Improving Subkilometer Modelling with GEM : case of Paris

Urban-scale modelling has been performed at ECCC for experimental subkilometer systems for specific events. For example, the model was run realtime with grid spacing down to 250m over Toronto, Canada, in the context of 2015 PanAm sport games, and with grid spacing down to 100m over Paris in the context of 2024 Olympics games WMO-Research and Demonstration Project. 

The purpose of this study is to continue to explore the benefits and limitations of subkilometer modelling with the Global Environmental Multiscale (GEM) model by taking the opportunity of recent multi-source experimental datasets in the Paris region. For selected case studies of conditions favorable to the presence of strong UHI and thunderstorms, diagnostics are provided on the impact of the resolution on different surface and atmospheric indicators, such as the temperature heterogeneity and UHI intensity, the urban boundary-layer height and the respective contribution of the resolved and subgridscale turbulence.   

Sensitivity modelling experiments are conducted to progress on the collection of evidence of the importance of some elements of the configuration with 100m grid spacing. First, the robustness of the dynamical core is assessed by varying the time step in order to optimize the computational time. Second, the impact of the shallow convection scheme is investigated. The impact of the method for the description of the urban canopy is then highlighted. Method for the computation of surface energy budget and near-surface diagnostics is revised for dense built-up areas to compensate for the lack of advection in the street. Finally, a more advanced vegetation scheme is tested to improve the city’s surface energy budget but also the surrounding area’s temperature values encountering a persistent warm bias.  This study will help to produce multi-purpose reliable weather and environmental urban prediction including air quality and evaluation of urban adaptation scenarios.