Modelos de estimativa de radiação solar global e análise Estatística em dez cidades do Estado do Pará – Brasil

Abstract

Estimating solar radiation is important for various segments of society such as agriculture, meteorology and electricity generation (photovoltaic systems). This workaims to estimate the Global Solar Radiation in ten cities in the state of Pará - Brazil, using estimation models, Hargreaves-Samani and Bristow-Campbell, using meteorological data provided by Automatic Meteorological Stations, belonging to the National Institute of Meteorology (INMET), to compare the observed and estimated values and analyze which model best fits the region, through the parameters Coefficient of Determination (R²), Willmott Index (d), Pearson correlation coefficient (r) andRelative Root Mean Squared Error (rRMSE). The results showed that the models are satisfactory and can be used to estimate the Global Solar Radiation, however for the less rainy period there was an overestimation, between june and november. The Castanhal station had the lowest average global solar radiation compared to the other stations with an average daily radiation equal to 15.06 M/m².day, while the city of Redenção had the highest daily average, 18.33 MJ /m².day. The city of Tomé Açu presented better fit of the models referring to the observed values, since for the HS model, the confidence index was considered very good, c equal to 0.78, and the relative RMSE was considered excellent, with rRMSE equal to 9.93%. For the B-C model for this station, the confidence index and the relative RMSE were considered good, c equal to 0.71 and rRMSE equal to 19.14%.