Scholarlink

<p>Like many inverse problem-solving techniques, seismic inversion aims to create a model that describes petrophysical properties to replicate observed seismic data. These inversion problems are inherently ill-posed, lacking a single definitive solution. Traditional seismic inversion methods often yield a solution with limited and narrowly defined uncertainty measures. This study proposes a new method based on niching genetic algorithms (NGAs) that generates a range of optimal solutions in a multimodal space, thus addressing the major challenge of uncertainty in final models. In the proposed method, each iteration of the optimization process involves calculating niches of petrophysical property models using a machine learning clustering technique that evaluates similarity (proximity or distance) between models. The evolutionary process of the NGA produces multiple solutions that are close to the real seismic data, thereby establishing a multi-solution seismic inversion methodology. This approach is demonstrated through a case study aimed at risk assessment during the early stages of 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

Uncertainty Assessment by Using Geostatistical Seismic Multimodal Inversion

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • Joao Lucas de Olivera Alves,
  • Joao Felipe Coimbra Leite Costa,
  • Amilcar Soares

摘要

Like many inverse problem-solving techniques, seismic inversion aims to create a model that describes petrophysical properties to replicate observed seismic data. These inversion problems are inherently ill-posed, lacking a single definitive solution. Traditional seismic inversion methods often yield a solution with limited and narrowly defined uncertainty measures. This study proposes a new method based on niching genetic algorithms (NGAs) that generates a range of optimal solutions in a multimodal space, thus addressing the major challenge of uncertainty in final models. In the proposed method, each iteration of the optimization process involves calculating niches of petrophysical property models using a machine learning clustering technique that evaluates similarity (proximity or distance) between models. The evolutionary process of the NGA produces multiple solutions that are close to the real seismic data, thereby establishing a multi-solution seismic inversion methodology. This approach is demonstrated through a case study aimed at risk assessment during the early stages of exploration.