Arun Singh

and 1 more

This study introduces a novel method for performing 3D inversion of magnetotelluric (MT) data. The proposed method, referred to as the radiation boundary scheme, employs a two-step simulation strategy for the computation of both forward and adjoint responses. One of the key advantages of the scheme is its ability to handle arbitrarily shaped inversion domains, thereby optimizing the number of unknown model parameters by discarding model parameters that are not constrained by the data. Consequently, it significantly improves accuracy and computational speed as compared to traditional inversion algorithms. The effectiveness of the developed algorithm is demonstrated through a comprehensive analysis of 3D inversion using synthetic and two continental-scale (SAMTEX and USArray) MT data. The method’s efficiency facilitates a detailed analysis of large-scale MT data inversion. Through numerical experiments, it is observed that using a coarse mesh for inversion, the resolution is compromised in the shallower part, resulting in inferior imaging and, consequently, affecting the estimation of resistivity value in the deeper subsurface. The detailed numerical experiments indicate that performing a high-resolution inversion on a small portion of the survey data utilizing a coarsely inverted model may run into a local minimum. Hence, caution should be exercised in employing such an approach. Instead, the investigations suggest simultaneously executing a high-resolution inversion on the entire data set. The forward/adjoint problem can be solved with a two-order higher tolerance as compared to the conventional finite-difference-based inversion algorithm. Therefore, the proposed algorithm holds significant value for the MT inversion of large data sets.