Table Of Content(cid:1)(cid:2)(cid:3)(cid:4)(cid:5)(cid:3)(cid:6) (cid:7)(cid:8)(cid:3)(cid:6)(cid:9)(cid:10)(cid:5)(cid:10) (cid:3)(cid:8)(cid:11) (cid:12)(cid:13)(cid:14)(cid:15)(cid:14)(cid:16)(cid:2)(cid:17)(cid:4)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8)
(cid:18)(cid:3)(cid:8)(cid:19)(cid:20)(cid:13)(cid:11) (cid:18)(cid:21) (cid:22)(cid:5)(cid:10)(cid:23)(cid:24)(cid:13)(cid:20)
(cid:1)(cid:2)(cid:3)(cid:4)(cid:5)(cid:3)(cid:6) (cid:7)(cid:8)(cid:3)(cid:6)(cid:9)(cid:10)(cid:5)(cid:10)
(cid:3)(cid:8)(cid:11) (cid:12)(cid:13)(cid:14)(cid:15)(cid:14)(cid:16)(cid:2)(cid:17)(cid:4)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8)
(cid:1)(cid:13)(cid:6)(cid:13)(cid:23)(cid:4)(cid:13)(cid:11) (cid:25)(cid:10)(cid:10)(cid:3)(cid:9)(cid:10)
(cid:26)(cid:5)(cid:4)(cid:24) (cid:27)(cid:28) (cid:22)(cid:5)(cid:29)(cid:17)(cid:20)(cid:13)(cid:10) (cid:3)(cid:8)(cid:11) (cid:30)(cid:31) (cid:3)!(cid:6)(cid:13)(cid:10)
(cid:1) (cid:2)
"(cid:20)(cid:14)(cid:19)(cid:21) #(cid:20)(cid:21) (cid:18)(cid:3)(cid:8)(cid:19)(cid:20)(cid:13)(cid:11) (cid:18)(cid:21) (cid:22)(cid:5)(cid:10)(cid:23)(cid:24)(cid:13)(cid:20)
$(cid:5)(cid:13)(cid:8)(cid:8)(cid:3) %(cid:8)(cid:5)&(cid:13)(cid:20)(cid:10)(cid:5)(cid:4)(cid:9) (cid:14)(cid:19) (cid:25)(cid:23)(cid:14)(cid:8)(cid:14)(cid:16)(cid:5)(cid:23)(cid:10) (cid:3)(cid:8)(cid:11) ’(cid:17)(cid:10)(cid:5)(cid:8)(cid:13)(cid:10)(cid:10) (cid:7)(cid:11)(cid:16)(cid:5)(cid:8)(cid:5)(cid:10)(cid:4)(cid:20)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8)
((cid:8)(cid:10)(cid:4)(cid:5)(cid:4)(cid:17)(cid:4)(cid:13) (cid:19)(cid:14)(cid:20) (cid:25)(cid:23)(cid:14)(cid:8)(cid:14)(cid:16)(cid:5)(cid:23) (cid:12)(cid:13)(cid:14)(cid:29)(cid:20)(cid:3)(cid:2)(cid:24)(cid:9) (cid:3)(cid:8)(cid:11) (cid:12)((cid:1)(cid:23)(cid:5)(cid:13)(cid:8)(cid:23)(cid:13)
)(cid:14)(cid:20)(cid:11)!(cid:13)(cid:20)(cid:29)(cid:10)(cid:4)(cid:20)(cid:3)*(cid:13) +(cid:27),(cid:30),(cid:7)
+(cid:31)-(cid:31) $(cid:5)(cid:13)(cid:8)(cid:8)(cid:3). (cid:7)(cid:17)(cid:10)(cid:4)(cid:20)(cid:5)(cid:3)
((cid:1)’)/+(cid:31) (cid:28)/(cid:27)(cid:30)(cid:31)/(cid:28)(cid:27)01-/0(cid:1)(cid:2)(cid:20)(cid:5)(cid:8)(cid:29)(cid:13)(cid:20)’(cid:13)(cid:20)(cid:6)(cid:5)(cid:8) 2(cid:13)(cid:5)(cid:11)(cid:13)(cid:6)!(cid:13)(cid:20)(cid:29))(cid:13)3 4(cid:14)(cid:20)5
((cid:1)’)/+(cid:28) -06/(cid:28)/(cid:27)(cid:30)(cid:31)/(cid:28)(cid:27)01-/1(cid:1)(cid:2)(cid:20)(cid:5)(cid:8)(cid:29)(cid:13)(cid:20)’(cid:13)(cid:20)(cid:6)(cid:5)(cid:8) 2(cid:13)(cid:5)(cid:11)(cid:13)(cid:6)!(cid:13)(cid:20)(cid:29) )(cid:13)34(cid:14)(cid:20)5
(cid:15)(cid:3)(cid:4)(cid:3)(cid:6)(cid:14)(cid:29)(cid:5)(cid:8)(cid:29)/(cid:5)(cid:8)/"(cid:17)!(cid:6)(cid:5)(cid:23)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8)#(cid:3)(cid:4)(cid:3)
