Authors : Duccio Rocchini and Chiara Cateni
Abstract: In this study we deal with the measure of spatial centroid of polygon objects, defined as vectors joining points (e.g. vertices) in a vector space V. In particular, since geographical objects could be viewed as tuples containing both non-spatial (z(i)) and spatial (s(i)) information, we focus on the z(i) and s(i) related measures of centroid and to problems arising from their improper use. In the first case, no significance is given to the real (i.e., statistical) spatial location of the centroid but only on the possibility to directly (e.g. automatically) assign data from polygons to centroids (data storage). On the contrary, in the second case, centroid is a real statistically derived point (i.e., real centre of mass). The mostly used GIS softwares rely the centroid calculation to algorithms only based on a within-polygon assumption, by substantially deriving a z(i) related centroid. The aim of this study is to warn researchers to avoid this type of algorithms when dealing with point pattern analysis, strictly related to the location part s(i) of the object tuple.
Duccio Rocchini and Chiara Cateni , 2006. On The Measure of Spatial Centroid in Geography. Asian Journal of Information Technology, 5: 729-731.