Available with Network Analyst license.
A service-area index generally provides improvements to a service-area analysis by reducing processing time and improving the quality of the output polygons.
Processing time
Generating service-area polygons on a network dataset that has an index can be much faster than one without an index. In many cases it is 10 to 14 times faster. But an index reduces processing time only when the service-area analysis you configure meets the following criteria:
- Hierarchy is not used. (See the Hierarchy section below.)
- The output polygons are generalized. (Detailed polygons cannot use an index.)
- Lines are not generated.
Building a network dataset that has an index takes longer than it would otherwise; therefore, if you generate service areas more often than you build the network, including the index is a worthwhile option.
Hierarchy
When generating a large service area, using a hierarchy instead of an index tends to be faster. What constitutes a large service area depends on the density, or how many edges are in the network dataset, as well as other factors. In general though, a large service area for a road network is greater than or equal to a 60-minute drive time or 60-mile (97-kilometer) driving distance.
Hierarchical service areas are primarily measured along higher-order roads, like highways, and their output polygons reflect that. If polygon quality is more important than processing time, you should choose to generate the service areas using the index.
Polygon appearance
Service areas measure the reach of a facility along the network, so the most accurate output is line features, which follow along the network edges that are within reach. Creating polygons from the traversed edges is somewhat subjective because the network analysis doesn't measure off the network. The following series of graphics show how output polygons differ.
Hierarchical polygons—Measurements largely ignore lower-order roads and focus instead on higher-order roads. Some of the local roads covered by the polygon cannot truly be reached.
Generalized polygons (no index)—These polygons are more detailed than hierarchal ones but less detailed than the generalized polygons created on a network with an index. Generalized polygons without using an index takes longer to generate than both hierarchical polygons and generalized polygons that use an index.
Generalized polygons using an index—The processing time to generate these polygons is shorter than the ones generated without an index; moreover, these polygons have somewhat more detail. The quality improvements are especially noticeable on curved roads—the polygons follow the curves better—and near the boundaries, where the index prevents the polygon from inadvertently overlapping roads that couldn't be reached.
Detailed polygons—The obvious difference between detailed polygons and the others is that they contain holes where network edges exist but couldn't be reached due to a long, circuitous access road or, more often, due to restrictions that prohibit travel on those edges. For example, enabling a restriction to prohibit travel on private roads would cause the resultant polygon to have holes around private roads.
Trimming
The polygon trim settings don't affect the processing time of the analysis when it is referencing the service-area index; however, your trim distance is always rounded up to the nearest 25th meter. If you set your trim distance to 40 meters, for instance, the actual trim distance will be 50 meters.