Compensation of inaccurate meteorological data in source term determination problem using Bayesian methods

0489184 - ÚTIA 2019 DE eng A - Abstrakt
Tichý, Ondřej - Šmídl, Václav - Hýža, M. - Šindelářová, Kateřina - Ulrych, Lukáš - Belal, Alkomiet
Compensation of inaccurate meteorological data in source term determination problem using Bayesian methods.
Geophysical Research Abstracts. Munich: EGU, 2018. EGU2018-15902.
[EGU General Assembly 2018. 09.04.2018-13.04.2018, Vídeň]
Institucionální podpora: RVO:67985556
Klíčová slova: linear inverse problem * non-Gaussian statistics * ETEX
Obor OECD: Statistics and probability
http://library.utia.cas.cz/separaty/2018/AS/tichy-0489184.pdf

In this contribution, we propose two probabilistic models of the residue statistics. The first model is based on the assumption of non-Gaussian statistics of the residue. In particular, we choose Student-t model of the residues and design a Variational Bayesian estimation procedure for the source term. The second model is based on Gaussian model of the residues with non-diagonal covariance matrix.
Trvalý link: http://hdl.handle.net/11104/0283742