Giacco, Giovanni (2024) Empowering Sustainability through Artificial Intelligence and Earth Observation: Learning from the Past, Monitoring the Present, Building the Future. [Tesi di dottorato]

Anteprima

Testo Giacco_Giovanni__36.pdf Download (12MB) | Anteprima

Abstract

Sustainability is a concept that has gained significant attention in recent years due to the growing concerns about the impact of human activities on the environment. At its core, sustainability is about meeting the needs of the present without compromising the ability of future generations to meet their own needs. While the data-driven model has traditionally prioritized the accumulation and availability of vast datasets in the past, now we need a shift from a data-driven Earth Observation (EO) to a User-Centric Earth Observation to focus on user's needs. This thesis grounds its approach on the pillars of Learning from the Past, Monitoring the Present, and Building the Future. Learning from the Past explores retrospective analysis through applications like Land Consumption Assessment and Aboveground Biomass Estimation, integrating Remote Sensing data with novel deep learning models to generate historical and contemporary maps. These applications demonstrate the ability to monitor temporal changes in urban expansion and ecological biomass, providing a foundational understanding for future sustainability measures. Monitoring the Present addresses the need for continuous assessment of ongoing sustainability initiatives through accessible, shareable, and usable data. The proposed GeoAI Processing Block, embodies this concept by enabling the serverless execution of AI algorithms on EO Data within a Data Space ecosystem. The efficacy of this block is showcased through a case study on monitoring Thermal Comfort for Soft mobility. In Building the Future, the thesis illustrates two methodologies for user-centric action planning. The first leverages historical insights to guide future urban interventions, while the second employs scenario thinking to foresee the effects of urban interventions before their actual implementation by utilizing Generative Adversarial Networks to generate synthetic multispectral satellite imagery. These methods equip users with the tools not only to conceptualize but also shape the future strategically, ensuring that actions taken today are informed by their long-term implications. This work contributes to the field by demonstrating a multifaceted approach that combines the use of Earth observation data and Artificial Intelligence, facilitating a comprehensive process that spans from understanding historical patterns to real-time monitoring and strategic planning for sustainable development.

Downloads

Actions (login required)

Modifica documento