Fast automated airborne electromagnetic data interpretation using parallelized particle swarm optimization

A parallelized implementation of the particle swarm optimization algorithm is developed. We use the optimization procedure to speed up a previously published algorithm for airborne electromagnetic data interpretation. This algorithm is the only parametrized automated procedure for extracting the three-dimensionally varying geometrical parameters of conductors embedded in a resistive environment, such as igneous and metamorphic terranes. When compared to the original algorithm, the new optimization procedure is faster by two orders of magnitude (factor of 100). Synthetic model tests show that for the chosen system architecture and objective function, the particle swarm optimization approach depends very weakly on the rate of communication of the processors. Optimal wall-clock times are obtained using three processors. The increased performance means that the algorithm can now easily be used for fast routine interpretation of airborne electromagnetic surveys consisting of several anomalies, as is displayed by a test on MEGATEM field data collected at the Chibougamau site, Québec.