Influence of Frequency and Training Models on Physics-Guided Velocity Model Building

Abstract

Summary Velocity model building (VMB) is subsequently used to provide subsurface velocity model for workflows such as seismic imaging and interpretation. As two widely used velocity model building techniques, ray-based tomographic approaches are not very effective in complex geological settings; and Full waveform inversion (FWI) approaches are computationally extensive and sensitive to initial model. The physics-guided deep learning based velocity model building, that involves deterministic, physics-based modelling and data-driven deep learning components, is designed to capture the subsurface salt body shapes and locations, with a small amount of training models. In this work, we further discuss the influence of dominant frequency and training models on the velocity prediction by using H-PGNN method. Our results show that, the higher the dominant frequency, the more accurate the prediction accuracy of the salt body shapes and background information. For more complicated velocity models and real datasets, simple synthetic training models are not capable of capturing the salt body shapes, nor the background information. A more practical synthetic training set with much more smoothed background layered structures is more suitable to predict complicated models.