Back to search

Harokopio University

Mineral mapping with the application of machine learning techniques on data from passive sensors and interferometric coherence on data from active satellite sensors

Abstract

dc:description

The mapping of hydrothermal alteration zones in volcanic environments is of high interest. The hydrothermal alteration zones are often linked with ore deposits of high economical interest to the private and public exploration sector. Hydrothermal alteration mapping is a demanding task as hydrothermal alterations are usually located in inaccessible regions, and this can create restrictions to in-situ geological exploration methods. For that reason, new methods have been implemented for the investigation of hydrothermal alteration zones that combine concepts from Remote Sensing, signal processing and machine learning for the extraction of information from the available satellite datasets. The aim of this PhD thesis is the mapping of hydrothermal alteration zones via the application of machine learning and signal processing techniques on satellite data from both passive and active sensors. The case study of this PhD is the hydrothermal alteration zones of the Nisyros volcano caldera. In the framework of this PhD thesis, new methods for the extraction of spectral information are presented, which are applied on data from both passive (ASTER, Sentinel-2/MSI and Landsat-8/OLI) and active (Sentinel-1Β) sensors. Concerning the methodologies related with the processing of the data from passive sensors, the first of them relies on linear spectral unmixing, which is applied to the images collected in the visible near infrared (VNIR) and shortwave infrared (SWIR) regions of the electromagnetic spectrum from the above three satellite sensors, for the area of the hydrothermal alteration zones in the caldera of Nisyros’ volcano. The second methodology utilizes information from the same part of the spectrum relies on clustering (Online Generalized Possibilistic C-Means - O-GAPCM) on the first three Principal Components (PCs) on Sentinel-2/MSI, Landsat-8 (OLI) and ASTER satellite images for the mapping of the mineral groups that participate into the hydrothermal alteration zones of the caldera. The third methodology for the processing of images from the passive sensors, which is applied on ASTER data collected from the spectral region of thermal infrared (TIR), relies on clustering (Gaussian Mixture Model – GMM) for the detection and mapping of thermal differentiation zones into the hydrothermal alteration zones. In addition. a fourth methodology that is based on Spectral Indices and utilizes both the visible-near infrared and the shortwave infrared part of the electromagnetic spectrum, is proposed. Finally, combinations of the aforementioned methods are also investigated. Concerning the data from active sensors, corresponding to the microwave spectral region, satellite data from the active satellite sensor Sentinel-1B of the European Space Agency (ESA) is collected for the detection of differences in the texture of altered and unaltered lithologies within the region of the caldera of Nisyros. The outcome of the present study is a variety of thematic maps that has resulted from the proposed methodologies providing information on: a) the hydrothermal alteration degree and corresponding mineralogy of the different lithologies that are linked to the hydrothermal alteration field, b) the detection and mapping of temperature variations within hydrothermal zones and c) the differentiation in texture between altered and unaltered formations within the hydrothermal alteration zones of the Nisyros caldera. The results of the abovementioned methods for the mapping of the hydrothermal alteration zones within the caldera of Nisyros’ volcano are satisfactory as: a) a broad part of the electromagnetic spectrum is utilized, b) various satellite sensors are used and their capabilities are exploited, c) supervised and unsupervised machine learning methods are applied and d) the potential of the traditional technique of differential SAR interferometry (DInSAR) is investigated, focusing mainly on the coherence products.

Degree

thesis:*
Grantor dc:publisher
Harokopio University
Year dc:date
2023

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Tompolidi, Athanasia-Maria
  • Τομπολίδη, Αθανασία-Μαρία

Subjects

dc:subject × 18

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/53579

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Tompolidi, Athanasia-Maria; Τομπολίδη, Αθανασία-Μαρία. Mineral mapping with the application of machine learning techniques on data from passive sensors and interferometric coherence on data from active satellite sensors. Harokopio University, 2023. http://hdl.handle.net/10442/hedi/53579