Machine learning approach to hydrocarbon zone prediction from seismic attributes over “GEM” field, Niger-Delta, Nigeria

A computer programme (in Python language) was developed for the generation and performance assessment of predictive models, capable combining relevant seismic attributes reliable hydrocarbon zone prediction ahead drilling. It attempts to resolve problem making accurate efficient interpretations from a large database derived attributes. The research utilized post-stack 3D volume delineation structures Six faults were identified across mapped horizons. Five generated exported data as numerical values machine learning analysis. binary cross-entropy classification metric used evaluate models while an individual attribute (Maximum Amplitude Extract value) map validate models. correlation well depth-to-top selected horizons with depth slice surface further model validation. Multi-Layer Perceptron (MLP) results enhanced visibility other five zones. MLP gave higher precision predicted zones over Self-Organising Map (SOM) model, thus reinforcing confidence level former. Â