7(cid:5)!(cid:20)(cid:3)(cid:20)(cid:9)(cid:14)(cid:19)(cid:15)(cid:14)(cid:8)(cid:29)(cid:20)(cid:13)(cid:10)(cid:10)(cid:15)(cid:14)(cid:8)(cid:4)(cid:20)(cid:14)(cid:6))(cid:17)(cid:16)!(cid:13)(cid:20)81(cid:31)(cid:31)9-1-(cid:28)+(cid:30)
(cid:24)(cid:5)(cid:10) 3(cid:14)(cid:20)5 (cid:5)(cid:10) (cid:10)(cid:17)!:(cid:13)(cid:23)(cid:4) (cid:4)(cid:14) (cid:23)(cid:14)(cid:2)(cid:9)(cid:20)(cid:5)(cid:29)(cid:24)(cid:4)(cid:21) (cid:7)(cid:6)(cid:6) (cid:20)(cid:5)(cid:29)(cid:24)(cid:4)(cid:10) (cid:3)(cid:20)(cid:13) (cid:20)(cid:13)(cid:10)(cid:13)(cid:20)&(cid:13)(cid:11). 3(cid:24)(cid:13)(cid:4)(cid:24)(cid:13)(cid:20) (cid:4)(cid:24)(cid:13) 3(cid:24)(cid:14)(cid:6)(cid:13) (cid:14)(cid:20) (cid:2)(cid:3)(cid:20)(cid:4) (cid:14)(cid:19)
(cid:4)(cid:24)(cid:13)(cid:16)(cid:3)(cid:4)(cid:13)(cid:20)(cid:5)(cid:3)(cid:6)(cid:5)(cid:10)(cid:23)(cid:14)(cid:8)(cid:23)(cid:13)(cid:20)(cid:8)(cid:13)(cid:11).(cid:10)(cid:2)(cid:13)(cid:23)(cid:5)(cid:19)(cid:5)(cid:23)(cid:3)(cid:6)(cid:6)(cid:9) (cid:4)(cid:24)(cid:13)(cid:20)(cid:5)(cid:29)(cid:24)(cid:4)(cid:10)(cid:14)(cid:19)(cid:4)(cid:20)(cid:3)(cid:8)(cid:10)(cid:6)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8).(cid:20)(cid:13)(cid:2)(cid:20)(cid:5)(cid:8)(cid:4)(cid:5)(cid:8)(cid:29).(cid:20)(cid:13)(cid:17)(cid:10)(cid:13)(cid:14)(cid:19)(cid:5)(cid:6)(cid:6)(cid:17)(cid:10)/
(cid:4)(cid:20)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8)(cid:10). (cid:20)(cid:13)(cid:23)(cid:5)(cid:4)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8). !(cid:20)(cid:14)(cid:3)(cid:11)(cid:23)(cid:3)(cid:10)(cid:4)(cid:5)(cid:8)(cid:29). (cid:20)(cid:13)(cid:2)(cid:20)(cid:14)(cid:11)(cid:17)(cid:23)(cid:4)(cid:5)(cid:14)(cid:8) (cid:14)(cid:8) (cid:16)(cid:5)(cid:23)(cid:20)(cid:14)(cid:19)(cid:5)(cid:6)(cid:16) (cid:14)(cid:20) (cid:5)(cid:8) (cid:3)(cid:8)(cid:9) (cid:14)(cid:4)(cid:24)(cid:13)(cid:20) 3(cid:3)(cid:9). (cid:3)(cid:8)(cid:11)
(cid:10)(cid:4)(cid:14)(cid:20)(cid:3)(cid:29)(cid:13) (cid:5)(cid:8) (cid:11)(cid:3)(cid:4)(cid:3)!(cid:3)(cid:8)5(cid:10)(cid:21) #(cid:17)(cid:2)(cid:6)(cid:5)(cid:23)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8) (cid:14)(cid:19) (cid:4)(cid:24)(cid:5)(cid:10)(cid:2)(cid:17)!(cid:6)(cid:5)(cid:23)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8) (cid:14)(cid:20) (cid:2)(cid:3)(cid:20)(cid:4)(cid:10) (cid:4)(cid:24)(cid:13)(cid:20)(cid:13)(cid:14)(cid:19)(cid:5)(cid:10)(cid:2)(cid:13)(cid:20)(cid:16)(cid:5)(cid:4)(cid:4)(cid:13)(cid:11) (cid:14)(cid:8)(cid:6)(cid:9)
(cid:17)(cid:8)(cid:11)(cid:13)(cid:20) (cid:4)(cid:24)(cid:13)(cid:2)(cid:20)(cid:14)&(cid:5)(cid:10)(cid:5)(cid:14)(cid:8)(cid:10) (cid:14)(cid:19)(cid:4)(cid:24)(cid:13) (cid:12)(cid:13)(cid:20)(cid:16)(cid:3)(cid:8)(cid:15)(cid:14)(cid:2)(cid:9)(cid:20)(cid:5)(cid:29)(cid:24)(cid:4) 7(cid:3)3(cid:14)(cid:19)(cid:1)(cid:13)(cid:2)(cid:4)(cid:13)(cid:16)!(cid:13)(cid:20) -.+-9(cid:27). (cid:5)(cid:8) (cid:5)(cid:4)(cid:10) (cid:23)(cid:17)(cid:20)(cid:20)(cid:13)(cid:8)(cid:4)
