Parallel evolutionary modelling of geological processes

This paper illustrates a parallel computing methodology for modelling and simulating geological processes by means of Cellular Automata and Parallel Genetic Algorithms. Two models, concerning lava and debris flows, have been implemented using the Cellular Automata development environment CAMELot, and calibrated by means of Genetic Algorithms through the PGAPack library. Experiments have been carried out on two different distributed memory machines, namely a Beowulf cluster and a HP Alphaserver SC supercomputer. Results have demonstrated the goodness of both considered geological models and of the genetic algorithm employed for their calibration. High computational performances have been achieved. In particular, results obtained for the calibration phase demonstrated that even low-cost parallel machines can be fruitfully employed for the construction of reliable simulation models for geological processes. (C) 2007 Elsevier B.V. All rights reserved.