A Study of Alternative Land Use Forecasting Models

1
A Study of Alternative Land Use Forecasting Models

• Soon Chung and Fang Zhao
• Lehman Center for Transportation Research
• Florida International University
• Miami, Florida
• 11th TRB National Transportation Planning
  Applications Conference
• Daytona Beach, Florida
• May 8, 2007

2
Acknowledgements

• This project was funded by the Florida Department
  of Transportation Systems Planning Office
• Terrence Corkery, AICP Project Manager
• Mr. Michael Neihart, Volusia County MPO

3
Outline

• Introduction
• Objectives
• UrbanSim
• Study Area
• Design of Simulation
• Validation
• Test Scenarios
• Simulation Results
• Findings

4
Introduction

• Transportation models need good land use
  forecasts
• Many land use models have been developed
• Simple to complex
• Integrated or not integrated
• With/without economic theory basis
• Support community visioning
• GIS platform

5
Objectives

• Understanding the state-of-the-art and
  state-of-the-practice of land use models
• Determining the data requirements
• Identifying application issues
• Investigating need for data processing and
  interfacing FSUTMS
• Identifying future research and implementation
  issues

6
What Is UrbanSim?

• A land use microsimulation model
• Developed by the University of Washington
• Provide new land use forecasting and analysis
  capabilities
• Based on economic theories
• Model the interactions of markets and policies,
  including dynamic disequilibrium
• Design to interface activity-based models
• Open Source software source code free to use,
  modify, and redistribute (available at
  www.urbansim.org)

7
UrbanSim Users

• US Users
• Seattle
• Eugene-Springfield
• Houston
• Honolulu
• Salt Lake City
• Phoenix
• Detroit
• Europe Users
• Amsterdam
• Paris
• Zurich
• Middle East Users Tel Aviv
• Potential Users - Downloaded from 80 Different
  Countries

8
UrbanSim Model Structure
9
Sub-Models

• accessibility-model
• household-transition-model
• employment-transition-model
• household-relocation-choice-model
• employment-relocation-choice-model
• household-location-choice-model
• employment-non-home-based-location-choice-model
• employment-home-based-location-choice-model
• scaling-procedure-for-jobs-model
• land-price-model
• developer-model

10
Data Required

• Grid Cells
• Parcel Data
• Property Tax Data
• Employment Data (Info/USA)
• Environmental Layers
• Water
• Wetlands
• Floodplains
• Parks and open space
• National forests
• Steep slopes (DEM)
• Stream buffers (riparian areas)
• Planning and Political Layers
• Traffic Analysis Zones (TAZs)
• Cities
• Urban growth boundaries
• Military
• Major public lands
• Tribal lands

11
Land Price Model

• Linear Regression Model
• Dependent Variable natural logarithm of the
  total land value within a grid cell
• Independent Variables
• Site characteristics
• Development type
• Land use plan
• Environmental constraints
• Regional accessibility
• Access to population and employment
• Urban design-scale
• Land use mix and density
• Proximity to highways and arterials

12
Household Location Model

• Discrete Choice Model
• Variables
• Housing Characteristics
• Prices (cost to income ratio)
• Development types (density, land use mix)
• Housing age
• Regional accessibility
• Job accessibility by auto-ownership group
• Travel time to CBD and airport
• Urban design-scale (local accessibility)
• Neighborhood land use mix and density
• Neighborhood employment

13
Employment Location Model

• Discrete Choice Model
• Employment Home-Based Location Model
• Employment Non-Home-Based Location Model
• Variables
• Real Estate Characteristics
• Price
• Development type (land use mix, density)
• Regional accessibility
• Access to population
• Travel time to CBD, airport
• Urban design-scale
• Proximity to highway, arterials
• Local agglomeration economies within and between
  sectors center formation

14
Developer Model

• Discrete Choice Model
• Variables
• Site characteristics
• Existing development characteristics
• Land use plan
• Environmental constraints
• Urban design-scale
• Proximity to highway and arterials
• Proximity to existing development
• Neighborhood land use mix and property values
• Recent development in neighborhood
• Regional
• Access to population and employment
• Travel time to CBD, airport

15
Possible Scenarios

• Macroeconomic Assumptions
• Household and employment control totals
• Development constraints
• Can select any combination of
• Political and planning overlays
• Environmental overlays
• Land use plan designation
• Determine which development types cannot occur
• Transportation infrastructure
• User-specified events

16
Study Area Selection Criteria

• Have recent household survey data
• Up-to-date GIS data, including parcel-level
  property data
• Being relatively self-contained

17
Volusia County

• 1,263 square miles
• Population 443,343 in 2000
• Surrounded by Flagler, Marion, Lake, Seminole,
  and Brevard counties (most rural)

18
Volusia County Planning Regions
19
Simulation Process
1
2
3
4
20
Validation

• Model Output was compared with
• Model results adopted in the LRTP
• 2005 InfoUSA Employment Data

21
Cumulative Percentage of TAZs vs. Differences in
Zonal Households and Population
22
Cumulative Percentage of Employment Differences
between UrbanSim and the 2005 InfoUSA Data
23
Spatial Distribution of Differences between
UrbanSim and 2005 InfoUSA Data
24
Five Scenarios

• 3 alternatives from Volusia County 2020 LRTP
• Alternative 2
• Alternative 3
• Final Plan
• 3 projections (low, mid, high) of population from
  Bureau of Economic and Business Research

25
Scenarios
26
Comparison of Volumes in 2020LRTP - Scenario 1
27
Spatial Distribution of Households from Scenario
12020 Base Year
28
Spatial Distribution of Employment from Scenario
12020 Base Year
29
Comparison of Households between the 2020 LRTP
and Scenario1
30
Comparison of Employment between the 2020 LRTP
and Scenario1
31
Comparison of Alternative 2 and Final Plan
32
Comparison of Volumes in 2020Scenario 2
Scenario 1
33
Spatial Distribution of Households from Scenario
22020 Base Year
34
Spatial Distribution of Employment from Scenario
2 2020 Base Year
35
Comparison of Households between Scenario1 and
Scenario 2
36
Comparison of Employment between Scenario1 and
Scenario 2
37
Comparison of Alternative 3 and Final Plan
38
Comparison of Volumes in 2020Scenario 3
Scenario 1
39
Spatial Distribution of Households from Scenario
3 2020 Base Year
40
Spatial Distribution of Employment from Scenario
32020 Base Year
41
Comparison of Households between Scenario1 and
Scenario 3
42
Comparison of Employment between Scenario1 and
Scenario 3
43
Comparison of Volumes in 2020 Scenario 4
Scenario 1
44
Comparison of Households between Scenario1 and
Scenario 4
45
Comparison of Employment between Scenario1 and
Scenario 4
46
Comparison of Volumes in 2020Scenario 5
Scenario 1
47
Comparison of Households between Scenario1 and
Scenario 5
48
Comparison of Employment between Scenario1 and
Scenario 5
49
Findings

• High data requirements
• Impute missing data
• Join property TAX data with parcel layer
• Join employment data with parcel layer
• Data mining and synthetic data cleaning tools,
  currently being designed, will facilitate working
  with messy data
• Model estimation requires knowledge of discrete
  choice models
• Consultations with local government agencies are
  desirable in developing model specifications and
  estimating model parameters