ORBi: Hermans Thomas - Incorporation of prior information in the regularized inversion of ERT data with CRTomo

Reference : Incorporation of prior information in the regularized inversion of ERT data with CRTomo


	Document type :	Scientific congresses and symposiums : Unpublished conference
	Discipline(s) :	Engineering, computing & technology : Geological, petroleum & mining engineering
	To cite this reference:	http://hdl.handle.net/2268/111181

Title :	Incorporation of prior information in the regularized inversion of ERT data with CRTomo
Language :	English
Alternative title :	[fr] Incorporation d'information a priori de le problème inverse d'ERT régularisé avec CRTomo
Author, co-author :	Hermans, Thomas [Université de Liège - ULg > Département Argenco : Secteur GEO3 > Géophysique appliquée >]
	Caterina, David [Université de Liège - ULg > Département Argenco : Secteur GEO3 > Géophysique appliquée >]
	Martin, Roland [ > > ]
	Kemna, Andreas [ > > ]
	Nguyen, Frédéric [Université de Liège - ULg > Département Argenco : Secteur GEO3 > Géophysique appliquée >]
Publication date :	3-Feb-2012
On invitation :	Yes
Audience :	International
Event name :	Workshop CRTomo Day 2.0
Event date :	le 3 février 2012
Event organizer :	University of Bonn
Event place (city) :	Bonn
Event country :	Germany
Keywords :	[en] prior information ; electrical resistivity tomography ; inversion ; regularization ; geostatistics ; structural constraint
Abstract :	[en] The electrical resistivity tomography (ERT) inverse problem is ill-posed leading to non-uniqueness of its solution. A possible way to reduce the amount of mathematical solutions to more geologically plausible models is to incorporate priori information in the inversion process. In this presentation, we describe the implementation of three different techniques ways to go beyond standard Occam’s inversion for ERT with the code CRTomo : a simple reference model weighted by a user-defined parameter, a structural constraint defined by modifying the construction of the roughness matrix compared to standard Occam’s inversion and a geostatistical constraint based on a vertical correlation length deduced from borehole data to build the model covariance matrix and use it to regularize the inversion. With several examples, we show that the traditional smoothness constraint yield small contrasts of resistivity, far from the reality revealed by borehole measurements. Incorporating prior information from boreholes clearly or from another geophysical technique (like GPR or seismic refraction) improves the misfit with logging data. The choice of the prior information incorporation technique highly depends on which king of information is available and further developments are needed to extend the techniques to more complex cases.
Funders :	Fonds de la Recherche Scientifique (Communauté française de Belgique) - F.R.S.-FNRS
Target :	Researchers
Permalink :	http://hdl.handle.net/2268/111181

There is no file associated with this reference.

All documents in ORBi are protected by a user license.

Dspace v1.5