Validação de simulação para cenário futuro da microrregião da campanha ocidental, RS, Brasil

• Ano de defesa: 2018
• Autor(a) principal: Mello, Eliziane Pivoto
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• dARK ID: ark:/26339/001300000c0hg
• 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: Dinamica EGO; Modelagem; Paisagem; Bioma Pampa; Modeling; Landscape; Pampa biome; CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL
• Link de acesso: http://repositorio.ufsm.br/handle/1/18953

Dynamic landscape modeling has been used to simulate scenarios, quantifying and tracking changes in use and land cover through time. Based on these simulations, scenarios have been constructed projecting conversions into the future. However, there are no studies that validate these results insofar as the projected periods be achieved in the present. The goal of this research was to evaluate the effectiveness of a dynamic modeling that simulated future scenarios for the Campanha Ocidental Microregion of Rio Grande do Sul in order to validate or reject its prediction capacity. OLI / LANDSAT 8 satellite images generated real data for the year 2015, thus allowing comparison with data from simulations carried out in the past for the same period. The quantification and spatialization of the observed and simulated data for 2015 were tested at the pixel level by means of cross tabulation and Spatial Language analysis for Algebraic Geoprocessing in SPRING software, and at window level of pixels from 3x3 to 11x11, through of fuzzy logic in the Dinamica EGO software. The results obtained using Spring showed that, in a quantitative way, the simulated indexes overestimated native field areas in 4.74%, forest in 4.59% and sandy areas in 0.24%. For the agriculture and water classes, there was an underestimation of the data by 9.05% and 0.52%, respectively. Spatially, considering pixel level matching, the observed and simulated data were similar in 60.33% of the area and discordant in 39.67%. Considering the total covered by the reference area by class (real map for 2015), it was observed that the native field class presented the highest maintenance index, being correctly distributed in 68.52% of the study area, followed by forest in 59,60%, agriculture 52,46%, sandy areas 49,74% and water 49,44%. The similarity based on fuzzy logic of the EGO Dynamics presented indexes ranging from 0.8165 to 0.8469 for 3x3 and 11x11 windows, respectively, by exponential decay. Considering that the purpose of the simulation for 2015 was to investigate the coverage of land and its breadth, it is recognized of pixel window analysis is best suited on interpreting the expected patterns. Thus, the high indexes of fuzzy similarity by pixel windows validated the simulation for the Campanha Ocidental Microregion of Rio Grande do Sul. Such validation represents greater credibility for dynamic modeling, and the results may support appropriate environmental and economic management in the study area, especially to limit the gradual conversions of native fields to agricultural crops, projected to occur in the region.