Abstract

In oil and gas exploration industry, precise subsurface imaging is essential for visualizing underground structures, which ultimately reduce the uncertainties in the selection of drilling sites for exploration wells. However, the task of achieving high-resolution imaging becomes challenging in subsurface regions characterized by complex geological structures and lateral velocity variations. This type of complex structures contributes to a low signal-to-noise ratio (S/N ratio) in seismic data. Complicated geological conditions pose challenges for conventional imaging methods due to the physics on which they are based. Full waveform inversion (FWI) emerges as a powerful technique capable of generating high-resolution images, with a theoretical resolution of half the dominant wavelength, even in the presence of this type of geological feature. Our team has successfully developed a time-domain acoustic full waveform inversion (FWI) code specifically designed to handle the inversion of data for complex subsurface structures. The primary challenge associated with the FWI algorithm is its computational cost apart from non-linearity. In our software package, our central focus has been on optimizing the algorithm through parallelization, achieved by automatically allocating processors and memory to individual shots based on the availability of processors. This dynamic allocation makes the algorithm adaptable to a wide range of computing machines, enhancing its versatility.