Annex

Table 3: Household life cycle model for residential relocation behaviour [SE_1]

Name of model	Household life cycle model for residential relocation behaviour
Sources	Nijkamp et al. (1993 )
Technical data
Application area	Covered area, physical boundaries	Case study: Greater Amsterdam Area	Extent of area	350 square miles / About 800,000 people

	Spatial units	20 zones	Size or grain of grids/zones	–
Time horizon	Time step	1 year	Duration of model run	1971 – 1984
Modelling approach	Simulation technique	Spatial economics	Qualitative or quantitative	Quantitative
Contents
Main purpose	Modelling household life cycles and their impact on residential relocation behaviour and the urban housing market for a European capital city.
Main variables with relationships	(1) households, (2) migration, (3) occupancy, (4) housing demand, (5) dwelling supply in zones and dwelling types, (6) allocation of households.
Human decision making	Domain	Not explicitly	Temporal range	–

	Typology (classes) of agents?	Allocation of household	→ if yes: what types?	Households: single, 2-person household, 3-person household, 4+ person household, non-household

	Decision algorithm	Rational choice, maximum utility	Input into decision	Population and household data
Goals	Authors’ opinion	Successful runs, validation and scenarios.
Model development process	Concept	Given	Quantification of relationships	Empirical data

Table 4: Simulation of demographic changes and the housing market [SE_2]

Name of model	Simulation of demographic changes and the housing market
Sources	Mankiw and Weil (1989 )
Technical data
Application area	Covered area, physical boundaries	U.S. cities (census)	Extent of area	– / 203,190 people / 74,565 households

Application area	Spatial units	U.S. cities (census)	Size or grain of grids/zones	–
Time horizon	Time step	1 year	Duration of model run	1970 – 2007 or 2020
Modelling approach	Simulation technique	Spatial economics	Qualitative or quantitative	Quantitative
Contents
Main purpose	Simulation of demographic changes (baby boom and baby bust) and its influences on the housing market in the U.S.
Main variables with relationships	(1) population, (2) households, (3) housing market (demand, prices), (4) economy (GNP)
Human decision making	Domain	Not explicitly	Temporal range	–

	Typology (classes) of agents?	Allocation of household	→ if yes: what types?	Dummy household

	Decision algorithm	Rational choice, maximum utility	Input into decision	Census data
Goals	Authors’ opinion	Successful runs, validation and scenarios.
Model development process	Concept	Given	Quantification of relationships	Empirical data

Table 5: A System Dynamics Approach to Land Use / Transportation System Performance Modeling [SD_2]

Name of model	A System Dynamics Approach to Land Use / Transportation System Performance Modeling
Sources	Haghani et al. (2003a ,b )
Technical data
Application area	Covered area, physical boundaries	Varies with application area; Case study: Montgomery County	Extent of area	– / About 800,000 people

Application area	Spatial units	U.S. cities (census)	Size or grain of grids/zones	–
Time horizon	Time step	1 year	Duration of model run	C: 1970 – 1980 V: 1980 – 1990
Modelling approach	Simulation technique	Spatial economics	Qualitative or quantitative	Quantitative
Contents
Main purpose	Integrated land-use and transportation model for estimating scenarios regarding transport policies
Main variables with relationships	Seven sub-models: (1) population, (2) migration, (3) household, (4) job growth, employment and commercial land development, (5) housing development, (6) travel demand and (7) congestion.
Human decision making	Domain	Not explicitly	Temporal range	–

	Typology (classes) of agents?	Cohorts within population sub-model	→ if yes: what types?	Persons: age 0–17, 18–44, 45–64, 65 male and female / Households: single, married with children, married without children, male or female with children, other

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	First step is achieved, successful validation and scenarios.
Model development process	Concept	Not stated	Quantification of relationships	Empirical data

Table 6: CLUE-s (Conversion of Land Use and its Effects) [CA_1]

Name of model	CLUE-s (Conversion of Land Use and its Effects)
Sources	Verburg and Overmars (2007 )
Technical data
Application area	Covered area, physical boundaries	User-specified / Several examples published	Extent of area	User-specified

