|
> > Workshops > Workshop 7
Workshop 07 - 09:00 am to 12:30 am
Natural and technological risks
Chaired by Laurent Boisnard, Sous-Directeur Missions et Données d’Observation de la Terre & Applications, CNES and Gilles Grandjean, Directeur de programme, BRGM
Accounting for uncertainties in air pollutant dispersion predictions in urban areas
By Mélanie Rochoux, Chercheuse géophysicienne, CERFACS
Air quality can be severely degraded by wildland fires, volcanic eruptions and even industrial accidents, which release substantial amount of toxic gas and particles into the atmosphere, which are then carried and dispersed by the wind. In urban areas, buildings have a blocking effect on the flow, which can induce localized pollution peaks and expose people to toxic doses. Models that resolve the interactions between the atmospheric boundary layer and the built environment are useful to obtain an accurate map of the spatial variability of pollutant concentrations. At Cerfacs, we have designed and validated a modeling workflow that takes advantage of large-eddy simulation ensembles to represent this spatial variability, estimate modeling uncertainties and characterize the envelope of plausible dispersion scenarios at the scale of an event. This system relies on statistical learning methods to build a statistical emulator of the physical model response and then produce a robust uncertainty quantification.
|
Biography: Mélanie Rochoux is a geophysicist, expert in micrometeorological modeling issues, with a focus on wildfires and air pollutant dispersion. She obtained her PhD at the École Centrale Paris in 2014. She has been a researcher at CECI, a joint CERFACS/CNRS research unit, since 2015, working on detailed modeling of atmospheric boundary-layer processes, uncertainty quantification and their reduction through observational data assimilation. |
|
|