Complexity, Land use and Cellular Automata Modelling - PowerPoint PPT Presentation

Loading...

PPT – Complexity, Land use and Cellular Automata Modelling PowerPoint presentation | free to view - id: 21cca7-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Complexity, Land use and Cellular Automata Modelling

Description:

Land use in the floodable areas. jaar. MOLAND Flooding risks in Pordenone, Italy (EU-JRC) ... Urbanisation in the floodable area; ... – PowerPoint PPT presentation

Number of Views:134
Avg rating:3.0/5.0
Slides: 43
Provided by: guy54
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Complexity, Land use and Cellular Automata Modelling


1
Complexity, Land use and Cellular Automata
Modelling
  • Guy Engelen
  • VITO Flemish Institute for Technological
    Research
  • Centre for Integrated Environmental Studies
  • Boeretang 200, 2400 Mol, Belgium
  • guy.engelen_at_vito.be
  • RIKS bv
  • Abtstraat 2A, 6200 AL Maastricht, The Netherlands
  • http//www.riks.nl

2
Once upon a time …
Dynamic spatial interaction based models
Net growth gains - losses
GIS - Geographical Information Systems
3
… 15 years ago
Dynamic spatial interaction based models
High-resolution Spatially-dynamic
Net growth gains - losses
GIS - Geographical Information Systems
4
Morphogenesis, spatial organisation and spatial
interaction (micro macro)
5
Cellular Automata capture the Process behind the
morphogenesis
  • Cellular Automata are explicitly dynamic,
    representing change over time
  • Cellular Automata are explicitly spatial, since
    they are defined on an n-dimensional grid

6
Example of a Cellular Automata Conways Life
(Gardner, 1970)
7
Characteristics of CA models (1)
  • Spatially-dynamic model enabling extreme spatial
    detail.
  • Simple and intuitive. They start with local
    relationships and behavioural rules. Complexity
    without complication (Couclelis 1986).
  • Self-organising systems with emergent properties
    locally defined rules resulting in macroscopic
    ordered structures. Massive amounts of
    individual actions result in the spatial
    structures that we know and recognise.
  • Bottom-up approach to spatial modelling.
  • Open. They generate a range of possible futures.
  • But…They are not normative they show what
    could happen rather than what should happen.

8
Characteristics of CA models (2)
  • A universal Turing machine is formally equivalent
    to a Cellular Automata CA can mimic the action
    of any physical system
  • Straightforward linkage with (raster) GIS, easy
    integration of additional spatial attributes
  • CA can be modelled using computers with no loss
    of precision
  • CA are irreversible and irreducible . There are
    no computations or algorithms to speed up the
    cellular automata simulation. There is no way
    but simulation to predict their evolution.
  • Cellular Automata models confirm that modelling
    is about exploration rather than prediction.
    Reality is simply too complex to be captured in a
    model. (sharpened intuition, informed
    speculation, and educated guess (Couclelis,
    1997)).

9
Generalisation of the CA Concept
  • Practical applications in the spatial sciences
    have involved the following amendments to the
    basic CA principles
  • CA-space
  • from infinite to finite
  • from homogeneity to non-homogeneity
  • regularity (cells) to irregularity (polygons).
  • Neighbourhood
  • from stationarity to non-stationarity
  • From isotropy to non-isotropy.
  • Transition function
  • From universal (all cells all states) to
    non-universal
  • from invariance to time variance
  • from deterministic to probabilistic.
  • Time steps
  • from regularity to non-regularity.
  • System closure
  • from closed to open.

10
Generalisation of the CA Concept
  • Practical applications have involved the
    following amendments to the CA principles
  • CA-space
  • from infinite to finite
  • from homogeneity to non-homogeneity
  • regularity (cells) to irregularity (polygons).
  • Neighbourhood
  • from stationarity to non-stationarity
  • from isotropy to non-isotropy.
  • Transition function
  • From universal (all cells all states) to
    non-universal
  • from invariance to time variance
  • from deterministic to probabilistic.
  • Time steps
  • from regularity to non-regularity.
  • System closure
  • from closed to open.

