ML-based 3D seismic clustering technology for prospect reserves evaluation: A case study from fields in South Capsian Basin

This paper presents an innovative algorithmic method for analyzing seismic data, focusing on automating spectral decomposition interpretation and clustering distinct 3D seismic facies. By leveraging Self-Organizing Maps (SOM) and K-means clustering, the proposed approach effectively detects and categorizes geological structures. Incorporating a broad frequency spectrum and a combined unsupervised machine learning strategy, it enhances feature identification beyond conventional techniques and facilitates the construction of 3D geobodies. With practical applications in reservoir characterization, hydrocarbon exploration, and reserve assessment, this method minimizes human bias and offers valuable insights for industry professionals and researchers working in seismic data analysis and exploration.