Transportation and Spatial Modelling: Lecture 16

828 views
729 views

Published on

Published in: Business, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
828
On SlideShare
0
From Embeds
0
Number of Embeds
249
Actions
Shares
0
Downloads
37
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Transportation and Spatial Modelling: Lecture 16

  1. 1. CIE4801 Transportation and spatial modellingSpatial modelling: Land use and transportationmodelsRob van Nes, Transport & Planning17/4/13 Delft University of Technology Challenge the future
  2. 2. Content•  Land use and transportation models (LUT) •  UrbanSim •  TIGRIS XL •  TIGRIS XL Applications•  Choice modelling •  Firm location behaviour •  Household location behaviour•  Circle of Wegener revisited CIE4801 Land use and transportation models 2
  3. 3. 1.Land use and transportation models CIE4801 Land use and transportation models 3
  4. 4. LUT models: Framework government land commercial market houses buildings developers real-estate housing infrastructure market market labor firms households market transport costs, accessibilities demography transport model CIE4801 Land use and transportation models 4
  5. 5. 1.1UrbanSim CIE4801 Land use and transportation models 5
  6. 6. UrbanSim: Modelling approach •  Models choices of decision-makers in a theoretic framework •  Models processes using multinomial or nested logit •  Microsimulation of individual decision-makers •  Dynamically simulate annual time steps •  Models market interactions •  Uses very disaggregrate spatial units CIE4801 Land use and transportation models 6
  7. 7. CIE4801 Land use and transportation models 7
  8. 8. UrbanSim: Inputs and outputs(per grid cell, 150m x 150m) INPUT OUTPUT •  inhabitants (census data) •  households •  land ownership parcels •  employment by type •  firms •  real-estate development •  household travel survey •  real-estate prices •  land use plan •  environmental constraints •  transportation system CIE4801 Land use and transportation models 8
  9. 9. UrbanSim: Model structure Accessibility model determines the accessibility level, depending on traffic model outcomes Economic & demographic transition model determines creation or loss of households and jobs Household & employment mobility model determines if households and jobs are moving within the regiont=t+1 Household & employment location model determines the location choices of households and jobs from available vacant real-estate Real-estate development model determines the type, location, and quantity of new construction by developers Land price model determines the price of land at each location CIE4801 Land use and transportation models 9
  10. 10. UrbanSim: some results (1/3) Employment, 1980Eugene-Springfield (Oregon, USA) CIE4801 Land use and transportation models 10
  11. 11. UrbanSim: some results (2/3) Employment, 1994Eugene-Springfield (Oregon, USA) CIE4801 Land use and transportation models 11
  12. 12. UrbanSim: some results (3/3) Employment change 1980 - 1994Eugene-Springfield (Oregon, USA) CIE4801 Land use and transportation models 12
  13. 13. 1.2TIGRIS CIE4801 Land use and transportation models 13
  14. 14. TIGRIS•  Transport Infrastructure Land-use (‘Grondgebruik’) Interaction Simulation•  Developed in the 90’s•  Primarily based on expert judgement •  Model structure as well as parameters•  Meant to be a sketch-planning model •  GIS-based, incremental development (year by year), dashboard CIE4801 Land use and transportation models 14
  15. 15. t-1 t-1TIGRIS: Flow chart attraction living Figure 2. Attractions submodel Land use is characterised by the number of population, jobs and amenities per zone (Figure national and regional developments, and the m determined on the basis of attractiveness of a z Dwelling construction per zone is determined o minimum and maximum number of dwellings. T CIE4801 Land use and transportation models 15
  16. 16. TIGRIS: Submodels•  Attractiveness of a zone •  Living and working•  Land use To be used in year t+1 •  Dwellings and industrial/office sites•  Mobility•  Congestion ‘equilibrium’ •  Car•  Travel impedance •  Car and PT•  Accessibility •  Amenities, population, jobs CIE4801 Land use and transportation models 16
  17. 17. TIGRIS: Applications•  Four applications •  Randstadrail •  Leiden-Haarlem-Amsterdam •  Arnhem-Nijmegen •  Randstad: Urbanisation beyond 2030•  Primarily used as an explorative model •  Unclear role in planning processes•  Evaluation •  Not state of the art (no choice modelling) •  Not tailored to the questions in practice (e.g. link with economic analyses) CIE4801 Land use and transportation models 17