11
CA model for Land use Change (1992-2005)
  • Cellular space
  • 2-D space, consisting of equally sized square
    cells (size limited by capacity of the pc).
  • States are the actual land uses. 3 classes
  • Active states or Function states
  • Passive states
  • Features.
  • Cells may have additional attributes
  • Suitability (static / dynamic)
  • Accessibility (static / dynamic)
  • Zoning status (quasi static).
  • Maximum 32 states
  • Grid size 25 - 1000 m

12
CA model for Land use Change (1992-2005)
  • Circular Neighbourhood, maximum radius 8 cells.
  • Circular because square neighbourhoods have
    a bias towards diamond shaped spatial
    patterns (Li and Yeh, 2000, Hagen 2000)
  • Larger than Moore because socio-economic interacti
    ons take place over longer distances than next
    neighbours (more than diffusion alone)
  • Spatial agents perceive of their immediate
    neigh- bourhood as anything between 50 and 5000
    m (real estate studies) (Urban Studies, 2001,
    Vol. 38, No. 12)
  • Interactions change with distance (Geoghegan,
    2002 green areas on residential)
  • Information passing in the CA. Relation between
    time and spatial expansion in the CA.
  • Practical choice 8 23.
  • enables studying the importance of resolution
    (Grass, 1996)
  • multi-raster applications and linkages with other
    modelling approaches (Lorek, Sonnenschein)
  • provides better solutions for Patch problems.

13
CA model for Land use Change (1992-2006)
  • Transition rules representing
  • Locational preferences of spatial agents in
    competition for space
  • Appreciation of the proximity of other competing
    or befriended activities and static elements in
    the immediate neighbourhood
  • Willingness to develop or give-up activity in a
    particular location.

Commerce
Water
Housing
Forest
Industry
  • Push and pull forces, agglomeration benefits,
    inertia, etc
  • The rule set consists of all the significant
    rules (2-3/function suffices)
  • Interaction weights have relative value
    within the model only.

Commerce
Industry
Housing
14
Transition rules
  • Rules express the locational preferences and
    locational behaviour of spatial agents in a
    survival of the fittest situation
  • Rules express their willingness to produce or
    give up floor space (see also Webster and Wu,
    2001)
  • Rules express (on a relative scale) how they
    value the presence of other functions and land
    uses / land covers in their neighbourhood.
  • Effect at distance 0 of the function on itself
  • Inertia expressing the strength with which the
    existing land use will stick to its present
    location.
  • Effect at distance 0 of any other function on
    the function
  • Ease of re-conversion the ease with which a new
    land use will take over from the existing land
    use.

15
Transition rules
Effects at distance gt 0 No interaction Attracti
on positive agglomeration benefits diminishing
with distance. Repulsion negative agglomeration
benefits diminishing with distance Change in
type of interaction from attraction to repulsion
or/ and vice versa. Strong interaction with far
neighbours, abruptly falling Gradual
decay, Strong interaction with immediate
neighbours, gradually falling Sphere of
influence short tail the interaction is limited
to short distances long tail the interaction
effect works over longer distances.
16
Transition Rule Change cells to land-use for
which they have the highest transition potential
untill the demands are met.
Time Loop
17
Bottom-up Agent based approach to morphogenesis
18
Simulating the growth of Cincinnati from 1840
till 1960
19
Simulation (left) vs. Reality (right)
20
Transition Rule Change cells to land-use for
which they have the highest transition potential
untill the demands are met.
Time Loop
21
Constrained Cellular Automata
  • The Cellular Automata dynamics evolve in a
    non-homogeneous geographical space defined by GIS
    attributes and layers (see also most of the
    others)
  • Their overall dynamics are not determined by the
    micro Cellular Automata transition rules, but
    by processes at a larger macro scale (see also
    most of the others)
  • Cellular Automata models have been integrated
    with more traditional dynamic models, which in
    the most general case are regionalised (spatial
    interaction based) (Engelen et al., 1993).

22
EU-JRC MOLAND Modelling Framework (Example
Greater Dublin)
  • Scenarios on Demographic growth and on job growth
    in 3 economic sectors (Industry, Commerce and
    Services) for the whole area.

