Abstract
Indications of short and long range correlation are important basis for stochastic modeling of geological reservoirs. This paper examines the merits of various statistical methods for recognizing rock properties correlations in vertical columns. Specifically the persistence of patterns analyzed by the R/S method were analyzed with the application of semi-variogram and power spectral density methods. From synthetically generated patterns (fGn, fBm), short and long range correlations for large data sets corresponding to pure, layered and composite homogeneous and heterogeneous sets was analyzed. Additionally, the impact of the overlap averaging on the estimation of correlation were evaluated. Concurrent analysis using the three techniques is shown to enhance the interpretation of the correlation type.
