Comparison between simulated and measured solar irradiance during a desert dust episode

Abstract

The aim of this study is to analyze the reliability of the libRadtran v. 1.7model in the estimation of irradiance in the shortwave spectral range (285-2800 nm) during a desert dust episode. For that purpose downward irradiance measurements at the surface and corresponding model simulations have been compared during the three days of a desert dust event (9-11 August 2012) observed over Évora, Portugal. The comparison between measured and simulated values shows a highly significant correlation, with a correlation coefficient of 0.999 and a slope very close to unity (0.998±0.006), supporting the validity of the model in the estimation of global irradiance in the shortwave spectral range. Relative differences between the simulated and measured irradiances have also been calculated and indicated that the libRadtran model slightly underestimates the experimental global irradiance, with a mean relative difference equal to 1.28 %. These small differences could be associated with the experimental errors of the measurements, as well as with uncertainties in the input values given to the model, namely those related with the actual aerosol properties. The notably good agreement between simulated and measured irradiances guarantees that the libRadtran model can be used to estimate clear sky irradiance when no radiation measurements are available during desert dust events. In order to obtain accurate estimations of the irradiance, the model must be fed with reliable values of the aerosol properties.