A multi-world agent-based model working at several spatial and temporal scales for simulating complex geographic systems

Interest in the modelling and simulation of complex systems with processes occurring at several spatial and temporal scales is increasing, particularly in biological, historical and geographic studies. In this multi-scale modelling study, we propose a generic model to account for processes operating at several scales. In this approach, a 'world' corresponds to a complete and self-sufficient submodel with its own places, agents, spatial resolution and temporal scale. Represented worlds can be nested: a world (with new scales) may have a greater level of detail than the model at the next level up, making it possible to study phenomena with greater precision. This process can be reiterated, to create additional scales, with no formal limit. Worlds' simulations can be triggered simultaneously or in cascade. Within a world, agents can choose destinations in other worlds, to which they can travel using routes and inter-world 'gates'. Once they arrive in a destination world, the agents 'fit' the new scale. An agent in a given world can also perceive and interact with other agents, regardless of the world to which they belong, provided they are encompassed by its perception disc. We present and discuss an application of this model to the issue of the spread of black rats by means of commercial transportation in Senegal (West Africa).