When an emergency occurs, hazard evolution simulators are a very helpful tool for the teams in charge of making decisions. These simulators need certain input data, which defines the characteristics of the environment where the emergency is taking place. This kind of data usually constitutes a big set of parameters, which have been previously recorded from observations, usually coming from remote sensors, pictures, etc. However, this data is frequently subject to a high degree of uncertainty, as well as the results produced by the corresponding simulators. Hence, it is also necessary to pay attention to the simulations' quality and reliability. In this work we expose the way we deal with such uncertainty. Our research group has previously developed a two-stage prediction methodology that introduces an adjustment stage in order to deal with the uncertainty on the simulator input parameters. This method significantly improves predictions' quality, however, in order to be useful, a good characterization of the adjustment techniques has to be carried out so that we are able to choose the best configuration of them, given certain restrictions regarding resources availability and time deadlines. In this work, we focus on forest fires spread prediction as a real study case, for which Genetic Algorithms (GA) have been demonstrated to be a suitable adjustment strategy. We describe the methodology used to characterize the GA and we also validate it when assessing in advance the quality of the fire spread prediction. © 2012 Published by Elsevier Ltd.
|Journal||Procedia Computer Science|
|Publication status||Published - 1 Jan 2012|