Graphics are a means by which the cartographer communicates selected information clearly and easily to the map reader or data user. The information can be contained in various graphics such as maps, charts, diagrams and graphs, and aided by related textual material.
The following characteristics are necessary to produce effective graphics:
i) Simplicity: The graphic should be easy to learn, remember, apply and interpret;
ii) Conciseness: The graphic should provide concise identification of the data portrayed;
iii) Uniformity: Identical graphics should be provided for similar data within the limits of the study;
iv) Informativeness: The graphics should convey enough information concerning an area to allow an immediate evaluation of its characteristics;
v) Reproducibility: The graphic should be designed for easy and economical reproduction;
vi) Talkability: The graphic units must be distinctive, easy to identify and easy to use in conversation and publications. They thus communicate an immediate and clear idea of the subject matter.
The following graphic design guidelines should be considered:
i) Measured lines should be shown by continuous (solid) line symbols in contrast to interpolations (estimates) which should be shown by interrupted (broken) line symbols;
ii) Symbols, which may be defined as a point, a line or an area (refer to Section 9.7), should be selected for ease of identification on the graphic, rather than in the legend or colour chart. This is particularly important in statistical presentations which use a single symbol repetitively, changing only one aspect, such as size, to represent a different level of data;
iii) A single graphic should not include more than three variables (refer to Section 9.5);
iv) Information that does not contribute directly to the message should be removed from the graphic;
v) The subject should be separated clearly from the background;
vi) A single colour graphic should contain a total amount of black ink that ranges from 5–10% of the figure area. This prevents the creation of designs that are too light or too dark;
vii) In general, cross hatch patterns and similar dominant designs should be avoided. Dominant patterns, common in commercially available transfer films, have a tendency to create visually disturbing and “vibrating” graphics (Figure 9.1);
viii) Dark or solid tones and intense colours should be reserved for small areas of the most significant information. Lighter tones and colours should be used for secondary information;
ix) In any statistical graphic the data may be designed for its perceived visual effect or for the ability to take accurate physical measurements from the graphic;
x) In the graphical representation of data the visual thresholds should be taken into consideration: perception, separation and differentiation (Figure 9.2).
Figure 9.1 The vibration effect obtained when graphic elements are in a 50% relation to the white background. (After International Cartographic Association, 1984)
Figure 9.2 The thresholds of graphic representation: (a) perception; (b) separation; (c) differentiation. (After International Cartographic Association, 1984)
Well designed graphics are easy to look at and easy to obtain information from. The following are some of the characteristics of “User Friendly” and “User Unfriendly” designs:
|User Friendly||User Unfriendly|
|Words are spelt out. Unusual and elaborate encoding avoided.||Abbreviations abound, i.e. the viewer must sort through text to decode abbreviations.|
|Words run from left to right, normal for occidental languages.||Words run vertically, especially on Y-axis. Words run in several different directions.|
|Short messages help decode data.||Graphic is cryptic, requires repeated references to text material.|
|Elaborately encoded patterns, shades, tones and colours are avoided. Labels are placed on the graphic itself. No legend required.||Obscure codings require going back and forth between legend and graphic.|
|Graphic attracts viewers and provokes curiosity.||Graphic is repellent, i.e. filled with redundant decoration and other embellishments.|
|Colours are chosen so that the colour-deficient and colourblind can make sense of the graphic (blue can usually be distinguished from other colours).||Design insensitive to colour deficient viewers. Red and green used for essential contrasts.|
|Type is clear, precise, modest. Lettering may be done by hand.||Type is heavy, complex and unclear. The effect is over- bearing.|
|Type is upper and lower case.||Type is all capitals.|
Modified from “The Visual Display of Quantitative Information” by Edward R. Tufte, 1983.
Graphics can be subdivided into three visual levels of organization: the qualitative, ordered and quantitative levels. These levels are not mutually exclusive and many graphics can use two or even all three of the visual levels available:
i) The qualitative level includes illustrations of two or more different concepts - two different species, colours, products, trades, etc. These can be displayed as either “this is similar to that” or “ this is different from that”;
ii) The ordered level is a systematic arrangement of elements into a well recognized ranking, such as temperature from hot to cold or size from small to large;
iii) The quantitative level illustrates measurable or counted information, “This is twice that”, “This is exactly what we measured here”, etc.
