Samir Aghayev

Research

I work on teaching machines to understand the subsurface. My research explores how deep learning models can use seismic, well, and geological information to estimate properties, interpret structure, and quantify uncertainty from sparse and indirect data. I apply this to reservoir characterization and CO₂ storage, where the goal is to make property predictions more reliable and geologically grounded.

Publications

Inversion with stratigraphy-guided deep learning

Property estimation from seismic data is an ill-posed inverse problem. This work introduces stratigraphy-guided deep learning, where geologic units are used as model features to better constrain subsurface property prediction.

Resolving Thin Beds Using AI-Based Petrophysical Property Prediction

This work applies stratigraphy- and rock physics-guided deep learning to predict elastic and petrophysical properties from seismic data, improving thin-bed resolution and geological consistency for reservoir characterization, exploration, and CO₂ storage.

Seismic Fluid Detection with Deep Learning Elastic Property Prediction and Rock Physics

This work evaluates deep learning for seismic fluid detection by predicting fluid-sensitive elastic properties. On Volve data, DL captures blind-well fluid trends, separates hydrocarbon-bearing from dry wells, and shows seismic as the key input.

Språkvelger

Samir Aghayev

Samir Aghayev

Research

Publications

Inversion with stratigraphy-guided deep learning

Resolving Thin Beds Using AI-Based Petrophysical Property Prediction

Seismic Fluid Detection with Deep Learning Elastic Property Prediction and Rock Physics