Applied Spatial Modelling and Planning

John R. Lombard, Eliahu Stern, Graham Clarke, John R. Lombard, Eliahu Stern, Graham Clarke

• Publisher: Routledge
• Year: 2016
• Print ISBN: 9781138925700
• eBook ISBN: 9781317406723
• Edition: 1
• Topic: Negocios y empresa
• Subtopic: Negocios en general

This book highlights the extraordinary range of areas to which geographical analysis and spatial modelling can bring lessons and insights. It shows how these techniques have been used to address 'real world' issues that are of concern to international organisations, public agencies and businesses, as illustrated by actual funded projects that geographers have developed collaboratively with end-users.

Applied Spatial Modelling and Planning shows how much geographical research is policy relevant to a wide variety of agencies through the use of GIS and spatial modelling in applied geography. The book's chapters contain a cross-section of innovative applications and approaches to problem solving within five major domains of the dynamics of economic space, housing and settlements, population movements and population ageing, health care, and the environment. Using a number of case studies on the use of GIS and spatial modelling, this book demonstrates the fact that much of what is done by quantitative geographers is not only relevant within academia, but also has use in policy work.

This book will appeal to an international audience interested in cutting-edge spatial modelling to better understand the processes involved in solving real problems.

1 Introduction

John Lombard, Eliahu Stern and Graham Clarke

This is the fourth volume of applications of geographical modelling and planning produced by members and friends of the International Geographical Union’s Commission on ‘Applied Geography’. Each year the Commission holds two conferences and invites contributions from those geographers (and related disciplines) who work actively with public or private sector planners on real-world issues. We believe these contributions help to show how much geographical research is policy relevant to a wide variety of agencies. In this particular volume we explore the use of GIS and spatial modelling in applied geography. Although still heavily criticised by many ‘critical’ human geographers, the field of GIS and spatial modelling has been transformed since the quantitative revolution of the 1960s. The field is now much more policy relevant and the 19 chapters in this book show research undertaken with many different types of policy makers.

Before the individual applications are presented, Sir Alan Wilson gives an overview of applied spatial modelling in a short paper relating to his keynote address to the Commission in 2013. He has been a leading figure not only of spatial modelling in geography and planning but very much of applied modelling work, setting up the consultancy company ‘GMAP’ with colleagues in the early 1990s. He reflects on the progress we have collectively made in applied modelling through a variation on the traditional SWOT analysis. He concludes that urban modellers need to tap into the current policy environment around big data, arguing that we need ‘big science’ to tackle effectively many of the biggest problems in our cities and regions. The paper provides a great deal of food for thought.

The first main section that follows provides a set of three studies dealing with the dynamics of economic space in order to find, or to formulate, the best policy for their given situations. Starting with the analysis of spatial disparity of the rental housing market, Jitendra Parajuli and Kingsley E. Haynes examine the distribution of new firm formation in New England from 1999 to 2009 using entropy and entropy decomposition. Their study provides an understanding of the distribution of single-unit firm births in a region, over a period in which the US experienced major economic disruptions resulting from recession, terrorism and global financial crisis. In addition, their study also provides the distributional patterns of new firms in various industry sectors as well as comparing the distribution of new firm births with other economic indicators. The findings are useful to regional planning agencies that are formulating policies to attract new businesses, to create jobs and to promote economic growth. At the same time, the study also demonstrates the application of the entropy-based method to examine spatial variation of single-unit firm births, which could be replaced by other spatial variables such that their distributions could also be studied.

The second study in this first section investigates the human capital implications of a policy change resulting from the introduction of the 485 graduate visa in Australia in 2007. The authors, Jonathan Corcoran, Francisco Rowe, Alessandra Faggian and Robert Stimson, provide an exploratory analysis of a specific aspect of the skilled immigration phenomenon in Australia resulting from this recent visa policy change. This change precipitated a sharp increase in the number of international students who then stayed-on post-graduation and entered the workforce. The study enabled an analysis and comparison of the ‘before’ and the ‘after’ patterns of regional distribution of those immigrants as they transit from higher education to employment. It thus sheds light on the fields of employment, working conditions and inter-regional migration of patterns of those stay-on graduates and their potential impacts on Australian regions. The main interested party has been the Australian Government’s Immigration team.

