Thematic data are typically represented as single layers of information stored as image files and containing discrete classes. Classes are simply categories of pixels which represent the same condition. An example of a thematic layer is a vegetation classification with discrete classes representing coniferous forest, deciduous forest, wetlands, agriculture, urban, and so forth.
A thematic layer is sometimes called a variable, because it represents one of many characteristics about the study area. Since thematic layers usually have only one band, they are usually displayed in pseudo color mode, where particular colors are often assigned to help visualize the information. For example, blues are usually used for water features, greens for healthy vegetation, and so forth.
See Image Display for information on pseudo color display.
Class Numbering Systems
As opposed to the data file values of continuous raster layers, which are generally multiband and statistically related, the data file values of thematic raster layers can have a nominal, ordinal, interval, or ratio relationship (Star and Estes, 1990).
- Nominal classes represent categories with no particular order. Usually, these are characteristics that are not associated with quantities (for example, soil type or political area).
- Ordinal classes are those that have a sequence, such as poor, good, better, and best. An ordinal class numbering system is often created from a nominal system, in which classes have been ranked by some criteria. In the case of the recreation department database used in the previous example, the final layer may rank the proposed park sites according to their overall suitability.
- Interval classes also have a natural sequence, but the distance between each value is meaningful as well. This numbering system might be used for temperature data.
- Ratio classes differ from interval classes only in that ratio classes have a natural zero point, such as rainfall amounts.
The variable being analyzed, and the way that it contributes to the final product, determines the class numbering system used in the thematic layers. Layers that have one numbering system can easily be recoded to a new system. This is discussed in detail under Recoding.
Thematic layers can be generated from remotely sensed data (for example, Landsat TM, SPOT) by using the ERDAS IMAGINE Image Interpretation, Classification, and Spatial Modeler tools. A frequent and popular application is the creation of land cover classification schemes through the use of both supervised (user-assisted) and unsupervised (automatic) pattern-recognition algorithms contained within ERDAS IMAGINE. The output is a single thematic layer that represents specific classes based on the approach selected.
See Classification for more information.
Vector Data Converted to Raster Format
Vector layers can be converted to raster format if the raster format is more appropriate for an application. Typical vector layers, such as communication lines, streams, boundaries, and other linear features, can easily be converted to raster format within ERDAS IMAGINE for further analysis. Spatial Modeler automatically converts vector layers to raster for processing.
Use Vector to Raster options to convert vector layers to raster format, or use the vector layers directly in Spatial Modeler.
Other sources of raster data are discussed in Raster and Vector Data Sources.
Both continuous and thematic layers include statistical information. Thematic layers contain the following information:
- histogram of the data values, which is the total number of pixels in each class
- list of class names that correspond to class values
- list of class values
- color table, stored as brightness values in red, green, and blue, which make up the colors of each class when the layer is displayed
For thematic data, these statistics are called attributes and may be accompanied by many other types of information, as described in Attributes.
Use Image Metadata option to generate or update statistics for image files.
See Raster Data for more information about the statistics stored with continuous layers.