Influência e análise da cobertura florestal na modificação do albedo com o uso de inteligência artificial e sensoriamento remoto

• Ano de defesa: 2019
• Autor(a) principal: Alba, Elisiane
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Recursos Florestais e Engenharia Florestal; UFSM; Programa de Pós-Graduação em Engenharia Florestal; Centro de Ciências Rurais
• Palavras-chave em Português: Linguagem R; Análise de tendência; Mata Atlântica; R linguage; Trend analysis; CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL
• Link de acesso: http://repositorio.ufsm.br/handle/1/20707

The effects of climate changes influence in different scale, necessitating meansures aimed at reducing of emission of polluting gases, as well as meansures mitigating for this problem. In this context, the albedo determines the fraction of incident solar radiation that returns to the atmosphere and therefore corresponds to a key parameter in the radiant energy budget of the earth. The aim of this study was to investigate the influence of forest cover on surface albedo variations in areas of the Atlantic Forest biome during the period from 1987 to 2017. In addition, using spatial data and new data processing technologies to identify land cover physiognomies and their relationship to surface albedo. The albedo was obtained by the Surface Energy Balance Algorithm for Land method, while the land use and land cover mapping was performed by the Maxver classifier algorithm, and four thematic classes were identified. Finally, albedo was crossed with thematic classes identifying the variation of albedo as a function of changes in land cover. Machine learning algorithms were used to identify different stands in the study area, which were also associated with albedo variation. Albedo modifications were identified by trend analysis considering a 30-year time series obtained from TM/Landsat 5 and OLI/Landsat 8 images in the summer period. The influence of forest cover on albedo was related through the normalized difference vegetation index. Surface albedo ranged from 6 to 22%, with 1987 concentrating higher albedo values than 2017, due to the lower vegetation concentration. The Support Vector Machine (SVM) algorithm presented the best results in forest stand identification, however, it did not differ significantly from the Artificial Neural Networks (ANN) algorithm. The stands of coniferous species presented lower albedo than the hardwood species. Implementation of forest cover reduced albedo by approximately 60%, while changes in the structure of this cover resulted in an increase of approximately 20% in albedo values. Thus, the study demonstrates that the native forest cover of the Atlantic Forest biome, as well as the forest stands have a large participation in the energy balance, being important for the local and regional microclimate maintenance. Albedo, in turn, is a potential variable applied in the study of several structural parameters in forest cover.