19 September 2014 – Seminar
When: 16:00 on Friday, 19 September 2014
Where: DIAS, Geophysics Section, 5 Merrion Square, Dublin 2, (library)
Speaker: Nicola Piana Agostinetti (DIAS)
Title: Trans-dimensional Monte Carlo sampling for structure decoupling: an application to geophysical inverse problems
Abstract:
Trans-dimensional (trans-D) algorithms have been recently introduced to the earth sciences to solve inverse problems without having to impose a fixed spatial structure to the model parametrization (e.g. a fixed number of layers in a 1D structure). Trans-dimensional algorithms can be easily used for joint inversion of different data-set using a “Hierarchical Bayes” approach. However, some difficulties arise if the two observables display very different resolving power. In this case, the structure of the target solution might be twisted toward one observable, introducing non resolved (i.e. over-complex) structure for the other investigated parameters. In this study, we develop a trans-D algorithm for joint inversion of two different data-sets to reconstruct the 1D structure of two different physical parameters. The “parsimony” of the trans-D algorithm produces common discontinuities (a “coupled” structure) for portions of the 1D profile where the two observables displays similar resolving power, while the two reconstructed 1D structures will be different (i.e. the two structure are “decoupled”) where the two observables have different resolving power.
Leave a Comment
Last Updated: 22nd March 2016 by Anna
2014-09-19 – Seminar: Nicola Piana Agostinetti
19 September 2014 – Seminar
When: 16:00 on Friday, 19 September 2014
Where: DIAS, Geophysics Section, 5 Merrion Square, Dublin 2, (library)
Speaker: Nicola Piana Agostinetti (DIAS)
Title: Trans-dimensional Monte Carlo sampling for structure decoupling: an application to geophysical inverse problems
Abstract:
Trans-dimensional (trans-D) algorithms have been recently introduced to the earth sciences to solve inverse problems without having to impose a fixed spatial structure to the model parametrization (e.g. a fixed number of layers in a 1D structure). Trans-dimensional algorithms can be easily used for joint inversion of different data-set using a “Hierarchical Bayes” approach. However, some difficulties arise if the two observables display very different resolving power. In this case, the structure of the target solution might be twisted toward one observable, introducing non resolved (i.e. over-complex) structure for the other investigated parameters. In this study, we develop a trans-D algorithm for joint inversion of two different data-sets to reconstruct the 1D structure of two different physical parameters. The “parsimony” of the trans-D algorithm produces common discontinuities (a “coupled” structure) for portions of the 1D profile where the two observables displays similar resolving power, while the two reconstructed 1D structures will be different (i.e. the two structure are “decoupled”) where the two observables have different resolving power.
Category: Geophysics Section News & Events
Recent Posts
Irish scientists are part of groundbreaking discovery with James Webb Space Telescope
Dr Pauline Gagnon (formerly of CERN) to deliver two talks at DIAS
DIAS Professor announced as next President of the European Southern Observatory’s Council
Quake Shake: New programme encourages people to get involved in monitoring earthquakes
DIAS announces programme for Samhain agus Science festival 2023
Language switcher