It is very often convenient with matrices based on zones (of which there can be very many) to represent them using “aggregated” spatial areas. For example, with a national model one might wish to combine all zones within a county into a single spatial entity in order to obtain matrices where each cell corresponds to county-to-county movements.
Most commonly aggregation is based on geographically self-contained areas for which several standard terms may be applied. Thus we use “groups” as the most general terms but “sectors”, “traffic boroughs”, and ”districts” are commonly used as well with certain general assumptions as to their size. Thus sectors are generally very large – and few in number – whereas at the other extreme districts may contain, say, only 4 or 5 zones each.
Very often the different aggregates may sit inside one another, like Russian dolls, although this is not essential. Thus a group of adjacent zones might be aggregated into districts, adjacent districts into boroughs and adjacent boroughs into sectors.
However aggregates of zones need not be entirely spatial; one might also wish to define groups based on common characteristics, e.g., all zones which are predominantly industrial, all high household income zones, etc. etc.
The following sections described the properties generally ascribed to groups, sectors etc. followed by a discussion of how the “mapping” from one level to another is set within SATURN.