Abstract

The radiative transfer models allow calculating the spectral UV irradiance from some set of measured input quantities linked with the surface reflectivity, the solar zenith angle, the ozone column and the characteristics of clouds and aerosols. The spectral irradiance yielded by a model is influenced by errors in the measurement of the input quantities. In this paper, the influences of these errors are characterized and compared with other systematic effects through an uncertainty analysis. We evaluated the uncertainty of the spectral UV irradiance rendered by the UVSPEC model, under cloudless sky conditions. In order to express the uncertainty of the output quantities (the global, direct and diffuse irradiances) in terms of the standard uncertainties of the input quantities, we used a Monte Carlo-based uncertainty propagation technique. We found that the uncertainty of the irradiance in the UV-B part of the spectrum was strongly influenced by the uncertainty attributed to the ozone column datum. Moreover, the uncertainities associated with the aerosol parameters accounted for most of the UV-A global irradiance uncertainty; the latter increased from about 4% under low aerosol conditions, up to about 14% in case of polluted air. We conclude that the UV irradiance evaluation through radiative transfer models requires paying special attention to the assessment of the aerosols properties. © 2007 Elsevier B.V. All rights reserved.