Probability of surface fire spread in Brazilian rainforest fuels from outdoor experimental measurements

• Autor(a) principal: Filho, Guenther Carlos Krieger
• Data de Publicação: 2017
• Outros Autores: Bufacchi, Paulo, Santos, José Carlos, Veras, Carlos Alberto Gurgel, Alvarado, Ernesto C., Mell, William, Carvalho, João Andrade [UNESP]
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1007/s10342-016-1023-2; http://hdl.handle.net/11449/170058
• topic: Brazilian rainforest; Flammability; Logistic regression model; Surface fire

This paper describes the development of a logistic model to predict the probability of surface fire spread in Brazilian rainforest fuels from outdoor experimental measurements. Surface fires spread over litter composed mostly of dead leaves and twigs. There were 72 individual outdoor experiments in eighteen sites. The fire propagated in 49% of the experiments. In each experiment, the litter height, litter temperature, unburned litter mass, wet and dry litter mass, soil temperature, wet and dry soil mass, ambient wind velocity, ambient air temperature, ambient air relative humidity and duration of fire spread were measured. Using these data, the rate of fire spread, litter bulk density, litter and soil moisture content, litter load and litter residue fraction were determined. For the sake of analysis, experimental results were classified into two groups: one for which the fire propagated and the other one for which the fire self-extinguished. Analyses of a logistic regression model showed that the relevant parameters for fire propagation are litter height and litter moisture content. Concerning the probability of successful fire propagation, the model showed a true positive rate of 71% and a true negative rate of 84%. The outdoor experiments also served to gather data to improve the understanding of surface fires and to provide input data for future computer simulations.