There are eight variables (dimensions that can be manipulated) which are available to the cartographer to differentiate aspects of the graphic: the two dimensions of the plane, plus variations in size, value, texture, colour, orientation and shape. These variations are used to create visible marks, signs or symbols on graphics and maps. An effective design will use more than one of these variables; several if not all will be incorporated into complex products. Used alone variations in value usually prove to be the most effective. The commonly used variations in orientation and shape, often with colour, are less efficient in communicating information. To emphasize a message, “graphic redundancy” may be employed by using two graphic variables simultaneously. For instance, in traffic signals the vital stop light is larger in size in addition to being differentiated by colour. A more detailed explanation of the eight variables is as follows:
This refers to the x and y dimensions along a sheet of paper, with information being plotted against one or both of these axes. This is the common graph and its many variants such as matrices.
Any graphic or map symbol can be created in a variety of sizes using length, area or volume measurements. These various sizes can be ranked or ordered. A cartographic convention dictates that bigger symbols represent larger or more important features, regardless of whether the symbol is a simple point, a line or a complex symbol. Many thematic products apply a scale to the sizes of symbols shown to indicate detailed quantitative information.
Value, also known as “lightness”, refers to the variation in intensity of light as seen by the eye. On black and white drawings, this ranges from white to black as shades of grey (Figure 9.3). As this is a measurement of the amount of light reflected from a surface the notion of value applies equally to coloured graphics and maps. The colours, or hues, yellow and yellow-green reflect more light than red, brown or purple and are thus seen as lighter in value. This variation and its control are important in most graphic applications.
This effect is also called grain (Figure 9.4). It is closely related to value and is a reflection of the size of repeated symbols. When a pattern is photographically enlarged or reduced the value (relationship of white to dark elements) remains unchanged but the texture is affected. varying the amount of ink in the pattern or changing the frequency of the elements will also change the value, a useful effect in many graphics.
Colour, more accurately termed hue, traditionally has been a much used graphic tool (refer to Section 10). The colour or hue refers to the dominant wavelength of light reflected from the graphic or map surface. Symbols produced in full-strength colours are seldom as obvious on the map or graphic as might be predicted from the colour samples used for their selection. Again it is the control of contrast which is vital; each colour must be visible against the actual background on which it will appear. Many dark colours are virtually indistinguishable from each other, particularly when overwhelmed by a strong background. Colour is a very useful tool but attention must be given to its application. If subdued tones or light tints are reserved for extensive areas, small areas of solid colour will become highly visible if superimposed on or adjacent to the light tints.
The angle of any distinct symbol (except fine dot or line patterns) can be rotated and that angle given a significance or change in meaning. This can be a useful design element, particularly when combined with another graphic variable. Used alone repeatedly, it can also produce graphics which are visually disturbing and extremely hard to understand. Extensive areas of fine ruled lines and similar patterns are usually not seen as aligned symbols but as a tone. The same symbol at various angles creates the same tone, hence visual confusion. Again it is contrast which is the desired dominant effect. For the same visual reasons ruled pattern lines should not be drawn through a line symbol because the boundary will disappear.
Ideally the shape of a symbol should clearly represent the feature being shown. This is seldom possible because so many maps and graphics are produced at small scales where the true size of the object would be close to that of a small point. Many symbols, therefore, must be exaggerated in size and simplified in design in order to be visible.
Figure 9.3 Variation in “lightness”. (After International Cartogarphic Association, 1984)
Figure 9.4 Variation in grain or texture. (After International Cartographic Association, 1984)
The most common error is to design a large number of small symbols, approaching points in size, which vary only in shape. Against most backgrounds these minor variations are not apparent to many users. Shape, therefore, should be combined with another element, most often a change in size.
Graphic or map data may be scaled in four levels of measurement. They are as follows, in increasing order of precision:
i) Nominal scaling: a division of data based on qualitative considerations, e.g., a wharf, a lighthouse;
ii) Ordinal scaling: a differentiation of data within a class on the basis of rank but without any numerical value, e.g., large and small ports;
iii) Interval scaling: a ranking of data in exact standard units with differences between classes expressed in multiples of that unit. The zero point is arbitrary as in the centigrade scale, where 20 C° is not twice as warm as 10°C, merely 10°C warmer;
iv) Ratio scaling: as with interval scaling classes are exactly defined but in contrast the zero point is absolute, e.g., weight, distance.
