Determination of daily solar ultraviolet radiation using statistical models and artificial neural networks

Abstract

Abstract. In this study, two different methodologies are used to develop two models for estimating daily solar UV radiation. The first is based on traditional statistical techniques whereas the second is based on artificial neural network methods. Both models use daily solar global broadband radiation as the only measured input. The statistical model is derived from a relationship between the daily UV and the global clearness indices but modulated by the relative optical air mass. The inputs to the neural network model were determined from a large number of radiometric and atmospheric parameters using the automatic relevance determination method, although only the daily solar global irradiation, daily global clearness index and relative optical air mass were shown to be the optimal input variables. Both statistical and neural network models were developed using data measured at Almería (Spain), a semiarid and coastal climate, and tested against data from Table Mountain (Golden, CO, USA), a mountainous and dry environment. Results show that the statistical model performs adequately in both sites for all weather conditions, especially when only snow-free days at Golden were considered (RMSE=4.6%, MBE= -0.1%). The neural network based model provides the best overall estimates in the site where it has been trained, but presents an inadequate performance for the Golden site when snow-covered days are included (RMSE=6.5%, MBE= -3.0%). This result confirms that the neural network model does not adequately respond on those ranges of the input parameters which were not used for its development.

Ultraviolet Radiation over Two Lakes in the Middle and Lower Reaches of the Yangtze River, China: An Innovative Model for UV Estimation

“…UV radiation transmissivity K UV is selected as a dependent variable, which has the great advantage because its changes can be exclusively attributed to atmospheric attenuation factors. Several authors have proposed different models that link the hourly UV radiation transmissivity K UV with its main attenuation factors, such as cloudiness (through global clearness index) and relative optical air mass (FoyoMoreno et al 1999;Martínez-Lozano et al 1999;Ogunjobi and Kim 2004;Varo et al 2005;Barbero et al 2006;Hu et al 2010;Mateos Villán et al 2010). Barbero's model has been used in this paper:…”

“…Great effort has been undertaken by national or international organizations for promoting the installation of UV monitoring networks with aim to obtain accurate data on surface UV radiation in different parts of the world, such as North America, China, Europe and Australia (Liu 2000;Gröbner et al 2006;Bhattarai et al 2007;Hu et al 2007;Cui et al 2008;Rieder et al 2008). However, existing UV observation stations are far too low to provide sufficient UV data because of the high cost and great difficulty in maintaining the sensors that register this spectral band (Barbero et al 2006;Janjai et al 2010;Mateos Villán et al 2010). Furthermore, most of these stations began operation in the last decade of the twentieth century (McKenzie et al 1992;Zerefos et al 1997;Gurney 1998;Bartlett and Webb 2000;Fioletov et al 2002;Junk et al 2007).…”

“…Thus, suitable modeling approaches should be an appropriate alternative for estimating UV radiation values when UV observation is not available. These models can be categorized mainly into two groups (Barbero et al 2006;Sabziparvar 2009;Láska et al 2010): (1) radiative transfer models which describe absorption and scattering processes of UV radiation in the atmosphere; and (2) statistical models that exploit the close relationship between UV radiation and atmospheric parameters commonly measured at radiometric or climatological stations. The radiative transfer models need extensive computation and detailed information of atmospheric parameters which are generally difficult to obtain from ground-based stations.…”

“…Previous research concentrated mainly on the relationship between UV and an individual factor, such as global solar radiation (Feister and Grasnick 1992;Ilyas et al 1999;Martínez-Lozano et al 1999;Kudish and Evseev 2000;Cañada et al 2003;Ramirez Santigosa et al 2003;Ogunjobi and Kim 2004;Robaa 2004;den Outer et al 2005;Escobedo et al 2009Escobedo et al , 2011Sabziparvar 2009), relative optical air mass (m) (Foyo-Moreno et al 1999;Cañada et al 2000;Dubrovsky 2000;Murillo et al 2003;Antόn et al 2009b) and ozone (McKenzie et al 1991;Madronich 1992;Bais et al 1993). Only a few studies analyze the integrated effects of main attenuation factors on UV radiation such as relative optical air mass and cloudiness (Foyo-Moreno et al 1999;Murillo et al 2003;Barbero et al 2006;Mateos Villán et al 2010); total column ozone, geometric factors and clouds (Díaz et al 2000); global solar radiation, total ozone, dew point temperature, and snow cover (Fioletov et al 2001); ozone, aerosols and clouds (Krzyścin et al 2001); daily sunshine fraction, the cosine of the daily minimum SZA and the daily total ozone column (de La Casinière et al 2002); ozone, airmass, sun-earth distance correction and solar zenith angle (Prasad and Niranjan 2005); global solar radiation, ozone and clouds (Foyo-Moreno et al 2007); ozone and reflectivity (Antόn et al 2005); ozone and cloudiness (Antόn et al 2009b) as well as ozone and cloudiness (Láska et al 2010). However, the...…”

Comparison of different UV models for cloud effect study

“…Different models have been proposed in literature with reliable results: Murillo et al [37] analyzed the correlation between UV irradiation and G in Valencia and Cordoba (Spain); Barbero et al [38] estimated daily UV radiation using statistical models and artificial neural networks; Bilbao et al [16] reconstructed erythemal UV levels in Spain during 1991e2010 using total ozone column. Several authors also studied the main factors influencing the UV radiation: Guarnieri et al [39] analyzed the ozone influences; Borkowski [40] investigated the relationship between UV and G and its dependence on total ozone column, cloudiness and aerosol at different scales.…”