Application area	Spatial units	CLUE: soft-classified data (large pixels with fraction of land-uses)	Size or grain of grids/zones	User-specified / CLUE: 7 to 32 km / CLUE-s: 20 to 1,000 m
Time horizon	Time step	Iterative process stops when demand for land-use meets allocated area	Duration of model run	–
Modelling approach	Simulation technique	Cellular automata	Qualitative or quantitative	Quantitative
Contents
Main purpose	Tool for understanding land-use patterns, possible future scenarios for given demand
Main variables with relationships	Input: Pre-defined change in demand for land by different sectors for whole simulation area → CLUE-s assigns new land-uses per grid Each cell: most preferred land use based on suitability of location and competitive advantage of different land use types (demand), check: is land use change allowed? If no: next most preferred land use is chosen
Human decision making	Domain	Not explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Case-study specific
Model development process	Concept	Not mentioned	Quantification of relationships	User-specified: empirical analysis, expert knowledge, spatial interactions, conversion elasticities

Table 7: CUF-2 (California Urban Futures) [CA_2]

Name of model	CUF 2 (California Urban Futures)
Sources	Landis and Zhang (1998a ,b )
Technical data
Application area	Covered area, physical boundaries	San Francisco Bay Area (California)	Extent of area	1.8 million ha

Application area	Spatial units	Grid cells	Size or grain of grids/zones	100 × 100 m
Time horizon	Time step	Econometric: 10 years / Probabilities for land use change: once per simulation	Duration of model run	C: 1985 – 1995 / S: ?
Modelling approach	Simulation technique	Cellular automata	Qualitative or quantitative	Quantitative
Contents
Main purpose	Simulating urban growth, scenarios for future development
Main variables with relationships	Top-down approach: future trends of population, household, jobs → are assigned to grid cells Econometric models predict future population, households, employment (10 year intervals) LUC-model: estimates probabilities for land use change out of historical data, and simulation engine assigns probabilities to cells Probability of land use change (multinomial logit models) for a cell from i to j = f (initial site use, site characteristics, site accessibility, community characteristics, policy factors, relationships to neighbouring sites) → probabilities are interpreted as bids for (re-)development → population and jobs are assigned to cells by bids 7 urban land-use categories: undeveloped, single-family residential, multifamily residential, commercial, industrial, transportation, public
Human decision making	Domain	Not explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Achieved
Model development process	Concept	Not mentioned	Quantification of relationships	Calibration using maps of land use change

Table 8: CURBA (California Urban and Biodiversity Analysis) [CA_3]

Name of model	CURBA (California Urban and Biodiversity Analysis)
Sources	Landis et al. (1998 )
Technical data
Application area	Covered area, physical boundaries	San Francisco Bay Area (California)	Extent of area	See CUF-2

Application area	Spatial units	Grid cells	Size or grain of grids/zones	100 × 100 m
Time horizon	Time step		Duration of model run
Modelling approach	Simulation technique	Cellular automata	Qualitative or quantitative	Quantitative
Contents
Main purpose	Development of policy scenarios of urban growth, impact on habitat change/biodiversity
Main variables with relationships	Two components: (1) urban growth model and (2) policy simulation and evaluation model / Urban growth model is based upon CUF-2 Policy simulation and evaluation: several growth scenarios → impact on habitat change and habitat fragmentation
Human decision making	Domain	No explicit decision making	Temporal range	-

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Achieved
Model development process	Concept	See CUF-2	Quantification of relationships	See CUF-2

Table 9: ILUMASS (Integrated Land-Use Modelling and Transportation System Simulation) [ABM_1]

Name of model	ILUMASS (Integrated Land-Use Modelling and Transportation System Simulation)
Sources	Strauch et al. (2003 ); Moeckel et al. (2006 )
Technical data
Application area	Covered area, physical boundaries	Dortmund and its 25 surrounding municipalities	Extent of area	About 2,000 km² / 2.6 million people

Application area	Spatial units	Statistical zones (total: 246) and grid cells	Size or grain of grids/zones	Grid cells: 100 × 100 m
Time horizon	Time step	One year	Duration of model run	S: 2000 – 2030
Modelling approach	Simulation technique	Coupled simulation system including agent-based simulations	Qualitative or quantitative	Quantitative
Contents
Main purpose	Dynamic simulation model with a focus on urban traffic flows, including activity behaviour, changes in land use, and effects on environment
Main variables with relationships	Five modules (+ integration module): 1. changes in land use, 2. activity patterns and travel demand, 3. traffic flows, 4. goods transport, 5. environmental impacts of transportation and land use Land use → demand for spatial interaction (work, shopping trips, etc.) → traffic → environmental impacts Feedbacks: (a) transport → accessibility of locations → location decisions of households, firms, developers. (b) environmental factors → location decisions (e.g., clean air, traffic noise) Land use module: moving households, location of firms, investment of developers, new industrial area
Human decision making	Domain	Various, e.g., transport, household location, daily activity plans	Temporal range	Depending upon domain (daily travel behaviour vs. moving)

