In recent years the availability and quality of UV radiation data have improved appreciably. However, high quality measurements from intercalibrated ground-based spectrometers are available at only a few sites worldwide. Satellite derived UV irradiances, on the other hand, offer the advantages of global coverage and avoid the problems of instrument intercalibrations when investigating geographic differences in UV. However, with the satellite sensors in common use, the retrievals can be subject to errors due to variability in the distribution of ozone and aerosols in the lower atmosphere. Previous studies have shown good agreement between satellite-derived UV and ground-based measurements at pristine locations, but with increasing overestimations of the surface UV at more polluted sites, which are characterized by larger concentrations of ozone and aerosols in the troposphere. Consequently, the contrast in UV
between pristine locations and more polluted locations may be underestimated by the satellite retrievals. Here we investigate the relative contribution to these differences due ozone profile differences and aerosol extinction. It is found that both factors contribute to the errors, but that the aerosol effect dominates. Tropospheric aerosols can result in satellite overestimations exceeding 30% in populated regions. Large concentrations of tropospheric ozone also lead to satellite overestimations. Firstly, the total ozone column would be underestimated, and secondly, the underestimated component is disproportionately important because of the increased path length and warmer temperatures in the troposphere. If tropospheric ozone were less than expected, as in more pristine locations, then there would be a tendency for satellite-derived UV to be too large by up to ~ 5%.