&(cid:13)(cid:20)(cid:10)(cid:5)(cid:14)(cid:8).(cid:3)(cid:8)(cid:11)(cid:2)(cid:13)(cid:20)(cid:16)(cid:5)(cid:10)(cid:10)(cid:5)(cid:14)(cid:8)(cid:19)(cid:14)(cid:20)(cid:17)(cid:10)(cid:13)(cid:16)(cid:17)(cid:10)(cid:4)(cid:3)(cid:6)3(cid:3)(cid:9)(cid:10)!(cid:13)(cid:14)!(cid:4)(cid:3)(cid:5)(cid:8)(cid:13)(cid:11)(cid:19)(cid:20)(cid:14)(cid:16)(cid:1)(cid:2)(cid:20)(cid:5)(cid:8)(cid:29)(cid:13)(cid:20)/$(cid:13)(cid:20)(cid:6)(cid:3)(cid:29)(cid:21)$(cid:5)(cid:14)(cid:6)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8)(cid:10)
(cid:3)(cid:20)(cid:13)(cid:6)(cid:5)(cid:3)!(cid:6)(cid:13)(cid:19)(cid:14)(cid:20)(cid:2)(cid:20)(cid:14)(cid:10)(cid:13)(cid:23)(cid:17)(cid:4)(cid:5)(cid:14)(cid:8)(cid:17)(cid:8)(cid:11)(cid:13)(cid:20)(cid:4)(cid:24)(cid:13)(cid:12)(cid:13)(cid:20)(cid:16)(cid:3)(cid:8)(cid:15)(cid:14)(cid:2)(cid:9)(cid:20)(cid:5)(cid:29)(cid:24)(cid:4)7(cid:3)3(cid:21)
(cid:1)(cid:2)(cid:20)(cid:5)(cid:8)(cid:29)(cid:13)(cid:20)(cid:5)(cid:10)(cid:3)(cid:2)(cid:3)(cid:20)(cid:4)(cid:14)(cid:19)(cid:1)(cid:2)(cid:20)(cid:5)(cid:8)(cid:29)(cid:13)(cid:20)(cid:1)(cid:23)(cid:5)(cid:13)(cid:8)(cid:23)(cid:13);’(cid:17)(cid:10)(cid:5)(cid:8)(cid:13)(cid:10)(cid:10)(cid:18)(cid:13)(cid:11)(cid:5)(cid:3)
(cid:10)(cid:2)(cid:20)(cid:5)(cid:8)(cid:29)(cid:13)(cid:20)(cid:14)(cid:8)(cid:6)(cid:5)(cid:8)(cid:13)(cid:21)(cid:23)(cid:14)(cid:16)
<(cid:1)(cid:2)(cid:20)(cid:5)(cid:8)(cid:29)(cid:13)(cid:20)’(cid:13)(cid:20)(cid:6)(cid:5)(cid:8)=2(cid:13)(cid:5)(cid:11)(cid:13)(cid:6)!(cid:13)(cid:20)(cid:29)1(cid:31)(cid:31)9
"(cid:20)(cid:5)(cid:8)(cid:4)(cid:13)(cid:11)(cid:5)(cid:8)(cid:12)(cid:13)(cid:20)(cid:16)(cid:3)(cid:8)(cid:9)
(cid:24)(cid:13)(cid:17)(cid:10)(cid:13)(cid:14)(cid:19)(cid:29)(cid:13)(cid:8)(cid:13)(cid:20)(cid:3)(cid:6)(cid:11)(cid:13)(cid:10)(cid:23)(cid:20)(cid:5)(cid:2)(cid:4)(cid:5)&(cid:13)(cid:8)(cid:3)(cid:16)(cid:13)(cid:10).(cid:20)(cid:13)(cid:29)(cid:5)(cid:10)(cid:4)(cid:13)(cid:20)(cid:13)(cid:11)(cid:8)(cid:3)(cid:16)(cid:13)(cid:10).(cid:4)(cid:20)(cid:3)(cid:11)(cid:13)(cid:16)(cid:3)(cid:20)5(cid:10).(cid:13)(cid:4)(cid:23)(cid:21)(cid:5)(cid:8)(cid:4)(cid:24)(cid:5)(cid:10)(cid:2)(cid:17)!(cid:6)(cid:5)(cid:23)(cid:3)/
(cid:4)(cid:5)(cid:14)(cid:8) (cid:11)(cid:14)(cid:13)(cid:10) (cid:8)(cid:14)(cid:4) (cid:5)(cid:16)(cid:2)(cid:6)(cid:9). (cid:13)&(cid:13)(cid:8) (cid:5)(cid:8) (cid:4)(cid:24)(cid:13) (cid:3)!(cid:10)(cid:13)(cid:8)(cid:23)(cid:13) (cid:14)(cid:19) (cid:3) (cid:10)(cid:2)(cid:13)(cid:23)(cid:5)(cid:19)(cid:5)(cid:23) (cid:10)(cid:4)(cid:3)(cid:4)(cid:13)(cid:16)(cid:13)(cid:8)(cid:4). (cid:4)(cid:24)(cid:3)(cid:4) (cid:10)(cid:17)(cid:23)(cid:24) (cid:8)(cid:3)(cid:16)(cid:13)(cid:10) (cid:3)(cid:20)(cid:13)
