Modeling Urban Growth using the CaFe Modeling Shell
Mantelas A. Eleftherios
Regional Analysis Division Institute of Applied and Computational Mathematics
Foundation for Research and Technology - Hellas
An urban growth modeling shell to:
Explore and map the urban growth dynamics
Simulate Urban Expansion
Support Decision and Planning
CaFe
simple, open, with visible mechanisms
extracts and reproduces spatial patterns of change
retains a extendible/reducible knowledge base
combines various knowledge sources
expresses extracted knowledge in a comprehensible way
little data limitations
tranferable
Design
no pre-defined formulas or functions
it does not exclude/require certain input
calculates mean values of each variable’s conditional frequency distribution function
the extracted patterns are space sensitive
scale free
knowledge base in natural language
Exploring & Mapping
parallel connection of each variable and calculation of suitability indexes for urbanization
combines statistical, empirical and theoretical knowledge
allocation of an urban “amount”
the growth is an exogenous parameter
Simulation
alternative scenarios population
population of inverse optima scenarios
scenarios may be based upon : • input data • suitability indexes
Decision Support
Stand alone C code supporting:
information management through Fuzzy Logic
application of Cellular Automata Techniques
basic raster file managements
a GIS is necessary for data pre-processing and results’ visualization
Cellular Automata – Fuzzy Engine
explicit space
implicit time through terms of urban growth
variables are described as fuzzy sets
location is given by a 2D fuzzy variable
Information Management
knowledge in IF – THEN rules
each rule has a certainty factor
each certainty factor is spatially sensitive
suitability rules have simple hypotheses and are accumulated using the Dempster-Shaffer theory of evidence:
Knowledge Management
n
i 1i )CF -1 ( - 1CCF
Structure of CaFe
1. Calculation of suitability per variable and overall suitability
2. Iterative CA-based urban cover allocation
Case Study
the broader Mesogia area in east Attica
635 s.km
11+7
municipalities
> 100.000
population
Available Data: Corine land cover for 1994, 2000, 2004 road network for 1994, 2000, 2004 DEM
the 1994-2000 period was used for knowledge extraction and model calibration the 2000-2004 was used for model evaluation
Application
Evaluation
Error Indexes: Model Map
overestimation error 0,11 0,023
underestimation error 0,08 0,015
total error 0,19 0,039
total error for results with 0,05 0,009
Certainty >70%
Error Accumulation
Map Error
Model Error
Overestimation Underestimation Total
Overestimation Underestimation Total
Results
Results ΙΙ
Results ΙΙI
Fuzzy Logic and Cellular Automata consist an advisable framework to describe and simulate urban growth
CaFe is capable to simulate is a satisfactory way the short term urban growth using little data
CaFe’s output refers to housing activities rather than the whole of the artificial surface
Conclusions
stochastic KBE module
spatially sensitive Dempster-Shaffer operator
unbinding the over- and under-estimation errors
applications and further evaluation
Future Directions
Modeling Urban Growth using the CaFe Modeling Shell
Regional Analysis Division Institute of Applied and Computational MathematicsFoundation for Research and Technology - Hellas
CaFe: Cellular Automata – Fuzzy Engine
Mantelas A. Eleftherios
e-mail [email protected]. +30 2810 391736
Top Related