Investigating the Subsurface of the Iranian Plateau using Adjoint Tomography: A 3D Seismic Velocity Model

A 3D seismic velocity model of the Iranian plateau was constructed using adjoint tomography. The initial model used in this study is taken from a previous adjoint noise tomography investigation in the region, and was refined through the inversion of waveforms recorded at seismic broadband stations from 320 local/regional earthquakes for which a Centroid-Moment-Tensor (CMT) has been computed. The synthetic waveforms were calculated using the spectral-element method (SEM) through a mesh of 12 million grid points representing the Iran region. The model parameters were iteratively updated by Newton's method, using misfit and Hessian kernels to minimize the difference between observed and synthetic waveforms. The forward and adjoint simulations were computed on the HRL GPU-cluster in Frankfurt, requiring a total of 6720 simulations and approximately 13,000 node hours to achieve the final model after 11 iterations. Comparison of the synthetic waveforms produced by the proposed model to the observed waveforms for a subset of selected earthquakes indicates improved fit in the period range of 10-50 seconds confirms the model's ability to accurately predict actual waveforms. The identification of previously undetected anomaly features beneath the Iranian plateau was also observed, which could be attributed to geological features with sufficient data coverage and resolution.