Scholarlink

<p>Seismic deconvolution is a critical process in geophysical signal processing, aimed at enhancing the resolution of subsurface imaging by recovering sparse reflectivity sequences from convolved seismic data. This study introduces an advanced deconvolution approach utilizing the Proximal Dykstra (PD) algorithm, a convex optimization technique that leverages proximal operators to address the ill-posed nature of seismic inverse problems. The proposed method integrates a quadratic data fidelity term with a sparsity-promoting ℓ₁-norm regularization, enabling robust reconstruction of reflectivity series in the presence of additive noise. The PD algorithm iteratively alternates between projections that enforce data consistency and sparsity, achieving superior resolution and noise suppression compared to the benchmark Iterative Shrinkage-Thresholding Algorithm (ISTA). Performance evaluations on synthetic seismic datasets, with signal-to-noise ratios ranging from 1.92 dB to noise-free conditions, demonstrate the PD algorithm’s ability to sharpen reflectors, enhance lateral continuity, and broaden spectral bandwidth. Application to real seismic data from southern Iran further validates its effectiveness, revealing improved delineation of geological features and fault structures. Despite its computational complexity and sensitivity to parameter tuning, the PD algorithm offers a flexible and scalable framework for seismic deconvolution, making it a valuable tool for high-resolution subsurface characterization in challenging geophysical environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An improved deconvolution method using proximal dykstra algorithm for seismic signal

  • 期刊论文

  • 发表日期：

    2026年3月5日

查看全文
  • Alireza Goudarzi,
  • Seyed Hadi Dehghan-Manshadi

摘要

Seismic deconvolution is a critical process in geophysical signal processing, aimed at enhancing the resolution of subsurface imaging by recovering sparse reflectivity sequences from convolved seismic data. This study introduces an advanced deconvolution approach utilizing the Proximal Dykstra (PD) algorithm, a convex optimization technique that leverages proximal operators to address the ill-posed nature of seismic inverse problems. The proposed method integrates a quadratic data fidelity term with a sparsity-promoting ℓ₁-norm regularization, enabling robust reconstruction of reflectivity series in the presence of additive noise. The PD algorithm iteratively alternates between projections that enforce data consistency and sparsity, achieving superior resolution and noise suppression compared to the benchmark Iterative Shrinkage-Thresholding Algorithm (ISTA). Performance evaluations on synthetic seismic datasets, with signal-to-noise ratios ranging from 1.92 dB to noise-free conditions, demonstrate the PD algorithm’s ability to sharpen reflectors, enhance lateral continuity, and broaden spectral bandwidth. Application to real seismic data from southern Iran further validates its effectiveness, revealing improved delineation of geological features and fault structures. Despite its computational complexity and sensitivity to parameter tuning, the PD algorithm offers a flexible and scalable framework for seismic deconvolution, making it a valuable tool for high-resolution subsurface characterization in challenging geophysical environments.