  18. 18. TIGRIS XL: Modelling approach•  Dynamic spatial allocation model•  Accessibility influences location choice•  Determines effects of infrastructure concepts on land use•  Determines effects of spatial planning on transportation•  Simulates annual changes•  Uses aggregrate zones, no detailed spatial data•  Used for policy development, not for evaluation. CIE4801 Land use and transportation models 18
  19. 19. TIGRIS XL: inputs and output INPUT OUTPUT •  zones •  households by type •  inhabitants •  employment by type •  car ownership •  real-estate development •  employment levels •  real-estate prices •  services •  trips •  captives and non-captives •  travel times •  population growth factors •  safety & environment •  spatial development policies •  coarse infrastructure network CIE4801 Land use and transportation models 19
  20. 20. COROP regions Functional design of TIGRIS XL Regional Labor regional Labor market Market demand workforce Firms / households / Transport market demography jobs personszones Real estate market Housing market Office space/ Land market houses Industrial zones CIE4801 Land use and transportation models 20
  21. 21. Modules TIGRIS XL•  Demography - module addressing basic demographic developments•  Land market - simplistic, excludes role of land-owner and project developers•  Housing market - behavioral choice model estimated on housing market survey•  Labor market - calibration on period 1986-2000•  Transport module - integration of land-use modules with LMS CIE4801 Land use and transportation models 21
  22. 22. Land market, role of government TIGRIS XL can model different levels of government influence on the spatial development:•  Regulated development (‘directed allocation’) spatial developments can only take place on planned locations•  Free market development (‘free allocation’) spatial developments following preferences of households and firms. The developments are restricted by availability of land CIE4801 Land use and transportation models 22
  23. 23. Household choices Household X SStay Move Intra COROP Other COROP Zones (1308) CIE4801 Land use and transportation models 23
  24. 24. Household move/stay choiceExplanatory variables•  Household size•  Employment•  Household income•  Age head of household•  Zone type classification (urban to rural)•  Vacant houses in region•  Accessibility current location CIE4801 Land use and transportation models 24
  25. 25. Household location choiceExplanatory variables•  Number of vacant houses in a zone•  Average price of houses in a zone•  Zone type classification (urban – rural, 5 categories)•  travel time (travel time between old and new location)•  Accessibility location•  Zone characteristics (water, services, green, populationdensity, etc.) CIE4801 Land use and transportation models 25
  26. 26. Firm choicesIdentification of economic sectors is important, becausethese sectors show different location preferences andresponses to transport measures.Sectors in TIGRIS XL are: •  Agriculture •  Industry •  Logistics •  Retail sector •  Other consumer services •  Business services •  Government CIE4801 Land use and transportation models 26
  27. 27. Firm location choiceExplanatory variables•  Accessibility employees•  Population in a region•  Accessibility business•  Accessibility freight•  Agglomeration•  Urbanization•  Relative share of sector in a region CIE4801 Land use and transportation models 27
  28. 28. Accessibility variablesUtility based accessibility measures for (so-called logsum measures) •  Accessibility by household type •  Accessibility of firms for commuters •  Accessibility of firms for business CIE4801 Land use and transportation models 28
  29. 29. 1.3TIGRIS XL: Applications CIE4801 Land use and transportation models 29
  30. 30. TIGRIS XL: Randstad What are the consequences Amsterdam on land use and traffic when there is development inner - in the outer region; or region - in the inner region Den Haag Utrecht Rotterdam CIE4801 Land use and transportation models 30
  31. 31. Randstad: results land useDeveloping in outer region Developing in inner region increase in houses 0 0 - 2500 2010 - 2030 2500 - 5000 5000 - 7500 7500 - 10000 > 10000 CIE4801 Land use and transportation models 31
