Overview and advances of statistical concepts used within a climate Detection and Attribution framework (joint work with Alexis Hannart and Aurelien Ribes)
Event details
| Date | 31.03.2015 |
| Hour | 16:15 › 17:15 |
| Speaker | Dr Philippe Naveau, LSCE (UMR-8212), CNRS, Université de Versailles (F) |
| Location | |
| Category | Conferences - Seminars |
Abstract:
In climatology, the field of statistics has become one of the mathematical foundations in Detection and Attribution (D&A) studies (Detection’ is the process of demonstrating that climate has changed in some defined statistical sense, without providing a reason for that change and ‘Attribution’ is the process of establishing the most likely causes for the detected change with some defined level of confidence, see the IPCC definition).
To assess uncertainties in the estimated magnitude of climate change, the classical paradigm is to infer regression coefficients when observed climate change is regarded as a linear combination of externally forced signals and residual internal climate variability. In contrast to standard regression models, the regressor inputs (the climate models) are not considered perfect, but tainted by an modelling error. Hence, we are in the case of an Error-In-Variable (EIV) model of large dimensions with non-proportional error matrices. In this talk, we propose and study a Bayesian framework to handle a variety of EIV situations.
In addition, we will discuss the question of causality, a key aspect mathematical aspect in attribution studies.
If time allows, we will also speak about the question of rare events and the often-used idea of Fraction of Attributable Risk.
Short biography:
After obtaining his PhD in Statistics at Colorado State University in 1998, Dr. Philippe Naveau was a visiting Scientist at National Center for Atmopsheric Research in Boulder, Colorado for three years. Then, he was an assistant professor in the Applied Math Dept of Colorado University (2002-2004). Since 2004, he is a research scientist at the French National Research Center (CNRS) and his research work has focused on environmental statistics, especially in analyzing extremes events.
In climatology, the field of statistics has become one of the mathematical foundations in Detection and Attribution (D&A) studies (Detection’ is the process of demonstrating that climate has changed in some defined statistical sense, without providing a reason for that change and ‘Attribution’ is the process of establishing the most likely causes for the detected change with some defined level of confidence, see the IPCC definition).
To assess uncertainties in the estimated magnitude of climate change, the classical paradigm is to infer regression coefficients when observed climate change is regarded as a linear combination of externally forced signals and residual internal climate variability. In contrast to standard regression models, the regressor inputs (the climate models) are not considered perfect, but tainted by an modelling error. Hence, we are in the case of an Error-In-Variable (EIV) model of large dimensions with non-proportional error matrices. In this talk, we propose and study a Bayesian framework to handle a variety of EIV situations.
In addition, we will discuss the question of causality, a key aspect mathematical aspect in attribution studies.
If time allows, we will also speak about the question of rare events and the often-used idea of Fraction of Attributable Risk.
Short biography:
After obtaining his PhD in Statistics at Colorado State University in 1998, Dr. Philippe Naveau was a visiting Scientist at National Center for Atmopsheric Research in Boulder, Colorado for three years. Then, he was an assistant professor in the Applied Math Dept of Colorado University (2002-2004). Since 2004, he is a research scientist at the French National Research Center (CNRS) and his research work has focused on environmental statistics, especially in analyzing extremes events.
Practical information
- General public
- Free
- This event is internal
Organizer
- IIE - EESS; seminar funded by EPFL Center on Risk Analysis and Governance (CRAG)
Contact
- Prof. Michael Lehning, CRYOS-EPF Lausanne & SLF Davos