Statistical Methods for Spatial Data Analysis

Abstract

Statistical Methods for Spatial Data Analysis Oliver Schabenberger and Carol A. Gotway (Chapman & Hall/CRC Press) 2005, 488 pages. ISBN 1-58488-322-7. Hard cover only. Statistical Methods for Spatial Data Analysis offers plenty of information for the analysis of spatial data in a variety of disciplines. It is clearly written and well organized. The chapters are highly topical and come at a time when the literature on statistical methods for spatial data analysis is steadily growing. Interesting and relevant to the readership of the URISA Journal, this book is a valuable resource for educators, students, geographic information system (GIS) practitioners, and spatial scientists from varying disciplines. The aim of the book is ambitious: comprehensive and illustrative compilation of the basic statistical theory and methods for spatial data analysis. Few books on the subject of statistical methods for spatial data analysis describe the methods in a thorough yet accessible manner. This text stands out because of its comprehensive coverage of a wide range of statistical methods and spatial analysis techniques. One of the book's main strengths is the clear organization of its chapters. Each chapter starts with an explanation of the theory with well-chosen examples explaining the statistical method. Most of the examples use simplified real-world datasets and sometimes hypothetical datasets with a few exceptions. For example, the woodpecker data, lightning-strikes data, rainfall data, and low-birth-weight data represent a variety of disciplines, which makes the book very useful for scientists across disciplines. Necessary equations are provided for each method with a wealth of informative figures, which contribute substantially to developing a better understanding of the methods described. As could be expected for a book of this nature, it includes a fair amount of mathematics. Each chapter ends with problems that encourage the readers/students to apply the statistical methods described to a specific problem. The book contains nine chapters. The introductory chapter provides the needed background on the characteristics and types of spatial data, and the nature of spatial processes and patterns such as autocorrelation functions and the effects of autocorrelation on statistical inference. Chapter 2 describes the theoretical framework of random fields necessary for subsequent chapters, particularly Chapters 4 and 5. Chapter 3 covers point-pattern analysis with a well-named title, "Mapped Point Patterns." The authors should be congratulated on doing such a solid job of including the relevant spatial processes and techniques applicable to point-pattern analysis. Chapter 4 primarily deals with semivariogram, estimation, and modeling of the covariance function. Chapter 5 covers spatial prediction and kriging. In this chapter, the authors elaborate on general details of the spatial prediction problem and give an extensive overview of kriging, with comparisons such as local versus global kriging. They also cover trend surface models with illustrations. Chapter 6 is a comprehensive coverage of spatial regression models, beginning with linear models with uncorrelated errors and ending with a succinct discussion of Bayesian hierarchical models for spatial data. Chapter 7 describes simulation of random fields, followed by Chapter 8 on nonstationary covariance. The final chapter on spatiotemporal processes primarily deals with separable and nonseparable covariance functions and spatiotemporal point processes. Each of the various statistical methods is described in considerable depth. The book's main strength is that it describes basic statistical concepts for spatial data analysis and explains them and their relevance clearly in a single volume in a consistent manner. …