The goal of this project was to introduce the concepts of Linear Referencing and to illustrate its use when analyzing a long linear feature, such as a highway. This example uses Linear Referencing to analyze traffic accidents along a stretch of highway with varying characteristics.
This project was part of the requirements for GIS 520, Advanced Spatial Analytics. The data was provided with the assignment and consisted of a linear shapefile of the Pittsburgh area, and .dbf files containing locations of traffic accidents on various roadways. The exercise examined the different road segments and locations of accident reports to determine any correlation between the accidents and the conditions surrounding their location.
Linear Features such as road, highways, rivers, etc. often span many miles and extend through various administrative jurisdictions (counties, cities, etc.). It is common that segments of these linear features will be constructed, measured, maintained, etc at different times and will therefore, have different attribute values associated with the feature over its full length.
Linear Referencing is a method that allows these attributes to be referenced from a single point or origin on the feature, instead of requiring separate x-y coordinate values for each location of interest along the feature. For example, the origin of an interstate highway can be referenced from the county boundary. The attribute values associated with the highway will likely change values throughout the country (speed limit, surface type, town limites, etc). Each of these unique attributes can be referenced by a single vector distance from the origin instead of a coordinate value. This allows multiple sets of attribute values to be associated with a single Linear Feature.
Linear Analysis is an important concept to understand when analyzing linear features that share common attributes, but have different values. By dividing these features ("Routes") into smaller segments, it is possible to accurately represent the occurrence of the phenomena in question. For example, a length of road may have various speed limits, pavement types, number of lanes, etc over the given length. Using linear analysis to study the rate of accidents on the segments that comprise the total length of the study, it would be possible to determine any statistical correlation between those accidents and the different attribute values (speed, pavement type, lanes, etc.).