Stochastic Modeling of Alternate Methods for the Generation of Event Patterns with Complete Spatial Randomness

HARDISTY, Frank (Hardisty.1@osu.edu), The Ohio State University, Department of Geography, Columbus, OH 43210

Key Words: CSR, stochastic, event patterns

Many spatial statistical tests call for comparison of a given event pattern to a pattern or patterns with the characteristic of complete spatial randomness (CSR). Other spatial statistical tests rest on claims about characteristics of patterns under CSR that we may wish to test; therefore, the generation of CSR patterns is therefore of some interest. Unfortunately, not all writers on spatial statistics agree on accurate methods for generating such patterns. Three algorithms for the generation of CSR patterns are described, and then simulated under 1,200,000 trials of the three-event generation processes, generating nine test statistics for each of the processes in each trial. A new test statistic, Frank's K, for characterizing event patterns, is described.