Skip to main content
main-content

Über dieses Buch

Geocomputation may be viewed as the application of a computational science paradigm to study a wide range of problems in geographical systems contexts.
This volume presents a clear, comprehensive and thoroughly state-of-the-art overview of current research, written by leading figures in the field.
It provides important insights into this new and rapidly developing field and attempts to establish the principles, and to develop techniques for solving real world problems in a wide array of application domains with a catalyst to greater understanding of what geocomputation is and what it entails.
The broad coverage makes it invaluable reading for resarchers and professionals in geography, environmental and economic sciences as well as for graduate students of spatial science and computer science.

Inhaltsverzeichnis

Frontmatter

GeoComputational Modelling — Techniques and Applications: Prologue

1. GeoComputational Modelling — Techniques and Applications: Prologue

Abstract
GeoComputation may be a research paradigm still in the making, but it has the potential to dramatically change current research practice in the spatial sciences. Different people, however, have different views of this research paradigm. For some it is synonymous with geographical information systems [GIS]. But GeoComputation is not GIS. There does exist a relationship with GIS, defined as geographical information science, but GeoComputation has other relationships too that are just as important, for example, with computer science, pattern recognition and statistics, neurocomputing and computational intelligence technologies.
Manfred M. Fischer, Yee Leung

Concepts, Modelling Tools and Key Issues

Frontmatter

2. Computational Neural Networks — Tools for Spatial Data Analysis

Abstract
The proliferation and dissemination of digital spatial databases, coupled with the ever wider use of Geographic Information Systems [GIS] and Remote Sensing [RS] data, 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 that ignores it.
Manfred M. Fischer

3. Evolving Computational Neural Networks Through Evolutionary Computation

Abstract
Computational neural networks [CNNs] have been used widely in many application areas in recent years. Most applications use feedforward CNNs and the backpropagation [BP] training algorithm. There are numerous variants of the classical BP algorithm and other training algorithms, but all these training algorithms assume a fixed CNN architecture. They only train weights in the fixed architecture that includes both connectivity and node transfer functions [see also Chapter 8 in this volume]. The problem of designing a near optimal CNN architecture for an application remains unsolved. This is an important issue, because there is strong biological and engineering evidence to support the contention that the function, i.e. the information processing capability of an CNN, is determined by its architecture.
Xin Yao

4. Neural and Evolutionary Computation Methods for Spatial Classification and Knowledge Acquisition

Abstract
Non-linearity, complexity and dynamics have become a focal point of research in spatial analysis, especially in the analysis of spatial data. Regardless of what we are dealing with, the need to handle systems with a high degree of complexity is now the rule rather than the exception. With our spatial systems becoming more and more complex and highly fluid, non-linearity prevails and evolution is full of surprises. It is essential to develop approaches which can effectively analyse complexity, non-linearity and dynamics in spatial systems in general, and in particular spatial classification and knowledge acquisition.
Yee Leung

5. Cellular Dynamics: Modelling Urban Growth as a Spatial Epidemic

Abstract
Contemporary urban growth consists of three interrelated problems of spatial dynamics: the decline of central or core cities which usually mark the historical origins of growth, the emergence of edge cities which both compete with and complement the functions of the core, and the rapid suburbanization of the periphery of cities — core and edge — which represent the spatially most extensive indicator of such growth. Our understanding of these growth processes is rudimentary, notwithstanding at least 50 years of sustained effort in their analysis. Our abilities to ‘control and manage’ such growth or ‘sprawl’ as it is colloquially and often pejoratively referred to, is virtually non-existent despite occasional but short lived successes through planning instruments such as green belts. The suburbanization of cities and methods for the control of such growth go back to the origins of cities themselves. Urban history reveals a succession of instruments used to separate the growing city from its suburbs. Documented examples refer to Ur in Sumeria, ancient Rome, to Elizabethan London, where edicts were in place to ensure the quality of life in the core city by restricting overbuilding and access (Morris 1979). However the concept of suburb has changed through history. Jackson (1985) sums this up quite cogently when he says: ‘... the suburb as a residential place is as old as civilization... However, suburbanization as a process involving the systematic growth of fringe areas at a pace more rapid than that of core cities... occurred first in the United States and Britain, where it can be dated from about 1815’ (p. 130).
Michael Batty

Spatial Application Domains

Frontmatter

6. Spatial Pattern Recognition in Remote Sensing by Neural Networks

Abstract
Neural networks are sophisticated pattern recognition tools. Over the last decade, following the early pioneering work of Benediktsson, Swain and Ersoy (1990) and Hepner et al. (1990), they have grown significantly in popularity as tools for the analysis of remote sensing data, primarily for the purpose of deriving spatial information used in environmental management. This popularity is due to several factors:
  • Remote sensing applications are data rich [they rely increasingly on high dimensional imagery].
  • Remote sensing data arise from complex physical and radiometric processes.
  • There has been dissatisfaction with more conventional pattern recognition algorithms.
  • There is a general desire to get maximum accuracy out of the data in remote sensing; high accuracy in some applications [e.g. agricultural subsidy fraud monitoring, mineral resource location] has high economic value.
  • Neural networks are suitable for data for which statistical properties are unknown or poorly understood.
  • Neural networks have potential scalability.
  • Neural networks are potentially adaptable to parallel machine architectures.
Graeme Wilkinson

7. Fuzzy ARTMAP — A Neural Classifier for Multispectral Image Classification

Abstract
Spectral pattern recognition deals with classifications that utilize pixel-by-pixel spectral information from satellite imagery. The literature on neural network applications in this area is relatively new, dating back only about six to seven years. The first studies established the feasibility of error-based learning systems such as backpropagation (see Key, Maslanik and Schweiger 1989; McClellan et al. 1989; Benediktsson, Swain and Ersoy 1990; Hepner et al. 1990). Subsequent studies analyzed backpropagation networks in more detail and compared them to standard statistical classifiers such as the Gaussian maximum likelihood (see Bischof; Schneider and Pinz 1992; Kanellopoulos, Wilkinson and Mégier 1993; Fischer et al. 1994).
Sucharita Gopal, Manfred M. Fischer

8. Neural Spatial Interaction Models

Abstract
One of the major intellectual achievements and, at the same time, perhaps the most useful contribution of spatial analysis to social science literature has been the development of spatial interaction models. Spatial interaction can be broadly defined as movement of people, commodities, capital and/or information over geographical space (see Batten and Boyce 1986). Such interaction encompasses such diverse behaviour as migration, travel-to-work, shopping, recreation, commodity flows, capital flows, communication flows [e.g. telephone calls], airline passenger traffic, the choice of health care services, and even the attendance at events such as conferences, cultural and sport events (Haynes and Fotheringham 1984). In each case, it results from a decision process in which an individual trades off in some way the benefit of the interaction with the costs entailed in overcoming the spatial separation between the point of departure and the destination. It is the pervasiveness of this type of trade-off in spatial behaviour which has made spatial interaction analysis and modelling so important, and the subject of intensive investigation in human geography and regional science (Fotheringham and O’Kelly 1989).
Manfred M. Fischer

9. A Neural Network Approach for Mobility Panel Analysis

Abstract
Currently, neural networks are being widely discussed in many research areas. But despite this clear interest, it is curious to find such a large gap between theoretically based research activities and real world applications. Especially in the field of social sciences, worthwhile applications of computational neural networks [CNNs] are the exception rather than the rule.
Günter Haag

Backmatter

Weitere Informationen