Model-based geostatistics. (Q863732)

Geostatistics is a sub-branch of spatial statistics but originally the term geostatistics was coined by Matheron. Despite the apparent integration with spatial statistics, much geostatistical practice still reflects its independent origins, and from a mainstream statistical perspective this has some undesirable consequences. In particular, explicit stochastic models are not always declared and ad hoc methods of inference are often used. The authors of this book describe an approach to geostatistical problems based on the application of formal statistical methods under an explicitly assumed stochastic model. This approach is called model-based geostatistics. The authors produce an applied statistical counterpart to \textit{M. Stein} [Interpolation of Spatial Data: Some Theory for Kriging, Springer (1999; Zbl 0924.62100)], who gave a rigorous mathematical theory of Kriging. The intended readership includes postgraduate statistics students and scientific researchers whose work involves the analysis of geostatistical data. Throughout the book, the statistical methods are illustrated by applying them in the analysis of real data-sets. Most of the book's chapters end with a section on computation using R software, especially the packages geoR and geoRglm. The first two chapters of the book provide an introduction and overview. Chapters 3 and 4 then describe geostatistical models, whilst Chapters 5 to 8 cover associated methods of inference (5. Classical parameter estimation 6. Spatial prediction 7.Bayesian inference 8. Geostatistical design). Due to the longstanding experience in geostatistics of the first author this book is a spirited performance and can be recommended for anyone interested in geostatistics.