Unsupervised neural clustering and seismic attribute analysis for gas channel delineation in the Scarab field

Abstract The western offshore Nile Delta of Egypt represents a significant hydrocarbon province, characterized by complex stratigraphy and challenging reservoir connectivity. This study investigates two vertically stacked, unconnected channel systems, Channels 1 and 2, which serve as the primary reservoirs in the Scarab Field. Conventional seismic interpretation techniques often fall short in delineating subtle stratigraphic features and reservoir architecture. To overcome these limitations, we employed an integrated machine learning workflow combining Principal Component Analysis (PCA) and Self-Organizing Maps (SOMs) to enhance reservoir characterization. After data conditioning and the calculation of seismic attributes, PCA was applied to reduce dimensionality and extract the most informative seismic attributes. The resulting principal components were input into SOMs, an unsupervised neural network approach used to classify seismic meta-attributes into meaningful lithological facies. The SOM successfully identified six distinct facies classes: siltstone, clayey sand, clayey limestone, gas-bearing sandstone, water-bearing sandstone, and interbedded siltstone, validated through well log calibration. In particular, the technique delineated gas-water contacts and subtle stratigraphic trends that are critical for both exploration and development planning. The results and quantitative analysis through classification accuracy and GWC misfit assessments using different data types demonstrate that unsupervised machine learning, supported by multi-attribute seismic analysis, offers important improvements over traditional interpretation methods in detecting Direct Hydrocarbon Indicators (DHIs), refining stratigraphic resolution, and reducing exploration risk.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-16
DOI
https://doi.org/10.1038/s41598-026-75298-2
Primary Topic
Seismic Imaging and Inversion Techniques
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Unsupervised neural clustering and seismic attribute analysis for gas channel delineation in the Scarab field

Amin Esmail Khalil, Amir Ismail, Ahmed Eleslambouly, Tarek Khalifa
5 links
Scientific Reports
Seismic Imaging and Inversion Techniques
article

Unsupervised neural clustering and seismic attribute analysis for gas channel delineation in the Scarab field

Amin Esmail Khalil, Amir Ismail, Ahmed Eleslambouly, Tarek Khalifa
article en

Abstract

Abstract The western offshore Nile Delta of Egypt represents a significant hydrocarbon province, characterized by complex stratigraphy and challenging reservoir connectivity. This study investigates two vertically stacked, unconnected channel systems, Channels 1 and 2, which serve as the primary reservoirs in the Scarab Field. Conventional seismic interpretation techniques often fall short in delineating subtle stratigraphic features and reservoir architecture. To overcome these limitations, we employed an integrated machine learning workflow combining Principal Component Analysis (PCA) and Self-Organizing Maps (SOMs) to enhance reservoir characterization. After data conditioning and the calculation of seismic attributes, PCA was applied to reduce dimensionality and extract the most informative seismic attributes. The resulting principal components were input into SOMs, an unsupervised neural network approach used to classify seismic meta-attributes into meaningful lithological facies. The SOM successfully identified six distinct facies classes: siltstone, clayey sand, clayey limestone, gas-bearing sandstone, water-bearing sandstone, and interbedded siltstone, validated through well log calibration. In particular, the technique delineated gas-water contacts and subtle stratigraphic trends that are critical for both exploration and development planning. The results and quantitative analysis through classification accuracy and GWC misfit assessments using different data types demonstrate that unsupervised machine learning, supported by multi-attribute seismic analysis, offers important improvements over traditional interpretation methods in detecting Direct Hydrocarbon Indicators (DHIs), refining stratigraphic resolution, and reducing exploration risk.

Scientific ReportsVol. 16(1)
Khalifa University of Science and Technology (AE)
Science and Technology Development Fund, Helwan University
Seismic Imaging and Inversion Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon