An efficient deep learning method for VSP wavefield separation: A DAS-VSP case

Abstract

The quality of wavefield separation on vertical seismic profiling (VSP) data directly affects subsequent imaging and inversion processes. However, the traditional methods have various defects in separating upgoing and downgoing waves. The application of deep learning brings another train of thought to solve related problems. The traditional methods, such as f-k filtering and Radon transform, produce results with spatial aliasing and inaccurate amplitudes. The median filtering method relies on accurate first-break picking and waveform consistency. Moreover, the results of traditional methods are subject to manual intervention. To overcome these problems, a deep-learning method is developed for the automatic wavefield separation of VSP data. First, the traditional Radon transform method is used to produce training data sets. Then, to ensure amplitude preservation and suppress spatial aliasing, new input data is formed by recombining the upgoing and downgoing waves extracted by Radon transform. The developed deep-learning network is highly efficient, and its accuracy exceeds that of the traditional methods only after one epoch training. A practical workflow of wavefield separation based on deep learning is established, and it is applied to synthetic data and field distributed acoustic sensing VSP data. The results indicate that our method is superior in terms of amplitude preservation and spatial aliasing suppression. The time consumption of our method is very acceptable and can be further minimized by training the network using downsampling data.