EESS talk on "Geophysics in environmental science and engineering"
Event details
Date | 20.04.2021 |
Hour | 12:15 › 13:00 |
Speaker | Dr Niklas Linde, Professor, Institute of Earth Sciences (ISTE), Faculty of Geosciences and Environment, UNIL |
Location |
ZOOM
Online
|
Category | Conferences - Seminars |
Abstract:
Geophysics offers remotely sensed (from the Earth’s surface, boreholes, space etc.) information on the Earth’s physical properties (electric, dielectric, magnetic, density, mechanical etc.) at depth scales ranging from centimeters to hundreds of kilometers and at similarly variable lateral scales. For a given environmental setting, the success of geophysics hinges on (1) the properties, processes or predictions of interest, as well as on their spatial and temporal scales, resolution needs and prior knowledge; (2) to what extent geophysically-sensed properties can be linked to the former; (3) the resolving power of the geophysical method(s) and the experimental design; (4) data quality and accuracy of the (numerical) forward models used to link geophysical properties and observables and (5) computing resources. These aspects need to be formalized and integrated to avoid random success stories and failures resulting in over-confidence and deception, respectively. Bayesian statistics offers a suitable framework to achieve realistic expectations and proper uncertainty quantification. Relying on advances in mathematical statistics and deep learning, I use hydrogeophysical examples to showcase the strengths of such a Bayesian approach, the main obstacles and future directions. Since one person or research group is unlikely to possess all the necessary knowledge and skills, I argue that successful quantitative use of geophysics for environmental applications is best achieved by close collaborations among environmental scientists and engineers, geophysicists and statisticians.
Short biography:
Niklas Linde is a professor of environmental geophysics at the University of Lausanne (UNIL). He studied environmental engineering at Uppsala University before completing a PhD in Solid Earth Geophysics at Uppsala University in 2005. During his PhD, he spent one year at Lawrence Berkeley National University, California. After a one-year post-doc at Aix-Marseille University in Aix-en-Provence and a two-year postdoc at ETHZ, he joined the faculty at UNIL. His main research interests revolve around integration of geophysical data in hydrology, inverse theory, uncertainty quantification and Bayesian model selection with environmental applications.
Geophysics offers remotely sensed (from the Earth’s surface, boreholes, space etc.) information on the Earth’s physical properties (electric, dielectric, magnetic, density, mechanical etc.) at depth scales ranging from centimeters to hundreds of kilometers and at similarly variable lateral scales. For a given environmental setting, the success of geophysics hinges on (1) the properties, processes or predictions of interest, as well as on their spatial and temporal scales, resolution needs and prior knowledge; (2) to what extent geophysically-sensed properties can be linked to the former; (3) the resolving power of the geophysical method(s) and the experimental design; (4) data quality and accuracy of the (numerical) forward models used to link geophysical properties and observables and (5) computing resources. These aspects need to be formalized and integrated to avoid random success stories and failures resulting in over-confidence and deception, respectively. Bayesian statistics offers a suitable framework to achieve realistic expectations and proper uncertainty quantification. Relying on advances in mathematical statistics and deep learning, I use hydrogeophysical examples to showcase the strengths of such a Bayesian approach, the main obstacles and future directions. Since one person or research group is unlikely to possess all the necessary knowledge and skills, I argue that successful quantitative use of geophysics for environmental applications is best achieved by close collaborations among environmental scientists and engineers, geophysicists and statisticians.
Short biography:
Niklas Linde is a professor of environmental geophysics at the University of Lausanne (UNIL). He studied environmental engineering at Uppsala University before completing a PhD in Solid Earth Geophysics at Uppsala University in 2005. During his PhD, he spent one year at Lawrence Berkeley National University, California. After a one-year post-doc at Aix-Marseille University in Aix-en-Provence and a two-year postdoc at ETHZ, he joined the faculty at UNIL. His main research interests revolve around integration of geophysical data in hydrology, inverse theory, uncertainty quantification and Bayesian model selection with environmental applications.
Practical information
- General public
- Free
- This event is internal
Organizer
- EESS - IIE
Contact
- Prof. Anders Meibom, LGB