Elements of Cartographic Style

Maps are a vehicle that may transfer your ideas about a place into the mind of a map reader. This is a powerful yet delicate art. If you understand how a map communicates, you may be very effective in framing a discussion of critical aspects of a place around the ideas that you feel are important. Despite the complicated data and analysis that may underly it, the best map is one that communicates its specific message to the viewer with a minimum of effort on the viewer's part, and yet demonstrates that you have understood and portrayed the important details in the data in their true relation. This page covers some of the essential properties of maps that communicate effectively and with credibility:

Establish Trust
Be Concise
Use Intuitively Engaging Graphical Conventions

Topographic and Thematic Maps

The subject of evry map is a place. Topographic maps are designed especially to support a general exploration and discussion of the essential physical and cultural components of a place and its pertinent surroundings and their relationships with eachother. Thematic Maps maps are designed to communicate more abstract quantitative or qualitative observations of enitities or areas within and surrounding the subject. Thematic maps should always include a reference overlay that reveals the essential topographic framework of the place and its context.

The discussion that follows assumes a familiarity with the ideas presented in the lecture presentation: Good Scholarship with GIS Models and Fundamentals of Map Projections.

Elements Every Map Should Have

Every map is an exercise in selective emphasis and clarity. You are concerned with communicating key concepts and relationships of a n area of interest, its components, and the components of the surroundings. Data are chosen to represent the concepts; and the data are transformed for portrayal in planimetric scale, and with a graphic hierarchy that makes it intuitively easy for the reader to discover the key concepts and relationships that you intend to emphasize. The art of selecting, transforming, and portraying information on a map involves the delicate balance of anticipating and answering reasonable questions related to your subject, while not overwhelming your reader's attention with needless detail, or forcing the reader to work in order to figure out what you are saying.

Include a Title that indicates the purpose for the map. This is a matter of being concise. Don't make me guess what you are trying to communicate with this map.
Put your name, your institutional context, and the date on each map. E.g. Paul Cote, GSD6322 Assignment 1, February 11, 2008. A savvy map user expects a map to be biased, depending on the circumstances of its creation. Why should anyone invest much creidibility on an anonymous map?
Include a caption that explains the critical concepts and relationships you are trying to illustrate.
Label Key Elements on the Map. Certainly, any feature that you mention in your caption should be clearly portrayed and labeled on your map. Other reference features that might be part of a discussion of the place, ought to be labeled as well.
Cite the primary sources for your data and their dates. Am I supposed to believe that you actually measured all of these data yourself? If you don't know the name of the authority responsible for the content of the data, or the time period that the data are intended to represent, then say so.
Cite Projection Method and Case All maps below a scale of 1:500,000 should have planimetric scale properties -- that is a scale that is constant in all directions and portions of the map. Understanding this requires a knowledge of the projection method used to transform the data for portrayal. Therefore your choice of map projection method and case should be stated on your map.
Put a grapical scalebar on the map Most maps these days are intended to be viewed on computer screens or projected against a wall. In these cases, a scale expressed as a fraction, eg. One Inch to One Mile or 1:63,000, is almost guaranteed to be wrong. In all cases a map should include a graphic scale bar. Only include fractional scales if you never to share your map in any other way than printouts that you make.
Incorporate a Graphical Hierarchy The key concepts as discussed in the caption should be given emphasis with a bright color and bold lineweights and labels. Key relationships may be portrayed with diagramatic graphics. At a lesser level of emphasis you should provide a frameork of reference for named places and circulation. There may be a hierarchy of emphasis among reference elements, such as lineweights and colors to portray different grades of roads. When color portrayal is an option, the color white should be reserved for non-map areas, such as margins, and the background of legend and text boxes. Other aspects of graphic hierarchy are discussed in the sections on topgraphic and thematic mapping.
A Concise Legend, if Necessary The map legend should be reserved for making key distinctions that are important for understanding the points you are making in your caption. Not every symbol used on the map needs to be in the legend. When the symbology on the map is self-explanitory, or if the distinctions being symbolized are not an aspect of the key concepts being described, then the map symbols should speak for themselves. When legends are included, the headings and descriptions should always be in plain english, avoiding cryptic file names and attribut codes. More tips on legends are discussed in the section on thematic maps, below.

