Scholarlink

<p>Accurate mapping of hydrothermal alteration zones using remote sensing data remains a challenge due to spectral complexity and overlapping mineral signatures. This study proposes a hybrid approach that integrates Principal Component Analysis (PCA) with decision tree classification to enhance alteration zone detection using ASTER imagery. The method was applied to the Darrehzar porphyry copper deposit in Iran, using statistical data from the visible–near infrared (VNIR) and shortwave infrared (SWIR) bands. PCA was used to extract the first principal component (PC1) from VNIR + SWIR (Dataset 1) and SWIR-only (Dataset 2), and statistical thresholds based on the mean (µ) and standard deviation (σ) were applied to construct four decision tree models. The resulting classifications accurately delineated key alteration types—propylitic, phyllic, and argillic—as well as sub-zones and transitional areas. Validation using confusion matrix analysis and field data showed that the highest classification accuracy (91.46%) and kappa coefficient (0.80) were achieved using Dataset 1.1 (15&#xa0;m resolution, VNIR + SWIR). These results demonstrate the method’s effectiveness and potential for interpretable, statistically driven mineral 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

Evaluation of decision tree on ASTER statistical data for segregating alteration zones in Darrehzar deposit, Iran

  • 期刊论文

  • 发表日期：

    2025年8月4日

查看全文
  • Aliyeh Seifi,
  • Saeed Yousefi,
  • Ahmad Aryafar,
  • Mahdieh Hosseinjanizadeh,
  • Arash Salajegheh,
  • Hadi Farhadian

摘要

Accurate mapping of hydrothermal alteration zones using remote sensing data remains a challenge due to spectral complexity and overlapping mineral signatures. This study proposes a hybrid approach that integrates Principal Component Analysis (PCA) with decision tree classification to enhance alteration zone detection using ASTER imagery. The method was applied to the Darrehzar porphyry copper deposit in Iran, using statistical data from the visible–near infrared (VNIR) and shortwave infrared (SWIR) bands. PCA was used to extract the first principal component (PC1) from VNIR + SWIR (Dataset 1) and SWIR-only (Dataset 2), and statistical thresholds based on the mean (µ) and standard deviation (σ) were applied to construct four decision tree models. The resulting classifications accurately delineated key alteration types—propylitic, phyllic, and argillic—as well as sub-zones and transitional areas. Validation using confusion matrix analysis and field data showed that the highest classification accuracy (91.46%) and kappa coefficient (0.80) were achieved using Dataset 1.1 (15 m resolution, VNIR + SWIR). These results demonstrate the method’s effectiveness and potential for interpretable, statistically driven mineral exploration.