LandUseModel CUSIM M.ppt · Land Use and Economic Policies Model. Land Use Modeling Frameworks From Waddell 2005 LOWRY: Gravity Model Leontieff: ... Vector grid zonal system example

CUSIM MCUSIM-M:Competing-destinations Urban Spatial

I t ti M d l i MATLAB Interaction Model in MATLAB

A Forecasting Tool for Spatial Distribution– A Forecasting Tool for Spatial Distribution of Employment and Population –

David Jung-Hwi LEE, Senior Regional Planner

George Washington Regional CommissionOFredericksburg Area MPO

GWRC/FAMPO

DC

RichmondRichmond

Virginia: 21 Planning District CommissionsVirginia: 21 Planning District CommissionsPD16 GWRC PDC vs. FAMPO (Mid/small size MPO)

Stafford, Spotsylvania, King George, Caroline, and Fredericksburg City

9I-95

Population: 309,763 (2006)

Employment: 113,501 (2006) Employment: 113,501 (2006)

TAZ: 802 zones

2000 GWRC/FAMPO Work Trip Patterns

Work T i

Work T iTrip

67148 59.1% To FAMPO Workplace

46471 40.9% To outside Workplace

113619 100.0% From FAMPO Residence

Trip

67148 82.3% From FAMPO Residence

14477 17.7% From outside Residence

81625 100.0% To FAMPO Workplace

2000 GWRC/FAMPO Work Trip Patterns (Intra Only)

Integration of

Trip Generation Trip DistributionCapacity-

Constrained TripAssignment

Mode-SplitAnalysisLand Use

and Transportation

Future (Design)Network

Specification

Evaluation ofCharacteristics of

Loaded

TransportationModel

Transportation

Transport Services (Congested)Future Network

Integration:Forecasting Data

Integration:Initial Data Integration:

Forecasting Data

Specification(Characteristics) of Future

Transportation Systems andFacilities

INTEGRATION: Policy Feedbacks

INTEGRATION: Evaluation Feedbacks

Facilities

Description ofBase Year Spatial

Distributions ofActivities

Forecasts ofSpatial

Distributions of

Land Use (Residence andEmployment Location)Forecasting Procedure

Local / Regional Review andEvaluation of Land Use and LocationPatterns and Policy Consequences

Region-wideForecasts of

Population andEconomy

ActivitiesForecasting Procedure Patterns and Policy Consequences

Land Use

Descriptions of ProposedLand Use and Economic

Policies

Use Model

Land Use Modeling Frameworks From Waddell 2005

LOWRY:Gravity Model

Leontieff:Input-Output Model

Alonso/Mills/Muth:Urban EconomicBid-Rent Theory

Orcutt:Microsimulation

1960

Spatial InteractionLILTDRAM/EMPALHLFM II+

Spatial Input-OutputMEPLANTRANUS

McFadden:Discrete-Choice

1970

HLFM II+LUTRIM Models

EquilibriumC

DynamicC

Geographic f

1980Discrete Choice Land Use Models:METROSIMMUSSAM t

Discrete Choice Land Use Models:NBER/HUDSMASTERIRPUD

Information Systems

Spatially Detailed1990Metroscope IRPUD

Spatially DetailedDynamic Land Use

Spatially DetailedRule-based Planning Tools:INDEXPlaces2000

Hybrid Spatial I/OPECAS/ODOT 2nd

Gen

Models:DELTAILUTEUrbanSim

PlacesUPlanCUFMWhatIf?

2000

• Land Use

Integration of Land Use and Transportation• Land Use

– Macro: Spatial Interaction Models• TELUM, Metropilus, ITLUP, DRAM/EMPAL/LANCON (S.H. Putman) •• CUSIMCUSIM--M / CUSIMM / CUSIM--C++ / CUSIMC++ / CUSIM--G (David JG (David J--H. LEE)H. LEE)

– Linear Programming Models• TOPAZ

– Input-Output– Input-Output• TRANUS (De La Barra)• MEPLAN (Echenique)• PECAS (J.D. Hunt)

– Micro: Economic Based Approaches• MUSSA (Francisco Martinez C.)• METROSIM (Alex Anas)• METROSCOPE (Portland Metro)• METROSCOPE (Portland Metro)• UrbanSIM (Paul Waddell)• DELTA (D. Simmonds)

• Transportation

– Four Step Modeling/ l d d l ( )/ l d d l ( )•• CUBE/Voyager Travel Demand Model (TDM)CUBE/Voyager Travel Demand Model (TDM)

• TransCAD Travel Demand Model (TDM)– Activity Based Approach

MODEL SEARCH :Aggregate s Disaggregate Zonal S stemAggregate vs. Disaggregate: Zonal System

• TAZ: small enough?

