Systematic deviations between surface solar radiation obtained from ground site radiometers and IGMK satellite model predictions for Brazil between 1985 and 1986 were evaluated by using statistical tests. The relatively high mean bias error of 1.2 MJ.m-2 was originally attributed to a large off-nadir position of the satellite relative to Brazil. Some of the ground stations, however, presented deviations that could not be entirely accounted for by this explanation. In the present work we applied statistical tests to the results and found that the altitude and the size of the city where the ground station is located can explain these deviations, with a 99% confidence level. The tests have shown that the mean bias deviation is 4 times higher for validation sites located above 700 meters than for sites below this altitude. Furthermore, mean bias found for the larger cities is also 4 times higher than the value found for sites located in small cities and remote areas.
Solar radiation; Solar energy; Radiative transfer; Atmosphere; Aerosols
