Covariance is a measure of dispersion. An analogous measure for geostatistical data is a covariogram, which is the same as covariance with the only difference being that observations are spatially indexed and covariance between points is measured at a fixed separation distance. Similar to the case with correlation, we have a correlogram for spatial data where, again, observations are spatially indexed and we take correlation between pairs of points at a given separation distance. Now, if we plot a correlogram for all possible separation distances, or lags, we get a correlogram plot.
We need second-order stationarity to explain local variation using covariograms. Second-order stationarity conditions stipulate that the ...
