AVO inversions are carried out on synthetic data using a pseudo-linear form of the exact Zoeppritz equations in order to demonstrate the significance of various types of errors in the input data. Noise is represented by random Gaussian data added to the reflectivities, and this is shown to be one of the key sources of error. Error in the background parameters (Β/α, Δα/α, ΔΒ/Β, Δρ/ρ) is represented by adding deviations to their exact values. Error in the incidence angle may be either random or systematic. Random errors are modeled with noise added to the assumed values. Systematic errors are represented by linearly scaling the assumed values. Some input errors are seen to result in contrast errors which are correlated with the value of ΔΒ/Β. This method allows direct and quantitative comparison of the effect of input errors with the effect of approximations to the Zoeppritz equations. In this study comparisons are made with the effect of the Aki-Richards approximation.
View full article as PDF (0.33 Mb)