skip to primary navigation skip to content

Department of Geography


Discrete Simulation Systems for Environmental Modelling

Many systems that are encountered in environmental modelling consist of set of discrete entities that we might be able to represent directly in making computational models of the environment.

Examples include physical systems, for example particle-based flows, such as debris avalanches, or the transport of grains in a river-bed, ecological systems, including forests, animal herds and predator-prey interactions, and social systems, including the effects of farming and land-use change. In addition to these intrinsically discrete aspects of the environment, there are nearly continuous things that we are forced to model discretely simply because of computational restrictions. Here the primary example is fluid flow, including the atmosphere and ocean, but also flow in streams and rivers. Arguably we need to try to model all of these aspects together in order to be able to form a clear view of how the environment will evolve over time, including interactions between physical, ecological and social systems.

As a way of achieving such a goal we aim to build up a suite of discrete simulation models that span the disciplines in a way that gives us a unified environmental modelling system. The system will include:

  • Discrete Element Particle models
  • Smooth-Particle Hydrodynamic Flow models
  • Individual-based ecological models
  • Agent-based Social Simulation models

An example of the last approach can be found on the CLOUD project page, and of the last two coupled together and including a physical model of hydrological flow on the Integrating page. The ideas behind the project can be made more concrete with the aid of the following diagram:-

Diagram as described adjacent

The focus of the diagram here is an individual agent, show embedded in an environment that provides resources for the agent’s survival, but also acts as a source of hazards that need to be dealt with. These resources and hazards are themselves dynamical systems that require modelling in order to be able to understand the system as a whole. So one source of hazard might be flooding, which requires modelling to gain an insight into its frequency and magnitude, but does not operate independently of land-use, itself a dynamic process involving the evolution of the resource-base that provides human-agents with their means of survival. An agent in this context , may represent any kind of biological or social entity, and the degree to which the internal structure of the agent needs to be elaborated varies with context. For example, trees do not need the arrow labelled “cognitive”, but are purely reactive agents, interacting with the environment through their physiology, and with other agents in the form of competition for light, spreading of seed, or response to damage by animals or people. People on the other hand may need to be represented using a more detailed set of rules governing behaviour, modified by perception of the environment and of other agents, including the cultural and economic context that provide the perceptual filter through which they gain environmental information. A key issue here is that these human-agents are limited both in their ability to gather accurate information, and their ability to carry through plans to achieve their goals. They are thus unlikely to be able to act in an optimal fashion. Finally agents can include larger bodies such as households or social institutions each with their own sets of rules for behaviour, and perhaps made up from sets of interacting smaller scale agents.


  • Richards, K, Bithell, M, Dove, M and Hodge, R (2004) Discrete element modelling: methods and applications in the environmental sciences. Phil. Trans. Roy. Soc. London, Ser A 362, 1-20
  • Richards, K, Bithell, M and Bravo, M (2004) Space, time and science: towards a geographical philosophy. In Common Heritage, Shared Future: Perspectives on the Unity of Geography; eds Matthews, JA and Herbert, D T, London, Routledge