• Zonal Scale- Broadbent’s Rule- Sikdar (1982, 1984)

S l d H l (1996)- Steel and Holt (1996)- Norner and Murray (2002)- Gitlesen (2004)as disaggregate as possibleas disaggregate as possible

• GIS & Microsimulation: models the behavior of individuals

Di t d t DBMS- Dissagregate data; DBMS- Fotheringham & Wegener (2000)- Spiekermann & Wegener (1996)- Landis & Zhang Landis & Zhang

• GWRC data: Limited parcel layer availabilityG di TAZ l l

Vector grid zonal system example

Geocoding: TAZ level accuracy Thiessen Polygon

MODEL SEARCH :Aggregate or Disaggregate ?Fotheringham and O’Kelly (1989)

Aggregate or Disaggregate ?

Aggregate FlowsSpatial Interaction

Disaggregate FlowsIndividual DecisionsMore BehavioralGeo-psychologicalSpatial AwarenessSpatial Information Processing

Spatial Choice

Spatial Information Processing

MODEL SEARCH :S ti l i t ti d li (SIM)

(1) spatial interaction as social physics

Spatial interaction modeling (SIM)

Newtonian gravity modelLowry model

(2) spatial interaction as statistical mechanics

Extensions of Lowry model

Wilson’s entropy maximizing procedurefamily of spatial interaction modelsy pAlonso’s (1973; 1978) framework Additive Version of Family of SIM by Tobler

(3) spatial interaction as aspatial information processing

(4) spatial interaction as spatial information processingg

Fotheringham 2000

MODEL SEARCH :S ti l i t ti d li (SIM)

(1) spatial interaction as social physics

Spatial interaction modeling (SIM)

( ) p p y

(2) spatial interaction as statistical mechanics

(3) ti l i t ti ti l i f ti i(3) spatial interaction as aspatial information processing

discrete choice models by McFadden LOGIT f l ti (1974 1978 1980)LOGIT formulation (1974; 1978; 1980)Two undesirable properties - independence from irrelevant alternatives (IIA)

(4) spatial interaction as spatial information processing

- regularity

Nested LogitCompeting Destinations modelNo more IIA and RegularityNo more IIA and Regularity

Fotheringham 2000

MODEL SEARCH :Nested Logit s Competing Destinations

hierarchical processing

Nested Logit vs. Competing Destinations

search strategy

Nested Logit Competing Destinationsp g

assumes that the modeller has knowledge to individual choice

considers the likelihood of an alternative being o dg o d dua o

setsof an alternative being in the true choice set

Thi lik lih d i di t This likelihood is according to the similarity, or position, of that alternative relative to the other alternatives

Pellegrini and Fotheringham (1999)

to the other alternatives.

and Fotheringham (1999)

OPERATIONALIZATION[C++: CUSIM C++ 2005 10 Model][C++: CUSIM-C++-2005-10 Model]

OPERATIONALIZATION[MATLAB CUSIM M 2007 09 M d l][MATLAB: CUSIM-M-2007-09 Model]

Better Capability for Matrix Calculation

Faster Model Run TimeFaster Model Run Time

Simple Modular Programming

OPERATIONALIZATION[GAMS CUSIM G 2007 09 M d l][GAMS: CUSIM-G-2007-09 Model]

Good Optimization Tool

Various Solvers

Good Optimization Tool

COMPONENT :Economic Economic Base TheoryTheory

COMPONENT :Deterrence Function [Travel Time Matrix] : [ ]From FAMPO Travel Demand Model

