Summary characteristics play an important role in the analysis of spatial point processes. We discuss various approaches to estimating summary characteristics from replicated observations of a stationary point process. The estimators are compared with respect to their integrated squared error. Simulations for three basic types of point processes help to indicate the best way of pooling the subwindow estimators. The most appropriate way depends on the particular summary characteristic, edge-correction method and also on the type of point process. The methods are demonstrated on a replicated dataset from forestry.
point process, $K$-function, nearest-neighbour distance distribution function, non-parametric estimation, replication
60G55, 62G05, 62M30