Carbonari, Rolando (2018) Hydrothermal system monitoring by continuous magnetotelluric time series: sensitivity analysis and data denoising. [Tesi di dottorato]

[thumbnail of carbonari_rolando_31.pdf] Documento PDF
carbonari_rolando_31.pdf
Visibile a [TBR] Repository staff only

Download (6MB) | Request a copy
Item Type: Tesi di dottorato
Resource language: English
Title: Hydrothermal system monitoring by continuous magnetotelluric time series: sensitivity analysis and data denoising
Creators:
Creators
Email
Carbonari, Rolando
rolando.carbonari@unina.it
Date: 11 December 2018
Number of Pages: 131
Institution: Università degli Studi di Napoli Federico II
Department: Scienze della Terra, dell'Ambiente e delle Risorse
Dottorato: Scienze della Terra, dell'ambiente e delle risorse
Ciclo di dottorato: 31
Coordinatore del Corso di dottorato:
nome
email
Fedi, Maurizio
fedi@unina.it
Tutor:
nome
email
Di Maio, Rosa
UNSPECIFIED
Date: 11 December 2018
Number of Pages: 131
Keywords: Magnetotelluric; geothermal systems; monitoring; denoising; Campi Flegrei;
Settori scientifico-disciplinari del MIUR: Area 04 - Scienze della terra > GEO/11 - Geofisica applicata
Date Deposited: 10 Jan 2019 20:00
Last Modified: 17 Jun 2020 07:43
URI: http://www.fedoa.unina.it/id/eprint/12650

Collection description

The MT method is a good candidate for characterizing the dynamics of geothermal or hydrothermal systems as it is suitable for deep exploration of the subsurface (from one hundred meters to hundreds of kilometers) in terms of electrical resistivity values, which are strongly sensitive to variations in underground fluid temperature and gas saturation. However, at present, the potential of the MT method for monitoring purposes has not been completely assessed as the only studies in this field focus on monitoring fluid injections in Enhanced Geothermal Systems. The present PhD thesis aims to provide a contribution in this research field by presenting the first attempt for studying the sensitivity of the MT response to natural geothermal (or hydrothermal) system variations. The sensitivity of the MT method has been studied by simulating spatial and temporal evolution of temperature and gas saturation distributions in a hydrothermal system and by calculating the MT response at different time steps through continuous MT measurements. In particular, two possible scenarios have been considered: the first related to an increase in the fluid flow rate from the system source, the second associated to an increase in the permeability of the rocks hosting the hydrothermal system. For each scenario, the sensitivity has been analyzed by evaluating the time interval needed to observe significant variations in the MT response. This study has been applied to the hydrothermal system of the Campi Flegrei volcanic district (southern Italy) and it has shown that the MT monitoring is much more sensitive to changes in rock permeability rather than in the fluid flow rate emitted by the source. In general, long time intervals not useful for volcano monitoring purposes are found if only changes in fluid flow rate are assumed to govern the hydrothermal system dynamics. Conversely, by increasing the permeability of the hosting rocks up to about one order of magnitude, significant resistivity variations are observed over a period ranging from one year and a half to three months. Such findings are promising and encourage the use of the continuous MT measurements in active volcano-hydrothermal areas. Due to the high sensitivity of the MT data to presence of man-made noises that, if not properly detected, can lead to biased resistivity estimates, the present thesis focused also on the development of two non-standard denoising techniques for magnetotelluric data. Both the developed filters are based on the decomposition and analyses of the MT signal through the Discrete Wavelet Transform (DWT), whose use represents an innovative approach in MT processing. The DWT has been chosen considering the resolution that it provides in both time and frequency domains making it a powerful tool to deal with transient and non-stationary signals, as the man-made noise usually appears on the magnetotelluric recordings. The first proposed filter, called polarization filter, aims at detecting the presence of noise by analyzing the polarization of different portions of the electric components of the MT field and by removing those portions whose polarization is higher than a specific threshold. The second filter, called SOM filter, aims at discriminating noisy and clean impedance tensor estimates through a clustering performed with the Self-Organizing Map neural network. Both filters applied to synthetic and field MT data have proven effective in noise detection and in improving the quality of the impedance tensor estimates.

Downloads

Downloads per month over past year

Actions (login required)

View Item View Item