Implication of Geostatistical Inversion and Attribute Analysis for Thin Bed Prospect Detection in Lower Goru Formation, Badin area, Pakistan

Abstract

Efforts to explore and characterize oil and gas reservoirs encounter formidable challenges, especially when dealing with inherent heterogeneities and thin sand layers within shale strata. This complexity is exacerbated by the limitations of conventional reservoir classification approaches, struggling with low-resolution seismic data, missing well-log information, and inaccessible angle stack data. Seismic data, often concealing hidden features in the form of thin beds, require a broader frequency spectrum for delineation. These thin beds hold substantial implications for reservoir exploitation and should not be overlooked. This study focuses on the characterization of thin, heterogeneous potential Upper sand within the Lower Goru Formation (LGF), a proven gas reservoir in the Badin area. The reservoir sands with varying thicknesses have been assessed in detail for their optimized description and field productions by handling challenges including low seismic resolutions, heterogeneities, and missing data sets. Utilizing wireline logs and unsupervised fuzzy classification, lithologies, namely, gas-sand, shaly-sand, and shale electrofacies have been predicted within the LGF. Stochastic inversion was applied on post-stack 3D seismic data for estimating the P-impedance values and spatial distribution of lithologies to delineate the thin bedded reservoirs in the formation. The predicted lithologies were spatially distributed over the inverted section using the probability density function which clearly identifies the presence of thin gas sand beds. The extracted slices for two identified thin beds clearly indicate that the eastern part of the study area is favorable for hydrocarbon potential. Furthermore, the high values of sweetness and absorption of high frequencies validate the inversion results. The adopted workflow can be applied to detect thin beds with similar characteristics around the globe. An innovative solution was developed based on the integration of continuous wavelet transform (CWT) and machine learning techniques to predict the missing Shear slowness log and subsequently enhance the elastic and petrophysical properties. Furthermore, the improved properties have been augmented by the high resolution attained through CWT, capturing variability more profoundly through the implication of residual neural networks. The limitation of low-resolution seismic for characterizing thin sand had been attained by developing a novel approach, i.e., an integrating advanced ML technique (ResNet) with decomposed seismic traces of CWT. The incorporation of multiscale properties through the employed technique has the potential to revolutionize subsurface properties through the approximation of enhanced elastic and petrophysical attributes, benefiting hydrocarbon exploration and production efforts. The elastic attributes (P-impedance <7500 m/s*g/cm3, S-impedance <4000 m/s*g/cm3, RHOB < 2.3 g/cm3) and the petrophysical properties, such as Vsh < 0.3, PHIE ≈ 20, and Sw < 0.45, successfully demarcated the thin channelized sand body throughout the field along with the incorporation of the heterogeneities. This study transcends the confines of traditional reservoir characterization by showcasing the transformative potential of geostatistical inversion, machine learning, and high-resolution seismic analysis. The proposed workflow not only unlocks the secrets of thin beds in Badin but also presents an outline for future applications globally, heralding a future of enhanced resource exploration and development.