Network

4-Step Model

Visualized Travel Time Matrix

Travel Time Matrix

INPUT :Control TotalsControl Totals

2000 2005 2006 2010 2015 2020 2025 2030 2035

BAS 51,199 68,417 73,042 82,891 94,658 107,137 119,616 131,766 143,917

NBAS 33,414 32,566 37,742 42,831 48,911 55,359 61,808 68,086 74,364

Virginia Employment Commission

Virginia’s Electronic Labor Market Access

EMP 84,613 100,983 110,784 125,722 143,569 162,496 181,424 199,852 218,281

POP 243 505 301 355 309 483 351 213 401 070 453 945 506 819 558 302 609 784

US Census Bureau Estimates Annual Population Data

State Demographer Projections Population Data

Activity Rate a=TTP/TTE2004 2005 2006

a 2 878 2 984 2 794

POP 243,505 301,355 309,483 351,213 401,070 453,945 506,819 558,302 609,784 Population Data

Activity Rate a TTP/TTE

Population Serving Ratio b=TSE/TTPa 2.878 2.984 2.794

b 0.137 0.108 0.122

INPUT :Control Totals (Pop lation)Control Totals (Population)

INPUT :Basic s Non Basic (Location Q otients)

• e = employment in some

Basic vs Non-Basic (Location Quotients)

• eir = employment in some industry (i) in some region (r)

• Σeir = total employment in the region

eir

ei∑ region eir∑Ei

• Ei = national employment in some industry (i)

i

Ei∑• ΣEi = total national employmenti p y

INPUT :Basic s Non Basic (Location Q otients)Basic vs Non-Basic (Location Quotients)

• Location quotient = 1 – Local production can just satisfy local consumption.

• Location quotient > 1– Local production can satisfy local consumption, and p y p ,

the excess is exported. This is a “basic” industry.• Location quotient < 1q

– Local production can not satisfy local consumption, and the difference must be imported.

INPUT :Basic s Non Basic (Location Q otients)Basic vs Non-Basic (Location Quotients)

• Not all of a basic industry is “basic.”• Only that part of the industry that servesOnly that part of the industry that serves

the export market is considered basic.It i th t t f th i d t th t i th• It is that part of the industry that raises the location quotient above 1.0

INPUT :Basic s Non Basic (Location Q otients)

Calculating Basic Employment Using Location Quotients

Basic vs Non-Basic (Location Quotients)

PD16 United States

NAICS 236xx 2,050 1,721,661Total Private Sector Employment 105,167 131,640,225

Employment Ratio 019492 013078Employment Ratio .019492 .013078

Location Quotient = ( 019492 / 013078) = 1 49Location Quotient = (.019492 / .013078) = 1.49

Non-basic Employment = (.013078) * 105,167 = 1,375

Basic Employment = 2,050 – 1,375 = 675

INPUT :Zonal Basic Emplo ment (e ogeno s)Zonal Basic Employment (exogenous)

APPLICATIONS & FILE FORMAT

CUBE/Voyager: TDM MATRIX: ** . MAT

MATLAB: CUSIM-M ** . CSV

TransCAD MATRIX: ** . CSV ** . DBF

GAMS: CUSIM-G MATRIX: ** . DBF ** . GDX

** . DBF

Visualization: ArcGISCUBE/Vo ager

. DBF

Visualization: ArcGISCUBE/Voyager

CUSIM-M-2007-09 Model : Forecasting2035

Forecasting for Year 2009 2035 with “CUSIM M 2007 09 Model”Forecasting for Year 2009~2035 with CUSIM-M-2007-09 Model

Density C t i t Constraint

Density Constraint Wt_POP:

weighting factorA vector of the ratio of the zonal maximum constrained population A vector of the ratio of the zonal maximum constrained population over the zonal estimated population

Maximum PopulationCapacity

(Const_POP): 500

Maximum PopulationCapacityCapacity

(Const_POP): 2,000

Density Constraint

U Uneven capacity

ByJurisdiction

Maximum Maximum PopulationCapacity

(Const_POP): 12,00012,000

Density Constraint w/ Land Use

Current Land Use for George Washington Region, 2006

Density Constraint Zonal Specific Population Capacity Constraints

Zonal Population Capacity vs. Estimated Population

Hybrid Modeling Approach

MATLABMATLABSimulation (Descriptive) Model

characterizing the general operations and mechanics of land use changes and development patterns

Model

GAMS prescribing Optimization (Prescriptive) Model

p goptimal urban development patternswith minimal concerns associated

Land Use and Transportationin LRTP

Thank you very much!!Thank you very much!!

David J-H. Lee

GWRC/FAMPO540-373-2890 Ext.235 0 3 3 890 t [email protected]