A GDB-supported file can have any number of raster-image layers. A raster layer, also referred to as a channel or a band, consists of a rectangular grid (or array) of pixels. Each pixel has associated digital numbers indicating its display brightness. Digital values can also represent attributes other than radiometry.
A thematic raster layer can have a set of values, representing a crop type or a forest type, corresponding to the ground truth for a scene or an image. Raster layers can also contain values for continuous fields of data relating to temperature, proximity to fire stations, or probability of erosion. You can assign raster-layer values manually, automatically, or from the metadata associated with the raster layer.
All raster layers have a data type and must have the same width, measured in pixels, and height, measured in lines. Layers are presumed to correspond to the same real-world region.
8-bit unsigned (8U)
Each pixel is stored as one byte of data, and can have an integer value from 0 to 255.
16-bit signed (16S)
Each pixel is stored in two bytes of data, and can have an integer value from -32768 to 32737.
16-bit unsigned (16U)
Each pixel is stored as two bytes of data, and can have an integer value from 0 to 65535.
32-bit real (32R)
Each pixel is stored as four bytes of data, and can have an IEEE floating-point value between -3.4E38 and 3.4E38.
Complex 16-bit signed (C16S)
Each pixel is stored as four bytes of data represented by a real and an imaginary pair of 16-bit signed integers.
Complex 32-bit floating point (C32R)
Each pixel is stored as eight bytes of data represented by a real and an imaginary pair of 32-bit floating-point values.
Bitmaps can also be used as 1-bit raster layers. When bitmaps are used as graphic masks, they are stored as separate layers.
GDB-supported file formats are mapped to one of the four data types they resemble most closely. In some cases, mapping can lead to a loss of precision. For example, a file format that supports double-precision (64-bit) floating-point values is mapped to 32R with a corresponding loss of precision. A data type with 4-bit integers in TIFF files is mapped to 8-bit unsigned format with no loss of precision.
Raster pixels are addressed as pixel-and-line locations in the pixel array. The top-left corner is addressed (1,1) and appears at offset (0,0) from the origin. Pixel values increase to the right, along the x-axis, and line values increase downwards along the y-axis.
Some calculations require operations at a finer level. Pixel values are divided to allow for more data. For example, the upper-left corner of the upper-left pixel is at 0.0 and 0.0, and the lower-right corner of the upper-left pixel is at 1.0 and 1.0. The center of the upper-left pixel is at 0.5 and 0.5. The center of the lower-right pixel of a 1000 x 1000 image is at 999.5 and 999.5, respectively.
© PCI Geomatics Enterprises, Inc.®, 2026. All rights reserved.