Digitizing and distance estimation errors in spatial analysis: sinuosity and network configuration effects

In this paper, we examine facets of error that occur when parametric and digitizing approaches are used in distance estimation. First, Euclidean, Manhattan and parametric distance metrics are briefly described. Next, the distance estimation procedures are examined and compared to a normative measure in study areas selected on the basis of road network configuration. The relative efficacy of the different estimation procedures in each area is described. We also examine a series of lines with varying amounts of sinuosity to determine variation in distance estimates derived from manual digitizing.