From Cities to Countries: High-Resolution (100m) Climate Services Supporting Early Warning Systems

The ‘Early Warnings for All’ initiative provides a framework for multi-hazard warning systems that aim to protect people from the negative consequences of environmental events. Due to climate change, we see these events occurring more frequently and with higher magnitudes than before.

One of the main bottlenecks in early warning systems is the lack of high-resolution meteorological information, restricted by the mesoscale resolution of most climate models. This impedes the correct representation of e.g. temperatures and heat stress in cities, which can be significantly higher compared to rural environments (the so-called Urban Heat Island effect). Traditional models require multiple nesting steps and are therefore often not suited for early warning management systems. The UrbClim model partially fills this gap by providing fast and reliable meteorological and climatological information at resolutions of up to 100m. The main bottleneck remains in the limited spatial domains (which are usually limited to the size of a city) at which the UrbClim model operates.

To tackle this issue, an AI-based model is designed, based on a 2D Neural Network, leveraging Copernicus ERA5 reanalysis drained and validated with UrbClim simulations. This model provides instant high-resolution (100m) meteorological information (temperature, humidity and heat stress) at daily and hourly frequencies for spatial domains extending to sizes of full countries. A modular Python package underpins the workflows, enabling automated data retrieval, processing, and integration into operational environments. The data feeds in directly in two operational climate services: i) vector borne disease modelling in Belgium & (ii) heat-health early warnings in the Arabian Peninsula. The added value of high-resolution and urban-resolving physics will be demonstrated on both operational and long-term time scales, showcasing its effectiveness in supporting decision-making by regional and federal (health) authorities on both short-term (e.g. issuing warnings) and long-term time scales (e.g. urban planning).