Land use

1
Land use Transportation Integrated models (The
competition to ILUTE)

• MAMAMIA - module 3
• Microsimulation experiences
• Ilan Elgar

2
Introduction

• A sort of definition- Integrated land
  use-transportation models attempt to predict the
  dynamics of land use patterns and travel
  patterns, and their interactions.
• The spatial location of land use constitutes the
  choice set from which each individual can choose
  where to perform his activities.
• Land use patterns influence activity travel
  patterns, through providing opportunities and
  constituting constraints.

3

• The choice of a destination is important for the
  feasibility of a particular function at a
  particular location. If there is a shift to a
  different activity location, there will be a
  change in land use patterns and as a result a
  second change to travel behavior.
• All the above should have made the field very
  attractive to research and practice. However
• Integrated land use-transportation models has
  always been a weak field of interest in
  comparison to transport demand, which usually
  treat land uses as exogenous.

4
Why the field is so under appreciated?

• It is inherently very complex and thus difficult
  to model
• Models require a tremendous amount of data, and
  time before any test can be performed.
• Models are expensive to implement - a big
  investment in people, data and equipment is
  required
• The models support mainly long term and strategic
  planning decisions. Such decisions are infrequent
  in most urban areas - the return on investment is
  potentially low.

5
Field Overview
6
The first wave aggregate spatial interaction
based models

• Aggregated models
• Without any or with Partial transportation side
• Gravity models
• Lowry, Garin, TOMM, PLUM, ITLUP/DRAM/EMPAL/METROPI
  LUS, IRPUD, LILT

7

• Lowry
• Three types of sectors - population, service
  employment and manufacturing employment.
• Three types of corresponding land uses
  residential, service and industrial.
• The model allocates activities to zones according
  to the potential of the zone, subject to capacity
  constraints and maximum density.
• Time treatment- all activities respond to changes
  in potential in a given period.
• TOMM- same model structure, but disaggregated
  population into a few socio economic groups. Time
  treatment- only a certain proportion respond in
  the first year, to account for inertia.

8
The second wave utility maximizing multinomial
logit based models

• Combinations of aggregate and disaggregate
• Modules involving logit and microsimulation
• Elaborated transportation microsimulation
• MEPLAN, TRANUS, BASS/CUF, MUSSA/RUBAN,
  CATLAS/METROSIM, DELTA, UrbanSim, IMREL, TILT,
  Uplan

9
The third wave towards activity based,
microsimulation models

• Mostly disaggregate
• Microsimulation oriented
• Very focused on activity
• ILUTE, Ramblas, ILUMASS, The Irvine simulation
  models

10

• Ramblas
• Simulates the whole Dutch population (16m).
• input to the model is the distribution of types
  of HH across the different kinds of dwellings per
  zone and the dwelling distribution in the zone.
  These variables and their changes are external.
• Attributes of zones are variables that can be
  manipulated by planning.
• The spatial distribution of activities and trips
  are treated as dependent variables. Thus policy
  decisions that will affect zone attributes will
  change activity patterns
• The micro simulation occurs in two steps. In the
  first agendas are chosen for each simulated
  individual. In the second step a schedule and
  location for each activity is being determined.

11
MEPLAN

• Treatment of space- the state of Oregon and a
  ring of 50 miles surrounding it, is divided to
  three space unit types. Each space definition is
  used in different parts of the model. Each of
  these forms covers the whole of the model area.
• Grid cells- (14.5m) are 30m30m in built areas
  and 300300 otherwise. Each cell represents only
  one land use.
• Link tributary areas- containing the O-D of trips
  feeding the links in the transport system
• Zones- roughly traffic zones (about 3200).
  Connected to the transport network using centroid
  connectors

12
MEPLAN-CONT.

• Treatment of time- the model evolves by yearly
  steps.
• The characteristics of the system in year t1 are
  partly influenced by the conditions in year t.
• Some of the simulation processes evolve through
  shorter than one year time steps, so they are
  performed a number of times within a one year
  period.

