Modelagem INAR(p) para previsão de índices de qualidade do ar

Abstract

This work is based on the INteger AutoRegressive model (INAR) for air qualityindices modeling and forecasting in the Grande Vit ́oria region (RGV). The analysisperiod comprehended 1 january, 2007 to 19 march, 2007 (summer of 2007) and itobtained forecasts on 20 march, 2007 to 25 march, 2007. The area of Grande Vit ́oriapossesses eight air quality monitoring stations which supplies daily data about airquality in Laranjeiras, Carapina, Jardim Camburi, Enseada do Su ́a, downtown ofVit ́oria, Ibes, downtown of Vila Velha, and Cariacica. In Laranjeiras pollutants asCOandNO2were investigated, in Carapina, thePM10, in Enseada do Su ́a,PM10,CO, andNO2, in downtown of Vit ́oria,PM10andCO, in Ibes,PM10andSO2,in downtown of Vila Velha,PM10andSO2, and in Cariacica,SO2,NO2,O3andCO. For each pollutant belonging to determined station, seven possible models weregenerated. The choice of the best model was based on the criterion of automaticselection for models INAR (p) and theAICCINAR, that selects the best orderpforeach model. All forecasts for air quality indices were classified as GOOD, accordingto the CONAMA 03/90 resolution. However, based on OMS 2005 guidelines, theforecast of the pollutantSO2verified in downtown of Vila Velha on march 20, 2007exceeded, in concentration terms (μg/m3), the value of 20μg/m3for an average of 24hs.