Geographic Data

Purpose of lecture:

Structure of lecture (horrors!): Big distinction | The geometry part | The attribute part | Scale

The BIG distinction

Cartographers might argue about which distinction is the biggest one, but in terms of viewpoints, there is a real division between thinking about the world as discrete things and as smooth continuous distributions. The first leads to making divisions, distinguishing OBJECTS. The second leads to an underlying model of SURFACES.

Objects

Can also call them "features", places, spatial units, counties, departments, provinces or whatever...

In the extreme form:

• sharp boundaries
• objects are just "different" from each other
• have a name
• can have whole tables of numbers attached to them (indirectly)

Surfaces

In the extreme form:

• mathematically pure
• smooth variation
• each point has a different value
• values on a real number line

Examples (why all distinctions have limits):

Objects:

• Soviet Industry: each city has a symbol for its principal product(s), size for output (no industry in between)
• Highways: each has a label, connects between specific destinations (could have traffic volume or expenditure on snow clearance attached)
• Population density of USA by county: here a concept that is a surface is shown through a set of discrete objects (the counties) - more on the borderlands...

Surfaces:

• the classic is always topographic relief (elevation); peaks, ridges, watersheds, slopes, valleys are ways of giving names to features on a surface (people like to name objects, even on a continuous smoothness.)
• Rainfall (averaged over many years, observed at discrete sampling points, interpolated to fill in and pretend you have a surface)
• Magnetic Field: [This is what the compass reacts to.] There is an direction (horizontal and vertical - dip) at every point, plus an intensity. This field changes slowly as the mantel moves, shifting the Magnetic Pole.
• Concentration of Carbon Monoxide in the atmosphere over LA after ten windless hot summer days (technically this is a volume, but you can get a surface of the concentration at various altitudes...)

Another big distinction structurally:

Geographic information is about Time, Space and Attribute.

Except for a few rare cases, maps emphasize the spatial distribution of the attribute, and leave the time part to be done with multiple maps (movies, etc.)

Example: Exxon Valdez event: contours of concentrations of oil on the water at twelve hour intervals

Geometric distinctions:

Objects can be assigned to dimensional classes

Surfaces don't really fit. They cover a two-dimensional region, extend into a third dimension... They have plenty of INTERNAL geometry.

Most cartographers would consider a land use map or a soil map to be a set of distinct area objects, but these collections of objects are intended to cover the whole region exhaustively. There is something surface-like that isn't there if you have isolated objects that are surrounded by the void. The boundaries on a soil map or a land use map are there to show a transition, and to keep the collection exhaustive... (Yes, all distinctions can oversimplify.)

Attribute distinctions

The "attribute" is the value of interest, the "theme", it covers any kind of science, commerce, humanity. Cartographers will make maps for anybody.

The cartographer doesn't look at the "content" of the attribute, but a few of its fundamental properties:

Scale

Whenever you want to make trouble with a geographer, you say, "yes, but at another scale..."

At a continental scale the categories "Desert" and "Forest" work to distinguish Sonora from the Olympics, but get down at the scale of meters, and the forest turns into a bunch of discrete objects... Both levels of scale are valid for some purposes.

In a given mountain valley you might be able to order the vegetation by hardiness (tolerance to cold, snow, etc.) but if you include a larger and larger region, you wouldn't be able to order everything completely. They would become just categories: cranberries and bananas are more different than apples and pears...

Rent is a discrete phenomenon, you pay $kkk for this object for this period; however, the regional trends in rental costs may approximate a smooth curve, adjusting for the gradients in desirability.

Version of 6 December 1999