13
MEPLAN-CONT.

• Mechanics of the system- the behavior of
  individuals and groups is based on utility
  values. A utility value is a dependent value in a
  utility function the utility function is
  constructed through alternative attributes and
  individual sensitivities.
• Five different categories of utility are defined
• Rutility- allocate aggregate quantities.
  Attribute variables are either average or zonal,
  while sensitivity values are typical values for
  the category of aggregate quantity being
  allocated

14
MEPLAN-CONT.

• Zutility- used for agent based microsimulation
  of individuals and households, the attribute
  values are average or zonal, but the sensitivity
  values are specific to the person or HH
• Two other utility values are used for traffic
  assignments where one is for aggregate zone to
  zone flows and the other is for path selection of
  individuals
• Cutility- used for microsimulation of land
  development decisions. The attribute values are
  cell specific while the sensitivity values are
  typical values assigned to the developers as a
  category

15
MEPLAN-CONT.

• Model structure- The model is constructed of
  seven separate modules. Each of The modules is
  connected to the data store component and through
  it to each of the other components.
• This design allows flexibility since it allows
  different approaches in each component
  equilibrium or disequilibrium, aggregate or
  disaggregate and statistical models or
  microsimulation

16
MEPLAN Modules

• Regional economics and demographics- provides
  regional control totals for production and
  import-export by economic sector, migration etc.
• The module requires exogenous forecasts as
  inputs and includes simple macroeconomic models
  to assist with trade flow predictions.
• The module is a combination of econometric and
  input-output models

17

• 2. Production allocations and interactions- The
  module determines the total regional production
  of all sectors and then allocates them.
• Commodities (goods, services, labor and space)
  flow from production locations to exchange
  locations to consumption locations.
• Exchange locations (fictitious) represent the
  places where prices for those commodities are
  being set. Prices for each commodity in each zone
  are updated based on demand and supply
• The allocation to locations is being done
  through logit models that take into account the
  exchange prices and the transportation costs.
• The ability of labor supply to change in each
  zone beyond certain limits in a given year is
  constrained.

18

• 3. Household Allocations Module- determines for
  each year
• Changes in HH and members characteristics
• Specific sensitivities
• HH actions regarding home location
• Car ownership change
• Employment and school status and location for
  each HH member
• The module works through microsimulation of each
  HH member in each time period

19

• 4. Land development Module- determines the
  changes from one year to the next.
• The supply of built space is fixed for a given
  year.
• Other modules determine the prices of different
  spaces in each zone based on supply the land
  development model has to adjust the quantity of
  space in the zone over time in response to price
  changes.
• The process is done for each grid cell.
• Logit models determine whether a cell will be
  redeveloped, if so into what type of space and
  the quantity of the new development in the cell.

20

• 5. Commercial movements Module- determines truck
  movement during a representative workday for each
  year.
• A fully disaggregate list of truck movement is
  synthesized providing
• vehicle type
• start and end links
• starting time
• commodity carried for each truck
• The origin and destination of each truck is
  randomly attributed to particular grid cell that
  have a consistent space type with the shipped
  commodity

21

• 6. Household travel Module- creates a list of
  specific individual trips during a representative
  workday for each year.
• For each trip
• start and end links
• starting time
• tour mode
• vehicle occupancy if auto mode
• The process starts with scheduling of activities
  for each household member (first activity pattern
  then duration).
• Each home based and work based tour is
  considered separately.

22

• 7. Transport supply Module- determines
  transportation network loadings given transport
  demands from household and commercial travel
  modules.
• It uses a micro assignment at the level of the
  individual vehicle and traveler.
• Primary mode of trip dictates which network links
  are available for the trip.
• In the first stage of the trip assignment an
  aggregated zone to zone assignment is done. The
  equilibrium set of links is the starting point of
  the micro simulation assignment.

23
UrbanSim

• Treatment of space-location is determined using
  grid cells of 150 by 150 meters, the cell size
  can be modified
• The location grid allows explicit cross
  referencing of other spatial features including
  planning and political boundaries such as city,
  county, traffic zones, and urban growth
  boundaries, as well as environmentally sensitive
  areas
• Each cell is classified according to its main
  development type and the density of the
  development in it

24
UrbanSim cont.

• The data store represents each household in the
  metropolitan area as an individual object, with
  the primary characteristics relevant to modeling
  location and travel behavior
• household size
• number of workers
• presence of children
• household income, etc
• The household list is synthesized by integrating
  census household-level data.
• Employment is represented in the data store as
  individual records for each job and its
  employment sector.

