ABSTRACT

The fields of geostatistics and spatial analysis are closely intertwined due to their joint emphasis on the use of traditional and novel approaches to describe, analyze, and visualize the spatial variability of naturally occurring phenomena. Both fields share analytical objectives that seek to uncover broad spatial patterns and relationships, pinpoint localized departures and anomalies in the data, and derive parameter estimates for predictive purposes. Like spatial analysis, geostatistics combines practical and conceptual thoughts on the modeling of spatial variability with mathematical and statistical principles. It can facilitate the analysis of spatial variability of an entire population or a sample. “Geostatistics,” which literally means statistics of the earth, is firmly rooted in traditional regression theory with past applications mostly in the natural and earth sciences. Pioneering work in the field began in the 1950s with inspiration from the South African Danie Krige’s work in geological mining (Krige 1951). This work later expanded in the 1960s under the French Mathematician George Matheron’s leadership and efforts to showcase the practical applications of the methods. Many disciplines, including engineering, hydrology, soil studies, medical geography, epidemiology, ecology, and environmental assessment now fully embrace geostatistical methodologies to solve spatial prediction and modeling problems (Goovaerts 1997, 1999, 2009; Haining et al. 2010; Barro and Oyana 2012; Birkin 2013; Noor et al. 2014). With the advent of GIS, spatial statistics and geostatistics have become virtually inseparable as computerized

analytical and visualization approaches are developed to handle and display the large volume and variety of datasets representing both natural and anthropogenic phenomena in spatial modeling. These approaches are now fairly well established and integrated into leading software packages and are used in many scientific endeavors due to their analytical rigor and robustness. In a GIS context, the geostatistical approaches can be used to successfully analyze and integrate the different types of spatial data, measure spatial autocorrelation by incorporating the statistical distribution and spatial relationships between the sample data, perform spatial prediction, and assess uncertainty. Several scholars have also used these approaches (especially Poisson kriging and p-field simulation) to account for small number/population problems (Goovaerts and Jacquez 2004; Goovaerts 2005, 2006), to account for uncertainty (Oyana 2004), and to perform spatial prediction, as they are known to accurately predict better local estimates (Goovaerts 1997; Guo et al. 2006).