GIS and Cartographic Modeling

Introduction

Now take the map of surface water. Place a piece of tracing paper over it and draw rings around all lakes and streams at a distance of 200 meters. If you will then shade the area within those rings, this tracing can be combined with the tracings of steep slopes, sensitive views, high land costs, and so on. To do so, align the sheets on top of one another and place the whole stack on a light table or up against a window. The lightest areas will be those that can best accommodate the proposed land development.

— Anon.

Years ago, this sort of advice could often be found in how-to publications on environmental planning. It refers to a process in which geographic data are organized and manipulated in the form of single-factor maps called overlays. In the typical case, a database initially included overlays of site characteristics relating to topography, land use, demographics, and so on. Derived either from direct observation or from sources such as published maps or aerial photographs, these were registered with respect to a common cartographic frame of reference. Existing overlays could then be retrieved and graphically transformed in order to generate new overlays of additional site characteristics. The process is called overlay mapping.

Although overlay mapping methods have been employed since early in the twentieth century, it was only in the 1960s that they began to gain wide acceptance. Since then, much of the interest in these methods has been associated with the evolution of geographic information systems. Indeed, GIS technology has presented a great opportunity to refine and extend traditional overlay mapping methods.

The field of geographic data processing is rooted in geography, computing, and application areas ranging from the natural and social sciences to urban planning and environmental management. This field has grown steadily since the 1960s, and it continues to expand in terms of the number of practitioners involved, the range of applications addressed, and the sophistication of tasks performed. The field has now developed to a point where fascination with tools has matured into greater concern for the ways in which these tools are put to use.

The cartographic modeling approach attempts to generalize and standardize the analytic and synthetic capabilities of geographic information systems. It does so by decomposing data, data-processing tasks, and data-processing control notation into elementary components that can then be recomposed with relative ease and great flexibility.

To understand the nature of this approach, consider a simple example. Suppose we are interested in mapping levels of connectivity among the eight ponds depicted in figure I-1. For any given location, we would like to determine the length of the shortest path through that location that connects the two nearest ponds.

Figure I-1 A map of ponds. EightPonds is a map on which each of eight ponds within a particular geographic area is distinguished from its surroundings and from the seven other ponds by name, number, and graphic symbolism.

To do so, we start by using the EightPonds map to generate the new map presented in figure I-2, a map entitled DistanceToNearestPond. The processing required to do this would be specified as follows:

AllPonds

= LocalClassification

of EightPonds setting 1 … 8 to 0

DistancetoNearestPond

= FocalProximity

of AllPonds

Figure I-2 A map of distance to the nearest of a group of ponds. DistanceToNearestPond is a map indicating each location’s distance to the nearest of the ponds shown in EightPonds (figure I-1).

In figure I-3 is a similar map entitled DistanceToBrownsPond. This map was also generated by way of an operation called FocalProximity, but only after a LocalClassification operation had been used to isolate a particular pond. Those two steps would be specified as follows:

BrownsPond = LocalClassification

of EightPonds

setting 1 … 8 to −0

setting Brown’s Pond to 0

DistanceToBrownsPond

= FocalProximity of BrownsPond

Figure I-3 A map of distance to a particular pond. DistanceToBrownsPond is a map indicating each location’s distance to one of the ponds shown in EightPonds (figure I-1).

Now suppose the technique used to generate DistanceToBrownsPond had also been used to generate seven more maps, each indicating distance to a different pond. If so, then another way to generate DistanceToNearestPond would be to set each location to the lowest of its eight pond-distance values. If those seven new maps were to be entitled 1stDistance, 2ndDistance, 4thDistance, 5thDistance, 6thDistance, 7thDistance, and 8thDistance, then this would be specified as

DistanceToNearestPond

= LocalMinimum

of DistanceToBrownsPond

and 1stDistance and 2ndDistance

and 4thDistance and 5thDistance

and 6thDistance and 7thDistance

and 8thDistance

If a similar technique were used to set each location to the second lowest of its eight pond-distance values, the result would be a map like DistanceTo2ndNearestPond, shown in figure I-4.

Figure I-4 A map of distance to the second nearest of a group of ponds. DistanceTo2ndNearestPond is a map indicating each location’s distance to the second nearest of the ponds shown in EightPonds (figure I-1).

Given DistanceToNearestPond and DistanceTo2ndNearestPond, we now know each location’s distance to the two nearest ponds. All that remains is to calculate the sum of those two values, and this can be done as follows to generate the map shown in figure I-5:

DistanceToNearestTwoPonds

= LocalCalculation

of DistanceToNearestPond

+ DistanceTo2ndNearestPond

The concepts and methods underlying this type of process are introduced in the following chapters. The text starts slowly, proceeds methodically, and ultimately approaches territory that has yet to be fully explored.

