Geostatistical analysis of spatial data
All data have (more or less) precise spatial and temporal labels associated with them. That is, a measurement is obtained from a particular location at a particular time, although that information may be lost by omission or made less precise by aggregation. For most of this chapter, it is assumed that only the data's spatial labels are important-hence the term spatial data. As a discipline, spatial statistics has components of all the classical areas of statistics, such as design, statistical methods (including data analysis and diagnostics), stochastic modeling, and statistical inference. Importantly, the spatial labels form an integral part of a spatial statistical analysis. Geostatistics is the area of spatial statistics that is concerned mostly with prediction of unknown values at given locations (or of aggregations over given regions). Typically, the prediction is based on univariate and bivariate distributions of the spatial values, and these distributions (or appropriate moments of them) are estimated from an initial analysis of the data.