(cid:13)>(cid:13)(cid:16)(cid:2)(cid:4) (cid:19)(cid:20)(cid:14)(cid:16) (cid:4)(cid:24)(cid:13) (cid:20)(cid:13)(cid:6)(cid:13)&(cid:3)(cid:8)(cid:4) (cid:2)(cid:20)(cid:14)(cid:4)(cid:13)(cid:23)(cid:4)(cid:5)&(cid:13) (cid:6)(cid:3)3(cid:10) (cid:3)(cid:8)(cid:11) (cid:20)(cid:13)(cid:29)(cid:17)(cid:6)(cid:3)(cid:4)(cid:5)(cid:14)(cid:8)(cid:10) (cid:3)(cid:8)(cid:11) (cid:4)(cid:24)(cid:13)(cid:20)(cid:13)(cid:19)(cid:14)(cid:20)(cid:13) (cid:19)(cid:20)(cid:13)(cid:13) (cid:19)(cid:14)(cid:20) (cid:29)(cid:13)(cid:8)(cid:13)(cid:20)(cid:3)(cid:6)
(cid:17)(cid:10)(cid:13)(cid:21)
(cid:15)(cid:14)&(cid:13)(cid:20)/#(cid:13)(cid:10)(cid:5)(cid:29)(cid:8)8(cid:25)(cid:20)(cid:5)(cid:23)(cid:24)?(cid:5)(cid:20)(cid:23)(cid:24)(cid:8)(cid:13)(cid:20).2(cid:13)(cid:5)(cid:11)(cid:13)(cid:6)!(cid:13)(cid:20)(cid:29)
(cid:1)"()++00-106 66,(cid:28)+(cid:31)(cid:31)/(cid:27) (cid:30) (cid:28) 1 + (cid:31)@"(cid:20)(cid:5)(cid:8)(cid:4)(cid:13)(cid:11)(cid:14)(cid:8)(cid:3)(cid:23)(cid:5)(cid:11)/(cid:19)(cid:20)(cid:13)(cid:13)(cid:2)(cid:3)(cid:2)(cid:13)(cid:20)
Preface
The dissemination of digital spatial databases, coupled with the ever wider use of
GISystems, is stimulating increasing interest in spatial analysis from outside the
spatial sciences. The recognition of the spatial dimension in social science
research sometimes yields different and more meaningful results than analysis
which ignores it.
The emphasis in this book is on spatial analysis from the perspective of Geo-
Computation. GeoComputation is a new computational-intensive paradigm that
increasingly illustrates its potential to radically change current research practice in
spatial analysis. This volume contains selected essays of Manfred M. Fischer. By
drawing together a number of related papers, previously scattered in space and
time, the collection aims to provide important insights into novel styles to perform
spatial modelling and analysis tasks. Based on the latest developments in estima-
tion theory, model selection and testing this volume develops neural networks into
advanced tools for non-parametric modelling and spatial interaction modelling.
Spatial Analysis and GeoComputation is essentially a multi-product under-
taking, in the sense that most of the contributions are multi-authored publications.
All these co-authors deserve the full credit for this volume, as they have been the
scientific source of the research contributions included in the present volume. This
book is being published simultaneously with Innovation,NetworksandKnowledge
Spillovers:SelectedEssays.
I would also like to thank Gudrun Decker, Thomas Seyffertitz and Petra Staufer-
Steinnocher for their capable assistance in co-ordinating the various stages of the
preparation of the book.
Manfred M. Fischer Vienna, May 2006
Contents
Preface v
1 Introduction 1
PART I Spatial Analysis and GIS
2 Spatial Analysis in Geography 17
3 Spatial Interaction Models and the Role of Geographic
Information Systems 29
4 GIS and Network Analysis 43
