identifying colors. Scientists have developed a number of different systems for specifying color, some of which are based on physical properties of colors (i.e., reflectance curves) and others that are based on human perceptual abilities. These systems may be device-dependent or device-independent. In this concept gallery item, we will discuss two major color spaces, the Munsell system and the CIE system, and discuss their limitations in the context of GIS cartography. Then we will describe the three color spaces that are commonly implemented in a GIS: CMYK, RGB and HSV.

The Munsell system is based on how users perceive color. Munsell recognizes three main components of color: hue, value and chroma. Color hue is determined by the wavelength of light that is emitted by a light source (in the case of additive colors) or the wavelength of light that is reflected from some surface (in the case of subtractive colors). For a review of additive and subtractive colors, see Lesson 1, Part IX: Color Specifications. Color hue is also the component of color that we most commonly refer to when we are talking about colors (i.e., “ the grass is green or the apple is red”). Light that is reflected or emitted at different wavelengths produces different color hues (see Figure, below). Color value is the perceived lightness or darkness of a color. You can think of this component as a measure of how much light (at any wavelength) is reflected or emitted (see figure below). One interesting aspect of color value is that people’s perception of a color’s value is not proportional to the actual amount of light that is reflected. The third component of color is color chroma. Chroma is a measure of the purity of a color. High chroma colors are quite pure, like candy apple red or emerald green, while low chroma colors are muddy and greyish. High chroma colors are emitted over very narrow portions of the electromagnetic spectrum, while low chroma colors are emitted at a larger portion of the spectrum (see Figure, below).

A diagram to show differences in color hue, color value, and color chroma explained in text above
Figure This diagram depicts differences in color hue (red, green and blue peaks), color value (light and dark green peaks) and color chroma (pure blue and greyish-blue).
Credit: Adrienne Gruver

The Munsell system is based on human perceptual capabilities and was created after extensive user testing with subtractive color (i.e., pigments applied to a surface). It is a graphical representation of what humans are capable of seeing and the colors we are capable of creating with pigments. This basis is important because we are better at perceiving some color hues (i.e., we can perceive more variation within that hue) than other color hues. This unevenness in our perceptual capabilities is reflected in the uneven shape of the solid (see Figure, below).

A diagram to show color space in a 3D solid, explained in text above
Figure Color value is represented along the vertical axis in this solid, with high value colors (i.e.. light colors) towards the top and low value colors at the base. Hues are arranged in circular fashion around this vertical axis, while chroma increases as you go from the middle of the solid to its edges.
Credit: Wikimedia Commons

For example, we are able to perceive more shades of light yellow than light blue, but on the other hand, we are able to perceive more shades of dark blue than dark yellow. This is a result of the fact that yellow is an inherently high value (i.e., light) color, and even the darkest yellows will not be as dark as the darkest blues (see Figure, below).

Graphic illustration to show a sample slice from a Munsell solid explained in text above
Figure This figure is a ‘slice’ from the Munsell solid. In it you can see the difference in how much variation we can perceive within a color hue.

One reason the Munsell system is useful for cartography is that it divides up the universe of colors made from pigments (e.g., inks or paint) into perceptually equal steps. In other words, the difference in color value, color hue or color chroma between one color and its neighbor in the color space is the same for all neighbors. This is helpful in cases where we are trying to design a sequential color series (e.g., ranging from light to dark or from grey to vibrant color). We will discuss why this is important in more detail in the Color Ramps concept gallery item in Lesson 7.

For now, imagine that you want to create a color scheme that shows differences in the kind of features that are represented on the map (e.g., as you might in a land use map). For this map, you might want to choose color hue as your visual variable (see the Symbolization concept gallery item for more discussion of visual variables). In your color scheme, you will want to choose color hues that have similar color values (i.e., colors that are equally dark). In other words, you would not want to use navy blue, sky blue, yellow, light green and forest green as your colors, as the difference in lightness between the colors in your color scheme might imply to the map reader that you are trying to communicate something about the amount of something in the map. Choosing colors with different hues but similar values is quite easy in the Munsell space: as long as you choose colors from the same value step, your color scheme will fulfill this criteria (see Figure, below).

Graphic illustration to show a qualitative color scheme derived from the Munsell space, explained in text above
Figure To create a color scheme that shows differences in kind, you could choose colors that intersect the outer portion of the color ring in this depiction of the Munsell color space. These colors would all have similar color value and color saturation, but different color hue.

Although the Munsell color space is quite helpful for designing color schemes, it is not implemented in commercial GIS software. For this reason, you would have to translate the Munsell colors into another color space. One helpful resource for doing this is a paper by Cindy Brewer that lists equivalent CMYK values for a set of Munsell colors.