  32. 32. Randstad results transportationDeveloping in outer region Developing in inner region congestion 2010 - 2030 CIE4801 Land use and transportation models 32
  33. 33. TIGRIS XL Trend scenario Working population (15-64) CIE4801 Land use and transportation models 33
  34. 34. Results trend scenario Jobs Industry Logistics Retail % per jaar 0.72 of meer 0.48 tot 0.72 0.27 tot 0.48 0.00 tot 0.27 -0.31 tot 0.00 -0.58 tot -0.31 -0.98 tot -0.58 minder dan -0.98Other consumer Business services Land useGovernment models CIE4801 and transportation 34 services
  35. 35. Accessibility of jobs for households(by car) morning peak 206 - 250000 rest of the day 206 - 250000 km 0 10 20 40 60 80 250001 - 500000 km 250001 - 500000 0 510 20 30 40 500001 - 750000 500001 - 750000 750001 - 1000000 750001 - 1000000 1000001 - 1250000 1000001 - 1250000 1250001 - 1500000 CIE4801 Land use and transportation models 35
  36. 36. Accessibility of jobs for households(by public transport) 462 - 100000 km 0 10 20 40 60 80 100001 - 200000 200001 - 300000 300001 - 400000 400001 - 500000 CIE4801 Land use and transportation models 36
  37. 37. Accessibility of employees for firms(by car) morning peak rest of the day 206 - 250000 192 - 250000 km 250001 - 500000 250001 - 500000 0 10 20 40 60 80 500001 - 750000 km 500001 - 750000 0 10 20 40 60 80 750001 - 1000000 750001 - 1000000 1000001 - 1250000 1000001 - 1250000 1250001 - 1500000 1250001 - 1500000 1500001 - 1917866 CIE4801 Land use and transportation models 37
  38. 38. Accessibility of employees for firms(by public transport) 462 - 100000 100001 - 200000 km 0 10 20 40 60 80 200001 - 300000 300001 - 400000 400001 - 500000 500001 - 639338 CIE4801 Land use and transportation models 38
  39. 39. Application concentration scenario Appointed concentration areas CIE4801 Land use and transportation models 39
  40. 40. Results concentration scenario Population CIE4801 Land use and transportation models 40
  41. 41. Results concentration scenario Population – concentration vs. trend CIE4801 Land use and transportation models 41
  42. 42. Results concentration scenario Arbeidsplaatsen per jaar Jobs – concentration vs. trend 508 of meer 200 tot 508 38 tot 200 0 tot 38 -68 tot 0 -164 tot -68 -437 tot -164 minder dan -437 CIE4801 Land use and transportation models 42
  43. 43. 2.1Choice modelling: firm locationbehaviour CIE4801 Land use and transportation models 43
  44. 44. Factors influencing location choiceof firms•  Characteristics of the firm •  Size •  Growth •  Age •  Sector•  Characteristics of the location •  Vicinity of infrastructure •  Accessibility of the market and of employees •  Neighborhood of other firms CIE4801 Land use and transportation models 44
  45. 45. Aggregated approach: Disaggregated approach:All firms in a zone Individual firms SHPmigration92 less then 5 employees over 5 employees firms92.txt Events 0-5 6 - 25 26 - 50 51 - 100 101 - 978 CIE4801 Land use and transportation models 45
  46. 46. Data available in the Netherlands•  LISA National information system of employment, contains information on the whole firm population•  LMS National model system, contains information on accessibilities•  GIS Geographic information system, contains information on location of railways, freeways, etc. CIE4801 Land use and transportation models 46
  47. 47. Vicinity of infrastructure CIE4801 Land use and transportation models 47
  48. 48. Accessibility Logsum business trips: Logs / <None> < 6.5 6.5 - 7.0 7.0 - 7.5 7.5 - 8.0 > 8.0 railway motorway Study area0 1.5 3 6 9 12 Kilometers CIE4801 Land use and transportation models 48
  49. 49. Neighborhood of other firms specialization / diversity CIE4801 Land use and transportation models 49
  50. 50. Specialization Railway HighwayPS Business Services Rb <7.5 minutes < 0.5 0.5 - 1.0 1.0 - 1.5 > 1.5 CIE4801 Land use and transportation models 50
  51. 51. Diversity Railway HighwayDiversity Index Range Band <7.5 minutes 0.750001 - 1.060000 0.500001 - 0.750000 0.440001 - 0.500000 0.030000 - 0.440000 CIE4801 Land use and transportation models 51
  52. 52. Firm location choice model Firm move probability Stay Move location probability Location 1 …… Location N CIE4801 Land use and transportation models 52
  53. 53. Move probability stay V fi = 0 move V fi = β 0 + β1Grf + β 2 (1/ Age f ) + β3 Sec f + β 4 Acci + ... firm characteristics location characteristics Gr = firm growth f – firm index Age = firm age i – location index Sec = firm sector Acc = location accessibility move move exp(V fi ) 1 P fi = move stay = move exp(V fi ) + exp(V fi ) 1 + exp(−V fi ) CIE4801 Land use and transportation models 53