	Typology (classes) of agents?	Yes	→ if yes: what types?	Not mentioned

	Decision algorithm	Various (Markov, Logit, Monte-Carlo)	Input into decision	Depending upon domain, feedbacks included
Goals	Authors’ opinion	Time of report: work in progress, later papers all focus on single modules
Model development process	Concept	Not mentioned	Quantification of relationships	Not mentioned

Table 10: ILUTE (Integrated Land Use, Transportation, Environment model) [ABM_2]

Name of model	ILUTE (Integrated Land Use, Transportation, Environment model)
Sources	Salvini and Miller (2005 ); Miller et al. (2004 )
Technical data
Application area	Covered area, physical boundaries	Tests for Toronto area	Extent of area	5 million people

Application area	Spatial units	Two versions: grids and buildings	Size or grain of grids/zones	2 parallel approaches: Grid: 30 × 30 m / Buildings as objects
Time horizon	Time step	Varying with sub-models	Duration of model run	V: 1986 – 2001 / S: 10 – 20 years into future
Modelling approach	Simulation technique	Agent-based simulation	Qualitative or quantitative	Quantitative
Contents
Main purpose	Evolution of an entire urban region with emphasis on transport
Main variables with relationships	Land development → location choice → activity schedule → activity patterns → back to land development and all other variables in chain transportation network → automobile ownership → travel demand → network flows → back to transportation network and all other variables in chain influences
Human decision making	Domain	Activity/travelling scheduling, route choice, real estate market, behaviour of economy, land development, household ownership	Temporal range	Depends upon domain. E.g.: typical travel day is computed once per simulation year per agent type.

	Typology (classes) of agents?	Yes	→ if yes: what types?	For households, individuals, firms

	Decision algorithm	Rule-based: reducing number of choices / logit model for selecting the “best” option	Input into decision	Not mentioned
Goals	Authors’ opinion	Work in progress
Model development process	Concept	Not mentioned	Quantification of relationships	Empirical data

Table 11: Modelling biodiversity and land use [SD_3]

Name of model	Modelling biodiversity and land use
Sources	Eppink et al. (2004 )
Technical data
Application area	Covered area, physical boundaries	No explicit representation of a specific area. Urban region with surrounding area including wetlands	Extent of area	–

Application area	Spatial units	No spatial resolution	Size or grain of grids/zones	–
Time horizon	Time step	1 year	Duration of model run	S: 100 years
Modelling approach	Simulation technique	System dynamics	Qualitative or quantitative	Qualitative
Contents
Main purpose	Assessing the impact of urban sprawl on wetland biodiversity and social welfare
Main variables with relationships	Population growth within city → higher population density and more need for agricultural land → expansionists attempt to buy surrounding area → change of wetland area to urban area & more agriculture decrease wetland biodiversity → conservationists’ valuation of remaining biodiversity increases → conservationists buy wetland area for nature protection
Human decision making	Domain	Human decision making is represented within system dynamics equations	Temporal range	1 year

	Typology (classes) of agents?	Yes	→ if yes: what types?	Expansionists, conservationists (see above) and owners of land

	Decision algorithm	Land is sold to the highest bidder	Input into decision	Prices offered by conservationists and expansionists.
Goals	Authors’ opinion	First step for improving relationship between economic development and biodiversity
Model development process	Concept	Not mentioned	Quantification of relationships	Not mentioned

Table 12: MOLAND [CA_4]

Name of model	MOLAND
Sources	Engelen et al. (2007 )
Technical data
Application area	Covered area, physical boundaries	Several examples across Europe and elsewhere	Extent of area	User-specified

