2005 Volume 13 Issue 1 Pages 33-41
This paper proposes Inverse Distance Weighted (IDW) interpolation on a network for analyzing the phenomena that occur on a network. This method interpolates an unknown attribute value at an arbitrary location on a network using observed values at some other points in its vicinity, which are called sample points. The shortest-path distances from the sample points to the unknown point are used in two separate stages. First, they are used for selecting the fixed number of nearest sample points. Second, the method discretely estimates the attribute value at an unknown point as a weighted mean of a function of their negative power. This is followed by the comparison of the difference in the results derived from the network interpolation and the planar interpolation. Cross validation is applied to the same data set of road networks and elevation data on a network.