Global Greater Dublin
Macro
  • Dynamic Spatial interaction-based model.
    Allocates and re-allocates the population and the
    economic activities among the 9 counties.
  • Takes regional information on population and jobs
    as well as local information on the quantity and
    quality of space as an input.

Regional 9 Counties
  • Cellular Automata based model. Takes the regional
    figures as an input and allocates the population
    and jobs to cellular units.
  • Returns aggregate information to the regional
    model relative to the quantity and quality of the
    space available in the counties.

Local 630,000 cells of 4ha each
Micro
23
Transition Rule Change cells to land-use for
which they have the highest transition potential
until Regional demands are met.
Time Loop
24
Macro-level - Regional Dynamic spatial
interaction based
f ( )
All economic activities, jobs, population,… zoning
, suitability, accessibility, … in zone and at a
distance
For each of the 9 counties, this model calculates
on a yearly basis the changing number of
inhabitants and jobs in the selected economic
sectors.
25
Macro-level - Global Growth scenarios
Scenarios on demographic and job growth in the
selected economic sectors (Industry, Commerce and
Services) for the whole modelled area
26
Building a new application
27
(No Transcript)
28
Assessing planning alternatives Dynamic
indicators at several geographical levels
29
MOLAND Flooding risks in Pordenone, Italy (EU-JRC)
Pordenone 2002 floods
(source Brezger, 2004)
30
MOLAND Flooding risks in Pordenone, Italy (EU-JRC)
Floodable areas in 2000
Urbanised in 2000
F River bed. Frequently flooded P3 Very high
risk. Areas close to dikes P2 High risk.
Critical area with 100-year event discharges P1
Moderate risk. Areas flooded in 2002 and 1996.
31
MOLAND Flooding risks in Pordenone, Italy (EU-JRC)
Population density (inh/ha) in 2000
Land use in 2000
32
MOLAND Flooding risks in Pordenone, Italy (EU-JRC)
Urbanised in 2000
Urbanised in 2020
33
MOLAND Flooding risks in Pordenone, Italy (EU-JRC)
Floodable areas

Land use in the floodable areas
jaar
34
MOLAND Flooding risks in Pordenone, Italy (EU-JRC)
Flooding risk
  • Dangerous cocktail
  • Urbanisation in the floodable area
  • Increased frequency and importance of extreme
    weather events caused by climate change.

Urbanised area per risk category
jaar
35
What else other CA-work is there at RIKS?
  • Calibrations, validations of various applications
    have shown that the Constrained Cellular Automata
    modelling methodology and type of models works
    fine in general
  • That these models can be readily and usefully
    applied for practical planning purposes
  • Models have been developed that integrate
    physical processes, represented at the cellular
    level, enabling to work with a dynamic
    suitability
  • Models have been developed that have a so-called
    two-step land use allocation
  • Models have been developed that combine
    Individual Based Models (IBM) with Cellular
    Automata (Beach Plover Model and Lobster model)

36
Variable grid Cellular Automata the way ahead?
  • Applications coupling CA with spatial interaction
    based models work fine for areas consisting of
    reasonably uniform, equally sized zones (ex. The
    Netherlands)
  • They work less well for areas consisting of
    widely different zones (ex. Larger Dublin
    Metropolitan Area)

37
(No Transcript)
38
The solution Eliminate the regions?
  • Each cell is allocated a certain amount of the
    activity corresponding to its land use. Real
    cell densities can be worked with, rather than
    regional densities
  • The cell neighbourhood is expanded to include the
    entire area modelled.

39
(No Transcript)
40
(No Transcript)
41
The solution Eliminate the regions?
  • Each cell is allocated a certain amount of the
    activity corresponding to its land use. Real
    cell densities can be worked with, rather than
    regional densities
  • The cell neighbourhood is expanded to include the
    entire area modelled.
  • The neighbourhood weighting functions now capture
    the distance decay effects previously represented
    in the macro model.
  • Each cell is now effectively its own region,
    competing with all others for activity.
  • Cell figures are aggregated per administrative
    entities to enable usage at the macro-level, but
    also calibration and validation.
  • First results seem promising, but more research
    work is required, starting next Monday -)

42
The END
About PowerShow.com