Características temporais da concentração de material particulado na atmosfera da cidade de São Carlos - SP

• Ano de defesa: 2009
• Autor(a) principal: Pozza, Simone Andréa
• Orientador(a): Coury, José Renato
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Engenharia Química - PPGEQ
• País: BR
• Palavras-chave em Português: Ar - poluição; Análise de séries temporais; Material particulado; São Carlos (SP)
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA QUIMICA
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/3860

The city of São Carlos, of medium size of São Paulo State, can be considered as representative of the Brazilian Southeast area, with agricultural activities practically equivalent to the industrial ones. As part of that profile, it possesses a series of sources of atmospheric pollution that include a considerable fleet of vehicles, agricultural burns and several industrial emissions. The time evolution of this environmental pollution has been reason of concern, once it can cause damages to the human health and to the environment. The present work studied these time characteristics, specifically for the case of particulate matter (PM) suspended in the atmosphere, from experimental data and through statistical models. The concentrations of PM10, PM2.5 and PM10-2.5 in the city of São Carlos were collected and analyzed. The period of study comprises the years from 1997 to 2006 for PM10 and from 2001 to 2006 for PM2,5 and PM10-2,5. Monthly averages were used for obtaining the models and the 30 (thirty) or 6 (six) final data were used for the validation of the forecast. The selection of the best model was made considering the smallest value of AIC (Akaike Information Criterion), through the successive variation of the parameters. Of the 10 models with smaller value of AIC, was chosen the one of smaller value and other with a smaller number of parameters. In the statistical analysis, the models that did not comply with the requirements of the Shapiro-Wilk tests (normality) and of steadiness were eliminated. In the three cases the model SARIMA represented the groups of studied data better, when compared with the Holt-Winters correlation. The model found for the series of data of PM10, was the SARIMA(1,0,0)×(1,0,1)6. In the second case, the concentration of PM2.5 the best model was the SARIMA(2,1,3)×(1,0,1)6. In the third case, the concentration of PM10-2.5, the best model was the SARIMA(2,0,2)×(1,0,0)6.