Pre and Post Stack Seismic Inversion: Implications in Seismic Reservoir Characterization

Abstract

The mapping and characterization of thin gas sand layers have always been a challenge through conventional seismic interpretation procedures. It is a known fact that hydrocarbon and reservoir properties can only be accurately defined by using integration of seismic and well data. Geological predictions become challenging in areas revealing structural and lithological complexity, as well as in units located away from exploration wells. Seismic inversion for assessing inverted impedances has been frequently employed from many decades using integrated data sets. Seismic inversion techniques have been effectively applied to evaluate the hydrocarbon reservoirs with high heterogeneity and deep burial depth having thicknesses below seismic resolution. In this study, gas sand layers of the Cretaceous and Paleocene reservoirs of the Mehar gas field in the Kirthar Fold Belt (KFB), Pakistan, have been characterized via the application of rock physics analysis, post-stack model-based seismic inversion (PSMBI), pre-stack simultaneous seismic inversion (PSSI) and geostatistical Bayesian inversion. These sands are producing in the nearby fields; however, their potential is still untapped in the study area, which needs to be explored for further development. These sands are widely distributed in the southwestern part of the basin and are enormously heterogeneous, which makes it difficult to distinguish facies and fluid content in the reservoir intervals. Mehar structure possess a pop-up anticlinal geometry with frontal (eastern) and back (western) thrusts. The back limb is still unexplored having same hydrocarbon facies and sand anomalies. In order to evaluate the gas potential of Mehar block, three phases of studies have been adopted. Gas sand facies have been demarcated in reservoir (Lower Ranikot and Pab) formations on the basis of cut-offs defined by detailed wireline log (petrophysical) analysis. A careful rock physics analysis was carried out to synthesize the 𝑉𝑠 log and to improve the quality of log data especially in washout zones. This helped in prediction of petro-elastic models (𝜆𝜌 versus 𝜇𝜌, impedance versus 𝑉𝑝⁄𝑉𝑠 ratio) at true reservoir conditions/parameters. Advanced seismic inversion algorithms (PMBSI, PSSI and geostatistical Bayesian inversion) were applied for properly calibrated detailed characterization of gas sands utilizing integration of improved log quality data analysis. 2 | P a g e Petrophysical analysis demarcated three gas bearing zones within the Lower Ranikot Formation named as LMS1, LMS2 and LMS3 along with detailed evaluation of Pab Sandstone. The spatial extent of these identified sand packages has been evaluated in detail by the application of the above selected seismic inversion techniques. Initially, a robust petro-elastic relationship obtained from rock physics models has led to more precise discrimination of pay and non-pay facies in the sand intervals of the study area. The calibrated rock physics model shows good consistency between modelled and measured logs. The cross plots clearly separated and demarcated the litho-fluid classes (wet sand, gas sand, shale and limestone) with specific orientation/patterns which were randomized in conventional petrophysical analysis. Although, the results of PSMBI depicted low impedances within the sand layers of the Lower Ranikot Formation, however, PSSI efficaciously separated the thin gasbearing and non-bearing sand intervals. Furthermore, the impedance-derived porosities predicted via probabilistic neural network (PNN) show excellent agreement with the computed porosities from interpreted well logs and at producing blind wells. PSSI along with its attributes i.e., volumes of 𝑍𝑝, 𝑍𝑠, 𝑉𝑝⁄𝑉𝑠 ratio, are applied to highlight the presence of tight and potential sand intervals, whereas Lambda-MuRho (LMR) seismic inversion is applied to determine fluid content and the lithology of reservoirs. The integration of PSSI and LMR inversion was successful in delineating new hydrocarboncharged sands in the western part of the study area. Moreover, the volumes inverted using both techniques allowed the successful discrimination between fluids and lithology, not only at the blind well locations, but away from these wells. Importantly, the results show that high values of MuRho (𝜇𝜌) indicative of clean sands while low values of LambdaRho (𝜆𝜌) correlate with areas containing hydrocarbons. The application of Bayesian stochastic inversion further improved our efficiency of characterizing thin sand packages by utilizing Sequential Gaussian Simulation in linearized AVO inversion, which performs trace-by-trace decomposition of global posterior impedance into local posterior impedance. The techniques adopted to delineate and map the potential thin gas sands of the Mehar gas field may be very fruitful and adequate for basins with similar geological settings