Jorge Castel's thesis defense

Abstract:

Fire activity in the French Mediterranean region is influenced by meteorological conditions, as previous research has shown. Human factors play a crucial role due to high population densitý and land use and occupation. However, the combination of all fire factors has persistent limitations. Identifying the climatic and non-climatic factors governing fire regimes poses many challenges in southeastern France. The aim of this thesis is to define fire danger indicators for Mediterranean France that can be used by scientific and operational players. Three specific objectives are defined to meet the project's main objective. Firstly, to identify explicit spatio-temporal fire drivers by testing their contribution to recent fire regimes in south-east France. Secondly, to examine whether meteorological data, other than those currently used by operational services, can improve the prediction of fire activity on a daily scale. Thirdly, to determine whether optical remote sensing data can complement or replace meteorological and/or landscape variables in fire forecasting.  

The study integrates non-climatic variables into the Firelihood modeling framework, a Bayesian model of fire activity that incorporates spatio-temporal components. Then, using a new evaluation method, fire weather variables are analyzed as predictors of fire activity. Finally, new models, integrating optical remote sensing data, measure the contribution of fire weather, remote sensing and landscape variables to fire forecasts. The results show that adequate fire-weather indices improve fire activity forecasts, but spatio-temporal effects are essential to guarantee accurate forecasts. Several landscape factors contributed as much as fire weather indices to the distribution of fire occurrences. Optical remote sensing is particularly useful for forecasting large fires. Temporal fire patterns are mainly influenced by changes in fire weather and unexplained effects. These results will help to improve short- and long-term fire danger forecasting, refining operational hazard classes and creating specific regional models.

Members of jury :

Vincent MORON                   Professor – AMU                                                       President of jury

Florent MOUILLOT             Research Director – IRD                                            Reporter

Giada ADELFIO                     Professor  – Università degli Studi di Palermo   Reporter

Imma OLIVERAS MENOR  Research Director – IRD                                           Reviewer

Maxime SOMA                     Research engineer                                                    Reviewer

Jean-Luc DUPUY                   Research Director – INRAE                                     Thesis supervisor

François PIMONT                Research engineer – INRAE                                    Thesis co-supervisor - Guest

Thomas OPITZ                      Research Director – INRAE                                     Thesis co-supervisor - Guest