LET, Transport Economics Laboratory (CNRS, University of Lyon, ENTPE) International Conference of Territorial Intelligence...
Contents <ul><li>1. Introduction </li></ul><ul><li>2. UrbanSim </li></ul><ul><li>3.  Data </li></ul><ul><li>4. Model estim...
<ul><li>Understanding the relations between  </li></ul><ul><li>population, employment, land use and transportation  </li><...
<ul><li>The overview of the state-of-the-art  </li></ul><ul><li>of transportation-land use modelling frameworks:  </li></u...
<ul><li>Studies of residential location choice:  </li></ul><ul><li>Sermons and Koppelman (2001):  </li></ul><ul><li>female...
<ul><li>The Lyon Urban Area:  </li></ul><ul><li>- the second in France by population  </li></ul><ul><li>- the demographic ...
1. Introduction MOSART: Numerical Platform of Modelling Mod é lisation et Simulation de l'Accessibilité aux Réseaux et aux...
<ul><li>The aim of the study: </li></ul><ul><li>evaluation of the predictability  </li></ul><ul><li>of the residential loc...
<ul><li>UrbanSim is an urban simulation system developed at the University of Washington.  </li></ul><ul><li>Waddell (2002...
UrbanSim Model Components and Data Flow   (from UrbanSim Project, 2008) 2. UrbanSim
<ul><li>The main data components of UrbanSim: </li></ul><ul><li>zones (TAZs) </li></ul><ul><li>gridcells  </li></ul><ul><l...
<ul><li>utilities of locations  j =1..n:  probabilities: </li></ul>2. UrbanSim is equal to 1 if individual  i  made choice...
3. Data <ul><li>Grand Lyon  (INSEE 1999): </li></ul><ul><li>491 sq. km </li></ul><ul><li>1,1 million inhabitants </li></ul...
3. Data <ul><li>479 TAZs </li></ul>5296 ILOTs Grand Lyon : population density in thousands per sq. km
3. Data <ul><li>Data on ILOTs: </li></ul><ul><li>ILOT identifier </li></ul><ul><li>TAZ identifier </li></ul><ul><li>coordi...
3. Data <ul><li>Data on  households : </li></ul><ul><li>household identifier </li></ul><ul><li>ILOT identifier </li></ul><...
3. Data Travel times from transportation model from MOSART Road network (NAVTEQ) : More than 222000 street segments More t...
3. Data <ul><li>Annual relocation rate for households: 7.6% </li></ul><ul><li>Base year: 1999 </li></ul><ul><li>Subsequent...
4. Model estimation Likelihood ratio test: 4828  Number of observations: 49512 12.79 <0.001 Index of population access if ...
5. Simulation results Relative differences in simulated and actual population 2005
5. Simulation results Relative differences between  simulated and actual population 2005 53 36 ±20 30 20 ±10 18 12 ±5 Perc...
5. Simulation results Differences in simulated and actual population 2005
5. Simulation results Differences in simulated and actual population 2005 in Lyon and Villeurbanne    Difference   Relativ...
6. Conclusion General tendencies: - The most significant attributes are the number of households    and residential units ...
6. Conclusion <ul><li>Population is mainly over-predicted in central Lyon  - Population is mainly under-predicted in the f...
Upcoming SlideShare
Loading in...5
×

Marko KRYVOBOKOV, Nicolas OVTRACHT, Valérie THIEBAUT: Analysis and prediction of household location choice in Grand Lyon with urban land use simulation tool UrbanSim

663

Published on

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

  • Be the first to like this

No Downloads
Views
Total Views
663
On Slideshare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Transcript of "Marko KRYVOBOKOV, Nicolas OVTRACHT, Valérie THIEBAUT: Analysis and prediction of household location choice in Grand Lyon with urban land use simulation tool UrbanSim"

