Assessment of four methods to estimate surface UV radiation using satellite data, by comparison with ground measurements from four stations in Europe

Abstract

[1] Four different satellite-UV mapping methods are assessed by comparing them against ground-based measurements. The study includes most of the variability found in geographical, meteorological and atmospheric conditions. Three of the methods did not show any significant systematic bias, except during snow cover. The mean difference (bias) in daily doses for the Rijksinstituut voor Volksgezondheid en Milieu (RIVM) and Joint Research Centre (JRC) methods was found to be less than 10% with a RMS difference of the order of 30%. The Deutsches Zentrum für Luft-und Raumfahrt (DLR) method was assessed for a few selected months, and the accuracy was similar to the RIVM and JRC methods. It was additionally used to demonstrate how spatial averaging of high-resolution cloud data improves the estimation of UV daily doses. For the Institut d'Aéronomie Spatiale de Belgique (IASB) method the differences were somewhat higher, because of their original cloud algorithm. The mean difference in daily doses for IASB was about 30% or more, depending on the station, while the RMS difference was about 60%. The cloud algorithm of IASB has been replaced recently, and as a result the accuracy of the IASB method has improved. Evidence is found that further research and development should focus on the improvement of the cloud parameterization. Estimation of daily exposures is likely to be improved if additional time-resolved cloudiness information is available for the satellite-based methods. It is also demonstrated that further development work should be carried out on the treatment of albedo of snow-covered surfaces.

Modelling natural surface UV radiation with satellite data: examples of applications

“…The difference decreases with time averaging (~5% rms on monthly averages). Similar results have been obtained when comparing the satellite-derived estimates to measurements at Garmisch-Partenkirchen (D), Brussels (B), Bilthoven (NL), Jokioinen (FIN) 6 , Thessaloniki (GR) and Tromsø (N). …”

Temperature dependence of the Brewer global UV measurements

“…Fountoulakis et al: Temperature dependence of the Brewer global UV measurements B irradiance that reaches the Earth surface (Fioletov et al, 2001;Kerr and McElroy, 1993;McKenzie et al, 1999) and quantified the interaction between the solar UV irradiance, the Earth surface and the atmospheric components which mainly control its levels, such as ozone, sulfur dioxide, aerosols and clouds (e.g., Arola et al, 2003;Bais et al, 1993;Bernhard et al, 2007;Fioletov et al, 1998). Spectral measurements from Brewers have been used widely for climatological studies of biologically effective UV doses (e.g., Fioletov et al, 2003Fioletov et al, , 2009Kimlin, 2004), validation of satellite products (e.g., Arola et al, 2002;Bernhard et al, 2015;Kazadzis et al, 2009) and validation of radiative transfer models (Kazantzidis et al, 2001;Mayer et al, 1997). Lately, spectra from stations with long measurement records have been used for the study of the changes of the solar UV irradiance, showing that changes in air quality and climate have an important impact on its short-and long-term variability (De Bock et al, 2014;Fountoulakis et al, 2016a;Fragkos et al, 2016;Lakkala et al, 2017;Simic et al, 2011;Smedley et al, 2012;Zerefos et al, 2012).…”

On the use of quantitative diurnal cloud information for the calculation of daily UV dose maps over Europe

“…The CCF calculated with the standard processing (Option III) is 0.89 for July 2006. The bias with the Brewer for this month is quite significant, and it has been only slightly reduced by the introduction of the cloud optical thickness (Options IV and V) in combination with equation (2). Below we investigate the differences in the monthly means UV doses further on a day-to-day basis.…”

“…This gives some confidence in daily UV dose estimates that are based on polar orbiting satellites only (cf. Arola et al, 2002). More months of Meteosat-8 data will be analyzed, also going back in time.…”