FLAME 1.0 - a novel approach for modelling burned area in the Brazilian biomes using the Maximum Entropy concept

Aug 27, 2026

Upload by Maeve

As fire seasons in Brazil extend and intensify, improving fire simulations and clarifying the drivers of burning is essential. Determining which factors operate across different Brazilian biomes remains difficult due to a highly uncertain link between drivers and fire outcomes. FLAME (Fire Landscape Analysis using Maximum Entropy) combines Bayesian inference with Maximum Entropy to quantify uncertainty. It models fire using bioclimatic, land cover, and human driving variables, assessing burned area for ALL, NAT, and NON fire categories.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages47

Reading Time1 h 58 min

AccessFree to download

Abstract

Introduction

Challenges in identifying fire drivers

Quantifying uncertainty and relevant modelling frameworks

Maximum Entropy principle (MaxEnt)

[https://doi.org/10.5194/egusphere-2024-1775](https://doi.org/10.5194/egusphere-2024-1775)[ ](https://doi.org/10.5194/egusphere-2024-1775)Preprint. Discussion started: 30 August 2024 􀀍c Author(s) 2024 . CC BY 4.0 License. 1 FLAME 1.0 : a novel approach for modelling burned area in the 2 Brazilian biomes using the Maximum Entropy concept 3 Maria Lucia Ferreira Barbosa 1,2, *, Douglas I Kelley3, Chantelle A Burton4, Igor J M 4 Ferreira1,5, Renata Moura da Veiga 1, Anna Bradley4, Paulo Guilherme Molin², Liana O 5 Anderson6 6 7 1 National Institute for Space Research-INPE, Avenida dos Astronautas, 1758. Jd. Granja- 8 São José dos Campos-São Paulo, 12227-010, Brazil 9 2 Federal University of São Carlos, Rodovia Lauri Simões de Barros, km 12-SP-189- 10 Aracaçu, Buri-São Paulo, 18290-000, Brazil 11 3 UK Centre for Ecology and Hydrology, Wallingford. OX10 8BB UK 12 4 Met Office Hadley Centre, Fitzroy Road, Exeter. EX1 3PB UK 13 5 Faculty of Environment, Science and Economy, University of Exeter, Exeter, UK 14 6 National Centre for Monitoring and Early Warning of Natural Disasters-Cemaden, Estrada 15 Doutor Altino Bondensan, 500-Distrito de Eugênio de Melo, São José dos Campos-São 16 Paulo, Brazil 17 18 * Correspondence to: Maria Lucia Ferreira Barbosa ([malucsp@gmail.com](malucsp@gmail.com)) 19 20 Abstract 21 22 As fire seasons in Brazil lengthen and intensify, the need to enhance fire simulations and 23 comprehend fire drivers becomes crucial. Yet determining what drivers burning in different 24 Brazilian biomes is a major challenge, with the highly uncertain relationship between drivers 25 and fire. Finding ways to acknowledge and quantify that uncertainty is critical in ascertaining 26 the causes of Brazil’s changing fire regimes. We propose FLAME (Fire Landscape Analysis 27 using Maximum Entropy), a new fire model that integrates Bayesian inference with the 28 Maximum Entropy (MaxEnt) concept, enabling probabilistic reasoning and uncertainty 29 quantification. FLAME utilizes bioclimatic, land cover and human driving variables to model 30 fires. We apply FLAME to Brazilian biomes, evaluating its performance against observed data 31 for three categories of fires: all fires (ALL), fires reaching natural vegetation (NAT), and fires 32 in non-natural vegetation (NON) . We assessed burned area responses to variable groups. The 33 model showed adequate performance for all biomes and fire categories. Maximum temperature 34 and precipitation together are important factors influencing burned area in all biomes. The 35 number of roads and amount of forest boundaries (edge densities), and forest, pasture and soil 36 carbon showed higher uncertainties among the responses. The potential response of these 37 variables displayed similar spatial likelihood of the observations given the model, between the 38 ALL, NAT and NON categories. Overall, the uncertainties were larger for the NON-category, [https://doi.org/10.5194/egusphere-2024-1775](https://doi.org/10.5194/egusphere-2024-1775)[ ](https://doi.org/10.5194/egusphere-2024-1775)Preprint. Discussion started: 30 August 2024 􀀍c Author(s) 2024 . CC BY 4.0 License. 39 particularly for Pampas and Pantanal. Customizing variable selection and fire categories based 40 on biome characteristics could contribute to a more biome-focused and contextually relevant 41 analysis. Moreover, prioritizing regional-scale analysis is essential for decision-makers and fire 42 management strategies. FLAME is easily adaptable to be used in various locations and periods, 43 serving as a valuable tool for more informed and effective fire prevention measures. 44 45 Keywords: Burned Area. Brazilian biomes. Maximum Entropy. Bayesian Inference. Climate. 46 Fragmentation. Land Use. 47 48 49 1 INTRODUCTION 50 51 The complexity of the interactions and feedbacks between fire, climate, people, and other earth 52 system components makes it challen

FAQs

What problem does FLAME address in Brazilian fire research?

FLAME targets the challenge of identifying and quantifying fire drivers across Brazilian biomes when the relationship between drivers and fire is highly uncertain.

How does FLAME handle uncertainty when modelling burned area?

FLAME integrates Bayesian inference with the Maximum Entropy (MaxEnt) concept to enable probabilistic reasoning and uncertainty quantification.

Which fire categories and inputs does the model use for evaluation?

FLAME is applied to Brazilian biomes and evaluated for ALL, NAT, and NON fire categories using bioclimatic, land cover, and human driving variables.

Aug 27, 2026

Upload by Maeve