MASS: From Social Science to Environmental Modelling

1
MASS From Social Science to Environmental
Modelling

• Hazel Parry

http//www.geog.leeds.ac.uk/groups/mass/
2
Outline

• Background
• Connections between social and ecological
  modelling
• Advantages and logic of using MASS in ecology
• Multi-agent systems as a unifying methodology for
  environmental modelling in geography?

3
Established modelling techniques in ecology and
physical geography

• Differential Equations
• Lotka-Volterra (predator-prey)
• Navier-Stokes equations (fluid-flow)
• Horton equation (infiltration)

4
Background The complexity paradigm

• Complexity
• Neither random nor regular, when it is hard to
  formulate overall behaviour of a system, despite
  individual-scale information.
• Self-organization
• The process by which autonomous agents interact
  in a seemingly chaotic manner, resulting in
  global order.
• Emergence
• Simple units, when combined, form a more complex
  whole. For example, ecosystems are a synergy of
  individuals. The ecosystem is greater than the
  sum of its parts (Odum).
• Complex systems
• Made up of agents interacting in a non-linear
  fashion. The agents are capable of generating
  emergent behavioural patterns, of deciding upon
  rules and of relying upon local data.

5
Social vs. Economic vs. Ecological worlds
Social Sciences Economics Ecology
Society Economic Interaction Ecosystem
World of (social) interactions Game/Puzzle World of (ecological) interactions
Interdependence Interaction Interdependence/ interaction
Dependence, value Utility Dependence, utility, need
Action Strategy/ Move Action
Dependence theory Game Theory Ecosystem Theory
Interference, Influence, Exchange Strategy Competition, predation, parasitism
6
Object based models in ecology and social science

• Individual-based models
• Large collection of interacting organisms.
• Cellular Automata
• Cells on a grid of specific dimension, undergo
  transition by global rules.
• Multi-agent simulation
• Intelligent agents, with ability to learn about
  their environment and adapt their behaviour
  accordingly.

7
Cellular Automata

• discrete models of spatio-temporal dynamics
  obeying local laws (Randy Gimblett, 2002, pp2)
• Grid-based formed by identical cells
• Interaction of cell with its neighbours
• Time advances in steps
• State of cell determined by global rules

8
Example - diffusion
















Von Neumann
t0
t1
9
Cellular Automata in ecology
Le Page and Bousquet Cellular Automata model for
the spread of forest fire
10
Cellular Automata in physical geography
Murray-Paola model of sediment transport in rivers
Baas Model of sand dune landscape formation
11
Multi-Agent Systems and Simulation (MASS)

• Similar to CA
• Less rigid structure
• Interactions between distant individuals at a
  variety of scales
• Facilitate investigation of lower level
  mechanisms leading to global structural and
  dynamical features

12
MASS a logical ecological modelling strategy
13
The advantages of a MASS approach

• Reduced randomness
• Increased flexibility
• Increased realism perception, communication,
  rationality, goals, interactions, autonomy,
  mobility and collaboration all possible.
• Can handle complex systems
• Agents have the capacity to evolve or adapt their
  behaviour.
• Dont need to throw the baby out with the bath
  water!
• Integration of landscape models with ecological
  and social models

14
A unifying methodology?

• Environmental management needs to be more
  integrated and flexible.
• Ecological models benefit from an integral
  dynamic environmental model to produce realistic
  simulations.
• They also benefit from a consideration of the
  social structure and dynamics where decisions
  impact the entire system.
• For example
• SIMDELTA
• MODULUS

15
SIMDELTA
Village

• The artificial world of SIMDELTA (Bosquet and
  Cambier)
• Dynamics of fish population
• Biological and topological factors affecting the
  evolution of the fish
• Decision making of the fishermen

16
Discussion

• Contributions of social science to agent-based
  simulation in ecology.
• Potential to use multi-agent simulation in other
  areas of physical geography.
• Multi-agent systems as a unifying methodology for
  environmental modelling in geography?