Hyperspectral

This project explores the use of hyperspectral imaging and machine learning to assess the environmental sustainability of oil well sites in the Vaca Muerta shale formation, Argentina — supported by Vista Energy.

Project Overview

Hyperspectral is a research initiative that applies airborne hyperspectral image analysis to monitor and assess the environmental footprint of oil exploration and production activities in the Vaca Muerta region (Neuquén Province, Argentina). By capturing hundreds of spectral bands simultaneously, hyperspectral sensors reveal surface composition, vegetation stress, soil contamination, and emission signatures that standard RGB imagery cannot detect.

The project combines remote sensing, computer vision, and machine learning to build a data-driven sustainability assessment pipeline — directly applicable to the regulatory and operational needs of the oil & gas industry in one of the world’s most significant unconventional hydrocarbon plays.

Scope

The project investigates whether hyperspectral imagery can serve as a scalable, objective tool for sustainability compliance monitoring across oil well pads. The core research questions are:

• Which spectral features are most predictive of environmental impact indicators around well sites?
• How do machine learning models trained on hyperspectral cubes generalise across different seasons, lighting conditions, and geographic sub-regions within Vaca Muerta?
• Can the pipeline support near-real-time sustainability reporting for industry partners?

Technologies: Hyperspectral image processing, dimensionality reduction (PCA, MNF), spectral unmixing, deep learning for remote sensing, Python, GDAL, spectral libraries

Milestones

1. Project inception & Vista Energy partnership — Research agenda defined jointly with Vista Energy;
2. Baseline processing pipeline — Radiometric correction, atmospheric compensation, and spatial registration pipeline implemented and validated on acquired datasets.
3. CONEXPLO 2024 paper — First results presented at CONEXPLO; methodological framework and preliminary findings published (paper PDF).
4. Model development (ongoing) — Machine learning classifiers and spectral analysis models under active development and benchmarking.
5. PhD thesis (in progress) — Full dissertation under preparation; expected to cover the complete sustainability assessment framework.

Contributions

• Hyperspectral data acquisition and pre-processing pipeline for Vaca Muerta overflight datasets
• Spectral feature extraction and analysis for oil well sustainability indicators
• Machine learning models for land cover classification and anomaly detection from hyperspectral cubes
• Industry collaboration framework between academia (ITBA) and the energy sector (Vista Energy)
• First peer-reviewed contribution linking hyperspectral remote sensing to sustainability assessment in Argentine shale exploration

Collaborators

• Esteban Roitberg, PhD

Funding & Partners

This project is funded and supported by Vista Energy, one of the leading independent oil & gas companies operating in the Vaca Muerta formation.

Publications

• CONEXPLO 2024 — Hyperspectral Analysis for Sustainability in Vaca Muerta (paper PDF)