Structures for Storing Geographic Data - Introducing Geographic Information Systems with ArcGIS - page 250

Raster-Based Geographic Data Sets—Logical Construction
In GIS a raster is a set of equally sized squares 11 that cover a rectangular surface. Rasters are used in two
basic ways in GIS:

As rasters that store attribute information about the area covered by the square (such as elevation, or
soil type), in which case individual squares are called cells.

❏

As images such as orthophotoquads, where the squares are picture elements (referred to as pixels),
containing values that prescribe the intensities of visual colors (e.g., red, green, blue) as well as IR
( infrared ), UV (ultra violet) , and thermal spectral elements.

❏

Rasters (Grids)
The raster data model, which can be very useful in spatial analysis, 12 comes in two flavors: those in
which the cells contain integer numbers and those in which the cells contain floating-point numbers.
In each case, each raster cell may contain a single number. If that number is in integer form, then the
raster represents categorical or discrete data. Each different integer represents a type of object or a
condition. If the numbers in cells are floating point—that is, they contain numbers that may have decimal
fractions—then the raster may represent continuous data such as an elevation surface over the area of
interest. Integer rasters usually 13 have associated with them a value attribute table (VAT). Floating-point
rasters usually do not have a VAT.

Each record in the VAT of an integer raster has a minimum of three fields: an ID field, a Value field, and a
Count field that indicates the number of cells in the data set with the given value. For example, Table 4-1
shows a part of a Kentucky land use dataset you examined earlier.

You may recall that value “5” indicated water and “14” indicated transportation, communication, and
utilities. A set of cells that contained the same number is called a “zone,” whether the cells are adjacent
to one another or unconnected. The Count column is useful because, when it is multiplied by the area
covered by a single cell, the total area of the zone is obtained.

TABLE 4-1

ObjectID Value Count

1 5 2,387,059

5 14 2,606,086

Each cell in the raster can contain a number (value) or can contain an indicator that says no number is
assigned to the cell: NODATA.

11 Or almost square—some rasters are based on fractions of a geographic degree.
12 Use of grid rasters will be discussed in detail in Chapter 8. Here we want only to acquaint you with the data
structure and how it is stored in the computer.
13 If there is an immense number of different integers, a VAT might not be created, due to storage limitations.