  1. 1. LET, Transport Economics Laboratory (CNRS, University of Lyon, ENTPE) International Conference of Territorial Intelligence University of Salerno, 4-7 November 2009 Marko Kryvobokov, Nicolas Ovtracht, Val é rie Thiebaut Analysis and prediction of household location choice in Grand Lyon with urban land use simulation tool UrbanSim
  2. 2. Contents <ul><li>1. Introduction </li></ul><ul><li>2. UrbanSim </li></ul><ul><li>3. Data </li></ul><ul><li>4. Model estimation </li></ul><ul><li>5. Simulation results </li></ul><ul><li>6. Conclusion </li></ul>
  3. 3. <ul><li>Understanding the relations between </li></ul><ul><li>population, employment, land use and transportation </li></ul><ul><li>is a necessary precondition of </li></ul><ul><li>efficient urban planning and management </li></ul><ul><li>aiming at sustainable development </li></ul><ul><li>Urban simulation models are important tools </li></ul><ul><li>for decision-makers and researchers </li></ul><ul><li>in urban planning and transportation </li></ul>1. Introduction
  4. 4. <ul><li>The overview of the state-of-the-art </li></ul><ul><li>of transportation-land use modelling frameworks: </li></ul><ul><li>Wegener (2004) , Hunt et al. (2005) </li></ul>1. Introduction Wegener (2004) : “it is difficult to empirically isolate impacts of land use on transport and vice versa, and mathematical models… are the only method by which the effects of individual determining factors can be analysed by keeping all the other factors fixed.” Source : Wegener et Fürst (1999)
  5. 5. <ul><li>Studies of residential location choice: </li></ul><ul><li>Sermons and Koppelman (2001): </li></ul><ul><li>female and male commute behaviour in San Francisco </li></ul><ul><li>Bürgle (2006): UrbanSim in Zurich </li></ul><ul><li>Pinjari et al. (2008): </li></ul><ul><li>integrated simultaneous choice of residential location, vehicle and </li></ul><ul><li>bicycle ownership and commute tour mode in San Francisco </li></ul><ul><li>de Palma et al. (2005) : </li></ul><ul><li>integrated land use model from UrbanSim </li></ul><ul><li>with a dynamic traffic model METROPOLIS </li></ul><ul><li>in the Paris region </li></ul>1. Introduction
  6. 6. <ul><li>The Lyon Urban Area: </li></ul><ul><li>- the second in France by population </li></ul><ul><li>- the demographic dynamics in 1999-2005 was +0.8% per year (INSEE Rhône-Alpes, 2007) </li></ul><ul><li>Understanding the mechanism of </li></ul><ul><li>spatial distribution of population? </li></ul><ul><li>The study is part of project PLAINSUDD (PLAteformes numériques INnovantes de Simulation Urbaine pour le Développement Durable – Innovative Numerical Platforms of Urban Simulation for Sustainable Development) sponsored through French ANR </li></ul>1. Introduction
  7. 7. 1. Introduction MOSART: Numerical Platform of Modelling Mod é lisation et Simulation de l'Accessibilité aux Réseaux et aux Territoires (Modelling and Simulation of Accessibility to Networks and Territories)
  8. 8. <ul><li>The aim of the study: </li></ul><ul><li>evaluation of the predictability </li></ul><ul><li>of the residential location choice in Grand Lyon </li></ul><ul><li>with the UrbanSim application </li></ul><ul><li>The prospective of the further use of UrbanSim: </li></ul><ul><li>simulation of distribution of population </li></ul><ul><li>and other important attributes </li></ul><ul><li>in future years </li></ul>1. Introduction
  9. 9. <ul><li>UrbanSim is an urban simulation system developed at the University of Washington. </li></ul><ul><li>Waddell (2002), UrbanSim Project (2008) etc. </li></ul><ul><li>Free, open source, available at www.urbansim.org </li></ul><ul><li>Application of UrbanSim in Europe: Paris, Brussels, Zurich, Amsterdam, Rome, Turin, Lyon… </li></ul>2. UrbanSim
  10. 10. UrbanSim Model Components and Data Flow (from UrbanSim Project, 2008) 2. UrbanSim
  11. 11. <ul><li>The main data components of UrbanSim: </li></ul><ul><li>zones (TAZs) </li></ul><ul><li>gridcells </li></ul><ul><li>households </li></ul><ul><li>jobs </li></ul>2. UrbanSim <ul><li>Prediction of spatial distribution on a yearly basis of: </li></ul><ul><li>households </li></ul><ul><li>jobs </li></ul><ul><li>real estate development </li></ul>Different scenarios of urban development
  12. 12. <ul><li>utilities of locations j =1..n: probabilities: </li></ul>2. UrbanSim is equal to 1 if individual i made choice j , 0 otherwise Source: Ben-Akiva and Lerman (1985) Household Location Choice Model as Multinomial Logit
