ABSTRACT
In the spatial database domain, the data models currently used (relational, r-extended, object-oriented, etc.), are often soon outmoded by the ever-increasing needs of users. One of the main goals of a continuous field approach is to spread the information that has been measured at discrete sampling locations over the entire space of interest, whenever possible ignoring the boundaries of objects, if they do not affect the structure of the field. This means predicting the value of the field at all unsampled points, for which there are many methods. This chapter presents the main constraints on the estimation process. It describes the language specifications for defining and manipulating continuous fields. The estimation method must return the exact value of a sample point, and not an estimated one. Thus exact interpolation methods (returning exact values) are preferred to approximate ones (yielding estimates).
