|
|
Stochastic simulations and modelling in the magnetotelluric method
Klanica, Radek ; Pek, Josef (advisor) ; Růžek, Bohuslav (referee) ; Velímský, Jakub (referee)
In the thesis I deal with the development of a stochastic inversion procedure for the magnetotelluric method in 1D/2D isotropic and anisotropic cases, and its application to both synthetic and real data. The magnetotelluric method is a geoelectric inductive technique that utilizes variations of naturally occurring electromagnetic fields as a source of the electromagnetic induction for estimating the Earth's subsurface resistivity to depths of several tens of kilometres. The purpose of the inversion procedure is to estimate a real distribution of the electrical resistivity in the Earth's subsurface from surface measurements. Common inversion procedures in magnetotellurics perform a model optimization by minimizing the misfit between the data and the model response. Stochastic methods are based on the exploration of the model parameter space, and they pick models according to their probability, which makes them effective for the solution of high-dimensional problems which do not show a single pronounced minimum of the target function. The effective ways of mapping the parameter space are sampling algorithms based on Monte Carlo simulations which allow to sort models according to their probability. Results of these methods are obtained in the form of a fully probabilistic description of the...
|
|
|
Stochastic simulations and modelling in the magnetotelluric method
Klanica, Radek ; Pek, Josef (advisor) ; Růžek, Bohuslav (referee) ; Velímský, Jakub (referee)
In the thesis I deal with the development of a stochastic inversion procedure for the magnetotelluric method in 1D/2D isotropic and anisotropic cases, and its application to both synthetic and real data. The magnetotelluric method is a geoelectric inductive technique that utilizes variations of naturally occurring electromagnetic fields as a source of the electromagnetic induction for estimating the Earth's subsurface resistivity to depths of several tens of kilometres. The purpose of the inversion procedure is to estimate a real distribution of the electrical resistivity in the Earth's subsurface from surface measurements. Common inversion procedures in magnetotellurics perform a model optimization by minimizing the misfit between the data and the model response. Stochastic methods are based on the exploration of the model parameter space, and they pick models according to their probability, which makes them effective for the solution of high-dimensional problems which do not show a single pronounced minimum of the target function. The effective ways of mapping the parameter space are sampling algorithms based on Monte Carlo simulations which allow to sort models according to their probability. Results of these methods are obtained in the form of a fully probabilistic description of the...
|
|
|
Magnetotelluric sounding of the Earth's crust in a region with strong industrial electromagnetic noise
Klanica, Radek ; Pek, Josef (advisor) ; Velímský, Jakub (referee)
In this thesis, I am dealing with a re-processing of magnetotelluric data from the West Bohemia seismo-active region. By revealing zones of enhanced electrical conductivity in the Earth, the magnetotelluric method is potentially perspective for detecting fluids which may play a major role in the generation of local earthquakes. The magnetotelluric method employs the natural electromagnetic field of the Earth, which is, on the Earth's surface, relatively weak as compared to various cultural sources. In West Bohemia, the level of the cultural electromagnetic noise is very high. Therefore, I have tested various approaches suggested for reducing the cultural noise in the data. In particular, I used a data variance technique for automatically despiking magnetotelluric data by employing simultaneous records from a remote reference site. I also employed a noise separation technique for magnetotelluric single site data by applying a frequency domain data selection scheme. Due to an extremely high level of the cultural noise in the study area, both those approaches failed as standard pre-processing methods. High frequency data show a relatively lower level of the corruption by noise, and a subset of data acquired in 2001 could be modelled via magnetotelluric inversion procedures. Ten magnetotelluric sites...
|
|
|
Effects of cultural noise in natural magnetotelluric signals
Klanica, Radek ; Pek, Josef (advisor) ; Červ, Václav (referee)
Magnetotelluric method is an electromagnetic induction method which studies the dis- tribution of the electrical conductivity in the Earth by analyzing natural variations of the Earth's electromagnetic field of the external origin. A wide frequency range of the natural sources of the magnetotelluric field makes it possible to study the Earth's conductivity from near-surface structures down to depths of the upper mantle. Dependence on weak natural fields, however, causes the method to fail in case of data contaminated by electro- magnetic disturbances of cultural origin. This thesis summarizes basic principles of the statistical magnetotelluric data processing in the frequency domain, and shows the recent progress of the processing due to the application of robust statistical methods as well as due to employing reference data from remote stations. In some cases, the measurements are disturbed excessively, and even advanced statistics method fail in processing the data. In some of these cases a more thorough analysis of the noise field in terms of its directional and source characteristics may be useful. We present an example of simultaneous recor- dings of two magnetotelluric stations from the West Bohemian seismo-active region and show that the noise field is dominated by two types of strong electric...
|
guest ::
