Crosswell seismic tomographic inversion using NSGA II

TL;DR

This paper applies NSGA II, a multi-objective optimization algorithm, to seismic data inversion, effectively balancing misfit and stability constraints to produce reliable subsurface models.

Contribution

It introduces a multi-objective optimization approach using NSGA II for seismic inversion, incorporating Tikhonov regularization to improve stability and convergence.

Findings

Successful estimation of subsurface slowness from synthetic data

Effective trade-off between misfit and regularization achieved

Convergence on Pareto frontier demonstrated

Abstract

In the present paper, the solution of the seismic data inversion problem through multi-objective optimization with NSGA II is addressed. The seismic inversion consists of estimating the slowness of rocks in the subsurface from the travel times of the waves through the transmitters and receivers of the seismic waves. The inversion process uses the multi-objective optimization technique, in which it is necessary to estimate the trade-off between the misfit functional and the stabilizing constraint. The constraint used was the classic first-order Tikhonov. A synthetic model of inclined layers was chosen. The results obtained were acceptable. Based on the graph of the cumulative errors related to the solution, it was estimated correctly, that is, the optimization method used was effective to obtain acceptable results, in addition to the constraint solving the instability problem of the inverse problem treated. Another important result was the achievement of a convergence curve on the Pareto frontier, an important result for future research.