Keynote Lectures

Scaling Biophysical Reality: The Evolution of Operational National to Continental Monitoring
Peter Scarth, University of Queensland, Australia, Australia

From the Swiss Data Cube to Living Switzerland: Developing a Digital Twin of the Environment
Gregory Giuliani, University of Geneva, Switzerland, Switzerland

Available Soon
María Lucrecia Pettinari, Alcalá University, Spain, Spain


Dimitris Poursanidis, Foundation for Research and Technology - Hellas, Greece, Greece

(Cancelled)

Abstract
Effective global land management faces a critical tension: the necessity for consistent, long-term Earth observation monitoring extending back to the 1980s versus the rapid evolution and expansion of platforms, sensors and algorithmic paradigms. This keynote addresses the challenge of harmonising historical archives with modern high-dimensional data and analytics, using the Australian continent as a primary case study for operational biophysical monitoring of the land surface. With a focus on Australia, the keynote will outline the decadal evolution of a novel Fractional Cover methodology that has moved beyond static indices to a system that measures biophysical reality in landscapes defined by extreme climatic variability. The presentation details a progressive technical transition from traditional linear spectral unmixing to a modern Multilayer Perceptron (MLP) machine learning framework. Crucially, this Machine Learning (ML) approach maintains strict biophysical constraints—ensuring cover fractions (e.g., vegetation, soil, water) sum to 100% and remain non-negative—while offering superior computational efficiency compared to iterative spectral unmixing methods. This evolution enables seamless integration of diverse data sources (from Landsat to Sentinel-2) to achieve the consistency required for long-term trend analysis. Major focus will be placed on the validation challenges inherent to the mapping and monitoring of large land areas, including transitioning from subjective field-based estimates of biophysical attributes to more objective sensor-based methods and the automated generation of training data for ML algorithms from very high resolution (VHR) imagery acquired by sensors on platforms ranging from drones to satellites. Finally, and building on the case study, we demonstrate how the methods developed have transcended national borders. By underpinning the Digital Earth Australia and Digital Earth Africa platforms, this approach now supports diverse applications ranging from sediment management in the Great Barrier Reef (Reef 2050) to food security monitoring across Africa. The session will conclude with a discussion on the future of globalised algorithms and the harmonisation efforts required to apply these models across disparate ecological and spectral domains.

Abstract
Switzerland is advancing from the long-established Swiss Data Cube (SDC)—a national data infrastructure providing analysis-ready satellite Earth observation data—towards Living Switzerland, a next-generation digital twin of the environment. This transition represents a paradigm shift from static data access and analysis to dynamic, integrated environmental intelligence. Building on the SDC’s robust foundation of standardized, analysis-ready data and open science principles, Living Switzerland leverages the Living Earth framework to enable real-time integration of multi-source environmental data, modeling, and simulation. The system aims to create a continuously updated, semantically rich, and interoperable representation of Switzerland’s natural environment—encompassing land, water, atmosphere, and biodiversity. Through this transformation, Living Switzerland will support evidence-based decision-making, enhance environmental monitoring and forecasting, and foster collaborative research across domains.

Abstract
Available Soon

Abstract
Blue Carbon ecosystems, such as seagrass meadows, represent vital natural assets for climate mitigation, yet their distribution and condition remain poorly quantified in many regions. Here we presents recent advances in combining Earth Observation (EO) data with field missions to map and assess Blue Carbon habitats. Using multi-scale satellite imagery and machine learning algorithms, we developed reproducible workflows for habitat discrimination, bathymetry estimation, and seagrass extent mapping across coastal waters. Field observations collected through scientific diving protocols provided high-quality reference data for calibration and validation, ensuring accuracy and spatial robustness. The integration of EO analytics and in situ measurements allowed detailed quantification of seagrass cover and density, enabling refined Blue Carbon stock estimations at multiple spatial scales allowing transparency in the process of reporting in NDCs and under other MEAs. This fusion approach demonstrates the potential of cost-effective, scalable, and repeatable monitoring solutions to support restoration planning, conservation prioritization, and long-term carbon accounting. This work highlights how bridging remote sensing technologies with ecological expertise advances our understanding of Blue Carbon ecosystems and supports the implementation of evidence-based coastal management and climate adaptation strategies.