Fundamentals of Spatial Analysis and Modelling

Jay Gao

• Publisher: CRC Press
• Year: 2021
• Print ISBN: 9781032115764
• eBook ISBN: 9781000519891
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Environmental Science
• Index: Technology & Engineering

This textbook provides comprehensive and in-depth explanations of all topics related to spatial analysis and spatiotemporal simulation, including how spatial data are acquired, represented digitally, and spatially aggregated. Also features the nature of space and how it is measured. Descriptive, explanatory, and inferential analyses are covered for point, line, and area data. It captures the latest developments in spatiotemporal simulation with cellular automata and agent-based modelling, and through practical examples discusses how spatial analysis and modelling can be implemented in different computing platforms. A much-needed textbook for a course at upper undergraduate and postgraduate levels.

1 Introduction

DOI: 10.1201/9781003220527-1

1.1 What Is Spatial Analysis?

1.1.1 Definition

Although spatial analysis has been widely practised in the geographic information system (GIS) community for decades, it is not so easy to define it precisely. In the literature, spatial data analysis has been defined differently by various authors. Goodchild (1987) defined it as a set of techniques devised to generate a spatial perspective on data. It is distinctive from other forms of analysis in that the analytical results are dependent on the locations of features or events being analysed. It yields value-added products from existing datasets, which is not possible otherwise. The analysis involves both locations and the attributes of spatial entities, or just the locations themselves. This definition leans heavily towards the technical perspective, involving both locational and attribute information. Haining (1990) defined it as “a body of methods and techniques for analyzing ‘events’ at a variety of spatial scales, the results of which depend upon the spatial arrangement of the ‘events’”. Featured prominently in this definition are techniques that address scale and spatial patterns. Although it is more encompassing than Goodchild’s definition by incorporating scale in the analysis, this definition still fails to capture the new developments in spatial analysis, such as spatial modelling and simulation which are one step further than simple descriptive analysis.

Longley et al. (2005) considered spatial analysis as a set of methods that can produce results changing with the locations of the features under analysis. This definition is almost identical to that of Haining (1990). Basically, this definition reduces spatial analysis to a set of analytical tools that have a spatial component. The Environmental System Research Institute (Esri) GIS Dictionary expands the definition of spatial analysis as the “process of examining the locations, attributes, and relationships of features in spatial data through overlay and other analytical techniques in order to address a question or gain useful knowledge”. This rather narrowly focused definition emphasises the content of spatial analysis and its objectives. However, the field of spatial analysis has evolved to such a degree that spatial modelling and simulation have become the integral components of spatial analysis, but they are excluded from this definition.

All of the aforementioned definitions from different perspectives with different emphases and annotations are decades old. They can no longer reflect the field of spatial data analysis adequately. In this book, contemporary spatial analysis is defined as a series of techniques for describing, analysing, simulating, and predicting the spatial patterns and/or processes of geographic phenomena that may involve mobile agents. This analysis may include spatial statistic indices, spatial regression and adaptive modelling, spatial dynamic modelling, integrated spatial statistic and spatial mechanism modelling, and spatial complex system modelling. The spatial data can be either geo-referenced points, lines, or areas to identify their patterns and associations, and to predict unknown values of the concerned attribute in relation to its affecting variables. This definition has three unique characteristics: (1) it is rather comprehensive, in that it includes the nature of spatial data that must have geographic coordinates through which observations are associated with their locations on the ground; (2) it captures the spatial dimension of the entity; and (3) it includes the purposes of spatial analysis, namely, to characterise the observed spatial patterns and to predict their values/patterns in the future. What is missing from this definition is the temporal component. This is because time is treated as invariant in descriptive spatial analysis, and is always implicitly dealt with in spatial modelling in which the behaviour of variables in the model varies with time. What is unclear in this definition is the dimension of the geographic entity. At present, it is confined to only two-dimensional phenomena. In reality, some geographic phenomena such as air pollution are inherently three-dimensional. Such phenomena can still be studied by slicing the vertical dimension into certain height bands and then treating each as a single two-dimensional layer. It can be studied using the methods described in this book. This attribute can be expressed as a function of location and time, namely, Z(easting, northing, t).

Implicit in the above definition of spatial data analysis is the development of mathematical models of spatial distributions, the analysis of locational patterns, and the investigation and prediction of space–time dynamics. Spatial analysis embraces a wide cluster of techniques that apply formal, usually quantitative, structures to systems in which the prime variables of interest vary across space. Traditionally, spatial data analysis falls into the domain of quantitative geography, although ecology, urban studies, transportation, and a host of cognate disciplines draw from and have played an instrumental role in the development of this field. In turn, the applications of spatial analysis to some of these fields have enriched and expanded spatial analysis.

