Northeast Oil and Gas Branch, SinoPEC, Changchun 130026, China

Pre-stack inversion is now a common way to predict complex reservoirs, with random inversion or simulation as an addition to improve the resolution of thin reservoirs. In view of the difficulty of lateral control in stochastic lithology simulation constrained by pre-stack inversion, the author proposes a new method of pre-stack spatial variant inversion,in which a spatial variable factor is added to the inversion objective function based on Bayesian framework. It is applied to the tight reservoir with less than 5% porosity in Longfengshan area of Songnan, a more sensitive fluid factor is inverted by the new method.