Digital Holographic Microscopy and Machine Learning for Quantitative 3D Analysis and Automatic Classification of Volcanic Ash Particles

IF 3.8 2区 地球科学 Q2 METEOROLOGY & ATMOSPHERIC SCIENCES
A. C. Monaldi, J. I. Díaz, M. F. Martínez, N. Budini, W. A. Báez
{"title":"Digital Holographic Microscopy and Machine Learning for Quantitative 3D Analysis and Automatic Classification of Volcanic Ash Particles","authors":"A. C. Monaldi,&nbsp;J. I. Díaz,&nbsp;M. F. Martínez,&nbsp;N. Budini,&nbsp;W. A. Báez","doi":"10.1029/2024JD043283","DOIUrl":null,"url":null,"abstract":"<p>Determining the shapes, sizes and optical properties of volcanic ash presents a significant challenge in volcanology, the aviation industry and atmospheric models involving transport and dispersion of particles. Eruptive dynamics, including fragmentation mechanisms, magma viscosity and particle transport processes, among others, are encoded in the intricate shapes and sizes of these particles. Traditionally, the analysis of ash particles' morphology has relied on quantitative non-dimensional parameters, primarily derived from their 2D silhouette projected area, using conventional microscopy or particle analyzers. However, these fail to capture the 3D structure of their morphology. Additionally, atmospheric dispersion models often assume spherical particles with uniform refractive indices, introducing uncertainties in particle size estimations and dispersion calculations. In this study, we introduce a novel 3D characterization method for volcanic ash using digital holographic microscopy (DHM) combined with machine learning (ML). We implemented an off-axis interferometer to register holograms of volcanic ash samples. We show that segmented phase maps from the reconstructed holograms can be used to derive both 2D and 3D phase-based morphological parameters for individual ash particles or to estimate their refractive index. To illustrate the potential of this technique, we analyzed morphological differences between ashes acccording to their transport mechanism: fallout and flow. A ML algorithm based on support vector machine (SVM) was trained to classify particles into one of these two categories, achieving an average accuracy of 76%. These results show that the proposed approach serves as a valuable tool for monitoring volcanic eruptions providing insights on their characteristics and associated environmental impact.</p>","PeriodicalId":15986,"journal":{"name":"Journal of Geophysical Research: Atmospheres","volume":"130 13","pages":""},"PeriodicalIF":3.8000,"publicationDate":"2025-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Geophysical Research: Atmospheres","FirstCategoryId":"89","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1029/2024JD043283","RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"METEOROLOGY & ATMOSPHERIC SCIENCES","Score":null,"Total":0}
引用次数: 0

Abstract

Determining the shapes, sizes and optical properties of volcanic ash presents a significant challenge in volcanology, the aviation industry and atmospheric models involving transport and dispersion of particles. Eruptive dynamics, including fragmentation mechanisms, magma viscosity and particle transport processes, among others, are encoded in the intricate shapes and sizes of these particles. Traditionally, the analysis of ash particles' morphology has relied on quantitative non-dimensional parameters, primarily derived from their 2D silhouette projected area, using conventional microscopy or particle analyzers. However, these fail to capture the 3D structure of their morphology. Additionally, atmospheric dispersion models often assume spherical particles with uniform refractive indices, introducing uncertainties in particle size estimations and dispersion calculations. In this study, we introduce a novel 3D characterization method for volcanic ash using digital holographic microscopy (DHM) combined with machine learning (ML). We implemented an off-axis interferometer to register holograms of volcanic ash samples. We show that segmented phase maps from the reconstructed holograms can be used to derive both 2D and 3D phase-based morphological parameters for individual ash particles or to estimate their refractive index. To illustrate the potential of this technique, we analyzed morphological differences between ashes acccording to their transport mechanism: fallout and flow. A ML algorithm based on support vector machine (SVM) was trained to classify particles into one of these two categories, achieving an average accuracy of 76%. These results show that the proposed approach serves as a valuable tool for monitoring volcanic eruptions providing insights on their characteristics and associated environmental impact.

用于定量三维分析和火山灰颗粒自动分类的数字全息显微镜和机器学习
确定火山灰的形状、大小和光学性质对火山学、航空工业和涉及颗粒运输和分散的大气模型提出了重大挑战。喷发动力学,包括碎裂机制、岩浆粘度和颗粒传输过程等,都在这些颗粒复杂的形状和大小中进行编码。传统上,灰颗粒的形态分析依赖于定量的无量纲参数,主要来自它们的二维轮廓投影面积,使用传统显微镜或颗粒分析仪。然而,这些都无法捕捉到它们形态的3D结构。此外,大气色散模型通常假设具有均匀折射率的球形粒子,这在粒径估计和色散计算中引入了不确定性。在这项研究中,我们介绍了一种新的火山灰三维表征方法,该方法将数字全息显微镜(DHM)与机器学习(ML)相结合。我们实现了一个离轴干涉仪来记录火山灰样本的全息图。我们发现,从重建的全息图中分割的相位图可以用于推导单个灰颗粒的二维和三维相位形态学参数或估计它们的折射率。为了说明这项技术的潜力,我们根据灰烬的传输机制:沉降和流动分析了灰烬之间的形态差异。我们训练了一种基于支持向量机(SVM)的机器学习算法,将粒子分为这两类之一,平均准确率达到76%。这些结果表明,所提出的方法是监测火山爆发的一种有价值的工具,可以深入了解火山爆发的特征和相关的环境影响。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Journal of Geophysical Research: Atmospheres
Journal of Geophysical Research: Atmospheres Earth and Planetary Sciences-Geophysics
CiteScore
7.30
自引率
11.40%
发文量
684
期刊介绍: JGR: Atmospheres publishes articles that advance and improve understanding of atmospheric properties and processes, including the interaction of the atmosphere with other components of the Earth system.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信