clim4health

Project background

According to the IPCC, heatwaves are becoming more frequent and intense across Europe, with serious consequences for human health, particularly in cities where the Urban Heat Island effects may amplify impacts, while cold extremes remain a persistent contributor to excess mortality, particularly in vulnerable and energy-poor populations. These risks remain insufficiently represented in operational climate services and health policies.

International policy frameworks such as the Paris Agreement, the Global Goal on Adaptation, and the WHO–WMO Joint Programme on Climate and Health (https://www.who.int) highlight the need for actionable, locally relevant climate information to support early warning systems, preparedness, and adaptation planning. Yet, most existing heat- and cold-health warning systems rely on coarse spatial data, typically at city or regional scale, limiting their ability to identify neighbourhood-level vulnerability, prioritise targeted interventions, or address health inequities within cities. This gap undermines effective implementation of climate adaptation policies at the municipal level and constrains uptake of initiatives such as Destination Earth (DestinE) (https://destination-earth.eu) for decision support.

Science challenges

Major science challenges remain in linking climate hazards to health outcomes in a consistent, scalable and operational manner. These include the limited availability of long‑term, high‑resolution urban temperature data; insufficient integration of Earth Observation, reanalysis, and climate projections with epidemiological datasets; and a current lack of harmonised indicators for both extreme heat and cold, as highlighted by GCOS and WCRP research priorities (https://gcos.wmo.int; https://www.wcrp-climate.org).

Addressing challenges via clima4health

CLIM4health addresses the above challenges by developing EO-driven, AI-based climate–health risk algorithms that downscale temperature exposure to neighbourhood scale and link it with multi-decadal health records. This addresses the 3 of the European Commission's European Disaster Resilience Goals (EU-DRG), namely: anticipate, prepare and alert by offering enhanced resolution, precision and complete coverage of urban domains.

The platform will provide data-driven indicators on heatwave and cold extremes, facilitating risk mapping and assessment, as well contributing to enhance early warning systems at regional-to-local scales.

Group Use Cases

• Group 1: in Lisbon and Oporto, Portugal, we will be generating daily time series of Heat and Cold-related extremes, derived from the ML model outputs, downscaled to a 200x200m scale, replicating the methodology developed in CLIM4cities. This data will then serve as a predictor for Incidence Risk Models at the Municipal and Parish levels, using a 2000-2018 database of occurrences.
• Group 2: in Copenhagen and Odense, Denmark, the goal is again to deliver two (2) dose-response Heat and Cold-related impact indicators by downscaling hazard exposure and collocating with health impact data. The level of detail shall be similar to that of Use Case Group 1, i.e., the T2m downscaled to a 200x200m scale, replicating the methodology.

Project plan

CLIM4health is designed to respond directly to well-established user and policy needs for integrated urban weather, climate and health services, supporting human health impact prediction, early warning and climate adaptation at the local scale. The project addresses key scientific challenges related to sub-municipal interventions, such as the limited spatial resolution of existing climate–health indicators, gaps in urban exposure information, and the spatio-temporal incompleteness of weather and climate datasets.

The proposed approach integrates Earth Observation (EO), numerical weather prediction (NWP), reanalysis and in-situ observations through advanced Machine Learning (ML) and Artificial Intelligence (AI) techniques, ensuring both increased spatial resolution and improved accuracy of urban temperature exposure estimates. These are then linked with long-term health datasets to derive r climate–health risk algorithms suitable for operational uptake.

The work breakdown structure is organised around the following core activities:

• Climate scenario integration: Implementation of a DestinE-compatible coding framework to generate urban-scale climate change scenarios, linking Climate Digital Twin outputs with future population projections and health outcomes.
• Epidemiological validation: Systematic review and benchmarking of data and methods to ensure fitness for purpose for public health users, including transparent assessment of strengths and limitations.
• Scientific excellence and dissemination: Production of peer-reviewed communications and high-impact journal publications to contribute to best practices in EO- and ML-based urban climate modelling.
• Methodological robustness: Rigorous uncertainty management, including satellite data harmonisation, quality control of crowdsourced observations, cross-validation strategies, explainable AI diagnostics, and dedicated validation of extreme heat and cold events.
• Open science and interoperability: Adoption of FAIR principles, CF-compliant metadata standards, and open dissemination of data, code and documentation.
• Stakeholder engagement and uptake: Active liaison with ESA CCI ECV projects, WMO and GEO networks, and targeted communication towards scientists, decision-makers and citizens.