Application area	Spatial units	global: 1 zone / regional: zones, typically NUTS / local: grid cells	Size or grain of grids/zones	User-specified
Time horizon	Time step	annual	Duration of model run	C: last 40 – 50 years / S: user-specified, normally 30 years
	Simulation technique	Mainly rule-based cellular automata	Qualitative or quantitative	Quantitative
Contents
Main purpose	To monitor developments of urban areas and identify trends at the European level, focus is on growth scenarios
Main variables with relationships	Growth of economy and population (global level) → growth in competing regions (regional level), sets boundaries for all cells in a region → rules for land use change at the grid level: physical suitability, institutional suitability (e.g., planning documents), accessibility (via transport network), dynamics at the local level (land use functions attracting or repelling each other) Feedback from grid level to regional level: spatial distribution leads to quality and availability of space for different activities, which influences comparative attractiveness of a region
Human decision making	Domain	No explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Achieved
Model development process	Concept	Not mentioned	Quantification of relationships	Calibration with historical data

Table 13: PUMA (Predicting Urbanisation with Multi-Agents) [ABM_3]

Name of model	PUMA (Predicting Urbanisation with Multi-Agents)
Sources	Ettema et al. (2007 )
Technical data
Application area	Covered area, physical boundaries	North Dutch Ranstadt (including Amsterdam, Utrecht, Schiphol airport)	Extent of area	3.16 million inhabitants

Application area	Spatial units	Grid cells (and travel zones)	Size or grain of grids/zones	500 × 500 m
Time horizon	Time step	1 year / later: up to daily	Duration of model run	S: 2000 to approx. 2050
Modelling approach	Simulation technique	Agent-based simulation	Qualitative or quantitative	Quantitative
Contents
Main purpose	Predicting urbanisation using behavioural agents
Main variables with relationships	Demographic change → decisions of individuals → land use change / Not yet implemented: developers, authorities and firms/institutions (so far exogenous) [impact of household’s decisions on land use not described]
Human decision making	Domain	1. demographic events (no decisions, just stochastic) 2. residential relocation 3. job changes	Temporal range	Annual [Daily decisions in future work]

	Typology (classes) of agents?	Yes	→ if yes: what types?	Households: Number of adults and children; age of household head [dwellings are agents as well]

	Decision algorithm	Rational choice with utility maximisation	Input into decision	Residential relocation: characteristics of dwelling, commuting distance, socio-demographics / Job choice: salary, job type, distance to dwelling, personal preferences…
Goals	Authors’ opinion	Promising approach, still work in progress
Model development process	Concept	Empirical data	Quantification of relationships	Empirical data

Table 14: Rotterdam urban dynamics [SD_4]

Name of model	Rotterdam urban dynamics
Sources	Sanders and Sanders (2004 )
Technical data
Application area	Covered area, physical boundaries	Rotterdam	Extent of area	100,000 acres

Application area	Spatial units	16 grid cells called “zones”	Size or grain of grids/zones	Squares with 3,125 miles each side
Time horizon	Time step		Duration of model run	S: 250 years
Modelling approach	Simulation technique	System dynamics	Qualitative or quantitative	Quantitative
Contents
Main purpose	Redefining the model of urban dynamics by Forrester (1969 *), including: 1. spatial dimension (16 squares) and 2. disaggregation: different types of housing, industry, and people in zones
Main variables with relationships	Bi-directional causal loops between: population, housing availability, houses, land availability, business structures, and job availability (linked with population) / Two markets: labor market and housing market compete for land / (no transportation)
Human decision making	Domain	No explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Case of Rotterdam only as an example for generic results
Model development process	Concept	Not mentioned	Quantification of relationships	Out of statistical data and expert knowledge

Table 15: SCOPE (South Coast Outlook and Participation Experience) [SD_3]

Name of model	SCOPE (South Coast Outlook and Participation Experience)
Sources	Onsted (2002 )
Technical data
Application area	Covered area, physical boundaries	South Coast of Santa Barbara County	Extent of area	137,000 acres / Approx. 200,000 inhabitants

Application area	Spatial units	No spatial resolution	Size or grain of grids/zones	–
Time horizon	Time step		Duration of model run	V: 1960 – 2000 / S: 2000 – 2040
Modelling approach	Simulation technique	System dynamics	Qualitative or quantitative	Quantitative
Contents
Main purpose	Simulation model to provide scenarios for future land use in Santa Barbara, e.g., with restrictions to urban growth
Main variables with relationships	Five sectors: housing, population, business, quality of life, land use
Human decision making	Domain	No explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Achieved, but should still become more differentiated.
Model development process	Concept	Expert knowledge	Quantification of relationships	Assumptions and statistical data

Table 16: Simulation of polycentric urban growth dynamics through agents [ABM_4]