Small-Scale (broad area) Topography 1:100,000 to 1:10,000

Developing and communicating ideas about a place, you and your collaborators and clients will have predictable questions that should be addressed in a regional context map.

Questions Answered

Where is the area of interest within the network of places and geographic features that we may assume the audience is familiar?
What are the important routes to and through the area of interest?
What are the physical features, named places, and landmarks within and in the greater region that may be important for discussing the area of interest and the concepts you intend to emphasize.

Graphic Hierarchy

By regulating the brightness and lineweights in the portrayal of specific features, you help the reader to gather the ideas you are presenting, and to answer their own question in the proper order of emphasis.

Foreground:
- The boundary of your area of interest (labeled)
- Named places, and other features related to your argument (labeled)
Reference:
- Named places, large parks, water bodies,
- Transportation features involved with accessing your area of interest should be labeled. Other major roads and railroads should be shown using their own graphical hierarchy.
- Administrative boundaries of interest.
Background: transparency of these layers can be used to mix lots of information together whithout dominating the map. Note that in ArcMap, you can set a defalt background color of a dataframe.
- Parks and Preserves (green), Land Areas (buff), water areas (blue). YOu may be able to get away with adding a couple of other classes of land use: e.g. residential (buff), Commercial (pink) industrial (violet), and other (grey). But this amount of land use detail may be better portrayed on a separate thematic map (see notes on categorical maps, below.)
- Incidental transportation features: depending on the scale of your map, you will show a background of minor roads. Of course if at your scale, these all become colaesced together, then don't.
- Shaded relief. Note that shaded relief simulates the shading that would be seen if relief features were actually sticking out of the map and are illuminated by a light source overhead. This simulated shading should incorporate shades of grey. Beware that whatever other thematic colors you are using should not use shades of light and dark that may become indeciferable when displayed with shaded relief.

Example of a decent regional context map

Large Scale (Area of Interest) Topography 1:10,000 - 1:1,000

Topography at a closer scale should focus on the specific area of interest for your a design study, including those surrounding areas that directly impact or are directly impacted by the phenomena going on or proposed in your study.

Questions Answered

What defines your area of interest. Is it defined by actual barriers in the landscape, or distinctions that are discernable? or is it simply a line on the map?
What are the key places and features within and around your aoi?
Are there critical subdivisions of your area of interest? The answer to this question will depend on what your choice of key concepts and relationships. If these are complicated, you will want to show these on a separate thematic map.

Graphic Hierarchy

Foreground:
- The boundary of your area of interest
- Key places and features within and surrounding the AOI (labeled)
Reference:
- All roads, railways and pesestrian/bike easments should be portrayed and labeled.
- Civic places and parks (labeled)
- Places of interest to denizens of the place (labeled)
- Contours at an apropriate interval. Contours should be labeled
Background: Notes about background discussed regarding small scale topography also apply here. You may additionally consider adding these features if data avaliability and map scale permit:
- Building figure ground -- if data and scale permits.
- Aerial photographs are useful as background, but should be used with transparency in order to keep them from overwhelming the rest of the map.

Thematic Maps

Beyond an understanding of the current context of a place, many documents will include maps that portray data that helps to support some assertion that one may want to make about a place as it relates to other places (in terms of land use or demographics or some other theme.) These are known as thematic maps. Thematic maps symbolize features according to the value of their attributes. These attributes may be qualitative, or quantitative. In the case of quantitative maps, we make a distinction between attributes that represent raw quantities versus measures of intensity. The page, Critique of Data, Metadata and Referencing Systems for more discussion of referencing systems for qualitative data.

