Time-geographic density estimation for moving point objects

Abstract

This research presents a time-geographic method of density estimation for moving point objects. The approach integrates traditional kernel density estimation (KDE) with techniques of time geography to generate a continuous intensity surface that characterises the spatial distribution of a moving object over a fixed time frame. This task is accomplished by computing density estimates as a function of a geo-ellipse generated for each consecutive pair of control points in the object's space-time path and summing those values at each location in a manner similar to KDE. The main advantages of this approach are: (1) that positive intensities are only assigned to locations within a moving object's potential path area and (2) that it avoids arbitrary parameter selection as the amount of smoothing is controlled by the object's maximum potential velocity. The time-geographic density estimation technique is illustrated with a sample dataset, and a discussion of limitations and future work is provided. © 2010 Springer-Verlag.