Attributes are the (often) nongeographic, descriptive properties assigned to spatial entities. Attributes can be representative of any characteristic (e.g., physical, environmental, social, economic, etc.) of a spatial entity, and they can be assigned to any type of spatial data (e.g., points, lines, and areas for vector data; cells in a raster data set; and voxels or other volumetric spatial entities). Attribute information is used in making maps, distinguishing the characteristics of locations, and performing spatial analyses.
In the case of vector data, each feature in the data set is generally described using multiple attributes. Attributes are typically stored in attribute tables. Each row in the table contains information about individual geographic entities, and the columns in the table, usually called fields, contain information about attributes for all entities. Attributes in tables have defined characteristics, such as name, type, and length of the field. In a GIS, a field can contain only one type of data, and some data types, such as text, require additional specifications, such as length. In addition, it is possible for the GIS user to set whether or not there is an acceptable range of numeric values or a list of acceptable text entries for a field, as well as whether or not missing values are allowed; this is called a domain. Missing values are often indicated with an entry such as NULL or −9999, so that they can be identified quickly and removed from calculations.
Raster data can be either discrete (that is, they represent phenomena that have clear boundaries and attributes for qualitative categories, such as land use type or district) or continuous (that is, each cell has a unique floating point value for quantitative phenomena, such as elevation or precipitation). Continuous raster data have no attribute tables, but discrete raster data have a single attribute assigned to each cell in the raster that defines which class, group, or category the cell belongs to. Since there are frequently many cells in a raster that have the same value, the raster attribute table is structured differently than a vector attribute table. The raster attribute table assigns each row to a unique attribute value; a column contains the count of the number of cells with each value. This table may also have a column that provides a textual description of each of the unique attribute values (e.g., 1 is “urban,” and 2 is “rural”).
With both raster and vector data, the attribute tables can be related to external tables containing additional attributes. To create a relationship between a spatial attribute table (raster or vector) and an external attribute table, a common attribute field, called the key, must be contained in both tables. Most geographic attributes can be classified into one of four levels of measurement using Stevens’s scales of measurement: nominal, interval, ordinal, and ratio. The classification of attributes according to the level of measurement determines which mathematical operations, statistical methods, and types of visual representations are appropriate for the data. For instance, nominal attributes (i.e., categorical attributes such as urban and rural) are qualitative measures; therefore, it would be inappropriate to use mathematical operators for this type of attribute.