5 Expert Systems and Artificial Neural Networks for Spatial Analysis
and Modelling: Essential Components for Knowledge Based
Geographical Information Systems 61
PART II Computational Intelligence in Spatial Data Analysis
6 Computational Neural Networks – Tools for Spatial Data Analysis 79
7 Artificial Neural Networks: A New Approach to Modelling
Interregional Telecommunication Flows
with S. Gopal 103
8 A Genetic-Algorithms Based Evolutionary Computational Neural
Network for Modelling Spatial Interaction Data
with Y. Leung 129
PART III GeoComputation in Remote Sensing Environments
9 Evaluation of Neural Pattern Classifiers for a Remote Sensing
Application
with S. Gopal, P. Staufer and K. Steinnocher 155
viii Contents
10 Optimisation in an Error Backpropagation Neural Network
Environment with a Performance Test on a Spectral Pattern
Classification Problem
with P. Staufer 183
11 Fuzzy ARTMAP – A Neural Classifier for Multispectral Image
Classification
with S. Gopal 209
PART IV New Frontiers in Neural Spatial Interaction Modelling
12 Neural Network Modelling of Constrained Spatial Interaction Flows:
Design, Estimation, and Performance Issues
with M. Reismann and K. Hlavácková-Schindler 241
13 Learning in Neural Spatial Interaction Models: A Statistical
Perspective 269
14 A Methodology for Neural Spatial Interaction Modelling
with M. Reismann 283
Figures 311
Tables 317
Subject Index 321
Author Index 329
Acknowledgements 335
1 Introduction
Traditionally, spatial analysis is the domain of the academic discipline of geo-
graphy, especially of quantitative geography, although ecology, transportation,
urban studies and a host of other disciplines draw from and are instrumental in the
development of this field (Longley and Batty 1996). Spatial analysis is clearly not
a simple and straightforward extension of non-spatial analysis, but raises many
distinct problems: the modifiable areal unit problem that consists of two related
parts, the scale problem and the zoning problem (see Openshaw 1977); the spatial
association problem since the association between spatial units affects the inter-
pretation of georeferenced variables; the spatial heterogeneity problem, and the
boundary effects problem. By taking these problems into account, the spatial
analyst gives more meaning to the subject. The value of spatial analysis comes
from its ability to yield insights about phenomena and processes that occur in the
real world.
Spatial analysis, as it evolved over the past few decades, consists of two major
areas of research: spatial data analysis [in a more strict sense] and spatial model-
ling though the boundary is rather blurred (see Fischer and Getis 1997). Spatial
modelling lies at the heartland of regional science and includes a wide range of
different models (see Wegener and Fotheringham 2000), most notably models of
location-allocation (see, for example, Church and Revelle 1976), spatial inter-
