Forest fire are one of the most critical environmental risks in all the Mediterranean countries. The fight against these emergencies requires useful tools to predict the propagation and behaviour of forest fire in order to make the best decisions. This means it is necessary to know the propagation and behaviour of the forest fire in advance to act in the best possible way. Common to realistic models of time dynamic systems is their complexity, very often prohibiting numerical or analytical evaluation. Consequently, for these cases, simulation remains the only tractable evaluation methodology, making up an attractive alternative to conventional experimental tests. In this sense, a computer can be viewed in context as an “electronic wind tunnel.” Simulation requirements for these complex systems mean more and more computing power and storage capacity. As the volume of input (sensor) and simulation output data (visualisation) increases, large archival storage systems with rapid data retrieval play an increasingly important role. To accomplish the above objective “to predict the propagation and behaviour of forest fire” it is necessary to apply numerical methods and algorithms that solve the proposed models. This work implies direct cooperation between forest fire researchers and computer scientists. We can infer an important principle from this situation, namely, the necessity of yoking together the computer science and application science research communities. This collaboration defines a fundamental guiding principle: the combination of application pull and technology push (Karin and Graham, 1998).