Contract 07.1070 Project N930C MultiMode: A Multiscalar and Multiagent Modelling Framework for Assessing Sustainable Futures in a Globalised Environment Working Package 2 Development and Application of a Multi-scale Constrained Cellular Automata Model for Belgium Inge Uljee, Roger White, Anne Gobin, Guy Engelen Study accomplished under the authority of Belgian Federal Public Planning Service Science Policy Research programme: Science for a Sustainable Development (SSD) Contract SD/TA/01 July 2011 All rights, amongst which the copyright, on the materials described in this document rest with the Flemish Institute for Technological Research NV (―VITO‖), Boeretang 200, BE-2400 Mol, Register of Legal Entities VAT BE 0244.195.916. The information provided in this document is confidential information of VITO. This document may not be reproduced or brought into circulation without the prior written consent of VITO. Without prior permission in writing from VITO this document may not be used, in whole or in part, for the lodging of claims, for conducting proceedings, for publicity and/or for the benefit or acquisition in a more general sense. 2 Distribution List DISTRIBUTION LIST Public distribution VITO confidential information - Copyright 2011 VITO NV I Abstract ABSTRACT The key innovation proposed in MultiMode is its multi-scale and multi-agent integrative approach for assessing and forecasting the consequences of policies aimed at sustainable development. The project is organised into four work packages: Meta-Model of Policy Options and Scenarios (WP1), Multi-scale Constrained Cellular-Automata Land use model (WP2), Landscape Scale Agent-Based Model (WP3) and Stakeholder Dialogue and Feedbacks (WP4). This document reports on work carried out in WP2 with respect to the development of a Constrained Cellular-Automata Land use model. This model represents spatial dynamics in Belgium at three levels: 1. the National level, representing Belgium as one entity subjected to influences from abroad as quantified in scenario‘s developed in WP1; 2. the Regional level, representing Belgium in terms of its 43 arrondissements. The level of the arrondissements was chosen as a technical compromise providing sufficient regional differentiation while avoiding technical complication. As for the policy and planning relevance, the arrondissements are not the best choice as no policy making actually takes place at this level. 3. the Local level, representing Belgium as a regular grid of cells measuring 9 ha each (300 m by 300 m). From a technical point of view, this resolution is appropriate as the CA-algorithm applied enables working with cells ranging from 50 m to 1000 m. At both the National and the Regional level the population is represented as one cohort and the economy is represented by three aggregated sectors, namely Agriculture, Industry and Services. The latter are grouped on the basis of the NACE-codes. At the Local level 19 land uses are modelled, 7 of which are dynamic, 3 are passive and the remaining 9 are static. The model is equipped with a fairly simple transportation model. This is sufficient to analyse the complex interlinkages between transportation infrastructure and land use at both the Regional and the Local level of the model. However, it is too simple to deal with modal split, routing of traffic and detailed forecasting of congestion. The model developed is considered to meet closely the needs of the other WPs in MultiMode. Elementary robustness and consistency tests have been carried out with success. The scenarios developed in WP1 were implemented in the CCA modelling structure. A prerequisite was that the resolution of the cellular model and the land uses modelled were adequate to set the spatial constraints within which the agents of the ABM model of WP3 can operate. The focus of the ABM model is predominantly on agriculture and agri-environmental measures. A methodology has been developed to tune different spatial and statistical datasets of agricultural land uses in order to quantify the scenarios developed in terms of the parameters and variables of the model. The scenarios were computed and analysed by means of the CCA-model and the resulting land use time series were incorporated into the ABM-modelling of WP3. In parallel to the work on the above defined model and with precisely the same modelling output in mind, work has continued on an activity-based variable-grid cellular automata model (AB-CA). Its aim is to simplify the modelling approach primarily by reducing the need for embedding models of different kinds and enabling the integration with the agent-based model developed in WP3. The model was first extensively tested on a high-quality dataset available for Dublin, Ireland and applied to Belgium as soon as it met preset criteria with respect to performance, robustness and quality. The AB-CA model outperforms the CCA model in a number of aspects. It is now ready for additional development and tests relative to its practical application. VITO confidential information - Copyright 2011 VITO NV II Table of Contents TABLE OF CONTENTS Distribution List __________________________________________________ I Abstract ____________________________________________________ II Table of Contents _______________________________________________ III List of Figures ___________________________________________________ VI List of Tables ___________________________________________________ IX Chapter 1 Introduction ____________________________________________ 10 1.1 A Multiscalar and Multiagent Framework ________________________________ 10 1.2 Objectives of the Constrained Cellular Automata land use model _______________ 13 Chapter 2 Modelling concept ________________________________________ 16 2.1 National level __________________________________________________ 17 2.2 Regional level __________________________________________________ 17 2.3 Local level ____________________________________________________ 17 2.4 Interlinked dynamics structuring space _________________________________ 18 Chapter 3 Model description at the National level ___________________________ 20 3.1 Sectors and activities in the model ____________________________________ 20 3.2 Growth of activities at the National level ________________________________ 21 3.2.1 Population projections _________________________________________ 22 3.2.2 Employment projections ________________________________________ 22 Chapter 4 Model description at the Regional level __________________________ 25 4.1 Regional activities _______________________________________________ 27 4.2 Regional attractivity ______________________________________________ 30 4.3 Policy constraints on the regional demand _______________________________ 32 4.4 Transportation infrastructure ________________________________________ 32 4.5 Interregional distances ____________________________________________ 33 4.6 Intraregional distances ____________________________________________ 34 4.7 Density _______________________________________________________ 34 4.8 Demand for space _______________________________________________ 37 4.9 Total employment _______________________________________________ 38 4.10 Introduction of a new activity ________________________________________ 38 4.11 Synthetic cellular properties _________________________________________ 39 VITO confidential information - Copyright 2011 VITO NV III Table of Contents Chapter 5 Model description at the Local level ____________________________ 40 5.1 Land uses _____________________________________________________ 42 5.2 CA-neighbourhood _______________________________________________ 43 5.3 CA-transition rules _______________________________________________ 44 5.4 Transition potential _______________________________________________ 46 5.5 Land use allocation ______________________________________________ 47 5.6 Neighbourhood effect _____________________________________________ 48 5.7 Accessibility ___________________________________________________ 48 5.8 Suitability _____________________________________________________ 50 5.9 Zoning _______________________________________________________ 51 Chapter 6 Agricultural land use and linkages with both the ABM and CCA modelling frameworks ____________________________________________ 54 6.1 Introduction and objectives _________________________________________ 54 6.2 Comparison of Belgian databases on agricultural land use/cover _______________ 55 6.2.1 The Belgian Corine Land Cover (BE-CLC) ___________________________ 55 6.2.2 Belgian statistical data _________________________________________ 56 6.2.3 Belgian cadastral parcels _______________________________________ 57 6.3 Historical trends in agricultural area ___________________________________ 57 6.3.1 Proposed agri-environmental classification ___________________________ 57 6.3.2 Trends in agricultural surface area and future evolution __________________ 58 6.3.3 Trends in production units, agricultural production and farm decision making ____ 69 6.4 Implementation of scenarios ________________________________________ 70 6.4.1 Demand for agricultural area in Belgium _____________________________ 70 6.4.2 Spatial allocation of agricultural area at the arrondissement level ____________ 73 6.5 Conclusion ____________________________________________________ 74 Chapter 7 Simulating scenarios with the Land use Model _____________________ 75 7.1 Introduction: procedure to quantify scenarios _____________________________ 75 7.2 Output of scenarios ______________________________________________