Abstract

In a proposed methodology for probabilistic calculation of elastic parameters from seismic reflectivities, one can focus on bulk density and VP/VS ratios and how those change at a layer interface. The probabilistic approach helps to model uncertainties in the elastic parameters. Zoeppritz's equation for calculation of PP reflection coefficients (RCs) is used to create 2D conditional probability density functions (CPDFs) of model parameters which are constructed by systematic grid search over the entire space of the model parameters. The method accurately recovers even highly irregular (nonelliptical) PDFs with multiple maxima. Moreover, it finds a global maximum, as opposed to many other techniques in which the algorithms, if supplied by a poor initial set of parameters, can become “stuck” in a local maximum. No empirical relationships are built into the method. The final products include “most probable” as well as uncertainty maps of various rock properties. The methodology was tested in a North Sea reservoir, and results were validated at two well locations. The “best” estimates of inverted parameters are in very good agreement with parameters extracted from wells. More specifically, uncertainty recovery of density ratios is very small, whereas shear-impedance (SI) ratios are captured within one standard deviation. The combination of density and shear information reveals important reservoir characteristics. Diagenetic differences in sandstone reservoirs are revealed from shear-impedance ratios, whereas high-porosity sands saturated with hydrocarbons are detected from density ratios.

You do not currently have access to this article.