Surveillance systems for monitoring the development of spatial patterns

Statistical methods concerned with the identification of temporal patterns may be classified into those that examine retrospectively a set of observations, and those that constitute surveillance systems that monitor changes as new observations become available. A similar distinction applies to the identification of geographical patterns in spatial data. There has been a notable lack of attention given to the surveillance of spatial pattern. This paper concerns development of a cumulative sum statistic and procedure for the monitoring of spatial pattern, and its application to both simulated data and to data on Burkitt's lymphoma in Uganda.