The last example for the ‘dynamics of economic space’ section is provided by Robert Baker, who deals with the laws of spatial interaction modelling. He argues that Tobler’s law of distance decay can be deconstructed into two further ‘laws’ relevant to the digital economy, namely, that ‘all things can be, but not necessarily will be, connected globally’ and, secondly, ‘distance always matters, but time matters more’. The application of these new ‘laws’ to origin-destination trips are shown in a number of examples, such as the time-sensitivity in stock market transaction trades, satellite reception lags and filtering retail call-centre operators. In a more applied context Baker considers the implications of a new set of spatial interaction models (especially relating to e-commerce) on the traditional location models of Australian retailers.

The second section provides examples of applied spatial modelling and planning in the housing and settlements arena. Yan Liu, David Wadley and Jonathan Corcoran bring a new perspective to the analysis of the spatial disparity in the rental housing market. Its purpose is to advance understanding of affordability which often emerges as a major issue in developed countries. They bring a new perspective to the issue by linking graduated household income and rent outgoings data at the neighbourhood scale. A set of spatial metrics is developed to map and visualise the geographical disparity in the supply of, and demand for, private rental housing stocks. Principal Component Analysis, followed by a two-step cluster analysis method, is employed to define the spatial typology of the private market at a localised scale known as the State Suburb. The methodology is innovative since it brings a new perspective to the analysis of current, not previously linked, income and rent data at the neighbourhood scale to identify the spatial and structural grouping or clustering of suburbs. The results reveal four distinctive spatial clusters, representing a range of socio-economic and demographic outcomes at the local neighbourhood level. The divergence of the private rental market among the different suburbs indicates the pertinence of community and people-based policy responses and solutions to housing problems. It is also found that the extent of affordability over many suburbs can be reduced when households of more than sufficient income move down market in their choice of rental housing.

A deeper look into the housing market problems is provided by Paul Bidanset and John Lombard, who improve the performance and the predictability power of a mass real-estate appraisal model. They are using a geographically weighted regression across Gaussian and bisquare kernels with both fixed and adaptive bandwidths. They found that continuous kernels achieve superior city-wide dispersion levels, particularly exponential kernels with fixed bandwidths. Additionally, neighbourhood disaggregation of dispersion and price-related differential levels reveal that exponential kernels with fixed bandwidths significantly reduce the amount of sub-markets suffering from regressivity and relatively lower uniformity. The application of spatial modelling therefore aims to promote equity, uniformity and overall government accountability within the realm of ad valorem property tax.

A different tool of applied spatial modelling is presented by Eliahu Stern, who provides an example of the never-ending practice of spatial search. It is a GIS-based search aimed at finding locations for housing the growing ultra-orthodox population in Israel. Bounded by several sectorial restrictions like proximity, and/or high accessibility to Jerusalem, and low-rise buildings, the Ministry assigned the author to find feasible locations to accommodate a minimum of 50,000 dwelling units. Feasibility is defined by several different objective and subjective indicators. A paired-comparison analysis of location criteria is also used to examine the rank-orders of feasible locations by the leaders of the ultra-orthodox population and the community of urban planners. An agreeable solution was achieved. However, despite the comprehensive spatial search procedure, at the end, one should expect, to some extent, public objection to a near-by ultra-orthodox settlement, a common NIMBY phenomenon.

The final example in the housing and settlements arena is the joint work of Yair Grinberger and Daniel Felsenstein, who simulate the long-run impacts of an earthquake on the urban system. They present the first of a number of dynamic agent-based models in the book. This model simulates the disaster outcomes in two different urban contexts in Israel: Jerusalem and Tel Aviv. Attention is paid to the effectiveness of urban policies aimed at restoring the urban equilibrium. The policies relate to land use regulation, public provision of shelter and the restoration of damaged urban services. Results show that a similar shock in two different locations results in very different outcomes. Policy simulations imply that interventions directed at rebounding to pre-shock state do not do well and may even inhibit urban stability.

In the next section we have three very diverse papers but all relating to the broad topics of population movements and population ageing. In the first chapter, Philip Rees, Pia Wohland and Paul Norman review the context in which population projections are needed in applied and planning analyses. Countries experiencing below replacement fertility and high international immigration are experiencing an ethnic transition from a state of low diversity to high diversity. This process is illustrated by the authors, drawing on a set of ethnic population projections for the UK which show dramatic growth in most Minority Ethnic populations, contrasting with stagnation or decline in the White British population. However, because the inputs to such projections are difficult to estimate and because there is more uncertainty because of the multiplicity of groups, they argue that it is vital to evaluate their accuracy. They compare the ten-year projections from a 2001 base with 2011 Census results. The projections seem to over-estimate the White British and White Irish populations because of over-optimistic mortality assumptions. As the groups with the highest share of the elderly, these two groups gain most from better survival into old age. The projections also under-project the growth of Black and Asian Minority populations because the official projections of all group immigration were much lower than the actual inflows from 2008 to 2011. So the UK’s multi-ethnic future was arriving faster than the analysis suggested. The authors thus worked to revise the inputs and assumptions of new projections of the UK’s local ethnic populations, based on the 2011 Census. This is invaluable information for a whole swathe of Government offices and service providers.

