Joint Inversion of Seismic Data by Means of Multi-objective Evolutionary Algorithms

Any sort of data is able to cast light only onto a specific aspect of the investigated problem. The implementation of a joint inversion scheme is meant to proficiently integrate the information that can be extracted from one dataset with those coming from another one. If the two objectives depend upon the same variables we can obtain a better-focused solution, while if the two objectives pertain (even just partially) to different variables thus their joint use can lead to new considerations characterized by a higher so-to-speak added value. The proposed computational scheme allows the joint inversion of non-commensurable datasets by means of analysis of the Pareto front performed in the framework of a Multi-Objective Evolutionary Algorithm (MOEA). The adopted approach also allows the validation of the provisional interpretation. In fact, Pareto front symmetry proves to be a valuable tool to verify the coherency of the adopted interpretation as an incorrect number of layers, reflector/refractor attribution or assumed Poisson values determine non-symmetric Pareto front as well as a wider model distribution in the objective space.