A black hole particle swarm optimization method for the source parameters inversion: application to the 2015 Calbuco eruption, Chile

Abstract

The traditional genetic algorithm and simulated annealing methods have been widely used in geophysical modeling. However, these nonlinear inversion methods require a lot of calculations, many control parameters and are unstable. In this paper, a particle swarm optimization algorithm combined with black hole strategy (BH-PSO) is proposed to solve these problems. The comprehensive experiments show that the BH-PSO method consumes less time than the simulated annealing (SA) method and has a higher accuracy than the genetic algorithm (GA). It is more applicable to the inversion of parameters of volcanic magma chamber, and easier to be generalized to other kinematic source parameters inversion. Based on BH-PSO method, Sentinel-1 data, composite dislocation model (CDM), Yang model and Mogi model, the magma chamber parameters of Calbuco eruption in 2015 were retrieved. The results show that the RMSE of CDM model is 1.1 cm, which can better fit the surface deformation than the Mogi model and Yang model. The final results show that the magma chamber is located about 0.8 km northeast of the crater, about 9 km below the surface, and the total volume of the erupted volcanic material obtained with the CDM Model is of 0.209 km³, without considering dense rock equivalent.