Improved Distance Estimation Method for Location Modeling

Abstract

Use of Euclidean or rectilinear estimates of travel distance between pairs of points for location modeling can be inaccurate in a city with major barriers to travel. Euclidean distances are consistently lower than the actual travel distances. Rectilinear distances, on the other hand, could be higher or lower than actual travel distances. In order to improve distance estimation for location modeling, without resorting to the use of shortest-path algorithms based on the road network, a method based on Euclidean distance, that makes allowance for barriers, is proposed. This method considers the distance between two points, that are on different sides of a barrier, as the sum of the Euclidean distances: between the first point and the nearest barrier crossing, and between this barrier crossing and the other point. The method is tested for the city of Winnipeg, Manitoba, Canada, and found to produce improved distance estimates.