EN FR
EN FR


Section: New Results

Modelling of environmental impacts and natural hazards

Numerical simulation of the 30–45 ka debris avalanche flow of Montagne Pelée volcano, Martinique: from volcano flank collapse to submarine emplacement

Participant : Anne Mangeney.

In collaboration with M. Brunet, L. Moretti, A. Le Friant, E.D. Fernandez Nieto, F. Bouchut.

We simulate here the emplacement of the debris avalanche generated by the last flank collapse event of Montagne Pelée volcano (30–45 ka), Martinique, Lesser Antilles. Our objective is to assess the maximum distance (i.e., runout) that can be reached by this type of debris avalanche as a function of the volume involved. This result provides new constraints on the emplacement processes of debris avalanches associated with these collapses which can drastically change the related hazard assessment such as the generated tsunami, in a region known for its seismic and volcanic risks.

Global sensitivity analysis and uncertainty quantification of on-road traffic emissions

Participant : Vivien Mallet.

In collaboration with R. Chen, V. Aguiléra, F. Cohn, D. Poulet, F. Brocheton.

Road traffic emissions of air pollutants depend on both traffic flow and vehicle emission factors. Global sensitivity analyses, especially the computation of Sobol' indices, were carried out for the traffic model and the air pollutant emissions. In the process, the traffic model was replaced by a metamodel, or surrogate model, in order to reduce the high computational burden. The results identified the most important input parameters. Furthermore, the uncertainties in traffic flow and pollutant emissions were quantified by propagating into the model the uncertainties in the input parameters.

Uncertainty quantification in atmospheric dispersion of radionuclides

Participants : Ngoc Bao Tran Le, Vivien Mallet.

In collaboration with I. Korsakissok, R. Périllat, A. Mathieu, D. Didier.

In collaboration with IRSN, we investigated the uncertainties of the atmospheric-dispersion forecasts that are used during an accidental release of radionuclides like the Fukushima disaster. In order to quantify the uncertainties, Monte Carlo simulations and calibrations were carried out and coupled with ensemble meteorological forecasts from the European Centre for Medium-Range Weather Forecasts.

Simulation of air and noise pollution at high resolution and large scale

Participant : Vivien Mallet.

In collaboration with C. Pesin, P. Béal.

We developed fast surrogates for urban pollution models that they can be applied at global scale while preserving the street resolution, the main physical constraints and the performance against observational data. The surrogate models are based on the original models, machine learning algorithms and observational data.