Premium
Use of Bayesian dynamic models for updating estimates of contaminated material
Author(s) -
Gargoum A. S.
Publication year - 2001
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.499
Subject(s) - judgement , bayesian probability , computer science , matching (statistics) , prior probability , econometrics , operations research , risk analysis (engineering) , statistics , mathematics , artificial intelligence , business , political science , law
Experienced personnel and safety engineers have a great deal of expertise about the profile of the future emission of contaminated mass from the source after an accident (nuclear, chemical, …). This expert judgement, some quantitative and some qualitative, would be available from the time of the initial release (prior information) and can be accommodated into Bayesian uncertainty management in dispersal models such as puff models. To make full use of this expertise, we outline how prior qualitative information about the expected development of the emission can be modeled as a dynamic linear model (DLM). This prediction model will be used to provide estimates of the source term along with associated uncertainties. Details of matching expert judgement and expected emission profiles are discussed. Copyright © 2001 John Wiley & Sons, Ltd.