High-resolution forecasting models for predicting and capturing heat exposure contrasts in the Paris region

The intensity and spatial extent of urban heat island fluctuates nightly, resulting in variable heat exposure across neighbourhoods. During heatwaves, this variability can significantly affect the degree of temperature exposure experienced by different communities. To accurately evaluate the health impacts of heatwaves, it is essential to account for both the spatial distribution of heat exposure and the socioeconomic factors influencing vulnerability.

The French heat warning system relies on alert thresholds being exceeded at a reference weather station in each department (administrative division). This applies to the eight departments in the Paris region, with contrasts in the representativeness of the station in terms of the exposure conditions for the 12.3 million inhabitants (depending on department size and variability of the surface properties). In particular, the city of Paris uses a reference station located in a park, which may not adequately reflect the detailed spatial variability of urban heat exposure. Numerical weather prediction (NWP) models, such as AROME-France at 1.3 km resolution, represent the current operational status for weather forecasting and have potential as early warning tools for heatwaves, not only at the regional scale but also at the neighbourhood level.

In this study, we compare AROME-France with two higher-resolution models: AROME-500m, which includes enhanced urban representation, and Meso-NH, a research model with 300 m resolution and the most detailed urban parameterisation. By analysing the summers of 2022 and 2023 in the Paris region, we investigate whether numerical models with different levels of detail better reflect population heat exposure than sparse observations. These findings would highlight both the limitations and the potential of current models for further improving heatwave risk assessments.