action (see, for example, Sen and Smith 1975, Roy 2004, Fischer and Reggiani
2004), and spatial choice and search (see, for example, Ben-Akiva and Lerman
1985, Fischer et al. 1990, Fischer and Nijkamp 1985, 1987) and spatial dynamic
analysis (see, for example, Donaghy 2001, Nijkamp and Reggiani 1998). Spatial
data analysis includes procedures for the identification of the characteristics of
georeferenced data, tests on hypotheses about patterns and relationships, and con-
struction of models that give meaning to patterns and relationships among geore-
ferenced variables.
The breadth of interest in spatial data analysis is evident from earlier books and
edited volumes in the field: Ripley (1981), Upton and Fingleton (1985), Anselin
(1988), Griffith (1988), Haining (1990), Cressie (1991), Fischer and Nijkamp
(1993), Fotheringham and Rogerson (1994), Bailey and Gatrell (1995), Fischer et
al. (1996), and Longley and Batty (1996). The continued vitality of the field over
the last decade is illustrated by the increasing recognition of the spatial dimension
in social science research that sometimes yields different and more meaningful re-
sults than analysis that ignores it. The expanding use of spatial analysis methods
and techniques reflects the significance of location and spatial interaction in
2 M. M. Fischer
theoretical frameworks, most notably in the new economic geography as em-
bodied in the work of Krugman (1991a, 1991b), Fujita et al. (1999) and others.
Central to the new economic geography is an explicit accounting for location and
spatial interaction in theories of trade and economic development. The resulting
models of increasing returns and imperfect competition yield various forms of
spatial externalities and spillovers whose spatial manifestation requires a spatial
analytic approach in empirical work (Goodchild et al. 2000).
The technology of spatial analysis has been greatly affected by computers. In
fact, the increasing interest in spatial analysis in recent years is directly associated
with the ability of computers to process large amounts of spatial data and to map
data very quickly and cheaply. Specialised software for the capture, manipulation
and presentation of spatial data, which can be referred to as Geographical Infor-
mation Systems [GIS], has widely increased the range of possibilities of organi-
sing spatial data by new and efficient ways of spatial integration and spatial inter-
polation. Coupled with the improvements in data availability and increases in
computer memory and speed, these novel techniques open up new ways of
working with geographic information. Spatial analysis is currently entering a
period of rapid change characterised by GeoComputation, a new large-scale and
computationally intensive scientific paradigm (see Longley et al. 1998, Openshaw
and Abrahart 2000, Openshaw et al. 2000, Fischer and Leung 2001).
