Healing Seismic Data with Phase Corrections for Processing of Single-Sensor Data in the Desert Environment

Abstract

Abstract Acquiring data with single sensors or small arrays in a desert environment may lead to challenging data quality for subsequent processing. We present a new approach to effectively "heal" such data and allow efficient processing and imaging without requiring any additional acquisition. A novel method combines the power of seismic beamforming and time-frequency masking originating from speech processing. First, we create an enhanced version of the data with beamforming or local stacking. Beamforming effectively suppresses scattered noise and finds weak reflection signals, albeit sacrificing some higher frequencies. Next, we employ a seismic time-frequency masking procedure to fix the original data while using beamformed data as a guide. Time-frequency masking effectively fixes corrupt and broken phase of the original data. After such data-driven healing, prestack data can be effectively processed and imaged, while maintaining the higher frequencies lost during beamforming.