Pareto Joint Inversion of 2D magnetometric and gravity data- synthetic study

Abstract

Pareto joint inversion for two or more data sets is an attractive and promising tool which eliminates target functions weighing and scaling, providing a set of acceptable solutions composing a Pareto front. In former author’s study MARIA (Modular Approach Robust Inversion Algorithm) was created as a flexible software based on global optimization engine (PSO) to obtain model parameters in process of Pareto joint inversion of two geophysical data sets. 2D magnetotelluric and gravity data were used for preliminary tests, but the software is ready to handle data from more than two geophysical methods. In this contribution, the authors’ magnetometric forward solver was implemented and integrated with MARIA. The gravity and magnetometry forward solver was verified on synthetic models. The tests were performed for different models of a dyke and showed, that even when the starting model is a homogeneous area without anomaly, it is possible to recover the shape of a small detail of the real model. Results showed that the group analysis of models on the Pareto front gives more information than the single best model. The final stage of interpretation is the raster map of Pareto front solutions analysis.

An application of the NSGA-II algorithm in Pareto joint inversion of 2D magnetic and gravity data

“…Later, a magnetometric (MG) forward solver was implemented that replaced the MT module . A case study on synthetic data for MARIA 2.0 was performed (Danek et al 2019). The next step was the integration of the GV and MG forward solvers with the R environment (R is a free software environment as well as a programming language which can be used for statistical computing and graphics).…”

“…Each cell is considered a rectangle, to which some parameters of magnetic susceptibility and density are assigned (Danek et al 2019).…”

Pareto-Based Multiobjective Particle Swarm Optimization: Examples in Geophysical Modeling

“…[5] Despite this interest, very few researchers have studied MOPSO for joint modeling of geophysical data such as electromagnetic and gravity. [6,7] In fact, simultaneous optimization of multiobjective functions is also favored to increase uniqueness of model parameters in joint modeling of geophysical data that are generally sensitive to different physical phenomena. Multiobjective functions can be transferred into single-objective by combination of objective functions by using weighted-sum approach.…”