Coupling 3D modelling and forward-inverse modelling of potential field data (gravity and magnetic data).

Aurore Joly (1, 2) , Antonio Guillen (2) , Philippe Calcagno (2) , Gabriel Courrioux (2)

1 ISTO - Institut des Sciences de la Terre d'Orléans
2 BRGM - Bureau de Recherches Géologiques et Minières (BRGM)

The 3D modeling of geological objects is often decomposed in two steps: i) delimitation of the boundaries of the units corresponding to the various geological formations or ore types; and ii) verification and estimation of these boundaries using geophysical data. A new approach using potential-field interpolators addressing the 3D modeling problem is used here (Ch.Lajaunie et al. 1997). We will discuss how we can statistically estimate the validity of such 3D model taking in account various geophysical data. This estimation can be computed by inverting complementary datasets, provided (a) the data are a function of the 3D distribution of a source, (b) the response of a given 3D source distribution can be calculated, and (c) the source distribution shows some degree of correlation with the litho-regions. Gravity and magnetic potential field data generally satisfy these criteria. Unfortunately, these data do not allow source geometry to be uniquely resolved through inversion, nor is the source geometry likely to be perfectly correlated with the litho-regions. Even allowing for these limitations, we can see through the expression for the posterior probability density function (PPD) for a Bayesian inversion procedure how uncertainty in prior geological knowledge is modified by investigating the fit to observed potential field data for various models; (1) where is a normalizing constant, is the prior probability for the property model based on geological knowledge, and is the likelihood function that reflects the agreement between the observed potential field response and the predicted response of the model. Litho-models that have reasonable probability based on prior knowledge are downgraded if the likelihood deduced from the associated potential field response is very low. To reduce the non-uniqueness, we can add to the classical data, the tensor components of the field. The main goal with gradients measurements is to improve accuracy and spatial resolution of gravity and magnetic surveys. For those reasons, we propose to build a 3D forward modelling and inversion method for tensor data.

Domaines: Tectonique

Aurore Joly, Antonio Guillen, Philippe Calcagno, Gabriel Courrioux. Coupling 3D modelling and forward-inverse modelling of potential field data (gravity and magnetic data).. GIS in geology, Nov 2006, Moscou, Russia. 6p. ⟨hal-00107327⟩