1. 1 Multi-scale spatial models: linking macro and micro Michael Wegener Spiekermann & Wegener, Urban and Regional Research Dortmund, Germany Centre for Advanced Spatial Analysis University College London 09 January 2008

2. 2 Integrated land-use transport models Today's integrated land-use transport models suffer from several weaknesses: -Their classification of households and individuals is too crude; individual lifestyles cannot be represented. - Their transport models are not activity-based and cannot address "soft" transport policies. - Their spatial resolution is too coarse to take account of small-scale local policies. - Forecasting environmental impacts such as air pollution, land take and traffic noise is difficult, modelling environ- mental feedback is impossible. -Issues of spatial equity cannot adequately be addressed.

3. 3 Microsimulation New activity-based microsimulation models improve urban simulation models: -Individual lifestyles can be represented, households and individuals are disaggregated to the agent level. -Environmental impacts can be modelled with the required spatial resolution. -Environmental feedback between environment and land use and transport can be modelled. -Microlocations can be represented. Households affected by environmental impacts can be localised. PROPOLIS & ILUMASS

4. 4 However... To date, no full-scale microsimulation model of urban land use, transport and environment has become operational. There are still unresolved problems regarding the inter- faces between the submodels. The feedback between transport and location and environ- mental quality and location has not yet been implemented. Serious problems of calibration, stability and stochastic variation have not been solved. The computing time for existing models is calculated in terms of weeks or days, not hours. Conclusions

5. 5 How much micro is enough? Despite these problems, microsimulation modellers engage in ever more ambitious plans to further raise the complexity and spatial resolution of their models. The common belief among most microsimulation modellers seems to be: the more micro the better. This is the dream of the one-to-one Spitfire. Conclusions

6. 6 The Spitfire

7. 7 The one-to-one model of the Spitfire

8. 8 The one-to-one Spitfire "Simplifying assumptions are not an excrescence on model-building; they are its essence. Lewis Carroll once remarked that a map on the scale of one-to-one would serve no purpose. And the philosopher of science Russell Hanson noted that if you progressed from a five-inch balsa wood model of a Spitfire air plane to a 15-inch model without moving parts, to a half-scale model, to a full-size entirely accurate one, you would end up not with a model of a Spitfire but with a Spitfire". Robert M. Solow (1973)

9. 9 How much micro is enough? There seems to be little consideration of the benefits and costs of microsimulation: - Where is microsimulation really needed? - What is the price for microsimulation? - Would a more aggregate model do? For spatial planning models, the answer to these questions depends on the planning task at hand. For instance, for modelling the impacts of transport on land use, much simpler travel models are sufficient. Conclusions

10. 10 Macro or micro? These considerations lead to a reassessment of the hypothesis that eventually all spatial modelling will be microscopic and agent-based. Conclusions

11. 11 Adapted from Miller et al., 1998. Macro or micro? ??

12. 12 Conclusions (1) Only integrated microsimulation land-use transport models models permit the modelling of - "soft" and local planning policies - individual lifestyles - environmental impacts and feedback - microlocations and spatial equity. However, there is a price for the microscopic view in terms of data requirements and long computing times. There are privacy concerns and ethical issues involved. Conclusions

13. 13 Conclusions (2) Under constraints of data collection and of computing time, there is for each planning problem an optimum level of conceptual, spatial and temporal resolution. This suggests to work towards a theory of balanced multi- scale models which are as complex as necessary for the planning task at hand and as simple as possible but no simpler. Future urban models will be modular and multi-scale in scope, space and time. Conclusions

14. 14 Multi-Scale Modelling The Dortmund Example

15. 15 Model levels Dortmund Regions Dortmund Zones Dortmund City Raster cells Multi-level Multi-scale

16. 16 Level 1: Regions

17. 17 Model levels Dortmund Regions Dortmund Zones Dortmund City Raster cells

18. 18 SASI Model GDP Accessibility Production function Employment Migration function Population Income Labour force Transport policy Unemploy- ment

19. 19 The STEPs Project (2004-2006) The EU 6th RTD Framework project STEPs (Scenarios for the Transport System and Energy Supply and their Potential Effects) developed and assessed possible scenarios for the transport system and energy supply of the future. In the project five urban/regional models were applied to forecast the long-term economic, social and environmental impacts of different scenarios of fuel price increases and different combinations of infrastructure, technology and demand regulation policies. Here the model results for the urban region of Dortmund, Germany, will be presented.

20. 20 The STEPs Project: Scenarios The project developed a set of scenarios assuming different rates of energy price increases with three sets of policies: A-1 Reference Scenario

21. 21 Dortmund Cologne Hagen Düsseldorf Münster Essen Duisburg Bochum

22. 22 Economic impacts for the Dortmund region According to the SASI model, the fuel price increases and related policies of the scenarios have significant negative impacts on the economy of the Dortmund urban region:

23. 23 Level 2: Zones

24. 24 Model levels Dortmund Regions Dortmund Zones Dortmund City Raster cells

25. 25 Internal zones External zones Dortmund Hamm Hagen Bochum Study area

26. 26 E n v i r o n m e n t r b a n U Urban systems Networks Travel Population Workplaces Land use Housing Very fast Fast Slow Very slow Goods transport Employment Speed of change

27. 27 Dortmund model Microsimulation from SASI model Population households Nonresidential buildings Residential buildings Employment Workplaces Housing market Transport market Market for nonresidential buildings Land market Labour market

28. 28 Travel distance per capita per day (km)

29. 29 Share of walking and cycling trips (%)

30. 30 Share of public transport trips (%)

31. 31 Share of car trips (%)

32. 32 Average trip speed (km/h)

33. 33 Car fuel consumption per car trip per traveller (l)

34. 34 CO 2 emission by transport per capita per day (kg)

35. 35 Level 3: Raster Cells

36. 36 Model levels Dortmund Regions Dortmund Zones Dortmund City Raster cells

37. 37 The ILUMASS Project (2001-2006) The project ILUMASS (Integrated Land-Use Modelling and Transport Systems Simulation) embedded a microscopic dynamic simulation model of urban traffic flows into a comprehensive model system incorporating both changes of land use and the resulting changes in transport demand as well as their environmental impacts. For testing the land use submodels, the transport and environmental submodels were replaced by the aggregate transport model of the IRPUD model and simpler envir- onmental impact models (= reduced ILUMASS model).