In order to simplify their design or selection, symbols can be subdivided into three different classes: point, line and area. Variations of these classifications are made possible by subtle increments of the factors previously described in graphic variables (refer to Section 9.5).
i) Point symbols are individual signs such as dots, triangles, small crosses, etc., that are used to represent positional data such as towns, reefs, sample locations, soundings;
ii) Line symbols are variations on the theme of a single line representing a wide variety of data. The information can be both linear or non linear. Commonly, lines of communication (roads, railways), boundaries, rivers, etc. are drawn in this manner, in addition to flow lines, contours, depths and many forms of volumetric data;
iii) Area symbols indicate that an area has a common characteristic, which is commonly shown by the use of a tone, colour or pattern throughout the designated area.
Statistical graphics may be subdivided into two categories (Figure 9.5):
i) graphics showing relationships between quantities;
ii) graphics indicating the portion of the whole formed by several component parts.
i) Line graph (Figure 9.6): The independent variable is normally plotted on the horizontal scale and the dependent variable on the vertical scale;
ii) Bar graph (Figure 9.7): The bar graph is similar to the line graph, but directs attention to the actual quantities, whereas the line graph emphasizes the rise and fall of the values. Bar graphs are often “turned on their side” with horizontal bars; this provides more space for names, values, etc;
iii) Circular graph (Polar chart, Clock graph) (Figure 9.8): The circular graph can be used to denote a series of values which relate to a recurrent state of affairs, a continuum;
iv) Logarithmic graph (Figure 9.9): This graph is used to indicate a rate of increase rather than the amount of increase. It will show equal rates of change by lines of equal slope, whatever the numerical basis of that change may be;
v) Scatter graph (Scatter diagram) (Figure 9.10): The scatter graph is used to investigate the relationship which exists between two variables occurring over a wide area. The relationship can be expressed mathematically by means of regression analysis, with a straight or curved line drawn through the points on the graph.
Several of these graphics are modifications of those already described:
i) Compound line or bar graph (Figure 9.11): This is used to subdivide the area beneath the line, or contained within the bar, into any number of components if the vertical line commences at zero;
ii) Divided circle or pie graph (Figure 9.12): This is the commonest statistical diagram. The total quantity concerned is represented by a circle which is divided into segments proportional in size to the components. Comparison can be made between variations in these components in two or more examples if a circle, subdivided in this way, is drawn for each of the total quantities. Circles varying in size proportionally to the total quantity which they represent can also be used;
Figure 9.5 Graphical techniques for portraying statistical information. (After G.C. Dickinson, 1973)
Figure 9.6 Line graphs showing herring landings from 1948-78 for two statistical districts. (After S.N. Messieh et al.)
Figure 9.7 Bar graph showing the potential catch per unit effort using gillnets by Statistical Area. (After D.G. Reddin and P.B. Short, 1981)
Figure 9.8 Circular graphs showing wind characteristics on a monthly basis. (After MARTEC Limited, 1982)
Figure 9.9 A normal line graphs (a); and its logarithmic equivalent (b). (After G.C. Dickinson, 1973)
Figure 9.10 Scatter graphs showing the relationship between observed and estimated year-class size of cod. (After P.F.Lett, 1980)
Figure 9.11 Compound bar graph showing value of Canadian exports of fishery products 1955-72. (After Canada, Department of the Environment, 1974)
Figure 9.12 Divided circle showing percentage of total catch (weight) by major group.
iii) Divided rectangle (Figure 9.13): The rectangle, whose area may be proportional to the total quantity, is used in a manner similar to the circle and can be subdivided into layers, each representing one of the components;
iv) Triangular graph (Figure 9.14): This graph may show three variables; in both interpretation and use it has strong affinities to the scatter graph. The graph consists of an equilateral triangle with sides 100 units long, each carrying a scale running from 0° to 100.
The important element of position in statistical maps forces the cartographer to work within rather finer limits than with statistical graphics. The space available for display of any detailed information depends not only on the overall size of the whole map, but on the area of the map within which the information can be placed and still be associated with the appropriate geographic area or feature. The statistical information may be shown by non-quantitative or quantitative techniques (Figure 9.5):
These maps indicate the places or areas where features of interest occur, without the need to differentiate according to size or importance (Figure 9.15). Their merit lies in their ability to summarize a situation.
There are three main types of statistical techniques used to show quantitative distributions according to size or importance: a series of points, given areas and a series of lines.
