Facies prediction from seismic objects: an approach for multidisciplinary purposes
3D Seismic data is one of the most valuable products used for hydrocarbon exploration, since it provides an image of the subsurface that interpreted helps defining leads and prospects (and their play elements). The first commercial 3D survey was recorded in the 1970’s and since then 3D processing and migration haven’t stopped improving. Appearance of 3D seismic and its development in the last decades have led to a more reliable data that can be used in many different ways and for many different purposes: from purely stratigraphic and structural interpretation, to quantitative interpretation, that helps discriminating fluid types and/or lithology. In the middle, all sorts of attributes and tools that geoscientists have developed and that can be used to help the interpreter in many different ways. To obtain the maximum performance from the seismic data and reduce uncertainty in frontier or underexplored basins, attributes are generated and analyzed. Therefore, methodologies need to be defined in order to obtain a good final product that feeds other geoscience disciplines. The present study focuses in classification and neural network analysis that are natural solutions for extraction and identification of seismic objects. It stablishes a methodology for seismic attribute generation, facies prediction and model building using different Paradigm modules. The data used comes from the Plio-Pleistocene deep sea fan and their associated channel-levee complexes in the Western Black Sea region. First of all, available regional well information was studied and a combination of appropriate seismic attributes to emphasize stratigraphic response was selected. These seismic attributes were analyzed with neural networks to extract patterns and detect trends that are combined to obtain facies variation which was correlated with theoretical stratigraphic model. Based on seismic interpretation and petrophysical studies carried out in nearby wells, stratigraphic intervals were defined. The final product of this workflow is a structural model which contains the main interpreted horizon and facies information provided by the mentioned facies cube. The statistical analysis behind the 3D distribution of the facies identified in the model provides a semi-quantitative approach to lead/prospects generation and classification. The main objective of the facies model was the integration in a basin model in order to properly assess migration pathways. But during the process of developing the methodology, several ideas came up giving the product an added value. Among others, the new purposes of the facies model are: stratigraphic model, static model for volumetric calculation, net to gross estimation, etc.
AAPG Datapages/Search and Discovery Article #90325 © 2018 AAPG Europe Regional Conference, Global Analogues of the Atlantic Margin, Lisbon, Portugal, May 2-3, 2018