Scholarlink

<p>This study evaluates soil-emitted radon (<sup>222</sup>Rn) as a short-term earthquake precursor using neural network and regression on decade-long data from the Garhwal Himalaya. The neural network model yielded higher predictive accuracy (R ≈ 0.91–0.92) than the multi-layer regression model (R ≈ 0.85), reflecting its strength in capturing the nonlinear behaviour. Statistically defined radon anomalies—exceeding ± 2σ from the modelled baseline—often preceded moderate earthquakes (Mw ≥ 3.0) by several days. These findings suggest that radon monitoring, integrated with data-driven models, can provide valuable short-term precursory information. The work highlights potential for reliable, data-driven earthquake forecasting in seismically active regions.</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

Radon as an earthquake precursor: a machine learning case study from the Garhwal Himalaya, NW India

  • 期刊论文

  • 发表日期：

    2026年2月16日

查看全文
  • Vandana,
  • Priyadarshi Chinmoy Kumar,
  • Naresh Kumar,
  • Sanjay Kumar Verma

摘要

This study evaluates soil-emitted radon (222Rn) as a short-term earthquake precursor using neural network and regression on decade-long data from the Garhwal Himalaya. The neural network model yielded higher predictive accuracy (R ≈ 0.91–0.92) than the multi-layer regression model (R ≈ 0.85), reflecting its strength in capturing the nonlinear behaviour. Statistically defined radon anomalies—exceeding ± 2σ from the modelled baseline—often preceded moderate earthquakes (Mw ≥ 3.0) by several days. These findings suggest that radon monitoring, integrated with data-driven models, can provide valuable short-term precursory information. The work highlights potential for reliable, data-driven earthquake forecasting in seismically active regions.