Autocorrelation describes a set of data that is correlated with itself. When successive values ordered over time or space exhibit nonzero covariance, these data are said to be autocorrelated. Autocorrelation in time series data, often referred to as serial correlation, is frequently observed and has been widely studied and canonized. Examples are numerous: tomorrow’s temperature is often predicted by temperature today, and a county’s literacy rate next year will likely be well predicted by literacy this year.

While spatial autocorrelation remains an actively growing body of research, examples are also abundant: temperature in one county is often predicted by temperature in a neighboring county, and demographic makeup of a census block is likely similar to neighboring blocks. In both spatial and temporal data, autocorrelation has important ...