The principal driving forces behind this paradigm are four-fold: First, the in-
creasing complexity of spatial systems whose analysis requires new methods for
modelling nonlinearities, uncertainty, discontinuity, self-organisation and conti-
nual adaptation; second, the need to find new ways of handling and utilising the
increasingly large amounts of spatial information from the geographic information
systems [GIS] and remote sensing [RS] data revolutions; third, the increasing
availability of computational intelligence [CI] techniques that are readily appli-
cable to many areas in spatial analysis; and fourth, developments in high perfor-
mance computing that are stimulating the adoption of a computational paradigm
for problem solving, data analysis and modelling. But it is important to note that
not all GeoComputation based research needs the use of very large data sets or re-
quires access to high performance computing.
The present collection of papers is intended as a convenient resource, not only
for the results themselves, but also for the concepts, methods and techniques use-
ful in obtaining new results or extending results presented here. The articles of this
volume may thus serve usefully as supplemental readings for graduate students
and senior researchers in spatial analysis from the perspective of GeoCompu-
tation. We have chosen articles and book chapters which we feel should be made
accessible not only to specialists but to a wider audience as well. By bringing
together this specific selection of articles and book chapters and by presenting
them as a whole, this collection is a novel combination.
The book is structured into four parts. PART I sets the context by dealing with
broader issues connected with GIS and spatial analysis. The chapters included
have been written for more general audiences. Spatial analysis is reviewed as a
technology for analysing spatially referenced data and GIS as a technology com-
prising a set of computer-based tools designed to store, process, manipulate,
Introduction 3
explore, analyse, and present spatially identified information. PART II deals with
key computational intelligence technologies such as neural networks and evolu-
tionary computation. Much of the recent interest in these technologies stems from
the growing realisation of the limitations of conventional statistical tools and mo-
dels as vehicles for exploring patterns and relationships in data-rich environments
and from the consequent hope that these limitations may be overcome by the ju-
dicious use of neural net approaches and evolutionary computation. These techno-
logies promise a new style of performing spatial modelling and analysis tasks in
geography and other spatial sciences. This new style gives rise to novel types of
models, methods and techniques which exhibit various aspects of computational
intelligence. The focus of PART III is on neural pattern classification in remote
sensing environments. It provides the necessary theoretical framework, reviews
many of the most important algorithms for optimising the values of parameters in
a network and – through various examples – displays the efficient use of adaptive
pattern classifiers as implemented with the fuzzy ARTMAP system and with
error-based learning systems based upon single hidden layer feedforward net-
works. Anyone interested in recent advances in neural spatial interaction model-
ling may wish to look at the final part of the volume which covers the latest, most
significant developments in estimation theory, and provides a number of insights
into the problem of generalisation.
PART I Spatial Analysis and GIS
PART I of the present volume is composed of four contributions:
(cid:120) Spatial Analysis in Geography (Chapter 2)
(cid:120) Spatial Interaction Models and the Role of Geographic Information Systems
(Chapter 3),
(cid:120) GIS and Network Analysis (Chapter 4), and
(cid:120) Expert Systems and Artificial Neural Networks for Spatial Analysis and
Modelling (Chapter 5).
These four contributions largely drawing on the work done in the GISDATA re-
search network of the European Science Foundation [1993-1997] will now be
briefly discussed.
Chapter 2, a state-of-the-art review of spatial analysis that has found entry in
Elsevier's International Encyclopedia of the Social and Behavioral Sciences,
views spatial analysis as a technology for analysing spatially referenced object
data, where the objects are either points [spatial point patterns, i.e. point locations
at which events of interest have occurred] or areas [area or lattice data, defined as
discrete variations of attributes over space]. The need for spatial analytic tech-
niques relies on the widely shared view that spatial data are special and require a
specific type of data processing. Two unique properties of spatial data are
worthwhile to note: spatial dependency and spatial heterogeneity. Spatial depen-