1.1.2 Spatial Statistics

Spatial analysis has a number of closely related but not exactly meaningful terms, one of which is spatial statistics. In the Esri GIS Dictionary, it is defined as the:

study of statistical methods that use space and spatial properties (such as distance, area, volume, length, height, orientation, centrality and/or other spatial characteristics of data) directly in their mathematical computations. Spatial statistics may include a variety of analyses, such as pattern analysis, shape analysis, surface modeling and surface prediction, spatial regression, statistical comparisons of spatial datasets, statistical modeling and prediction of spatial interaction.

This definition encompasses spatial analysis. Besides, it goes one step further by spelling out the exact tasks of spatial analysis. According to this definition, spatial statistics applies statistical methods to the analysis of spatially referenced data. The topological, geometric, or geographic properties of spatial entities are the objects of study. Thus, all tools of traditional statistical analysis can be used to analyse spatial distributions, patterns, processes, and relationships of these entities without any modification. Spatial statistics differ from spatial analysis in that it focuses on the statistical properties of the attribute of a geographic phenomenon. These analyses can be descriptive, inferential, exploratory, and even geostatistical. What is missing from this definition is spatial modelling and simulation. Thus, spatial analysis has a much broader connotation than spatial statistics, which focuses on the quantitative description and exploration of spatial phenomena.

1.1.3 Geocomputation

Coined by Openshaw and Abrahart (1996), geocomputation started to appear in the literature in the mid-1990s at a specially themed conference dedicated to it. It refers to the application of computing technologies to solving spatial problems, including storage, analysis, and visualisation of spatial data and spatially modelled results (Esri GIS Dictionary). This definition emphasises the use of computers and the computing environment for running spatial analyses. It represents a marriage between computer science and spatial analysis. This definition is particularly suited to those spatial analyses that are memory- and CPU-hungry. Thus, geocomputation differs from spatial data analysis in that the latter has a much broader scope, including spatial modelling and simulation, whereas geocomputation aims at developing methods to analyse and model a range of highly complex, often non-deterministic problems with or without relying on GIS. As the field of spatial data analysis evolves, geocomputation has expanded to include principal component analysis, k-means clustering analysis, and maximum likelihood classification of satellite imagery, machine learning (e.g., artificial intelligence) decision trees, neural networks in identifying patterns and classifying satellite imagery data, and in mining the sheer volume of geo-referenced data for knowledge discovery. Some of these do not even have a spatial component. The analyses are considered spatial simply because the inputs are two-dimensional layers. Therefore, the spatial component of geographic data is not featured prominently in geocomputation.

1.1.4 Geostatistics

Geostatistics is a field of study applying statistical methods to analysing spatial data. It is a special branch of applied statistics initially developed by Georges Matheron, a French mathematician and geologist, in mining in the 1970s. This subset of spatial analytical methods aims at spatially depicting the observed attribute of a spatial entity quantitatively or statistically. When it was first developed, geostatistics found applications exclusively in the mining industry (e.g., to estimate ore reserves). It did not experience rapid development and wide uses until the 1990s, when advanced spatial analysis systems such as GIS became available, and the easy and wide availability of spatially referenced data in the digital format. In geography in general, and spatial analysis in particular, geostatistics can be regarded as a special branch of statistics applied to geographic data (e.g., data that have a spatial component attached to them). Since its invention, geostatistics has been widely used to analyse a wide range of spatial data in diverse disciplines, including petroleum geology, hydrogeology, oceanography, geochemistry, geometallurgy, geography, forestry, environmental control, landscape ecology, agriculture (especially in precision farming), and even medicine. These days, it has found even wider applications in more fields in natural sciences, including soil science, hydrology, meteorology, and environmental science. Common to all of these fields is that the data used have a spatial component, and the variable being analysed is predictable to a certain spatial extent. It is called quasi-statistics by Davis (1986) as some requirements of classical statistics are not met with the data. For instance, data independence is impossible to achieve wi...

Table of contents

• Cover
• Half-Title
• Title
• Copyright
• Dedication
• Contents
• Preface
• Author Biography
• Acknowledgements
• Chapter 1 Introduction
• Chapter 2 Space in Spatial Analysis
• Chapter 3 Spatial Data and Association
• Chapter 4 Descriptive and Inferential Spatial Analysis
• Chapter 5 Geostatistics and Spatial Interpolation
• Chapter 6 Spatial Modelling
• Chapter 7 Spatial Simulation
• Chapter 8 Time-explicit Spatial Analysis and Modelling
• Index