Satellite-Based Data-Driven Modelling of Inland Green-Blue Ecosystems

Abstract

Technological advances in Satellite Earth Observation (SEO), decision systems and geoinformation are revolutionizing the assessment of natural assets. The evolving integration of satellite-based Machine Learning (ML) techniques into Green-Blue Ecosystems (GBE) modelling demands a deeper understanding of applications across scales, and the identification of avenues for future research. To study the relationship between aquatic and terrestrial ecosystems, an assessment framework with methods, concepts, and models was developed. It includes a set of tools with reference data and guidelines, based on reviewing well-established and innovative techniques reported worldwide. The research is divided into three main phases: Conceptualization and Design (Stage 1), Proof of Concept (Stage 2), and Development and Implementation (Stage 3). The main output of Stage 1 is a systematic literature review and meta-analysis focused on studies that have developed ML models to estimate Ecosystem Services indicators based on high-resolution and open-access satellite data. Stage 2 demonstrates the feasibility, robustness, and replicability of research. The main outputs of this stage are modelling different spatial and temporal scales, and testing data and methods, providing guidelines to basis the development and implementation of research at the country level. At the development and implementation stage, the main outputs are the GBE capacity provisioning assessment at the national level and the spatial and temporal multiscale mapping of carbon storage and sequestration by forest ecosystems in Portugal's mainland. This dissertation contributes to advancing the knowledge of GBE, unveiling the potential of satellite-based data-driven models, understanding ecosystem dynamics, drivers of changes, modelling conceptualization, development, and implementation. Findings link the current knowledge to integrated models that include key aspects of GBE drawing on science-technology-innovation, presenting guidelines for the assessment of GBE from SEO, and filling the gap with advanced knowledge related to consistent GBE information across multiple scales.