Name of model	Simulation of polycentric urban growth dynamics through agents
Sources	Loibl et al. (2007 )
Technical data
Application area	Covered area, physical boundaries	Austrian Rhine valley with medium-sized centres and rural villages	Extent of area	7,330 hectares built-up area / 260,000 inhabitants

Application area	Spatial units	Grid cells	Size or grain of grids/zones	50 × 50 m cells
Time horizon	Time step	Simulation stops when certain household, population and workplace growth numbers are achieved	Duration of model run	V: 1990 – 2000 / S: user-specified
Modelling approach	Simulation technique	Agent-based simulation	Qualitative or quantitative	Quantitative
Contents
Main purpose	Development of built-up area in peri-urban region, driven by households and entrepreneurs; urban growth with different growth rates
Main variables with relationships	Initialisation: increase of household and workplace numbers is defined 1. Municipality choice depending on regional attractiveness criteria (numbers of people, households and workplaces in the start of the year, average travel time to district centres and capital city, average share of attractive land-use classes in the municipality (open space, forest area) → household growth and workplace growth per municipality → transformation of absolute values into relative search frequencies → agents choose municipality via discrete choice 2. Local target area search: start with random cell, choosing most attractive cell 3. land use change (new built-up area, higher density) → influencing local attractiveness
Human decision making	Domain	Causing the construction of new built-up area or the densification of existing area, no moving as ‘exchange’ of dwellings	Temporal range	Long-term (moving / start-up of companies)

	Typology (classes) of agents?	Yes	→ if yes: what types?	Four household types (1, 2, 3 or 4 persons) and two entrepreneurs (small and large)

	Decision algorithm	Discrete choice	Input into decision	Regional and local attractiveness
Goals	Authors’ opinion	Achieved
Model development process	Concept	Empirical data	Quantification of relationships	Empirical data

Table 17: SLEUTH (Slope, Landuse, Exclusion, Urban Extend, Transportation and Hillshade) [CA_5]

Name of model	SLEUTH (Slope, Landuse, Exclusion, Urban Extend, Transportation and Hillshade)
Sources	Clarke et al. (1997 ); Silva and Clarke (2002 ); Dietzel and Clarke (2007 )
Technical data
Application area	Covered area, physical boundaries	Numerous applications, mostly U.S.	Extent of area	User-specified

Application area	Spatial units	Grid cells	Size or grain of grids/zones	Input for model: 8-bit GIF (100 × 100 m cells can be converted)
Time horizon	Time step	1 year	Duration of model run	C: at least 4 time steps / S: User-specified
Modelling approach	Simulation technique	Cellular automata	Qualitative or quantitative	Quantitative
Contents
Main purpose	Modelling urban growth, scenarios for future development of an urban region
Main variables with relationships	Two components (use depends on available data): (1) Urban growth: cells have one of two states: urban or non urban (2) Urban land use change with different land-use types Four types of growth behaviour: spontaneous, diffusive (with new growth centres), organic (into surroundings) and road-influenced Five main coefficients: diffusion, breed, spread, slope, and road coefficient (need to be calibrated for each case study) Self modification rules: e.g., concerning the kind of exponential or S-curve growth; denser road network → road gravity factor increases; land availability decreases → slope resistance factor is decreased (more hilly areas); spread factor increases over time
Human decision making	Domain	No explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Achieved
Model development process	Concept	Not mentioned	Quantification of relationships	Calibration using historical maps

Table 18: Urban dynamics [SD_1]

Name of model	Urban dynamics
Sources	Forrester (1969 ); Alfeld (1995 )
Technical data
Application area	Covered area, physical boundaries	Either suburban or core area (Forrester 1969: 2) / Examples mentioned in Alfeld, 1995: Lowell, Boston, Concord, Marlborough, Palm Coast	Extent of area	User-specified

Application area	Spatial units	No spatial resolution	Size or grain of grids/zones	–
Time horizon	Time step		Duration of model run	S: Up to 250 years
Modelling approach	Simulation technique	System dynamics	Qualitative or quantitative	Quantitative
Contents
Main purpose	Modelling urban system in general, explicitly including “urban decline.” Examples: focus on a specific topic, e.g., rapid population growth, demolition, et cetera, and therefore need specific models.
Main variables with relationships	Original model by Forrester: Three subsystems: business, housing, population
Human decision making	Domain	No explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Achieved
Model development process	Concept	Expert knowledge	Quantification of relationships	Statistical data and own estimation

