Reconstruction of daily solar UV irradiation by an artificial neural network (ANN)

Abstract

Long-term records of solar UV radiation reaching the Earth's surface are scarce. Radiative transfer calculations and statistical models are two options to re-construct decadal changes in solar UV radiation from long-term records of measured atmospheric parameters that contain information on the effect of clouds, atmospheric aerosols and ground albedo on UV radiation. Based on earlier studies, where the long-term variation of daily solar UV irradiation was derived from measured global and diffuse irradiation as well as atmospheric ozone by a non-linear regression method⁽¹⁾, we have chosen another approach for the re-construction of time series of solar UV radiation. An Artificial Neural Network (ANN) has been trained with measurements of solar UV irradiation taken at the Observatories Potsdam and Lindenberg in Germany as well as measured parameters with long-term records such as global and diffuse radiation, sunshine duration, horizontal visibility and column ozone. This study is focused on the re-construction of daily broad-band UV-B (280-315 nm), UV-A (315-400 nm) and erythemal UV irradiation (ER). Due to fast changes in cloudiness at mid-latitude sites, solar UV irradiance shows an appreciable short-term variability. One of the main advantages of the statistical method is that it uses doses of highly variable input parameters calculated from individual spot measurements that are taken at short time steps, and thus do contain the short-term variability of solar irradiance. Our study has been supported by the European SCOUT-O3 project funding. The ANN model results have been evaluated within the European action COST726⁽²⁾.