In cartographic modeling, as in any game, you start by becoming familiar with the necessary equipment. Next, you try to understand the basic rules of play. Only then can playing skills and strategies be developed, often through years of observation and practice. In light of this, the text is organized in three parts. Respectively, they cover

  fundamental conventions,

  elementary capabilities, and

  illustrative techniques.

Figure I-5 A map of distance to the nearest two ponds. DistanceToTheNearestTwoPonds is a map indicating each location’s distance to the nearest two of the ponds shown in EightPonds (figure I-1).

Part I

Cartographic modeling conventions

Cartographic modeling methods can be implemented using any of a number of geographic information systems. While each has its own conventions, those presented here are not specific to any one GIS in particular. Quite the contrary, they are intended to relate to as wide a range of computing environments as possible. These conventions are introduced in three chapters that deal, respectively, with

  data,

  data processing, and

  data-processing control.

1-1 Cartographic models

1-2 Map layers

1-3 Titles

1-4 Metadata

1-5 Zones

1-6 Labels

1-7 Values

Null value

Ratio, interval, ordinal, and nominal values

Linear and cyclical values

1-8 Locations

Locations in columns, rows, and neighborhoods

Using locations to represent geographic space

Using locations to represent punctual characteristics

Using locations to represent lineal characteristics

Using locations to represent areal characteristics

Using locations to represent surficial characteristics

Locational precision and accuracy

Locations on multiple layers

1-9 Coordinates

Coordinate encoding

Raster and vector encoding schemes

1-10 Questions

Chapter 1

Data

Data are simply recorded facts. In the case of geographic data, these are facts pertaining (or like those pertaining) to locations on or near the surface of the earth.

Geographic data can be stored for digital processing in ways that can vary considerably and be of enormous practical consequence. In terms of the more general concepts and methods of cartographic modeling, however, specifics relating to data storage need not affect processing conventions.

In order to relate to as many actual computing systems as possible, we will establish conventions in terms of a highly generalized data construct. This construct does not dictate the specific manner in which data are actually stored, but it does indicate the general scheme by which data are organized (or appear to be organized) from the perspective of a typical user. The construct is intended to provide a common frame of reference that can easily be translated both to and from a variety of storage formats. Its major components are

  cartographic models,

  map layers,

  titles,

  metadata,

  zones,

  labels,

  values,

  locations, and

  coordinates.

A diagram of this data construct is presented in figure 1-1. Note that it is organized as a hierarchy in which certain components are composed of others. While several of these components might seem familiar, be aware that each is specially defined for use in the present context. Be aware too that other familiar terms might also have special meaning. Each of these special terms is presented in boldface when initially defined.

Figure 1-1 A general way of organizing geographic data. Geographic data can be organized as a hierarchy of component parts that correspond to familiar elements in traditional cartography.

1-1 Cartographic models

At the base of this hierarchy is the cartographic model. A cartographic model can be envisioned as a bound collection of maps much like an atlas. As suggested in figure 1-2, however, it is a collection organized in such a way that all of its data pertain to a common geographic locality. This locality is then referred to as the model’s study area.

It is important to note that a cartographic model conveys information about its study area in both explicit and implicit form. The data within such a model can explicitly record any number of facts pertaining to that study area.

Figure 1-2 Cartographic models. A cartographic model in its simplest form is a set of mapped data registered with respect to a common study area.

In the relationships among those data and the meanings attributed to them, however, there will also be additional information that is never explicitly recorded — at least not at first. Ultimately, the role of cartographic modeling is to convert such implicit information into explicit form.

The examples and questions presented throughout this text all relate to a particular cartographic model representing the study area depicted in figure 1-3. It is an area of some 13 square kilometers resembling an actual site located approximately 100 kilometers west of Boston in the village of Petersham (pronounced Petersham), Massachusetts. The study area takes its name from one of the more prominent bodies of water (and best swimming holes) in town, Brown’s Pond.

A cartographic model can include data indicating the size, geodetic position, history, or other distinguishing attributes of its study area. A model might also include data on how, when, or by whom it was created, how it is organized, and so on. While such information can be important, it need not conform to any particular conventions for cartographic modeling.

For our purposes, the essential components of a cartographic model exist as a set of map layers.

Figure 1-3 The Brown’s Pond study area. A conventional map depicts the site in central Massachusetts that is used throughout this text to illustrate cartographic modeling conventions, capabilities, and techniques.

