Determining Consistently the A Priori Information Weight in a Stratigraphie Inversion

Abstract Stratigraphie inversion of seismic amplitudes is a widely-used technique for estimating acoustic or elastic impedances, these impedances being related to reservoir properties in a second step. For inversion processes incorporating a priori information, it is very important to weight the a priori in the cost function, consistently with the a priori uncertainties. The a priori model is built from kriging impedance values at wells along stratigraphic correlation surfaces. The covariance of the kriging error is derived from a theoretical analysis. This covariance is approximated either by a stationary or by a locally-stationary exponential covariance. In both cases, the parameters (range and standard deviation) of the approximated covariance are a priori weights consistent with the true spatial structure of the a priori uncertainties. This methodology significantly changes the a priori weights, that were before simply user-defined, leading therefore to a different estimated impedance field, more coherent with all integrated data.