Global stochastic seismic inversion with variable prior knowledge for lithology prediction

Abstract

A global stochastic seismic inversion method has been used to evaluate the possibility of seismically resolving porous Zechstein carbonates in Southern Jylland, Denmark, using a variable degree of prior knowledge to guide the inversion. 3 inversions of one seismic section with two different wavelet estimations have been carried out: (1) the inversion is constrained at the well location by the acoustic impedance log using a wavelet estimated via the acoustic impedance log at the well location; (2) the inversion is constrained only by the acoustic impedance of 20 ms halite at one trace 6 km from the well location using the wavelet estimated at the well location in inversion 1; (3) the inversion is constrained by the acoustic impedance at one trace as in inversion 2 using a wavelet estimated from the frequency spectrum of the seismic section including the reflectivity of a reflector at one sample also 6 km from the well location. Each inversion generates three independent results: (a) a globally optimized acoustic impedance section, (b) a layer interface section, that indicates major acoustic impedance contrasts, and (c) a probability section, that indicates the likelihood of interpreted interfaces. The three inversions show, that the structural outline is consistentlymore » resolved along with specific layers as a porous carbonate layer even in the case where no well information is used. The resolution of each acoustic impedance section deteriorates only slightly with less prior knowledge included in the inversions. The relative acoustic impedances are consistently defined, however the absolute acoustic impedances partly drift.« less