Scholarlink

<p>The volume and complexity of volcanic seismic data present an ideal testbed for developing improved analytical tools to enhance interpretation accuracy, consistency, and decision-making in volcano monitoring. This review examines Artificial Intelligence (AI) techniques applied to the detection, segmentation, and classification of volcanic waveforms. Publications from January 2019 to July 2025 cover supervised and unsupervised methods and are analyzed by signal type, input representations, model architectures, and validation strategies. This review highlights the contributions, limitations, and practical implementations of these methods across multiple volcanic settings. It outlines promising approaches for improving signal processing accuracy and identifies emerging research directions in multivariable integration and data harmonization. Core challenges, such as model explainability, limited data availability, and the transferability of models across different contexts, are also discussed. Additionally, the need for a deeper understanding of how physical factors (e.g., attenuation and source effects) influence classification robustness is emphasized. By consolidating best practices and identifying critical gaps, this study offers a roadmap for enhanced resilient and dependable early-warning systems in operational volcano observatories and other rapid-response settings.</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

Artificial intelligence techniques for volcanic seismic signal analysis: a systematic review

  • 期刊论文

  • 发表日期：

    2026年2月19日

查看全文
  • Sergio Morales,
  • Andrés I. Avila,
  • Mailyn Moreno Espino

摘要

The volume and complexity of volcanic seismic data present an ideal testbed for developing improved analytical tools to enhance interpretation accuracy, consistency, and decision-making in volcano monitoring. This review examines Artificial Intelligence (AI) techniques applied to the detection, segmentation, and classification of volcanic waveforms. Publications from January 2019 to July 2025 cover supervised and unsupervised methods and are analyzed by signal type, input representations, model architectures, and validation strategies. This review highlights the contributions, limitations, and practical implementations of these methods across multiple volcanic settings. It outlines promising approaches for improving signal processing accuracy and identifies emerging research directions in multivariable integration and data harmonization. Core challenges, such as model explainability, limited data availability, and the transferability of models across different contexts, are also discussed. Additionally, the need for a deeper understanding of how physical factors (e.g., attenuation and source effects) influence classification robustness is emphasized. By consolidating best practices and identifying critical gaps, this study offers a roadmap for enhanced resilient and dependable early-warning systems in operational volcano observatories and other rapid-response settings.