MODELO DE TRANSMISSIVIDADE ATMOSFÉRICA: UM ESTUDO DE CASO PARA O MUNICÍPIO DE PETROLINA-PE

Abstract

The aim of this study was to develop and propose statistical models to estimate atmospheric transmissivity under conditions of clear and partly cloudy sky in the city of Petrolina, PE, Brazil. The new models were developed for simple and multiple linear regressions, with a database classified into nine days of clear skies and 9 days of partly cloudy sky, predominantly clear sky, in 2010. The input variables to fit the models for calculating the atmospheric transmissivity were actual atmospheric transmissivity, precipitable water, atmospheric pressure, zenith angle and atmospheric turbidity. The statistical software STATISTICA was used to fit the models. The meteorological data: solar radiation (W m2), direct radiation (W m2), atmospheric pressure (mb), relative humidity (%) and air temperature (° C) were used as input variables of the models, obtained by the minute on weather station Petrolina. The best model for the calculation of atmospheric transmissivity varied by day study. For a clear day or low cloud interference, the lowest average relative differences were obtained with the Reg.M2 model of 1.92% and 2.91%, however, the Reg.S3 model, where the zenith angle is the only input variable, becomes a more affordable model to find the atmospheric transmissivity on a clear day, with an average of 6.45% on difference and 5.26%. For days with clouds interference Reg.M1 both the model and the model obtained in Allen had the lowest average difference relative 0.58% and 2.29% and 1.83% and 1.68%, respectively.