TwinDryLands
Dryland ecosystems face mounting pressures from climate extremes, desertification, and unsustainable land use, while remaining critical for biodiversity, carbon sequestration, and human livelihoods. TwinDryLands develops a high-resolution Digital Twin that integrates Earth observation, artificial intelligence, dynamic vegetation modelling, and causal inference. The project uses very high- resolution satellite imagery and AI-based mapping to improve the characterisation of vegetation structure and functioning and to support dryland restoration planning, policy, and sustainable land management.
AI for Reforestation Monitoring
Reforestation can help combat land degradation and mitigate climate change, particularly in semi- arid regions, but its success is difficult to assess because field data are limited, vegetation patterns are highly variable, and large-scale monitoring is costly. This project develops AI-based approaches to derive indicators such as tree density, biomass, and biodiversity from high-resolution drone and satellite imagery. By combining Earth observation, machine learning, and field data, the project aims to create scalable and transferable models for spatially explicit assessment of reforestation outcomes.
IceGraph
Antarctic mass loss remains a major source of uncertainty in sea-level-rise projections, partly because complex physical processes are computationally demanding to represent in full-scale ice- sheet simulations. IceGraph develops a MeshGraphNet-based emulator that approximates complex ice- sheet models on irregular meshes. The project also incorporates damage physics into the Kori-ULB ice-flow model and examines how damage affects Antarctic mass loss and uncertainty in sea-level-rise projections.
MOAT
The future contribution of the Antarctic Ice Sheet to sea-level rise is strongly affected by processes that weaken the ice shelves surrounding much of the Antarctic coastline. MOAT investigates basal melting, surface melting, and ice-shelf rheology through radar and photonic sensors, airborne observations, and frequent high-resolution multi-source satellite imagery. The measurements will be used to improve the representation of surface- and basal-melt processes in regional climate and ice- sheet models, helping to reduce uncertainty in projections of Antarctic ice loss and sea-level rise.
ClimaVision
Understanding and projecting changes in key Antarctic climate processes, such as surface melt and surface mass balance (SMB), remains difficult due to the limited spatial detail of current climate models and sparse observational data. ClimaVision addresses this challenge by developing a novel, physically informed downscaling framework that combines remote sensing data with cutting-edge super-resolution deep learning techniques. By embedding physical constraints into machine learning models and leveraging diverse satellite datasets, the project will produce high-resolution climate outputs that better capture local variability across Antarctica. These improved datasets will support more reliable assessments of ice sheet behavior and its role in future sea-level change.
ACCU
Antarctica and the Southern Ocean influence the global heat balance and carbon uptake, but recent anomalies in sea ice, snowfall, and heat waves challenge current understanding of the region's response to climate change. ACCU investigates links between ice-sheet surface mass balance, sea ice, ocean temperature, and ice-shelf melt using observations, causal inference, and the coupled PARASO ice-sheet–ocean–sea-ice–atmosphere–land model. The project aims to clarify the direct and indirect connections among Antarctic climate components and assess their roles in recent and future change.
MicroSCOPE
The Arctic tundra is warming nearly four times faster than the global average, yet the fine-scale microclimatic conditions that govern soil temperature, moisture, and ecosystem respiration remain largely unresolved. MicroSCOPE develops a Gaussian-process emulator that learns mechanistic microclimate behaviour to generate fast, probabilistic, high-resolution predictions of soil temperature and moisture across the European Arctic tundra. The project will reconstruct four decades (1984–2024) of hourly microclimate conditions at 30 m resolution and use these data to assess how microclimate warming has diverged from macroclimate warming and to improve estimates of ecosystem respiration.
DeepVeeg
Accurately predicting how ecosystems will respond to future climate extremes has been a challenge due to the limitations of current vegetation models and the lack of precise data to constrain them. To overcome these obstacles, the DeepV project proposes the development of data-driven, spatially explicit ecosystem response models using remote sensing data and conditional Generative Adversarial Networks (cGANs). By harnessing the power of cGANs, DeepV aims to generate realistic satellite image time series of ecosystem response based on environmental conditions and hydro-meteorological data, enabling climate-to-ecosystem response rule establishment and ecosystem sensitivity assessment.
ECOGraph
Species distribution models support the analysis of biodiversity patterns and conservation planning, but existing methods struggle to represent complex predictor relationships, biotic interactions, irregular environmental data, and community networks. ECOGraph integrates graph theory and deep learning into species distribution models. It develops Graph Neural Networks and graph generative models to capture spatio-temporal relationships, represent species communities, and assess plant species distributions under different climate scenarios.
BRANCH
BRANCH advances how urban green spaces are mapped, understood, and related to public health. It develops a multi-scale, health-oriented typology that combines ecological structure, spatial composition, visibility, and accessibility. The project uses optical, hyperspectral, LiDAR, and SAR remote sensing; AI-powered image analysis; proximal sensing; and citizen science to produce high- resolution functional green-space maps. These maps will support research into relationships between urban greenery, health outcomes, and healthcare use, and inform planning and public-health practice.