A second major color space is the CIE color specification system. This is a device-independent system that is based on additive colors (i.e., emitted light). This system can be used to describe all possible colors (i.e., it is not limited to the colors we can mix with pigments, as is the Munsell system). Colors are specified in this system by making physical measurements of the light with a spectrophotometer. The main advantage of this system is that we can precisely describe colors based on their physical characteristics (independent of humans’ perceptual capabilities). The main graphical representation that is used to describe colors in CIE space is called a chromaticity graph (see Figure, below). The dark grey polygon in the diagram is a representation of all possible colors. One useful feature of the chromaticity diagram is that we can combine it with representations of other color systems to understand the extent of a particular color gamut (e.g., the RGB color gamut shown in the figure below). A color gamut is the range of colors that can be created within a particular system of producing color. You can see from the diagram below that the RGB system, where color is created from light emitted by electron guns in your CRT computer monitor, is not capable of producing many of the colors that we could potentially work with. Like the Munsell space, the CIE color space is not implemented in commercial GIS, and is mainly used for research purposes or cases where cartographers need to use very precisely defined colors.

A graphic illustration to show the CIE chromaticity diagram, explained in text above
Figure The CIE chromaticity diagram and a representation of the colors that can be produced within the RGB color gamut. The numbers along the edge of the color space represent the wavelength at which light is emitted.
Credit: Adrienne Gruver

Now that we have discussed two of the main color spaces that we can use for describing and specifying colors, we can begin to think about the color spaces that are implemented within a GIS, and that we, as practicing cartographers, can easily use. We briefly mentioned RGB and CMYK in Lesson 1, Part IX: Color Specifications, so we will begin with these spaces.

When you look at the Figure, below, you will notice that RGB and CMY are two sides of the same cube. This makes intuitive sense when you stop to consider that when you mix 0% cyan, magenta and yellow, you will have no color (i.e., white), and when you mix equal amounts of cyan, magenta and yellow, you will produce black (recall the color mixing figure from Lesson 1, Part IX: Color Specifications). Also, mixing the equivalent to 100% (i.e., 255 on a scale of 0 to 255) of red, green and blue light produces white (i.e., the opposite of mixing full strength CMY inks), while mixing the equivalent of 0% (i.e., 0 on a scale of 0 to 255) of red, green and blue light produces black. Also recall that if you mix two primaries from one color system (e.g., cyan and yellow), you will get a primary of the other color system (e.g., green). If we tip the color cube so that the white to black diagonal axis is in the vertical position (as in Figure, below), you can see that the hues are arranged around this axis in spectral order as they were in the Munsell color space (although they are not arranged in a perfect plane – the CMY primaries are all lighter than the RGB primaries). Although there are similarities in how the RGB/CMY color cube and the Munsell perceptual systems are spatially organized, the RGB/CMY space is not a perceptual one. In other words, an equal step through RGB/CMY space does not produce perceptually equal differences in colors. This difference between the systems is what makes specifying colors in CMY or RGB tricky.

A graphic drawing of the RGB/CMY color cube.
Figure The RGB/CMY color cube.
Illustration by Amy Griffin

One alternative to the RGB/CMY color space that is also implemented in several commercial GIS software packages is the HSV color space. The HSV color space uses the three components of color (hue, saturation (aka chroma) and value) to specify colors. While this is a good idea in principle, it is often implemented in the computer as a cone-shaped space, with all full saturation hues having the same value. However, we know from the Munsell solid, which was based on empirical data about human perception, that not all hues have equal values at full saturation (e.g., full saturation yellow is always lighter than full saturation blue), so again in this space, equal steps do not produce perceptually equal color differences.

A graphic drawing to show an HSV color cone.
Figure The cone-shaped HSV color space.
Illustration by Amy Griffin

So, given that we would often like to control the perceptual characteristics of color, and the fact that commercial GIS software has not implemented a perceptually based color space for specifying colors, what’s a cartographer to do? One option is to carefully perturb the CMY or RGB color specifications until you find a set of specifications that has the perceptual characteristics you want (e.g., equal value steps for an ordinal color scheme or constant value for a nominal color scheme). Alternatively, you can choose your colors using the Munsell system, and then use the Munsell/CMYK equivalence tables that Brewer developed to create a set of color specifications that your GIS can work with. Finally, you can pester your favorite GIS vendor for better support for perceptually-based color specification!

Recommended Readings

If you are interested in investigating this subject further, I recommend the following:

  • Brewer, C.A. 1989. “The development of process-printed Munsell Charts for selecting map colors.” American Cartographer. 16: 269-78.