Kriged Road-Traffic Maps

A kriging based on residuals is employed for spatial extrapolation of anisotropic directional road–traffic data. The set of data considered in the present paper includes, among others, the actual geographic positions and velocities of approximately 300 test vehicles in downtown Berlin. They are transmitted to a central station within regular time intervals. The main idea of the extrapolation technique is to interpret the recorded velocities as realizations of a random velocity field, which is sampled at selected points only. Structural properties of the resulting road–traffic maps are discussed combined with a statistical space–time analysis of polygonal traffic trajectories extracted from the original traffic data. Finally, a brief outlook to simulation and prediction of future traffic states is provided.