The following techniques are used to show quantities distributed at a series of points:
i) Repeated symbols (Figure 9.16): The symbols can be geometric, pictorial or descriptive in nature. The appeal of this quantitative method is its simplicity; quantities can easily be deduced by counting symbols and if the symbol used is representative in nature, the pictorial device makes its message easily understood;
ii) Proportional bars (Figure 9.17): The bars are simple to draw, flexible to arrange in congested areas and, because of the simple linear form, easy to estimate visually;
iii) Proportional circles (Figure 9.18): The area of the circle is proportional to the quantity represented. Since the area of the figure is proportional to the square of the radius, a symbol 100 times the amount of the other is only 10 times as large; hence, a great range of values can be represented if distinct visual steps in size are carefully chosen;
Figure 9.13 Divided rectangle showing percentage of total catch (weight) by major group and species composition within each major group.
Figure 9.14 Triangular graph showing the nomenclature of sediment types. (After F.P. Shepard, 1954)
Figure 9.15 Non-quantitative statistical map. (After Maritime Resource Management Service Inc., 1986)
Figure 9.16 Quantities distributed at a series of points using repeated symbols. (After Canada, Department of Fisheries and Oceans, 1981)
Figure 9.17 Quantities distributed at a series of points using proportional bars (simulated data).
Figure 9.18 Quantities distributed at a series of points using proportional circles (simulated data).
iv) Proportional spheres and cubes (Figure 9.19): By adding a third dimension the range of value that can be considered is increased tremendously. The value of these symbols is proportional to the cube of their radius or length of side; a symbol 10 times larger than another will represent a value 1,000 times greater. (N.B. the information contained in these graphics is extremely difficult to visualize and will create interpretation problems for many users);
v) A range of graduated symbols (Figure 9.20): Each symbol represents a specific group of values, the symbols increasing in size as the quantities they represent get larger. These are preferable to proportional circles for most uses.
This method of showing quantitative distribution is more common than any other type. The statistics simply indicate that within a given boundary line a certain number of features is to be found although the arrangement of features is not indicated. The following techniques are used to show quantities contained within given areas:
i) Dot maps (Figure 9.21): Although simple in principle, this technique raises design questions which need to be answered before mapping can begin:
a) How much or how many should each dot represent?
b) How big should each category of dot be drawn?
c) Should the dots overlap, coalesce or be otherwise differentiated in busy areas?
ii) Shading maps (Choropleth map) (Figure 9.22): Shading methods presuppose uniform distribution of the quantity throughout the given area which is often unjustified. Each shade will represent a different density per unit of area. If colours are available it is advisable to limit the number to two or three and build up variety by using tints of each colour;
iii) Proportional shading maps (Figure 9.23): This technique not only places a value within a range but actually represents it “true to scale”. Although these maps may be constructed to absolute statistical accuracy, they are usually inefficient at imparting their information. They often suffer from the common distracting defect of “visual vibration” caused by alternating black and white bands;
iv) Isoline maps (Figure 9.24): As with shading techniques, average densities are shown for each unit but this value is regarded as being typical of, rather than confined exactly to, the areas under consideration. It thus avoids the “unreal” effect which boundary lines produce on shading maps;
v) Repeated statistical graphics (Figure 9.25): Statistical graphics, previously described in Section 9.8, are used to illustrate variations in several factors throughout an area.
Figure 9.19 Quantities distributed at a series of points using proportional cubes (simulated data)
Figure 9.20 Quantities distributed at a series of points using graduated symbols. (After Nova Scotia Department of Development,1973)
Figure 9.21 Dot map showing unit area boundaries, and the geographical distribution and relative magnitude of scallop landings.(After G.S. Jamieson et al., 1981)
Figure 9.22 Shading map showing distribution of pelagic fish on the delta area of Burma. (After T. Stromme et al., 1981)
Figure 9.23 Proportional shading map. (After G.C. Dickson, 1973)
Figure 9.24 Isoline map showing abundance of herring larvae number under an area 10m² ) in Miramichi Bay, New Brunswick, Canada. (After S.N. Messieh et al., 1981)
Figure 9.25 Repeated statistical graphics showing lobster landings from 1947 to 1977 in selected counties of Nova Scotia, Prince Edward Island, and New Brunswick expressed as a percentage of the mean landings during that period. (After Canada, Department of Fisheries and Oceans, 1981)
The technique for showing quantities distributed along lines is commonly used in relation to traffic flows along route-ways of various kinds, e.g., line width is proportional to the quantity of traffic passing on the route (Figure 9.26). It can equally be applied to such topics as migratory routes, current flows, sediment transport, etc.
Figure 9.26 Quantities distributed along lines showing international fish product exports form Nova Scotia, Canada. (After Maritime Resource Management Service Inc., 1982)