Structures for Storing Geographic Data - Introducing Geographic Information Systems with ArcGIS - page 251

A floating-point raster has frequently has no VAT because, usually, almost all the values in the cells are
unique. Cell values would be numbers like 45.312 and 46.789. Thus, the number of records would come
close to the number of cells and the COUNT for most records would be “1,” since it is likely that very few
cells would contain identical numbers.

Image Rasters

Raster images come from scanning an area. The two primary types of area scanned are portions of
the Earth (scanned by satellites, piloted aircraft, and drones) and maps (scanned by various hardware
devices such as flatbed or drum scanners).

You saw the table of an image raster in your examination of the digital raster graphics file COLE_DRG.
That table looked partially something like Table 4-2.

The “value” here is simply a code for a color (or grayscale) with the intensity of red, green, and blue
shown by the floating-point numbers in the indicated columns.

TABLE 4-2

ObjectID Value Red Green Blue

0 0 0.00 0.00 0.00

3 3 0.79 0.00 0.08

4 4 0.51 0.25 0.14

12 12 0.80 0.64 0.55

Raster-Based Geographic Data Sets—Layout
in the Computer

The simplest storage of a raster band would be something like this:

66666666777777754444449999999

66666677777777777777444444999

66666777777777777744444445555

That is, the integers are stored in sequential locations in the memory of the computer or on a disk. In the
preceding case, the number of columns (i.e., the length of a row) is 29 and the number of rows is 3.

The nice thing about storing a raster band is that very little addressing needs to take place to know the
location of a raster cell in geographical space. If you know, for example (a) the location of the upper-left