Each grid value (or pixel) has a value normally indicating its brightness when displayed. However, the values can also have other meanings beyond just radiometry. A thematic raster layer might have a particular set of raster values, each indicating a particular crop type, or forest type. A raster layer might also be a continuous field relating to temperature, proximity to a fire station, or probability of erosion. The raster layers themselves don't indicate what the meaning of the raster values is. This is indicated by the user, the application, or in some cases metadata associated with the raster layer.
Within GeoGateway all raster layers in a particular database must have the same size. Size is normally reported as width in pixels, and height in lines. Furthermore, all raster layers are presumed to correspond to the same real world region, which will be discussed further in the georeferencing section.
Each raster layer has a datatype. Currently there are four supported GeoGateway raster data types:
1) 8-bit unsigned (8U): This is the most common datatype. Each pixel is normally stored as one byte of data on disk, and can have an integer value from 0 to 255; .
2) 16-bit signed (16S): Each pixel is normally stored in two bytes of data on disk, and can have an integer value from -32737 to 32736.
3) 16-bit unsigned (16U): Each pixel is normally stored as two bytes of data on disk, and can have an integer value from 0 to 65536.
4) 32-bit real (32R): Each pixel is normally stored as four bytes of data on disk, and can have an IEEE floating point value between approximately 3.4 x 10^38 and -3.4 x 10^38.
There is another kind of raster layer in GeoGateway called a bitmap, or graphic mask. In effect this is a 1-bit raster layer, but it is handled separately for historical reasons; they are covered separately in a later topic.
Where file formats supported by GeoGateway contain data not of these four datatypes, they are normally mapped to the closest datatype. In some cases this leads to a loss of precision. For instance: with a file format that supports double precision (64-bit) floating point values, the values will be mapped to 32R with the corresponding loss of precision. A data type like 4-bit integers in TIFF are mapped to 8U with no loss of precision.
The raster layers, or channels are numbered starting at one. For instance, a 24-bit TIFF file is represented within GeoGateway as a three channel raster file with channel 1 corresponding to the red, channel 2 to the green and channel 3 to the blue.
Pixels within a raster are addressed by their pixel/line location within the raster. The top left corner is the origin, with the pixel values increasing to the right (X axis), and line values increasing downwards (Y axis). The top left pixel is known as pixel (1,1), and appears at offset (0,0) from the origin.
Some calculations operate at a subpixel level. For these purposes the top left corner of the top left pixel would be at (0.0,0.0) and the bottom right corner of the top left pixel is at (1.0,1.0). The centre of the top left pixel is at (0.5,0.5). The centre of the bottom right pixel of a 1000x1000 image would be at (999.5,999.5).