Overview
Space2Health explores how satellite imagery can be transformed into meaningful measures of environmental exposures linked to health. The project relies on recent AI technologies to develop and evaluate satellite-derived indicators, examining how data choices and analytical approaches shape health-relevant evidence for research and decision-making.
About this opportunity
The global availability of high-resolution, high-frequency satellite imagery is transforming how environmental conditions are monitored and analysed at scale. Contemporary Earth Observation missions such as Sentinel-1 and Sentinel-2, Landsat-8/9 and commercial constellations deliver continuous, multi-sensor data streams describing urban form, green and blue infrastructure, air quality proxies, land surface temperature, land use dynamics and environmental change. In parallel, advances in geospatial artificial intelligence, including transformer architectures, foundation models, self-supervised learning and large-scale spatial embeddings, are redefining how complex spatial data can be processed and interpreted.
Despite these developments, a critical methodological gap remains. Many environmental health indicators rely on simplified or legacy remote sensing metrics that were not designed with health inference as their primary objective. Consequently, the analytical potential of modern satellite systems is not fully realised in public health research. Robust, interpretable and transferable approaches are needed to translate raw satellite data into meaningful, health-relevant spatial evidence.
This PhD will address that challenge by developing and evaluating next-generation satellite-derived indicators for health applications. The emphasis will be on methodological innovation: designing indicators that are explicitly fit for purpose for robust, scalable and interpretable spatial inference in health research. Rather than relying on conventional indices, the project will examine how design decisions influence the representation of environmental conditions and their linkage to health outcomes.
The candidate will investigate how modelling choices, including spatial resolution, temporal aggregation, feature engineering, multi-sensor data fusion (optical, SAR, thermal), and machine learning architectures, shape how environmental signals are extracted, structured and connected to health-relevant processes. Approaches may include convolutional neural networks, transformer-based models, graph neural networks, embedding-based representations of spatial structure and uncertainty-aware spatial statistical frameworks. Particular attention will be given to interpretability, reproducibility and uncertainty quantification, ensuring outputs are scientifically rigorous and policy-relevant.
Through empirical case studies, the project will examine how alternative methodological pipelines shape health-related inference, including analyses of environment–health relationships, spatial vulnerability and inequalities in environmental conditions. In doing so, it will develop a transferable framework for designing and evaluating satellite-based health indicators across spatial contexts. By bridging satellite science, geospatial AI and public health, the research will advance spatial health analytics and support equitable, data-driven decision-making.
Training, Collaboration and Project Structure
The candidate will gain advanced expertise in satellite data processing, geospatial AI, spatial epidemiology and uncertainty quantification. Training will include work with large-scale Earth Observation platforms (e.g. Google Earth Engine) and Python-based deep learning frameworks (e.g. PyTorch, TensorFlow), providing highly transferable technical skills.
Year one will focus on methodological design, advanced training and dataset development, followed by model development, empirical evaluation and journal outputs in years two and three. The final phase will synthesise findings into a coherent methodological framework with clear policy relevance. The project offers an opportunity to work at the forefront of satellite-enabled health research, equipping the candidate with interdisciplinary expertise increasingly in demand across academia, government and the geospatial sector.