  54. 54. Estimted parameters Firm mobility Move probability Parameter estimates-6.000 -5.000 -4.000 -3.000 -2.000 -1.000 0.000 1.000 2.000 CONSTANT FIRM CHARACTERISTICS Log of size Grow th rate 1 / age INDUSTRY SECTOR Finance Business services Government Education Health service General Services (ref.) ACCESSIBILITY α-location; near trainstation β-location; near trainstation & highw ay onramp γ-location; near highw ay onramp ρ-location; neither URBANISATION ECONOMIES Logsum business and commuting trips LOCALISATION ECONOMIES Diversity w ithin < 7,5 min. 54 Specialisation w ithin < 7,5 min. Significant Non significant CIE4801 Land use and transportation models 54
  55. 55. Location probability location Vsij = βs1Distij + βs 2 Acc j + ... Dist = relocation distance s – firm sector index Acc = location accessibility i – current location index j – new location index location location exp(Vsij ) Psij = location ∑ exp(V j sij ) CIE4801 Land use and transportation models 55
  56. 56. Parameter estimates Location probability (sector Business)-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 M IG R A T ION  A T T R IB U T E    D is tance  t o  o rig inal  loc.[km1/2 ] A C C E S S IB ILIT Y  A T T R IB U T E    near  s tation    near  s tation  &  o nramp    near  o nramp    neither U R B A N IS A T ION  E C ON OM IE S    L og s um  b us ines s  and  c ommuting  t rips LOC A LIS A T ION  E C ON OM IE S    D ivers ity  f irms  w ithin  7,5  min.    S pecialis ation  f irms  w ithin  7,5  min. C OM P E T IN G  D E S T IN A T ION S    C entrality  p arameter  T eta Significant Non significant 56 CIE4801 Land use and transportation models 56
  57. 57. Parameter estimates Location probability (sector Finance)-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 M IG R A T ION  A T T R IB U T E    D is tance  t o  o rig inal  loc.[km1/2 ] A C C E S S IB ILIT Y  A T T R IB U T E    near  s tation    near  s tation  &  o nramp    near  o nramp    neither U R B A N IS A T ION  E C ON OM IE S    L og s um  b us ines s  and  c ommuting  t rips LOC A LIS A T ION  E C ON OM IE S    D ivers ity  f irms  w ithin  7,5  min.    S pecialis ation  f irms  w ithin  7,5  min. C OM P E T IN G  D E S T IN A T ION S    C entrality  p arameter  T eta Significant Non significant 57 CIE4801 Land use and transportation models 57
  58. 58. Location choice companies(result of a more recent study)•  For the choice to consider another location characteristics of companies themselves are dominant•  For the location choice accessibility attributes play a role.•  Main accessibility attributes: •  Distance to original location •  Distance to freeway on-/off-ramp •  Distance to railway station •  financing, education, catering •  Accessibility by car •  business, financing, manufacturing, logistics, trading and retail M. De Bok, 2007 CIE4801 Land use and transportation models 58
  59. 59. 2.2Choice modelling: Household locationbehaviour CIE4801 Land use and transportation models 59
  60. 60. Household location choice model Household move probability Stay Move location probability Location 1 …… Location N CIE4801 Land use and transportation models 60
  61. 61. Location probability locationVhij = βh1Distij + βh 2 Acc j + βh3 Ethj + ... Dist = relocation distance h – household type index Acc = location accessibility i – current location index Ethn = similarity ethnical background j – new location index location location exp(Vhij )Phij = location ∑ exp(V j hij ) CIE4801 Land use and transportation models 61
  62. 62. Relocation distance 95% Old dwelling dmax dmax Workplace head CIE4801 Land use and transportation models 62
  63. 63. Location accessibility 63 CIE4801 Land use and transportation models 63
  64. 64. Other factors 64 CIE4801 Land use and transportation models 64
  65. 65. Location choice households(result of a more recent study)•  Characteristics of house and neighbourhood are dominant•  Accessibility plays a limited role: •  Distance to previous house •  Distance to work by car•  More advanced modelling provides more insight •  Distance to station (for household without a car) B. Blijie & De Vries, 2006 CIE4801 Land use and transportation models 65
