A model-based approach towards assessing landscape restoration activities in Watershed 263, Baltimore, MD

1
A model-based approach towards assessing
landscape restoration activities in Watershed
263, Baltimore, MD

• Brian Voigt
• University of Vermont - Spatial Analysis Lab
• brian.voigt_at_uvm.edu

2
Current Research - UrbanSim

• Modeling urban development patterns in Chittenden
  County, VT using UrbanSim
• simulate future land use and associated
  environmental impacts under baseline conditions
  and alternative scenarios
• Quantifying effect(s) of future urban development
  patterns have on
• water quality, habitat fragmentation, aesthetics,
  auto-dependency, energy consumption, etc.
• Intended to facilitate discourse not predict
  policy adoption or exact development locations

3
Project Collaborators

• Austin Troy, University of Vermont
• Morgan Grove, USFS
• Guy Hager George Friday, Parks and People
  Foundation
• Bill Stack, Department of Public Works
• Others
• Watershed council, community residents, BES
  collaborators

4
Research Questions

• How do we design a simulation modeling framework
  to facilitate learning about future landscape
  trajectories based on human interventions and
  watershed restoration activities?
• How will social and environmental conditions
  within Watershed 263 change as the City of
  Baltimore and the Parks and People Foundation
  strive to meet the urban forestry initiative
  goals?

5
Project Goals

• Use a participatory modeling approach to explore
  relationships among socio-economic and
  biophysical system characteristics of a complex
  natural human urban system
• Help residents and resource managers to consider
  the effects of human interventions associated
  with varying levels of green infrastructure
  investment
• Facilitate a learning process about the natural,
  biological and socio-economic components of the
  watershed and their collective interactions that
  define the current state and potential
  trajectories of watershed evolution

6
The Simile Modeling Environment

• Dynamic, spatially explicit, interactions
  feedback
• Stocks, flows parameters
• Visual modeling environment
• Sub-models can be used independently or grouped
  with other system components
• Use equation editor to formalize variable
  relationships and sub-model interactions

7
WS263 Model Framework

• Suite of sub-models that interact with one
  another
• Partition landscape into set of grid cells and
  define initial condition based on biophysical and
  socio-economic parameters
• Agent-based approach representing household level
  decision-making (e.g. relocation, rent v own,
  etc.)
• Scenario-based analysis to improve our
  understanding of the system and accommodate
  variations in data interpretation and relative
  effects of system components

8
Data Sources

• Demographics
• US Census Public-Use Microdata Samples (5
  sample), Summary File data tables (SF1 SF3)
• BNIA neighborhood indicators
• Biophysical
• BES land cover, topography, water quality, air
  quality
• Socio-economic
• BNIA employment and population control totals,
  forecasts
• BES employment sites, real estate transaction
  data, current land use / land use history,
  household surveys, PRISM classification
• Infrastructure
• Sewer system, road network, transit
• Landscape interventions
• PPF DPW list of completed, proposed,
  anticipated projects

9
Proposed Model Components 1

• Land use probability of transition from one type
  to another
• Land cover changes with interventions, aging
  vegetation, infrastructure addition / removal
  relationship to water quality and other ecosystem
  services
• Land price defined by a hedonic model at the
  cell level
• Employment allocate employment at the cell level
  based on externally derived control totals using
  a gravity model
• Residential location choice internal and
  external agents (households) synthesized from US
  Census, PUMS, and household survey data with a
  focus on tenure, length of residency, employment
  and income includes QOL attributes

10
Proposed Model Components 2

• Intervention location choice probability of
  success exogenous inputs define number and type
  of projects multiple sub-models for different
  types of interventions
• Landscape metrics land use mix, proximity to
  amenities / disamentities, fragmentation,
  residential and employment densities updated
  annually, these metrics will be used as variables
  in the other model components statistical
  analysis and existing literature will estimate
  relationships between metrics and system
  components
• Mechanism to integrate external models (e.g.
  UFORE, etc.)

11
Model Output

• Preliminary list of indicators
• land value, canopy cover, habitat fragmentation,
  residential relocation and vacancy rates, QOL,
  green infrastructure density and water quality
• refine list of indicators based on further
  collaboration with PPF and watershed council
  representatives
• Data visualization
• results depicted graphically as maps, overlaid
  with major streets and cultural landmarks, by
  joining the output to polygons bounded at
  alternative geographic scales (e.g., block group,
  neighborhood, etc.)
• convey findings and engage stakeholder discussions

12
Expected Products

• Fully documented model
• Detail assumptions, limitations, and future
  improvements
• Transferable to other urban sites
• Sub-models can be recycled for other
  applications
• Scenario analysis capability
• Foster discussion among stakeholders
• Useful for evaluating our knowledge of system
  components and understanding of system
  interactions
• Algorithms for computing indicators

13
Next Steps

• Explore relationships among diverse collection of
  data from multiple sources
• Define base year condition
• Create synthetic population at the household
  level
• Conceptual model development (early 2007)
• Work with project collaborators to identify
  appropriate indicators and techniques for
  conveying information / results to diverse
  stakeholder groups

14
Questions?