基于机器学习的深海多金属结核成因分类研究
DOI:
作者:
作者单位:

中山大学地球科学与工程学院

作者简介:

通讯作者:

中图分类号:

P744

基金项目:

本研究得到广东省引进:大数据-数学地球科学创新研发团队和极端地质事件团队(批准号:2021ZT09H399)资助


Genetic classification of deep-sea polymetallic nodules based on machine learning
Author:
Affiliation:

School of Earth Science and Engineering, Sun Yat-sen University

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    铁锰结核广泛分布于深海平原,储量巨大,具有商业开采潜力。本研究利用1128个铁锰结核样本的地球化学数据和8种地质与海洋要素,采用随机森林机器学习方法,探讨结核成因分类。通过Mn、Fe、Cu、Co、Ni、Mn/Fe和Fe/Co等元素比值作为训练数据,并基于海底沉积速率、海水底部溶氧量、海水表面生物初级生产力等地质-海洋特征建立预测模型,将结核划分为水成型、成岩型和混合型,结果显示,模型对水成型和成岩型结核的分类精度分别为91%和66%,对混合型的分类精度较低,仅为23%。应用该模型对全球4119个铁锰结核进行成因分类,结果表明,水成型结核占71.8%,混合型占21.8%,成岩型占6.2%。水成型结核广泛分布于各大洋,而成岩型和混合型则集中在大洋中纬度地区,如东太平洋的克拉里昂-克里帕顿断裂带和东南太平洋的秘鲁海盆等。这些地区的沉积物发育、海底生物量和含氧量显著影响结核分布。尽管基于地球化学数据的分类方法更可靠,研究表明,利用地质和海洋要素及机器学习方法也可有效分类。

    Abstract:

    Manganese nodules are widely distributed across deep-sea plains and have significant commercial mining potential due to their vast reserves. This study investigates the genesis classification of nodules using geochemical data from 1,128 manganese nodule samples and eight geological and oceanographic factors, employing the random forest machine learning method. Ratios of elements such as Mn, Fe, Cu, Co, Ni, Mn/Fe, and Fe/Co were used as training data. A predictive model was established based on geological-oceanographic characteristics including sedimentation rate, bottom water dissolved oxygen levels, and surface water primary productivity. The nodules were classified into hydrogenetic, diagenetic, and mixed types. The model demonstrated classification accuracies of 91% for hydrogenetic nodules, 66% for diagenetic nodules, and only 23% for mixed nodules. Applying this model to classify 4,119 global manganese nodules revealed that 71.8% were hydrogenetic, 21.8% were mixed, and 6.2% were diagenetic. Hydrogenetic nodules are widely distributed across all oceans, while diagenetic and mixed types are concentrated in mid-latitude regions of the oceans, such as the Clarion-Clipperton Fracture Zone in the Eastern Pacific and the Peru Basin in the Southeastern Pacific. The distribution of these nodules is significantly influenced by sediment development, seafloor biomass, and oxygen levels. Although classification based on geochemical data is more reliable, the study shows that using geological and oceanographic factors along with machine learning is also an effective method for classification.

    参考文献
    相似文献
    引证文献
引用本文

尹浩文,成秋明. 基于机器学习的深海多金属结核成因分类研究[J]. 科学技术与工程, , ():

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2024-05-27
  • 最后修改日期:2024-05-27
  • 录用日期:2024-05-28
  • 在线发布日期:
  • 出版日期:
×
亟待确认版面费归属稿件,敬请作者关注