Thematic Maps

Objectives of lecture

• Components of thematic maps
• Preparation for exercise
• Connection to GIS

Terminology

Muehrcke calls them "attribute maps" or "statistical" maps. It seems that any statisitic is a potential attribute, so the distinction isn't much help.

Thematic maps are designed to depict the spatial distribution of a particular attribute ("theme"). This distinction was designed to separate it from the maps more commonly used for navigation of various forms, BUT each component of a topographic map, a nautical chart, or a road map could be treated as a thematic map too.

From a design standpoint, all maps are composites of thematic maps.

Ways to organize thematic mapping

• Geometry of objects: point, line, area, surface, volume
• Level of measurement: nominal, ordinal, interval, ratio (raw and derived), counts (and so on)
• Graphic variables: size, shape, color (hue, value, intensity), pattern, etc.

These were introduced in earlier lectures (3,4,5)

Packaged solutions: Established techniques

How does each work? Read the legend!

Proportional Symbol maps

Use SIZE of a (usually) regular SHAPE to portray continuous variation of a POSITIVE ratio/ count value (raw data). Most common: proportional circles drawn for industrial production in cities, or population. [Think: would it work for negative numbers or an arbitrary zero?]

Coded Symbols

Use COLOR to portray points or lines (or even areas) that are in a particular class. The colors assigned to different classes can be distinctive or organized to show an order (or nominal or ordinal respectively). Lines (in particular) may also be SIZED (width) for additional (ordinal distinctions). [There is not a particular common name for this grouping...]

Area data that does NOT come from arbitrary zones: vegetation classes, land use, etc. These are "categorical coverages". The legend simply tells the classification for each category. May have to group classes together to make the map readable...

Choropleth

Area data (arbitrary collection zones), each area has an attribute. Primarly graphic variables: Color (particularly dark-light in black and white reproduction), possible use of pattern in moderation. If nominal or ordinal, it might be depicted directly. If continuous, it is CLASSED into ordinal classes. Lots of classification methods (equal intervals, quantiles, "natural breaks").

One big decision: How many classes? Graphic message works better with relatively few (5-7?). This DOES involve substantial loss of the message in the original distribution.

Due to unequal sized areas, technique works better with "derived" ratios (something that has adjusted for the potential differences in the "size" of the relevant areas)

Examples of constructing a derived measure:

• Property distributed over the area (? forestry/ agriculture): Divide by area DENSITY
• Property distributed among people: Divide by population PER CAPITA
• ... Divide by Gross Domestic Product, Divide by specific population groups, etc.

Classification methods

• Equal interval: Divide attribute range into equal sized distance along the number scale
• Quantiles: Place an equal number of objects into each class (then report the break points)
• "Natural Breaks": lots of ways to interpret... Gaps in distribution; minimize within-group variance

Relationship to GIS

Concept of "layers": organized groups of objects with a similar set of attributes (data table with same columns). NOT a computer-derived idea, actually dates from photographic approach (mid-twentieth century).

3-D metaphor: stacked themes, transparency, order of precedence... (ArcView interface as one example).

Version of 6 February 2000