The second chapter in this section, by Les Mayhew and David Smith, also looks at the future population in the UK, in this case exploring different scenarios around population ageing and life expectancies. Many applied policy issues surround population ageing. A major Government concern is how to pay for a larger retired population. The authors speculate that people will be required to take additional steps in planning for their own financial needs in old age and to become less dependent on the state. However, they argue that geographers and policy makers should be wary of taking official forecasts of the older population at face value. Given the fact that people will retire with very different financial resources, rapidly rising life expectancy will mean new coping mechanisms are required to help individuals and families understand the true costs of ageing for them. It will mean that available resources, for example pension savings, need to last longer. There will also be implications for service providers. Mayhew and Smith argue that the increasing uncertainty over how long a population or an individual will live will impact services affecting the whole care economy, types of employment and also health and social care services. This chapter argues that geography is well equipped to address the details but only if it is able to forecast ageing populations with reasonable accuracy and, in turn, the social and spatial ramifications are clearly understood.

In the next chapter by Nick Addis, population movements are captured through the application of agent-based models. As noted in the Grinberger and Felsenstein chapter, this is an important and growing spatial modelling technique (which will also be applied again in the chapters by Bithell and Fligg and Barros – see below). In this instance, the focus of enquiry is the movement of burglars through the urban landscape. The evolution of this new type of computational modelling approach has coincided with a culture whereby criminal justice agencies find themselves subjected to increasing levels of scrutiny over their performance. This has contributed to an increased awareness among these agencies of how such modelling approaches may be used to support crime prevention efforts. Addis’s chapter illustrates through case studies how agent-based models have been effectively applied to crime phenomena to help understand the underlying dynamics of existing crime systems, and how this technique may be supported through collaborations with criminal justice agencies.

The next section includes a series of chapters on health care planning and analysis. This is an area of geography which has always been very applied. Spatial simulation models and analysis can contribute hugely to the area of health care and health care planning.

The first chapter, by Holly Shulman, Graham Clarke and Mark Birkin, looks at the impact of opening new community hospitals in the West Yorkshire area of the UK. Since radical reforms were made to the UK National Health Service (NHS) in the 1990s, policy makers have tended to favour the promotion of economies of scale in health care provision and funding and have sought to concentrate surgical facilities in large, mainly urban hospitals. However, it is widely recognised that as service provision becomes more centralised, accessibility for patients becomes a more crucial consideration. Many geographical studies have shown that as distance from a service location increases, utilisation generally decreases. To counter increasing centralisation, the UK NHS is currently considering the policy of opening more local hospitals – smaller community hospitals. These are deemed to be especially important for serving older patients with greater mobility problems. Shulman, Clarke and Birkin explore access to hospital treatment for one major type of condition which could easily be treated in local (day care) community hospitals or ambulatory centres – cataracts (especially because surgery today has routinely fallen from a five-day inpatient stay to a day case). In order to evaluate the impact of new community hospitals they build two types of model. The first is a morbidity model for cataracts. This allows them to compare current geographies of hospital usage against potential demand (to explore whether there are areas of the city-region where they suspect patients are not getting the treatment they might need). Second, they then build a suite of spatial interaction models to reproduce the known flows of patients between an origin (patient residence) and destination (hospitals) to model the impacts on access of adding new community-style hospitals.

In the second chapter on health care, Melanie Tomintz and Victor Garcia-Barrios introduce simSALUD, a freely available and user-friendly web application that allows the simulation of health-related issues for small geographical zones. They demonstrate the power of this application through the estimation of small-area smoking rates. The application platform includes spatial microsimulation algorithms to allow users to produce synthetic data by combining different input data sources (normally a survey combined with small-area census-type data). Requiring no programming skills, the simulation model is run through a wizard-based web user interface, which guides the user through each step of model building. This allows non-experts in spatial microsimulation modelling to run powerful models for estimating a variety of health-related variables: that is, potentially smoking patterns, diabetes, obesity, and depression, variables which are not routinely available for small-area geographies. SimSALUD also provides a visualisation tool that allows the model results to be mapped, showing hot-spots of the simulated variable of interest. The visualisation tool is also designed in a way that health care planners, regional planners or non-experts from other sectors can access and use it easily, thus contributing to their future decisions on where to spend resources more effectively.

