DRCOG s Land Use and Travel Demand Models. Presentation to the Institute of Transportation Engineers Colorado/Wyoming Section 12 December 2014

Transcription

1 DRCOG s Land Use and Travel Demand Models Presentation to the Institute of Transportation Engineers Colorado/Wyoming Section 12 December 2014

2 The Dirty Secret Essentially, all models are wrong, but some are useful. Professor George Box University of Wisconsin Project Traffic Forecasting, NCHRP 255 Review 25 September 2008 Requoted in Travel Model Validation and Reasonableness Checking Manual, 2 nd Ed., 2010

3 Why Might Models Be Wrong? Errors in Input Values Errors in Relationships Errors in Trade-Offs Errors in Calibration Targets

4 DRCOG Modeling System Base Development Patterns Future Land Use Regional & County Control Forecasts Zoning Regulations Highway & Transit Networks Enplanement Counts & Forecasts External Station Counts & Forecasts School Points Intersection Density Zonal Summaries Housing & Job Points Household & Person Records Buffer Densities Land Values Detailed Employment Sector Forecasts VMT Mode Shares Impedances / Accessibilities

5 URBAN SIM

6 How does a region develop over time? UrbanSim represents how the regional real estate market is predicted to respond to a set of regulatory land use constraints and demand for new space, consistent with a regional forecast of population and employment.

7 950,000 parcels of data Buildings People Regulatory Environment Constraints

8 UrbanSim Theory UrbanSim combines urban economics, land use and real estate development theory into a series of models that can simulate individual choices. Discrete Choice models Pro-Forma models

9 There are 3 economic agents in the UrbanSim model

10 UrbanSim models the choices households make Rent or own? Single-family home, multifamily, apartment? Location choice within the region

11 UrbanSim models choices of businesses in the region Where will new employees locate? Which existing businesses will choose to relocate within the region? Where will those businesses locate?

12 Real estate developers choices are modeled Supply/demand interactions produce price signals Every parcel in the region is sampled as a viable alternative Land prices, zoning and existing buildings help determine where the NPV will lead a developer to build

13 UrbanSim can account for future scheduled developments UrbanSim will build known, future projects as specified Involvement by member governments is a crucial step in the modeling process

14 Interacts with Focus travel model UrbanSim and Focus can exchange information to reflect changes to the regional transportation plan over the forecast horizon Scenarios can be run to test various alternative transportation plans

15 Scenario analysis Transportation Subsidies, impact fees, financing TOD & Urban Centers

16 Questions?

17 FOCUS

18 What is a Tour? Escort Trip Shop Trip Home-Based Tour (Work) Work Trip Meal Trip Work Trip Shop Trip Work-Based Tour (Shop)

19 Choices Focus Models Long-Term Work, School, Autos Daily Schedule Activity Pattern, Number of Tours Tour Choices Destination, Mode, Time of Day Trip Choices Destination, Mode, Time of Day, Route

20 Focus v 4-Step Models More individual characteristics Household composition Realistic Travel Purposes Work School Escort Personal Business Shop Meal Social / Recreational Tour Consistency Constraints Additional Modes

21 Focus Doesn t Model Transportation Abilities / Disabilities Vehicle Type Choice & Allocation Second (Third ) Jobs Special Circumstances More Complicated Tour Structures Non-Motorized Assignment

22 DYNAMIC TRAFFIC ASSIGNMENT UPDATE

23 DynusT Timeline Summer 2012 Delivery Network review Fall 2013 Changing planning mandates lead to halts in further development Available for individual projects Ongoing discussions with CDOT

24 INTERPRETING FORECASTS

25 Error and Aggregation More Aggregate Regional VMT, VHT, PMT, etc. RTD System-Wide VMT Ridership etc. by Facility Type Ridership by Service Class Individual Freeway Volumes Individual Collector Volumes More Detailed Individual Choices Individual Route Ridership Individual Station Boardings Intersection Turning Movements 1% 10% 25% 100% Approximate Range of Error

26 Example of Simulation Error: Who Lives at 3356 Franklin St? Data Actual Focus 2014 Household Size 1 2 Household Income (1996$) $50-75k $15-25k Household Autos 0 1 Head of Household Gender Male Female Age Industry: 3-Digit NAICS 541 Professional, Scientific, and Technical Services Work Location Zone 1836 Denver CBD 230 Construction Zone 1868 Glendale School Location Not a Student Metro State U Child Gender Female Age None 6 School Location Wyatt Academy

27 Post-Model Corrections Based on Base Year Calibration Additive: Final = Raw + Base Obs. Base Model Ratio: Final = Raw * Base Obs. / Base Model Average of the Two

28 Precision Suggests Accuracy Forecast Volume Round to Nearest < to ,000 to 9, ,000 to 99, >99,999 1,000 Source: NCHRP 765 (2014) after Project Traffic Forecasting Handbook (2012)

29 OBTAINING & USING FOCUS

30 You Will Need TransCAD 6.0 SQL Server 2008 or later Visual Studio 2012 or later Powerful Server or Workstation A Free GitHub Account 100~150 GB Disk per Scenario

31 Process Contact DRCOG for Model Release Form Specify Project Details Forecast Years Areas of Interest Project-Specific Refinements Principal Signs DRCOG Provides Access to Code Repository Input Data Files Output Files if Requested

32 QUESTIONS?

33 THANK YOU!