Table 19: Urban transformation process in the Haaglanden region [SD_6]

Name of model	Simulating the urban transformation process in the Haaglanden region in the Netherlands
Sources	Eskinasi and Rouwette (2004 )
Technical data
Application area	Covered area, physical boundaries	The Haaglanden region, including the Hague and surrounding suburbs	Extent of area

Application area	Spatial units	No spatial resolution	Size or grain of grids/zones	–
Time horizon	Time step		Duration of model run	S: 1998 – 2010
Modelling approach	Simulation technique	System dynamics	Qualitative or quantitative	Qualitative
Contents
Main purpose	Assessing the impact of future policy interventions on the social housing market (specific: rate of building new dwellings)
Main variables with relationships	Four stocks: 1 Commercial housing stock 2 Social housing stock 3 Waiting families 4 Supply of available social houses Processes involved: Migration, demolition, construction
Human decision making	Domain	No explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Model is useful for its goal

	Validation	No (but impact of process on stakeholders is monitored)	Plausibility analysis	With stakeholders
Model development process	Concept	Participation of stakeholders, narrative approach	Quantification of relationships	Empirical data or expert guesses.

Table 20: Urban travel system [SD_7]

Name of model	A system dynamics model for the urban travel system
Sources	Raux (2003 )
Technical data
Application area	Covered area, physical boundaries	Hypothetical city	Extent of area	–

Application area	Spatial units	No spatial resolution	Size or grain of grids/zones	–
Time horizon	Time step		Duration of model run	S: 20 years into the future
Modelling approach	Simulation technique	System dynamics	Qualitative or quantitative	Quantitative
Contents
Main purpose	To simulate medium- and long-term effects of urban transport policies with reference to sustainable travel
Main variables with relationships	Seven major blocks: urbanisation, internal travel demand (trips within system), car ownership, external travel demand (inflowing, outflowing and through traffic), transportation (comparing supply and demand) and evaluation (socioeconomic and environmental appraisals)
Human decision making	Domain	No explicit decision making	Temporal range	–

	Typology (classes) of agents?	–	→ if yes: what types?	–

	Decision algorithm	–	Input into decision	–
Goals	Authors’ opinion	Work in progress
Model development process	Concept	Expert knowledge	Quantification of relationships	Expert knowledge and statistical values

Table 21: UrbanSim [ABM_5]

Name of model	UrbanSim
Sources	Waddell (2006 ); Waddell et al. (2003 )
Technical data
Application area	Covered area, physical boundaries	Several examples in the U.S., Europe and Asia	Extent of area	User-specified

Application area	Spatial units	Initially: mixture of parcels and zones / later: grid	Size or grain of grids/zones	User-specified / Cell: 150 × 150 m regarded as default
Time horizon	Time step	1 year	Duration of model run	User-specified
Modelling approach	Simulation technique	Coupled simulation models including agent-based simulations	Qualitative or quantitative	Quantitative
Contents
Main purpose	Link between transportation and land use; impact of different planning strategies
Main variables with relationships	Exogenous: (1) macroeconomics (population, employment) and (2) travel demand (travel conditions). Six models: 1 Accessibility (output: access to workplaces and shops for each cell) 2 Transition (output: number of new jobs and new households per year) 3 Mobility (output: number of moving (existing) jobs / households) 4 Location (output: location of new or moving jobs / households) 5 Real Estate Development (output: land use change) 6 Land price (output: land prices)
Human decision making	Domain	Mobility and location	Temporal range	Depends on issues

	Typology (classes) of agents?	Initially households / firms, later persons / jobs	→ if yes: what types?	User-specified

	Decision algorithm	Multinomial logit model	Input into decision	Land-use itself, socio-demographics, dwellings
Goals	Authors’ opinion	Achieved
Model development process	Concept	Not mentioned	Quantification of relationships	Out of empirical data

	Abstract
1	Introduction
	1.1	Urbanisation of landscapes
	1.2	The “ideal” urban land use model
	1.3	Existing reviews on urban land use models
	1.4	The purpose of this review
2	Evaluating urban land use simulation models
3	Models under review
4	Representation of urban landscapes
	4.1	Spidergrams
	4.2	Relationships between human sphere and land use
	4.3	Impact of land use on environment
	4.4	Feedback from environment to human sphere
	4.5	Feedbacks between local and regional scale
5	Conclusions
6	Acknowledgements
7	Annex
	References
	Figures
	Tables

7 Annex