Uncertainty in Seismic Inversion - Effect of Uncertainty in Wavelets and a Priori Model

Abstract

Seismic inversion for elastic parameters is prone to uncertainties in the inversion parameters such as wavelet and a priori vertical and lateral correlation lengths of elastic parameters. In current inversion tools, the inversion parameters are assumed to be given and their uncertainties are not taken into account in the inverted elastic properties. In this paper, we present a methodology based on Gaussian Mixture Models to estimate uncertainties in the elastic inversion result due to uncertainties in the inversion parameters. We applied this method on a data set from West Africa. Our results show that uncertainties about wavelet and a priori model induce large uncertainties on the elastic inversion results; these uncertainties are much larger than uncertainties estimated by a fixed set of inversion parameters.