There are three broad classes of spatial data:
Point-referenced data are values that are sampled at specific locations within an area of a predefined size. An example is air temperatures that are measured where weather monitoring instruments are located. The stochastic nature of spatial processes can be described by using spatial random fields (SRFs). A set of point-referenced data can be seen as a realization of a continuous SRF that takes values over the entire study area. The values at unsampled locations are unknown but can be predicted by means of geostatistical analysis. You can analyze point-referenced data by using the SAS/STAT procedures VARIOGRAM, KRIGE2D, and SIM2D.
Areal (lattice) data are values for a fixed number of areal units within a particular area. These data differ the point-referenced data in that one areal observation is assigned to a whole areal unit instead of to a specific location. An example is crime rates that are aggregated over counties within a state.
Point pattern data are a collection of locations of single events of a spatial process. In this category, the study area can have a variable size and observations might have associated covariates, but the main interest is in their spatial patterns of occurrence. Examples include locations of tree growth, locations of petty crimes, and so on. A set of point-pattern data can be seen as a realization of a discrete SRF that has values only at the event locations (Illian et al., 2008, p. 44). A collection of this type of data is known as a spatial point pattern. Point pattern analysis usually does not refer to the SRF concept. The applied techniques in point patterns differ from the geostatistical approach, although both types of analysis share corresponding measures to describe correlation among the data. You can use the SPP procedure to analyze point pattern data.