--> Application of Seismic Stratigraphy, Artificial Neural Networks and 3D Seismic Geomorphology to Improve Reservoir Understanding of Lower Goru Sequence in Central Indus Basin, Pakistan

International Conference & Exhibition

Datapages, Inc.Print this page

Application of Seismic Stratigraphy, Artificial Neural Networks and 3D Seismic Geomorphology to Improve Reservoir Understanding of Lower Goru Sequence in Central Indus Basin, Pakistan

Abstract

Central Indus Basin of Pakistan is one of the most prolific basins of Pakistan and has historically been producing from giant structures with Eocene carbonate reservoirs. Extensive exploration of Early Cretaceous reservoirs in 1990s added many new fields focusing mostly within Lower Goru shoreface sands of Aptian/Albian Age. This study follows the sequence stratigraphic analysis of Lower Goru and Sembar formations presented by Ahmad et.al. (2004). That study covered the geological facies and sequence stratigraphic analysis and was supported by seismic stratigraphic analysis. During the last decade numerous new 3D seismic datasets have become available in the producing areas as well as further west where the deeper potential of Lower Goru is still unknown. Apart from the western part, most of the new potential is believed to exist beneath the producing Eocene carbonate anticlinal pools such as Sui, Kandhkot, Qadirpur and Kandra etc. The methodology applied in this study includes the use of seismic dip attributes to auto-convert the whole Lower Goru section into hundreds of stratal slices, so that the surfaces follow the depositional trends. The key horizons are then extracted that represent local flooding surfaces. The benefit of stratal slicing unfolds in the Spectral Decomposition and Neural Network Facies analyses, which unveils the whole new vista of stratigraphic plays in the Lower Goru Sand/Shale sequences. In spectral decomposition the amplitude data is transformed into frequency cubes and three spectral bands are co-visualized using RGB color blending. Universal Vector Quantization (UVQ), a neural network segmentation process is then applied to the same interval to overlay the seismic waveform characteristics. These methods have been extensively used in Western Canada Sedimentary Basin (Baranova et.al., 2012) to locate the hidden stratigraphic reservoirs that are often masked by thin bed tuning amplitudes and high signal/noise ratios.