Seismic Signal Denoising Based on Surelet Transform for Energy Exploration

Abstract

Seismic signals are critical for subsurface energy exploration like oil, coal, and natural gas. Processing these signals while minimizing environmental impacts is crucial but lacking in several appropriate multi-scale geometric analysis (MGA) techniques. This study proposes using the Surelet transform, based on Stein’s unbiased risk estimate (SURE), for seismic denoising. The method combines SURE to find optimal thresholds and linear expansion for coefficient estimation. Experiments on two-dimensional (2D) and three-dimensional (3D) synthetic seismic data showed Surelet achieved higher peak signal-to-noise ratios (PSNR) and faster processing compared to wavelet, curvelet, and wave atom. For example, with 20% noise, Surelet improved PSNR by 6.11% and reduced time by 78.4% versus wave atom. The feasibility of the proposed technique for efficient seismic denoising was demonstrated, highlighting implications for enabling cleaner signals in energy exploration.

Keywords

• energy exploration
• multi-scale geometric analysis
• surelet transform
• seismic signal denoising
• peak signal to noise ratio