The Map is Not the Territory: It is very easy to make unsubstantiated claims with thematic maps. This is a bad thing to do, since it can damage your credibility. There are several traps that human beings are very tempted to fall into when making and discussing maps. The most tempting family of fallacies relates to data that are aggregated, either spatially or categorically. A census tract can be characterized as having a population density, but we know that in life that the population are spread unevenly in space (and in time.) Land use of a parcel may be referred to as "Commercial" but the aerial photograph may show that part of the parcel is lawn and part is parking, and the building on this parcel may have residential uses upstairs. Furthermore, the categorical referencing systems that we use to create discrete shades for choropleth maps (discussed below) or to designate a qualitive refereicning system that distiquishes "Industrial" from "Commercial" there are many potentially important distinctions that the data or the map simply ignores. The consequenses of this chunkiness of data may be inconsequential, but since we usually don't have any better information, it may be impossile to to tell. Just so your map readers know that you understand this, you should always speak litterally about what the data actually represent: observations made of particular classes of entities, given a pre-ordained referencing system with a stated precision and aggregate units. See the notes further down this page about the Modifiable Aerial Unit Problem and Ecological Inferences for more discussion of these fallacies. The page, Critique of Data and Metadata and Referencing Systems goes into this a little deeper.

Elements that Every Thematic Map Should Have

All of the requirements for maps, of portrayinbg a contextual framework, listed above, apply also to thematic maps. There are additional considerations that also apply when we are trying to portray other sorts of measurements and observations on out maps.

Contextual Framework Portraying data without a contextual reference overlay as discussed in the sections on topographic mapping is pointless.
Concise, evocative legend YOur thematic data should be recoategorized if necessary so that your readers are not challenged to keep track of more than 5 different classes. Seven, maximum.
Use plain terms in lagend headings and labels If you accept the software defaults for your legend labels and headings, people who understand maps will also understand that you simply don't care about communicating.
Try not to hide important information in arbitrarily broad categories The categories portrayed in the legend, whether qualitative or quanititative, should highlight distinctions that are useful.
Discuss the Aerial Precision of Mapping Units Whether the data are quantitative or qualitative, thematic data have a paricular granualarity. For example Census Data may be at a Block level or Tract. Land Use Data may only registere distinctions for patches of ground larger than a stated Minimum Mapping Unit (like 5 acres, or a 90 meter cell.)
Graphical Hierarcy the same ideas about graphical hierrchy that apply to topographic maps may also apply with thematic maps. This is especially true with regard to the foreground layer of key topographic features and a reference layers to provide context. You may decide to drop some of the labels used in your reference layer -- particularly when your map document includes separate maps for presenting the contextual framework. Typically, the thematic layer will be the background layer of the map but you may also use transparency and an aerial photo at large scales, or shaded relief at smaller (broader) scales. When mixing background layers with transparency you should be careful that whatever background layers you use -- particularly aerial photos and or shaded relief, to not make the key distinctions in your thematic layer more difficult to read.

Elements of Qualitative Thematic Maps

Be familiar with conventions for symbolizing land use. See Traditional Color Coding for Land Uses by Sanjay Jeer, AICP with Barry Bain, AICP.

Conventionnal Shades for Land Use

Land use data should always be portrayed with conventional shades as follows. A darker color value for each hue can be used to express a gradation of higher intensity development.

Residential: yellow -> orange
Commercial: pink -> Red
Openspace: Green
Water: Blue
Industrial: violet -> purple
Civic: grey

Quantitative Thematic Maps

Maps that portray quantitative measurements or summary statistics use tricks of graphics that cause the audience to visually weigh and compare aspects of places. Making effective quantitative maps and interpreting them requires an understanding the two major types of quantitative data: Intensive Statistics, versus Raw Counts; and how the intuitive computer of the eye/mind interprets symbol color intensity versus symbol size. intensive statistics (e.g. heat or concentration) versus extensive, count statistics (e.g. weights or counts).

