Stochastic modeling of gravity data for 2D basement reliefs without regularization coefficient

In this paper, a modified version of strength Pareto evolutionary algorithm SPEA (II) is used as a multi-objective optimization method in gravity data modelling. In this method, a two-dimensional gravity inversion problem is solved by iteratively random creation of forward models. It is shown that it can be used as a fast and effective inversion tool in the depth modelling of two-dimensional layer problems with applications in depth-to-basements, geometry of bedrocks and sedimentary basins modelling cases. Owing to the direct use of the regularization term as a separate objective function, smooth models have a high chance of being selected as final solutions, which makes the results more acceptable and easier to interpret. The most important advantages of this method are that it works independently of the regularization coefficient; thus, there is no need to run the algorithm so many times to find a proper regularization parameter. Furthermore, there is no need to directly deal with inverse formulations, and last but not least, by using a multi-objective algorithm as a global optimization method, convergence to a stable solution does not depend on the initial model, the way classical inversion methods do. For testing the algorithm, a synthetic model is used for layer boundary modelling and to assess the stability of this algorithm, white Gaussian noise is added to the synthetic model. To evaluate the validity of this method, real data from the Recôncavo basin in Brazil is considered for processing and inversion, and the results are compared to the ones from previous studies. All computations have been done in the GNU Octave 5.1.0 environment.