The third chapter in this section provides another example of the power of agent-based modelling. In this chapter Mike Bithell models infectious diseases at the scale of the individual, allowing him to include a number of factors, such as spatial structure and individual heterogeneity, which have often been neglected in more traditional aggregated approaches. In this chapter, an agent-based model of a primary school (children aged 4–11) is derived directly from observation of a real school. Laser-scanner measurements of the school building are used to determine its size and shape, the location of social spaces, and of obstacles to movement. Observations of classroom activity and the school timetable determine daily activity, and a fine-scale social-force model routes teachers and pupils through the school, while retaining the collective aspects that group together pupils in shared classroom spaces, or see them dispersed in the playground. All activities are set by a timetable that uses real-time units and real-space co-ordinates, so that the spread of a notional proximity-based infectious disease can be matched to observations. Preliminary results show how it is possible to separate disease-specific factors from social activity that mediates disease transfer.

In the fourth chapter in this health care section, Michelle Morris, Graham Clarke, Kimberley Edwards, Claire Hulme and Janet Cade (another multi-disciplinary team, this time at the University of Leeds) examine the growing problem of obesity. They provide a cross-sectional analysis using weight status data from a large cohort of UK women to explore the geographies of overweight and obesity at four different spatial scales, including a geodemographic classification, in order to identify areas with a higher likelihood of persons being overweight and obese. Statistical analysis is carried out using Stata statistical software and results are visualised using ArcGIS. Higher prevalence of obesity is observed in the North compared with the South of England and also in urban compared to rural areas. Significant differences are also shown to exist between the nine Government Office Regions in the UK. Once demographic characteristics are accounted for through the use of a geodemographic classification, it can be seen that those living in ‘Constrained by Circumstance’ and ‘Blue Collar Communities’ (the lowest income groups) have the highest odds of being overweight or obese. They conclude that such analysis (when undertaken for a combination of spatial scales) could be the best way forward when investigating the geographies of obesity and for producing meaningful results of use to policy makers.

The final chapter in this health care section is provided by a team of experts from a variety of different disciplines (Karyn Morrissey, Peter Kinderman, Eleanor Pontin, Sara Tai, Matthias Schwannauer) and addresses a growing concern in UK health planning – an increasing number of persons with mental health problems, particularly depression. There is considerable evidence of a social gradient in the prevalence of depression that may lead to lower levels of wellbeing among the population. Including the Index for Multiple Deprivation for England (2010) with the Stress Test, a survey launched on All in the Mind, a BBC Radio 4 programme, and developed by psychologists in Liverpool and Manchester Universities, Morrissey et al. explore the association between depression and area-level deprivation, controlling for demographic and socio-economic factors. This research shows that individuals in lower income categories have higher rates of depression, and within income categories, individuals with depression reside in relatively more deprived areas relative to those without depression.

In the final section we turn to environmental modelling and applied spatial modelling. In the first chapter Robert Fligg and Joana Barros present another agent-based model, this time for the Limpopo River Basin Area in Mozambique. More specifically, the study focuses on the evacuation procedure of a fast-flooding area in the vicinity of Xai-Xai, in the province of Gaza. First, Fligg and Barros simulate the phenomena of a fast-flooding area and evacuation behaviour; second, they develop a neural network designed to simulate an agent’s cognitive ability to sense, learn and adapt when travelling over a landscape during a flooding episode. The evaluation process undertaken at the end of the modelling exercise demonstrates there is potential for using this design of hybrid model of neural networks for further research about the evacuation behaviour of people in the fast-flooding area of Xai-Xai, which can be possibly extended to other geographic areas where models integrate human decision-making and land use/land cover.

Finally, Tal Svoray explores the issue of land degradation, an important topic in environmental management given that it is such a destructive phenomenon that causes damage to agricultural fields and neighbouring man-made infrastructures. Svoray provides a methodological framework to decrease land degradation based on the following three procedures: (1) the physical understanding of the processes by using GIS and spatial modeling; (2) identifying areas under risk; and (3) using the risk layer as an input to a spatial decision support system for prioritising actions. The framework is partially demonstrated on wh...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• List of plates
• List of figures
• List of tables
• List of contributors
• 1 Introduction
• 2 Applied spatial modelling: ‘big science’ and ‘best practice’ challenges
• PART I Dynamics of economic space
• PART II Housing and settlements
• PART III Population dynamics and population ageing
• PART IV Health care planning and analysis
• PART V Environmental modelling
• Index