Fire analysis in the Caatinga environment from Landsat-8 images, enhanced vegetation index and analysis by the main components

• Autor(a) principal: Silva Junior, Juarez Antonio
• Data de Publicação: 2021
• Outros Autores: Pacheco, Admilson da Penha
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Ciência Florestal (Online)
• Texto Completo: https://periodicos.ufsm.br/cienciaflorestal/article/view/43818
• topic: Vegetation Cover; Fires; Remote Sensing and Statistics; Cobertura Vegetal; Incêndios; Sensoriamento Remoto e Estatística

Fires generate negative environmental and socioeconomic impacts that directly and indirectly influence the Earth's regional and global climate changes. Forest fires and fires play a relevant ecological role as they affect the local biodiversity, soil properties and water supply. The Caatinga biome has a high level of degradation of human and natural activities, being extremely affected by fires that burn predominantly due to human activities. Remote orbital sensing, as it presents specific spatial, spectral and temporal characteristics, is an essential technological alternative in monitoring areas affected by fire on the Earth's surface. This work aimed to analyze, in a spatial, spectral and temporal scope, the behavior of a fire in a Caatinga environment from the multivariate statistical analysis of Landsat-8 Images data, Enhanced Vegetation Index and Analysis by theMajor Components. The quantification of characteristics of vegetation derived from the spectral index provides a better assessment of the physical condition of the earth's surface under the effects of fire. Remote sensing techniques and multivariate statistics were used to assess the spectral behavior of wildfires in the Caatinga biome. The results of the Kolmogorov-Smirnov Normality Test showed a significance level of 5%. The integration of the statistical methods of Simple Linear Regression and Analysis by the Principal Components enabled important diagnoses in the estimates and/or relationships between the random variables. The multivariate technique allowed 94% of the data variation to be assessed. The maps resulting from the tested methodology represent an important improvement in mapping the distribution of vegetation. This study generates indications for future scientific research related to the management of space concerning to vulnerability and recovery of vegetation landscapes from the semi-arid climate under fire situations generated by burnings.