Ana Patrícia Pires Marques Oliveira CTO for Space Applications, +Atlantic, CoLAB Atlantic, Portugal.

Title: From Citizen Science, Machine Learning and Earth Observation towards Urban Climate Services

Abstract: Climate change is reshaping patterns of human exposure to extreme weather, with profound consequences for health and urban resilience. Globally, rising temperatures are driving more frequent, and more intense heatwaves. In cities, these trends are further amplified by Urban Heat Island (UHI) effects, where modified urban surfaces intensify nocturnal heat retention, exacerbating thermal discomfort and contributing to excess mortality, increased energy demand, and wider socio-economic impacts. Understanding and anticipating these risks requires high-resolution environmental intelligence derived from Earth Observation (EO), climate modelling, reanalysis systems, and emerging Digital Twin of the Earth (DTE) capabilities. This work outlines how EO-driven innovation can strengthen climate–health risk assessment through three objectives: (1) assessing the role of EO in monitoring climate variables relevant to health; (2) exploring how DTE, reanalysis, and projections enhance understanding of climate–health linkages; and (3) demonstrating the application of these datasets in supporting adaptation and early warning. Within this context, CLIM4health—an ESA-funded project under the CLIMATE-SPACE activity—develops EO-informed, AI/ML-based Extreme Heat and Cold Climate–Health Risk Algorithms at 200 × 200 m resolution for four Functional Urban Areas (FUAs) in Portugal and Denmark. These indices integrate downscaled hazard exposure with health records to predict excess mortality and morbidity at sub-municipal scale. The methodology builds on a wider portfolio of initiatives. These include CLIM4cities, which demonstrated generalisable ML downscaling of reanalysis data across Denmark and the Urban Heat-Health Forecasting Service for ECMWF, which adapts data-driven downscaling for DestinE’s Extremes Digital Twin, as well as the Horizon Europe TerraDT project, which extends these methods to downscale climate projections and assess adaptation scenarios.