CamDeus: Cellular Automata Models to Design Environmental and Urban Systems
Creation date: 2002
The CamDeus program is a software inspired by a coupling: raster GIS technology and cellular automata. It is dedicated to the simulation of urban dynamics and can be used to test scenarios of territorial development, according to the past states of the territories or to political planning. It is a stand-alone software with user-friendly interfaces, conceived to be handled by persons who do not necessarily totally master the mathematical principles of the models used in the program.
1. Camdeus’ cellular space. The use of Camdeus necessitates geographical space to be cellular, i.e. composed by regular cells. Each cell is associated to one (and only one) land-use value. Such a configuration usually results from grid mapping (tessellation) and allows to count and classify the cells before applying mathematical routines on each one. These routine lead to simulations. The main originality of the program consists then in the succession of three modelling steps, using three specific and different models, allowing simultaneously to know the number, the [hyp209location] and the category of the cells that will change in the tested scenarios.
2. Step 1: quantification. The first step of the CamDeus modelling consists in answering the question «how many? » to determine the number of cells that will move from one land-use category to another one in the future. The model relies on the Markov chains’ principle, and automatically build transition matrices containing the probabilities of evolution from one category to another one. Primo, such transition matrices can be obtained by observing the past states and the evolution of the studied area. Secundo, they can be mixed with other matrices, built by the user to correspond to a scenario of possible or tested territorial development, relying on political wills or on global trends.
3. Step 2: location. When the cells are quantified, they must be located, so as to identify the areas where the changes (determined at the first step) will appear. The second step is then dedicated to locations and tries to answer the question « where?». It uses a potential model (spatial interaction model) based on two kinds of parameters: the mass of the cells and the distance between the cells. The mass could then be associated to a « coefficient of attractiveness » : the more a cell is near of cells with important masses, the more its potential to change will be great. All the possibilities are tested by the program so as to determine the areas with the strongest potential values, and to locate the cells quantified at the first step.
4. Step 3: differentiation. At the end of the second step, CamDeus determined the number and the location of the cells that will change in the future. But, it still does not precisely know the land-use category of these cells (buildings, facilities, forests, etc.). The third step consists then in answering the question « what? » to find these categories. Cellular automata are used. They associate transition rules to the cells so that each one moves to the corresponding category, according to what happens in its neighbourhood (principle of the spatial autocorrelation).
Further information about the CamDeus model
Antoni J.P., 2004, Modélisation de l’étalement urbain. Aspects conceptuels et gestionnaires. Application à Belfort, Thèse de doctorat, Faculté de géographie de Strasbourg, 524 pages. Click here for access.
Antoni J.P., 2003, Cam.Deus: a forecasting tool to anticipate urban development, 13th European colloquium on quantitative and theoretical geography, 5-9 september 2003, Lucca, Italy.
Antoni J.P., 2002, Urban sprawl modelling using GIS and grid mapping, Cybergeo, n°207, 9 pages. Consult the paper on the website of Cybergeo
Key word(s) linked to this articleDomain of application » urban
Domain of application » land use change
Type of organisation supporting the website » research centre
Types of modelling » operational
Temporal characteristics » discrete
Optimisation/Simulation » simulation
This article last updated Saturday 29 April 2006. by Jean-Philippe Antoni