1-Dimension magnetotelluric data inversion using MOEA/D algorithm

Abstract

Magnetotelluric (MT) data is used to derive resistivity imaging of subsurface. The subsurface resistivity is obtained by inversion of MT data. Generally, MT data contains two parts, namely: apparent resistivity and phase or real and imaginary parts. Inversion of MT data for reconstructing resistivity value of each layer is to minimize single objective (combination two parameters MT data) which used global or local optimization method. Nerveless, single objective optimization method has several disadvantages, such as; (1) weight value to combine two parameters of MT data is needed, where this weigh value depend on the amplitude of both MT data; (2) there is no validation of the inversion results. In this research, Inversion MT data to estimate 1D resistivity of subsurface uses multi-objective evolutionary algorithm based on decomposition (MOEA/D)to minimize root mean square error (RMSE) of calculated and observed data for apparent resistivity and phase data simultaneously. The algorithm has applied to synthetic and field data. This result shows that MOEA/D algorithm is robust and accurate to determine subsurface resistivity and lithology.