  13. 13. 3. Data <ul><li>Grand Lyon (INSEE 1999): </li></ul><ul><li>491 sq. km </li></ul><ul><li>1,1 million inhabitants </li></ul><ul><li>495 thousand households </li></ul><ul><li>Two levels of spatial resolution: </li></ul><ul><li>TAZs (IRISes) </li></ul><ul><li>‘ gridcells’ (ILOTs) </li></ul>
  14. 14. 3. Data <ul><li>479 TAZs </li></ul>5296 ILOTs Grand Lyon : population density in thousands per sq. km
  15. 15. 3. Data <ul><li>Data on ILOTs: </li></ul><ul><li>ILOT identifier </li></ul><ul><li>TAZ identifier </li></ul><ul><li>coordinates of centroid </li></ul><ul><li>total area </li></ul><ul><li>water area </li></ul><ul><li>wetland area </li></ul><ul><li>green area </li></ul><ul><li>residential land </li></ul><ul><li>number of residential units </li></ul><ul><li>average real estate price per square metre </li></ul>
  16. 16. 3. Data <ul><li>Data on households : </li></ul><ul><li>household identifier </li></ul><ul><li>ILOT identifier </li></ul><ul><li>number of persons </li></ul><ul><li>number of working persons </li></ul><ul><li>number of cars (0, 1, 2+) </li></ul><ul><li>dummy for home ownership status </li></ul><ul><li>age of head (rank of 1 to 5) </li></ul><ul><li>job status of head (rank of 1 to 5) </li></ul><ul><li>income group (rank of 1 to 3) </li></ul>
  17. 17. 3. Data Travel times from transportation model from MOSART Road network (NAVTEQ) : More than 222000 street segments More than 90000 nodes Public Transport network : More than 2300 stops Bus: more than 100 lines 4 lines of tramway 2 lines of funiculaire 4 lines of metro Regional train: more than 10 lines
  18. 18. 3. Data <ul><li>Annual relocation rate for households: 7.6% </li></ul><ul><li>Base year: 1999 </li></ul><ul><li>Subsequent years for simulation: 2000-2005 </li></ul><ul><li>Scenario: no new real estate development </li></ul><ul><li>Year for comparison: 2005 </li></ul>
  19. 19. 4. Model estimation Likelihood ratio test: 4828 Number of observations: 49512 12.79 <0.001 Index of population access if household does not have a car 12 -4.77 -0.171 Log of travel time to the CBD if household does not have a car 11 46.44 1.291 Log of number of households 10 -5.28 <-0.001 Population density 9 13.78 0.025 Percent of low income households wwd if low income household 8 11.67 0.028 Percent of middle income households wwd if middle income household 7 19.49 0.073 Percent of high income households wwd if high income household 6 -2.06 -0.009 Log of residential vacancy rate 5 -45.70 -1.275 Log of number of residential units 4 -9.72 -0.310 Log of average real estate price if low income household 3 -3.38 -0.081 Log of average real estate price if middle income household 2 3.22 0.173 Log of average real estate price if high income household 1 t-value Coefficient Variable Number
  20. 20. 5. Simulation results Relative differences in simulated and actual population 2005
  21. 21. 5. Simulation results Relative differences between simulated and actual population 2005 53 36 ±20 30 20 ±10 18 12 ±5 Percent of population Percent of ILOTs Relative difference, %
  22. 22. 5. Simulation results Differences in simulated and actual population 2005
  23. 23. 5. Simulation results Differences in simulated and actual population 2005 in Lyon and Villeurbanne Difference Relative difference
  24. 24. 6. Conclusion General tendencies: - The most significant attributes are the number of households and residential units - Low population density and vacancy rate are preferable Different behaviour of social groups: - Accessibility by public transport to the CBD and to population are especially important for households without a car - All income groups strongly prefer location closer to their income group - High income group prefers locations with expensive accommodation - Middle and low income households choose less expensive locations
  25. 25. 6. Conclusion <ul><li>Population is mainly over-predicted in central Lyon - Population is mainly under-predicted in the fringe - Considering the applied scenario of no real estate development, the overall predictability can be seen as acceptable - Perspective – development of other UrbanSim models (residential real estate development, employment location choice, real estate prices, land prices) </li></ul>

×