The cartographer should also understand two major classes of symbols for portrayiung quantitative properties: Proportional symbols change their visual weight according to a quantitative property. These are apropriate for extensive statistics. Chorpleth maps portray data collection areas (such as counties, or census tracts) with color. Color is best used to represent intensive statistics such as percentages or densities. When using color this way, obseve how the darkness and intensity (or value) of the color is evaluated by the eye as a measure of intensity or concentration.

Whenever you include a map portraying a proportion, such as Percent of housing units that are rentals you should include a map that shows the density of the total -- e.g. Total housing units per acre. It is often the case that areas that are near the ends of the scale in terms of proportion are ones that have very little actual activity in them. When interpreting the relative proportions on the first map, the density map will help answer the predictable question:

Whenever your legend involves quantities of any type, your legend title or labels should explicitly state the units! When normalizing for density, please use an areial unit that has an evocative scale. Can you create a picture in your mind of 10,000 people in a Square Kilometer? What about 100 People in a Hectare? (two soccer fields.) Convert your units if you have to!

Handy Conversion Factors
You Have:	Acres	Hectares
Square Miles	* 640	* 259
Square Meters	/ 4,047	/ 10,000
0.001 Square Kilometers	* 4.047	* 10

YOu may want to check your work with this handy online area conversion calculator

If you want to portray several statistics at the same time, avoid the temptation to 'normalize' your data more than once. These compund fractions are very difficult to interpret. It is much better to provide several maps in easy-glance distance from eachother. See example of a PDF document with several maps bookmarked.

Avoid Tempting and Fallacious Traps of Interpretation

There is a famous saying, The Map is Not the Territory It may be easy to explore the world through data and maps, but if a person confuses the data for the actual things in the world, there are many very common and embarrassing traps!

Modifiable Aerial Unit Problem: Because the block groups and tracts of the census are neither uniform in size or density of population, nor are the individuals and households all the same, the interpreter of census maps should not put too much stock in the patterns seen in data that has been aggregated to arbitrary aerial units.
Consider that a tract of unusually high density may be almost vacant except for one corner or along an edge that happens to be a major road.
A very good discussion of this problem can be found in How to Lie with Maps by Mark Monmonier.
Once again, it is important to be explicit about your spatial units on your map and in map text. Whether your aerial units be Parcels, Tracts, Blockgroups or Blocks, you should show their boundaries and be clear in your legend.
The fallacy of Ecological Inference: Aggregated data are generalizations that may tell us something about the character of an area or its people as a group, but we should studiously avoid taking these characteristics and applying them to sub-groups or individuals within the aggregation.
Two maps that show that low income tracts tend also to be tracts with relatively lower educational attainment may support our intuitive notions of class, income and privilege, but when interpreting these maps we should be aware that the low-income people in each tract are not necessarily the poorly educated ones.
A look at mapped census data may reveal that block-groups with lower than average rents, tend also to have a larger proportion of new units. This correlation at an aggregate level should certainly not be interpreted as evidence that new housing units have lower rents.
A good discussion of ecological inference can be found in Ecological Inference and the Ecological Fallacy by David Freedman. Harvard's own Gary King has been working out ways to understand the relative power and weakness of ecological inferences, recognizing that, under different conditions, some of these inferences may be less fallacious than others. See: Geography, Statistics and Ecological Inferences

The best defense against temptation into this embarrassing logical trap is to always be clear about your units of analysis. In the two examples above, you will avoid fallacious nuances if you speak in terms of census blockgroups. The problem is when you cross the line and begin making inferences about the people living there.

These fallacious traps are so tempting that, when interpreting census maps, people make them all the time without any recognition of potential problems. For better or worse, these interpretations reveal that the analyst would likely delude him/herself (but not us!) using data without understanding the logical limits of data as a representation of the world.

In my opinion, it isn't wrong to use data in a potentially fallacious way to support an argument, so long as the analyst is explicitly clear about what the units of analysis are, and calls attention to the potential errors of interpretation that the reader should avoid. And this is a good place to remind cartographers that this concept of units that need to be explicitly identified includes not only the units of your thematic statistic, but also the units of spatial aggregation in your geometry.