Poisson kriging: A closer investigation

This work revisits a simple geostatistical model for the analysis of spatial count data and describes some of its main second-order properties. This geostatistical model is simpler than an alternative hierarchical model, also used for the analysis of spatial count data, so it may be more appealing to practitioners and spatial data analysts. Geostatistical methods for trend parameter estimation, semivariogram estimation and prediction of the latent process are reviewed, and new estimators and predictors are proposed. Finally, a designed simulation experiment is carried out to investigate and compare the sampling properties of the different estimators and predictors.