38. 38 Reduced ILUMASS model Microsimulation from SASI model Microsimulation from SASI model

39. 39 0 < 20 < 40 < 60 < 80 < 100 < 120 < 140 < 160 < 180 < < 20 E/ha 40 E/ha 60 E/ha 80 E/ha 100 E/ha 120 E/ha 140 E/ha 160 E/ha 180 E/ha 200 E/ha

40. 40 Firms and households Start End Another household? Yes No Select a household Demographic events - Ageing - Death - Birth - Marriage/cohabitation - Divorce/separation - Persons leave household - Persons move together Economic events - Education - Place of work - Employment status - Income - Mobility budget of household Update lists - Persons - Households - Dwellings - Firms Start End Another firm? Yes No Select a firm Select alternative location Accept? No 1-10 11 Yes Update lists - Firms - Nonresidential floorspace - Employed persons Check satisfaction with present location Satisfied? No Yes

41. 41 Dwellings Another project? End No Update lists - Persons - Households - Dwellings - Construction Ye s Select zone and microlocation Select type and num- ber of dwellings Upgrading Select zone and microlocation Select type and num- ber of dwellings Demolition and result- ing moves Select zone and microlocation Select type and num- ber of dwellings Construction Start Expected demand - Demolition - Upgrading - Construction Select project - Demolition - Upgrading - Construction

42. 42 Moves Start End Another dwelling? Yes No Select dwelling - Area - Type Select household - Immigration - New household - Forced move - Move Accept? No Update lists - Households: - with dwelling - without dwelling - Dwellings - occupied - vacant 1-5 Housing supply Yes Start End Another household? Yes No Select household - Outmigration - Immigration - New household - Move Select dwelling - Area - Type Accept? No 1-5 6 Yes Housing demand Update lists - Households: - with dwelling - without dwelling - Dwellings - occupied - vacant 6

43. 43 Environmental Impacts and Feedback

44. 44 Model levels Dortmund Regions Dortmund Zones Dortmund City Raster cells

45. 45 Environmental feedback PROPOLISILUMASS Aggregate land-use transport model Zonal data Aggregate land-use transport model Zonal environmental impact model Aggregate land-use transport model Zonal data Aggregate land-use transport model Spatial disaggregation Zonal data Microsimulation land-use transport model Disaggregate environmental impact model Disaggregate environmental impact model No spatial disaggregation Spatial disaggregation of output Spatial disaggregation of input Few impacts Limited feedback All impacts Limited feedback All impacts All feedbacks Spatial disaggregation

46. 46 Net residential density (Raster) Net residential density (Zone) Zone v. Raster Density

47. 47 Percent open space (Raster) Percent open space (Zone) Zone v. Raster Open space

48. 48 Mean No 2 exposure (Raster) Mean No 2 emission (Zone) Zone v. Raster Air pollution

49. 49 Mean traffic noise exposure (Raster) Mean traffic noise (Zone) Zone v. Raster Traffic noise

50. 50 Reduced ILUMASS model Microsimulation from SASI model Microsimulation from SASI model Environmental impacts

51. 51 Typical Model Run

52. 52 Model dimensions 1.2 millionhouseholds 2.6 millionpersons 1.2 milliondwellings 80,000firms 92,000industrial sites 8,400public transport links 848public transport lines 13,000road links 246/54internal/external zones 209,000raster cells 30simulation periods (years) 90minutes computing time Conclusions

53. 53 Model parameters

54. 54 Micro data

55. 55 Travel flows

56. 56 Public transport flows Travel flows

57. 57 Travel flows

58. 58 Link loads

59. 59 Link loads Public transport speed

60. 60 Link loads Public transport speed

61. 61 Air quality NO 2

62. 62 Traffic noise dB(A)

63. 63 Firms

64. 64 Households

65. 65 HouseholdsDwellingsHouseholdsDwellingsHouseholdsDwellingsHouseholdsDwellings

66. 66 Vacant dwellings

67. 67 HouseholdsMoves

68. 68 Compression of micro data

69. 69 Aggregation to zones

70. 70 Micro data

71. 71 Simulation completed

72. 72 More information Moeckel, R., Schwarze, B., Spiekermann, K., Wegener, M. (2007): Simulating interactions between land use, transport and environment. Proceedings of the 11th World Conference on Transport Research. Berkeley, CA: University of California at Berkeley. Wagner, P., Wegener, M. (2007): Urban land use, transport and environmental models: Experiences with an integrated microscopic approach. disP 43(170):45–56. Wegener, M. (1998): The IRPUD Model: Overview. http://www. raumplanung.uni-dortmund.de/irpud/pro/mod/mod_e.htm. Wegener, M. (2007): After the Oil Age: Do we have to rebuild our cities? Presentation at the SOLUTIONS Conference, University College London, 11-12 July 2007. http://www.suburbansolutions. ac.uk/sitemapdocs.aspx.

73. 73