Scholarlink

<p>In ultra-deep oil and gas exploration, seismic waves generated at the surface suffer significant attenuation, resulting in substantial discrepancies between surface seismic data and vertical seismic profiling (VSP). Traditional calibration methods often yield waveform mismatches and calibration errors due to fixed frequency ranges. This paper presents a novel approach utilizing a Bidirectional Gated Recurrent Unit (BiGRU) and an attention mechanism within a neural network to dynamically match and tie surface and borehole seismic profiles. The method adapts to the time-varying characteristics of the VSP corridor stack and surface seismic profile, automatically generating optimal spectral distributions in real-time. By effectively extracting target frequency components and suppressing noise, this approach enhances flexibility and accuracy in borehole-surface seismic profile correlation. Applied successfully to the calibration of both synthetic data as well as field zero-offset VSP in Well ZH 2 in the Tarim Basin, the method significantly reduces calibration errors and improves waveform consistency, demonstrating its potential for broader applications in ultra-deep exploration.</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

Adaptive Surface-Borehole Seismic Profile Matching and Tying Using BiGRU: A Case Study of Ultra-Deep Zero-Offset VSP in the Tarim Basin

  • 期刊论文

  • 发表日期：

    2026年1月31日

查看全文
  • Teng-yu Wang,
  • Da-jun Li,
  • Zhen Zhang,
  • Duo-ming Zheng,
  • Bing Fang,
  • Pei Liu,
  • Jing-jing Zong

摘要

In ultra-deep oil and gas exploration, seismic waves generated at the surface suffer significant attenuation, resulting in substantial discrepancies between surface seismic data and vertical seismic profiling (VSP). Traditional calibration methods often yield waveform mismatches and calibration errors due to fixed frequency ranges. This paper presents a novel approach utilizing a Bidirectional Gated Recurrent Unit (BiGRU) and an attention mechanism within a neural network to dynamically match and tie surface and borehole seismic profiles. The method adapts to the time-varying characteristics of the VSP corridor stack and surface seismic profile, automatically generating optimal spectral distributions in real-time. By effectively extracting target frequency components and suppressing noise, this approach enhances flexibility and accuracy in borehole-surface seismic profile correlation. Applied successfully to the calibration of both synthetic data as well as field zero-offset VSP in Well ZH 2 in the Tarim Basin, the method significantly reduces calibration errors and improves waveform consistency, demonstrating its potential for broader applications in ultra-deep exploration.