25

• The data store maintains an explicit accounting
  of real estate and occupants, linking individual
  households to individual housing units, and
  individual jobs to job spaces.
• When jobs or households are predicted to move,
  the space they occupy is flagged as becoming
  vacant, and when they are assigned to a
  particular housing unit or job space, that space
  is reclassified as occupied.
• The integrated data store of households, jobs,
  land, and real estate is what the model
  components update over time.

26

• Some of the models, such as the economic and
  demographic transition models, are aggregate,
  nonspatial models. Other components, such as
  location choice, are discrete choice models of an
  agent making choices about alternative locations.
• Inputs to the model include
• base year data store
• control totals derived from regional economic
  forecasts from an external macroeconomic model
• travel access indicators derived from external
  travel demand models
• scenario policy assumptions regarding development
  constraints arising from land use plans
• environmental constraints etc.

27

• The individual modules are
• An accessibility module which predicts the
  pattern of accessibility by auto ownership level
• An Economic and Demographic Transition Module
  predicts the creation or loss of households and
  jobs by type
• Household and Employment Mobility module predicts
  the movement of households or jobs within the
  region
• Household and Employment Location Choice module
  predicts the location choices of households and
  jobs from the available vacant real estate
• Real Estate Development module, the location,
  type, and quantity of new construction and
  redevelopment by developers
• Land Price module determines the price of land at
  each location
• The Model Coordinator, manages the individual
  model components and handles the scheduling and
  implementation of events.

28
UrbanSim cont.

• The model interprets the cumulative impact of the
  policies by reflecting the most restrictive
  policies that apply to a given grid cell.
• In addition to development constraints, the
  scenario inputs include regional control totals
  from the external macroeconomic models and
  assumptions about the space utilization rates
  (such as square feet per employee for different
  development types).
• Transportation policy assumptions are
  incorporated in the external transportation model
  and are embedded in the travel time and utility
  outputs from the travel model that UrbanSim uses
  to calculate accessibility

29
UrbanSim cont.

• The link between the model and the Travel Demand
  Model System is two way. Different accessibility
  values from the travel model will influence the
  decisions agents make resulting in different
  travel demand. This new demand will feed back
  into the travel model.
• The external travel models provide travel times
  and utilities to the Accessibility Model.
• The travel model is run to simulate a typical
  day every 5 years. This is due to the long
  running time of the model and the slow change in
  its products.

30
Problems of integrated land use transportation
models

• Many models remain aggregate in nature and
  therefore are more of entropy maximization models
  than discrete choice utility maximization models.
• The field is behind in applying state of the art
  spatial choice models
• The behavioral and theoretical foundations of the
  field are still weak. Theory has been mostly
  borrowed instead of developed to specifically
  address the questions the field faces.
• Models are often criticized as being black-box
  models because of their complexity

31
Questions

• The behavior of search- does individuals are
  really utility maximizers? (full knowledge,
  variety searching etc.)
• Business establishments- Should there be
  microsimulation of firms? (How do firms appear
  and disappear? How can we model the variety in
  firm sizes, production functions, management?)
• Motivations- Activities do not arise of their
  own, they arise out of motivations to satisfy
  different needs. Should we model motivations?

32

• Perception of space- How do agents perceive
  location alternatives in space? How does the
  knowledge and accuracy of location vary across
  space? Is there a strong relation between
  distance and search cost (could it be used as a
  filter alternative)?
• Sensitivities- Does a long exposure to an
  alternative changes sensitivities to the
  attributes of that alternative? Do campaigns have
  effects on sensitivities in the long or short run
  and what are those effects?
• Public agencies- Hunts suggestion is to
  introduce an authority agent.

33

• The End

34
(No Transcript)
35
(No Transcript)