Exploring the Posterior Model Space of Seismic Inversion
Event name: Saint Petersburg 2018
Session: Seismic Attribute Analysis, Seismic Elastic Inversion, AVO
Publication date: 09 April 2018
Info: Extended abstract, PDF ( 782.54Kb )
Price: € 20
Providing the possible diversity of the reservoir model is a key issue for risk assessment during the life of a field. Elastic parameters, obtained from seismic inversion, which are a major source of information for building reservoir model, should cover as much as possible this diversity. Whatever type of seismic inversion used, it relies on a set of prior hyper parameters such as the wavelet, the parameter correlation, the various model constraint types (the lateral and vertical correlations…), the low frequency model, the signal to noise ratio… which all have their own uncertainties. In this paper we demonstrate that drawing realizations of elastic parameters around a posterior mean for a given set of hyper parameters is not enough to explore the posterior model space. We show that including in the seismic inversion the uncertainties associated with the hyper parameters enables to get closer to an extensive exploration of the posterior model space.