Abstract Two lithofacies and fluid discriminating seismic attributes are integrated using Artificial Intelligence (AI) via Unsupervised Artificial Neural Network (UANN) to characterize the architecture of deep water turbidite channels and submarine fan lobes across a hydrocarbon bearing reservoir within the Frem Field deep-water Niger Delta. A data-based approach including reservoir identification, environment of deposition prediction, seismic attribute analysis and finally UANN using the competitive learning algorithm (CLA) was used to match patterns from the two seismic attributes in order to reduce and capture uncertainties inherent with characterization of turbidite sands within the stratigraphic and structurally complex deep-water Niger Delta. One hydrocarbon bearing reservoir (Sand R001) with excellent reservoir quality was identified from the wireline logs interpretation after which gamma ray logs motifs as well as root mean square (RMS) amplitude and sweetness attributes imaging revealed the environment of deposition of the sand as an inner fan channel within a complex system of several channels and submarine fan lobes. Discreet facies map generated from the CLA enabled a better definition of the architecture, orientation and trend of the sand and lobate nature of the submarine fans lobes associated with the reservoir. The resulting output led to an enhanced characterization of the architectural patterns of the reservoir as well as associated deep-water facies in terms of reservoir architecture and orientation. The discreet facies map also revealed both northeast southwest and northwest southeast orientation of turbidite channels and submarine fan lobes and indicates the channels serves as feeders to the lobate submarine fan systems. The study has shown the efficacy of AI in enhancing deep water architectural patterns via pattern matching of facies and fluid related seismic attributes using CLA and thereby shows the method is effective in reducing uncertainties inherent with deep water reservoir characterization in the Niger Delta.

中文翻译：

---

使用人工智能表征深水浊流通道和海底扇叶；Frem Field 深海尼日尔三角洲案例研究

摘要 利用人工智能 (AI) 通过无监督人工神经网络 (UANN) 整合两种岩相和流体判别地震属性，以表征尼日尔 Frem Field 深水区含油气储层的深水浊流通道和海底扇叶的结构。三角洲。基于数据的方法包括储层识别、沉积环境预测、地震属性分析，最后使用竞争学习算法 (CLA) 进行 UANN 来匹配两种地震属性的模式，以减少和捕获浊积岩表征固有的不确定性地层和结构复杂的深水尼日尔三角洲内的砂岩。从电缆测井解释中确定了一个具有优良储层质量的含油气储层（砂R001），之后伽马射线测井图案以及均方根（RMS）幅度和甜度属性成像揭示了砂的沉积环境作为内部由多个水道和海底扇叶组成的复杂系统内的扇形水道。从 CLA 生成的谨慎相图可以更好地定义与储层相关的海底扇叶的沙质和叶状性质的结构、方向和趋势。由此产生的输出导致对储层结构模式以及相关深水相在储层结构和方向方面的增强表征。谨慎的相图还揭示了浊流通道和海底扇叶的东北西南和西北东南方向，并表明这些通道是叶状海底扇系统的馈线。该研究表明，人工智能通过使用 CLA 对相和流体相关地震属性进行模式匹配来增强深水建筑模式的功效，从而表明该方法可有效减少尼日尔三角洲深水储层特征固有的不确定性。