1-2 Map layers

A map layer (or simply, a layer) is much like a conventional map, a flat drawing indicating the identity, form, and position of selected conditions occurring within a bounded geographic area. In this context, a condition is a geographic object or occurrence such as a building or a particular height above sea level.

On a conventional map like the one shown in figure 1-3, a given location can be characterized in terms of two or more conditions or no conditions at all. As illustrated in figure 1-4, however, each location on a map layer is characterized in terms of no more and no less than a single condition. In this way, map layers are similar to what are variously called themes, overlays, coverages, and maps as well as layers in cartography; images or picture functions in image processing; and variables in statistics.

A typical cartographic model might include several dozen layers, each depicting its study area in terms of a different characteristic, each a set of conditions representing variations on a common theme. For example, one of the layers in a cartographic model might represent a characteristic such as topographic slope, while a second describes soil types, a third shows buildings, a fourth gives land cost, and so on. Significantly, characteristics that seem more abstract or interpretive, such as distance to the nearest road or suitability for land development, can also be included as map layers in a cartographic model.

The model of the Brown’s Pond study area initially includes just four layers. These describe the area’s topographic elevation, its surface water, its forest vegetation, and its pattern of development.

The essential components of a map layer are its

  title,

  metadata, and

  zone(s).

Figure 1-4 A map layer. A map layer is a set of data describing the spatial variation in one characteristic of a geographic study area. Here, a layer depicting the selected hydrological conditions (center) is shown with other layers (above) in relation to a diagram of the landscape they represent (below).

1-3 Titles

The title of a layer is its written name. It is an unbroken sequence of letters, numerals, and/or symbols used to represent the layer in various forms of communication. Hydrology, for example, is the title of the layer presented in figure 1-5.

Titles are specified when layers are initially created, and they can be selected arbitrarily as long as no two layers in the same cartographic model have the same title. It is common (and recommended) practice to select titles that clearly identify the data they represent. Typical examples include SoilType, Version-1.2, and AverageIncomePerCounty.

The titles of the base layers in the Brown’s Pond model are Elevation, Hydrology, Vegetation, and Development.

Figure 1-5 A map layer’s title, resolution, and orientation. The title of a layer is its assigned name. A layer’s resolution is a number indicating the relationship between its cartographic dimensions and the geographic distances they represent. Orientation is a number indicating the relationship between a layer’s cartographic and geographic directions. As highlighted here, a layer entitled Hydrology has been encoded at a resolution of 4 (meters) and an orientation of 45 (degrees).

1-4 Metadata

A map layer might include a set of metadata describing its subject matter, source, date, reliability, and so on. Importantly, a layer’s metadata will almost certainly include parameters describing how its cartographic depiction relates to geographic space. Two such parameters of particular importance for our purposes are the layer’s

  resolution and

  orientation.

The resolution of a layer is a number describing the relationship between distance as measured on the ground and distance as measured on paper. To that extent, it is much like a conventional cartographic scale. Unlike a conventional scale, however, this number does not relate to the physical size of any particular graphic image. Instead, it specifies the number of feet, meters, or other units of geographic distance that are to be associated with the elemental unit of spatial observation. This unit has yet to be described, but it is called a location and, as represented by each of the small circles in the lower right corner of figure 1-5, it is the smallest standard unit of space for which data can be recorded. Each of the layers in the Brown’s Pond model is at a resolution of 4 meters.

The orientation of a layer is a number much like its resolution. As illustrated in figure 1-5, however, orientation describes the relationship between geographic and cartographic directions rather than distances. It indicates the number of clockwise degrees from the direction depicted as true north to the direction facing the upper edge of the cartographic plane. The Brown’s Pond layers are oriented at 45 degrees as indicated by a north arrow that points toward the upper left.

1-5 Zones

The final major component of a map layer is a set of one or more zones. In common usage, a zone is defined as a geographic area exhibiting some particular quality that distinguishes it from other geographic areas. The same definition can be applied to our special use of the term. Here, a zone is the set of data pertaining to a specific geographic condition. The cartographic form of a zone can be large or small and in one piece or in a number of disconnected fragments. In figure 1-6, the Pond zone includes eight individual ponds.

Unlike the variations in theme depicted on a conventional map, the zones of a given map layer must be both all-inclusive and mutually exclusive in their spatial coverage. They must account for every part of a study area but must never overlap. In these terms, zones are similar to what are variously called categories, classes, or regions in other contexts.

The number of zones in a layer can vary considerably. A map layer indicating political election results, for example, usually has no more than