Hybrid Quantum–Classical Deep Learning for Satellite Image Analysis and Environmental Change Detection

PhD Project   Ongoing

Project Overview

This PhD project investigates hybrid quantum–classical deep learning approaches for satellite-based Earth observation, with a focus on analyzing and detecting environmental change across land, forests, oceans, rivers, climate systems, and urban environments. The project combines remote sensing, satellite data analysis, AI techniques for Earth observation, and statistical modeling to develop learning pipelines capable of handling large-scale, multi-spectral, and multi-temporal satellite data.

The research will develop and evaluate classical deep learning models alongside quantum and hybrid quantum–classical architectures, applying them to representative Earth observation tasks such as land-use change detection, deforestation monitoring, water-body dynamics, climate-related indicators, and urban growth analysis.

Hybrid architectures will be designed to integrate quantum components within classical workflows, enabling practical experimentation under current NISQ-era quantum computing constraints. Through systematic experimentation, the project will assess model performance, robustness, scalability, and uncertainty handling in real-world satellite analysis scenarios.

Research Context & Motivation

Earth observation systems increasingly rely on automated analysis to support environmental monitoring, climate studies, and urban planning. While classical deep learning methods have achieved strong results, their application to complex Earth observation data remains challenging due to high dimensionality, strong spatial–temporal dependencies, and the need for reliable statistical interpretation of results. These challenges motivate the exploration of alternative and complementary learning paradigms.

Recent advances in quantum machine learning suggest potential benefits in feature representation and optimization, but their relevance to real-world Earth observation remains largely unexplored. This project is motivated by the need to critically assess and contextualize quantum-enhanced learning within practical satellite data analysis, rather than treating it as a purely theoretical development. By comparing classical and quantum-based approaches within the same application domain, the research will clarify where hybrid quantum–classical models add value and where classical approaches remain sufficient.

There will be a dual contribution of this PhD: delivering applied solutions for environmental change detection, while also providing a clear justification for the study of hybrid quantum–classical deep learning as a future-facing approach for scalable and trustworthy Earth observation analytics.

Graduate Researcher

  • Firoz Hasan

Supervisory Team