Live fuel moisture content estimation using remote sensing and numerical modelling approach

Abstract

Climate change has led to an increase in wildfires, particularly on the Iberian Peninsula. In recent years, Portugal suffered several devastating wildfires, such as in 2003, 2005, and 2017. In 2024, the situation repeated itself, with several large-scale wildfires occurring in the Central region during September. Wildfires are directly related to the availability of combustible material, weather conditions, and ignition factors. Currently, land use and occupation management is a way to reduce wildfires. Identifying areas with higher fuel availability is essential for wildfire prevention in this context. This work aims to improve the live fuel moisture content (LFMC) representation across mainland Portugal, using remote sensing and numerical modelling through a machine learning approach. First, a product was developed to estimate LFMC for Portugal based on remote sensing, through satellite imagery, and machine learning (LFMC-RS). Next, numerical simulations were performed with the MESO-NH (research) and AROME (operational) models, non-hydrostatic mesoscale atmospheric models, producing forcing files to initialize the SURFEX surface model. All output variables from the SURFEX model were utilized as predictors in a machine learning classifier to estimate the LFMC (LFMC-SFX). These results are useful for understanding the FMC spatiotemporal variability in Portugal The research received financial support from the Foundation for Science and Technology, I.P. (FCT) through the PyroC.pt initiative (Ref. PCIF/MPG/0175/2019) along with a PhD Grant (2022.11960.BD). This work also was cofounded by the European Union through the European Regional Development Fund (FEDER) in the framework of the Interreg VI-A España-Portugal (POCTEP) 2021–2027, FIREPOCTEP+ (0139_FIREPOCTEP_MAS_6_E).