Geographic data tend to be spatially dependent. Spatial dependence is "the propensity for nearby locations to influence each other and to possess similar attributes" (Goodchild, 1992, p.33). In other words, to paraphrase a famous geographer named Waldo Tobler, while everything is related to everything else, things that are close together tend to be more related than things that are far apart. Terrain elevations, soil types, and surface air temperatures, for instance, are more likely to be similar at points two meters apart than at points two kilometers apart. A statistical measure of the similarity of attributes of point locations is called spatial autocorrelation.
Given that geographic data are expensive to create, spatial dependence turns out to be a very useful property. We can sample attributes at a limited number of locations, then estimate the attributes of intermediate locations. The process of estimating unknown values from nearby known values is called interpolation. Interpolated values are reliable only to the extent that the spatial dependence of the phenomenon can be assumed. If we were unable to assume some degree of spatial dependence, it would be impossible to represent continuous geographic phenomena in digital form.