ABSTRACT
Few would disagree that good-quality initial data i s needed to successfully produce an effective final solution from a geographical information system (GIS) . However, while we often think of the poor quality of spatial data as being associated with inaccurate observations or those made over too short a period of time, a more basic problem with much geographical data is the sparseness of locations at which samples are taken. Poss ible reasons for this may include the inaccessibility of some sites and the prohibitive costs that may be involved in carrying out a full survey. If, as on many occasions, we then wish to create a map or other realization showing the full spatial extent of a variable (based on the limited observations available), we may use one of a number of spatial interpolation methods. Spatial interpolation allows us to estimate the value of a property at a relatively large number of unsarnpled locations as a function of the relatively few observed point values.
