Outils pour utilisateurs

Outils du site

Panneau latéral

Liste de liens :


Task 1.1 - Characterization and spatial distribution of fuel variables

Goal: Derive fuel characteristics and distribution from operational surface model.

SPE (Perez Y, Filippi B, Ferrat L) With CNRM (Lemoigne, Masson) T6 to 24

The objective here is to obtain a high resolution map of fuel variables (with daily and intraday evolution) and fuel type based on a combination of IGN BDForet (for the combustible) and vector data maps (non-combustible, urban and water areas). The target is 50m and 3h frequency resolution for the combustible variables (of use for the risk estimation and weather model), with non combustible (roads, lakes, sea, rocks,…) vector based data used as a non-combustible mask for the on-demand higher resolution required to precisely track fire fronts.

Forefire fuel (Filippi et al. 2013) variables, as particle heat content or vegetation quantity and humidity, that evolve in time, are not available currently. The Surfex (Masson et al. 2013) system, used in operational Météo-France models, is able to describe the evolution of soil and vegetation related parameters at 1-km resolution, worldwide. Satellite soil humidity and temperature data over Europe is assimilated into the SURFEX model (Barbu et al. 2014), allowing to influence plant stress differently in time and space. During this task, the link will be made between fuel model and Surfex vegetation types (Faroux et al. 2013) defined at 1km resolution. New parameterizations or methods will be developed within Surfex platform to estimate variables that are not readily available, such as vegetation humidity from soil humidity, atmospheric conditions and plant modeling. The potential risk comes from the fact that the vegetation evapotranspiration simulated by Surfex corresponds to a live vegetation, since Surfex does not account for dead vegetation, whereas the fuel model distinguishes between dead and live fuels. There comes therefore the need to link this live fuel to the type and state of the vegetation. A modelling approach based on Surfex Isba-Ags model (Calvet et al. 1998) could be considered. This model aims at simulating the life cycle of the active biomass. In a second step, a downscaling method will be developed to go from 1km-gridded output to a 50m resolution that is more relevant to fire risk and spread simulation (see figure). Such downscaling will use the ability of the Surfex surface model to compute simultaneously 4 humidity levels (basically for Coniferous, Broadleaved, Small trees and Grass classes) within each 1-km resolution grid point. With the higher resolution of the BDForet, composed of 32 vegetation types, the actual spatial distribution of these four classes is known within this grid point. Vegetation humidity and state will then be disaggregated towards this high resolution (50m) by matching for each class (within four) the corresponding vegetation types (within 32). This will generate the high resolution map, at the same frequency as the Surfex surface model. Finally, an initiative is conducted at CNRM to improve the horizontal resolution of the vegetation maps. A new classification based on ESA-CCI Land Cover products will be derived and a 300-m resolution reached at the end of 2017. This can be a substantial improvement as compared to the existing Surfex vegetation classification. High resolution maps will be tested in this task on 5 fire reanalysis cases, comparing vegetation states to the actual measurements and fire behaviour on the day of the fire.

Overview of the super sampling efforts required to pass from the SURFEX model up to the wildfire simulation resolution.


             1.1.1 (T12) Code to match fuel characterization variables (from SURFEX). 5 cases will be selected and tested with standard 1 km resolution surfex data.
             1.1.2 (T24) Code related to the fuel variables spatial distribution on same cases.
             1.1.3 (T24) Report (journal publication of results).

Resources: A 1.5-year Pdoc will be devoted to this task. Computer resources of CNRM and SPE will be used. Fuel PDOC Risks: The risk associated to this task is limited since laboratories involved have experience in surface and fuel modeling. SURFEX is already operational; at most the model will have degraded resolution.

Task 1.2 - Test and validation of the method at the national level

Goal: Evaluate the fuel model over the whole south France for past meteorological situations.

CNRM (Lemoigne, Masson,) With SPE (Filippi) and Météo-France testing T18 to 36

The method developed in Task 1.1 will be evaluated over the whole South France for past meteorological situations. The simulation period will cover the years 2006/2016 (max). The Surfex simulations will be forced by meteorological analyses, including meteorological and satellite information. The fuel characteristics simulated by Surfex will be validated directly against the in-situ measurements and inventories already available from the Forest French National Institute (ONF) and field campaigns. Then, for each fire recorded in the Promethee database (publicly available) and a selection from EFFIS (European system, in the steering committee), the method will be tested as a parallel product by Météo-France South-West (quantity of false positives high risks; false negatives compared to old system).


              1.2.1 (T30) One year fire reanalysis.
              1.2.2 (T36) Report (journal publication of results).

Resources: Computer resources of CNRM and GENCI will be used. Risks: The risk associated to this task is limited since laboratories involved have experience in the preparation of scripts to run numerical simulations for long time periods. Nevertheless, the reanalysis and availability of model might delay results while not blocking the whole project.


Task 1.3 – Fuel smoke-plume composition

Goal: Field experimental burns in real conditions to provide detailed chemical speciation of pollution from a real vegetation fire.

LA (Leon, Mari, Liousse) with SPE (Cancellieri , Ferrat) T18 to 36

Impairment of visibility and air quality degradation near and downwind of vegetation fires are controlled by the emissions of aerosols and gases during the combustion processes. Direct measurements of Mediterranean wildland fire plume composition are very rare due to difficult access to regions near the fire front and plume. Assessing these emissions from real fires is crucial to estimate smoke exposure and health effect. Two prescribed burnings will be organized to obtain emission profiles for gases and aerosols for the improved coupled fire-atmosphere ForeFire/Meso-NH model (Task 2.3). The experimental strategy is based on in-situ mobile and fixed instrumentation to derive the aerosol and gaseous composition. In-situ measurements will be operated to estimate the fuel mass loss (with a mass sensor in the fuel) and the plume composition (FTIR, filters,…) including volatile organic carbon speciation, nitrous oxides, elemental and organic carbon (soluble and insoluble), CO and CO2 at different stages of the fire. Emission profiles for gases and particles will be calculated from emission factors based on CO/CO2 methodology and fuel mass estimates. Emission models for other chemical species will be derived from the CO and CO emission profiles and direct measurements when available.


             1.3 (T36) Data and report on gas and particle aerosols composition and emissions profile for open air vegetation fire.

Field campaign: UAV with multi-gas sensor (O2, H2S, CO and CH4), FTIR (volatile), mass sensor: (Cancellieri, Ferrat). Aerosol composition, properties: filters, aethalometer, particle counter, low pressure impactor (distribution between 0.03 and 10 µm): (Leon, Liousse). Risk: UAV not funded (separated project) or not operational for the field campaign, extremely bad weather conditions. With no UAV, standard sampling technique (pipe) will be used with gas and aerosol sampling probes at different altitudes in the smoke plume. If the weather is not clement with experiments, burns will be performed in large combustion chambers, with limited validation scope.

fuel.txt · Dernière modification: 2020/01/16 12:05 par