One thing must be pointed out here. The reader should not confound point objects with point measurements. Point objects are objects with no extent or of an extent that can be neglected. They may have attributes or not, but the main information is their location. Contrarily point measurements are measurements of fields at specific locations, so they are basically a representation of a field. The location itself is not so much of interest as the measurement itself. For this kind of data GIS provide a good collection of analytical methods. Variograms and kriging are two examples which interestingly were stimulating the idea for creating RDFs introduced in chapter 7.
At a first glance it is astonishing that only few analysis methods or operations are available in current commercial GIS for point objects. Until recently only three methods were available apart from the simple selection according to attribute data. The first methods is the intersection with polygons or fields. This enabled the user to identify in which polygon the point object is located or what value a continuous surface has at that location. The second method point objects can be used in is buffering. Point objects can be buffered to produce circular polygons around them. The third method available is the calculation of distances to other objects, whether they are also points, linear elements or polygon borders. In the last two years the situation was improved a little bit by implementations of simple density calculations of point objects (e.g. kernel density estimations).
To some extent this situation may be explained by the fact that one of the mainapplications of GIS today is in the area of administration and facility management. In these applications point objects exclusively appear as static, non-moving objects, for which the mapping itself is the main reason for its integration into a GIS. Although there is quite some activity in the scientific literature, the customer segment of researchers seems to be unattractively small for commercial GIS producers.
It may be argued that most GIS provide powerful tools for developing customized applications. To some extent this is true. But there are two difficulties with this approach. First and most important is that the dissemination of such 'user' developed methods is very difficult and often not of interest to the user due to financial or competitional reasons. The second difficulty in relying on 'user' developed methods is the speed and integration level at which it can be done. Especially in research computing speed is often very important, stimulating isolated applications.