  66. 66. Some overall conclusions•  Shift towards disaggregated approaches (individual firms and households)•  Relocation distance is preferably small•  Accessibility plays a role in the location choice behavior •  Accessibility is relevant for the location choice of firms, not in the decision to move •  Best accessible locations may not be preferred by households CIE4801 Land use and transportation models 66
  67. 67. 3.Circle of Wegener revisited CIE4801 Land use and transportation models 67
  68. 68. Choice processes Build Choice Choice location location investors users Attractiveness Move Accessibility Activities Travel time Ability to & costs travel Choice Choice route/time trip Choice Choice destina- mode tion CIE4801 Land use and transportation models 68
  69. 69. Empirical evidence?•  Databases of observed behaviour•  Observed or generated choice alternatives•  Formulating choice models (logit models)•  Estimating perception factors or weights•  Resulting models of attributes having significant weights CIE4801 Land use and transportation models 69
  70. 70. Location choice investors•  Different decision makers thus different objectives •  Project developers •  Authorities•  I’ve got no empirical evidence for this type of choice, however…•  Given the bi-level nature of the decision problem they should consider the choice behaviour of (future) users of the location CIE4801 Land use and transportation models 70
  71. 71. Location choice companies•  For the choice to consider another location characteristics of companies themselves are dominant•  For the location choice accessibility attributes play a role.•  Main accessibility attributes: •  Distance to original location •  Distance to freeway on-/off-ramp •  Distance to railway station •  financing, education, catering •  Accessibility by car •  business, financing, manufacturing, logistics, trading and retail M. De Bok, 2007 CIE4801 Land use and transportation models 71
  72. 72. Location choice households•  Characteristics of house and neighbourhood are dominant•  Accessibility plays a limited role: •  Distance to previous house •  Distance to work by car•  More advanced modelling provides more insight •  Distance to station (for household without a car) B. Blijie & De Vries, 2006 CIE4801 Land use and transportation models 72
  73. 73. Trip choice•  To travel or not to travel…..•  No impact of accessibility •  Recall constant number of trips per day•  Although, for some trip purposes an effect has been found, still having a minimal impact Willigers & De Bok, 2009 CIE4801 Land use and transportation models 73
  74. 74. Destination choice•  Distance/time has a substantial impact: the larger the distance (or the longer the time), the lower the probability of choosing the destination (assuming similar attractiveness of the destinations)•  Car accessibility is usually dominant, except for households without a car•  Alternative modes improve accessibility, however, net impact on total attractiveness is limited Ortúzar & Willumsen, 2001 CIE4801 Land use and transportation models 74
  75. 75. Mode choice•  Clear role of mode availability •  Recall impact of ‘captiveness’•  Clear role for quality (travel times) of each mode•  However, preferences for specific modes play a major role too! •  Recall impact of ‘captiveness’ Ortúzar & Willumsen, 2001 CIE4801 Land use and transportation models 75
  76. 76. Route choice•  Clear role of travel time•  In public transport different weights for trip time elements •  2.2*Ta+1.5*Tw+Ti+2.3*Twt+5.9*Nt+1.1*Te•  Train trips are in fact multi-modal (80% of train travellers uses another mode to travel to or from the station)•  Consequence of multimodality………. CIE4801 Land use and transportation models 76
  77. 77. Main components utility function multi-modal route choice 100%   Sta5on  type  indicators   90%   80%   In-­‐vehicle  5mes   70%   Frequency  related   60%   Parking  and  BTM  costs   50%   40%   Mode  and  service   indicators   30%   Ac5vity-­‐end   20%   10%   Train  part   0%   Home-­‐end   Trip  parts   U5lity  components   S. Hoogendoorn-Lanser, 2005 CIE4801 Land use and transportation models 77
  78. 78. Conclusion•  There is empirical evidence for the mechanism in Wegener’s circle•  However, in nearly every choice situation many other factors play a role, and often quite dominantly•  Often simple and more specific accessibility indicators are significant: •  distance to former location, distance to freeway•  Mechanism is stronger for car accessibility, however: •  Increasing car usage leads to congestion, thus making other locations more attractive CIE4801 Land use and transportation models 78

×