**2. Daily relative global irradiance** *G/Gclear* **and relative direct irradiance** *I'/I'clear*

The conventially measured daily energy amounts of global *G* and direct solar irradiance *I'* falling on a horizontal surface are presented in physical units of MJ/m2. For the direct irradiance perpendicular to solar rays, irradiance *I* is measured and its values transformed to the horizontal surface *I,'* are also made available for each day. Until 1996 the Yanishevski AT-50 actinometers and Savinov-Yanishevski M-115 pyranometers were used but were since replaced by the Eppley Labor. Inc. pyrheliometers and Kipp & Zonen pyranometers. The absolute accuracy of the ventilated Kipp & Zonen pyranometers is about ±2% and that of the pyrheliometers ±1%. In the case of older instruments these uncertainties usually were doubled. In the past intercalibration of sensors was regularly performed in Voeikov Main Geophysical Observatory (St. Petersburg, Russia), whereas now it is done in World Radiation Center (Davos, Switzerland). Between the comparison campaigns, the absolute radiometer PMO-6 No R850405 is used as a secondary standard for regular assurance of the calibration. The previous standard, Ǻngström pyrheliometer M-59-8 No. J-1981, has been in use during more than 20 years and the scales of old and new standards have been in agreement with the World Radiation Reference within ±0.1% (Russak and Kallis, 2003).

The results of statistical treatment of values measured in physical units are illustrative in the case of interannual variations over longer time intervals. Due to the annual cycle even in clear conditions, the summer daily maxima of global solar irradiance at the study site, in absolute scale, are about 17.5 times higher than these of winter minima. The intraseasonal differences of the relative availability of solar radiation are emphasized if the daily values are presented relative to their climatological clear-sky background as the ratios *G/Gclear* and *I'/I'clear*. The daily climatological clear-sky values *Gclear* and *I'clear* are the assumed clear-day values for each calendar day corresponding to the typical conditions of atmospheric

between summer and winter as well as the seasonal impacts of cloud cover and aerosols may be significantly different. Here, the variations of solar ground-level integral global and direct irradiance on seasonal and monthly scales are examined. The continuous record of pyranometer-measured daily global radiation extends back to 1953 and that of pyrheliometer-measured direct irradiance back to 1955. The study is based on this long-term data set for years 1955-2010 when both quantities are available. The data set is supported by

Much of information in meteorological and climatological studies is obtained from measurement data applying statistical methods. The aim of exploratory data analysis (EDA) is to get an insight into the possible processes behind the variations in the collected data. Often the seasonal or monthly data are analyzed for their trends in time. In EDA, mainly the numerical summary measures of collected data sets, characterizing central tendency, spread and symmetry of data samples during their time evolution are used (Wilks, 2006). Quite often the conventional mean is used as a central tendency measure without checking how adequate it is. To get realistic insights into the processes the chosen characteristics must be robust. Robustness means insensitivity to deviations from the assumptions made. Suitability of different central tendency and spread characteristics of the recorded daily sums is compared in the case of skewed probability density distributions and the appropriate characteristics of seasonal and monthly relative solar radiation are found. Major features of

variation and trends in the availability of solar radiation in 1955-2010 are studied.

**2. Daily relative global irradiance** *G/Gclear* **and relative direct irradiance** *I'/I'clear* The conventially measured daily energy amounts of global *G* and direct solar irradiance *I'* falling on a horizontal surface are presented in physical units of MJ/m2. For the direct irradiance perpendicular to solar rays, irradiance *I* is measured and its values transformed to the horizontal surface *I,'* are also made available for each day. Until 1996 the Yanishevski AT-50 actinometers and Savinov-Yanishevski M-115 pyranometers were used but were since replaced by the Eppley Labor. Inc. pyrheliometers and Kipp & Zonen pyranometers. The absolute accuracy of the ventilated Kipp & Zonen pyranometers is about ±2% and that of the pyrheliometers ±1%. In the case of older instruments these uncertainties usually were doubled. In the past intercalibration of sensors was regularly performed in Voeikov Main Geophysical Observatory (St. Petersburg, Russia), whereas now it is done in World Radiation Center (Davos, Switzerland). Between the comparison campaigns, the absolute radiometer PMO-6 No R850405 is used as a secondary standard for regular assurance of the calibration. The previous standard, Ǻngström pyrheliometer M-59-8 No. J-1981, has been in use during more than 20 years and the scales of old and new standards have been in agreement with the World Radiation Reference within ±0.1% (Russak and Kallis, 2003).

The results of statistical treatment of values measured in physical units are illustrative in the case of interannual variations over longer time intervals. Due to the annual cycle even in clear conditions, the summer daily maxima of global solar irradiance at the study site, in absolute scale, are about 17.5 times higher than these of winter minima. The intraseasonal differences of the relative availability of solar radiation are emphasized if the daily values are presented relative to their climatological clear-sky background as the ratios *G/Gclear* and *I'/I'clear*. The daily climatological clear-sky values *Gclear* and *I'clear* are the assumed clear-day values for each calendar day corresponding to the typical conditions of atmospheric

the conventional meteorological data and visual cloud inspection data.

characteristics (Eerme et al., 2006; Eerme et al., 2010). Also they could be defined as those for the assumed dry atmosphere (Eerme et al., 2010). The clear-sky climatological daily sums could be calculated, using radiative transfer codes inserting realistic aerosol optical depth (AOD) data as well as realistic vertical profiles of temperature, water vapor and aerosol content. The clear-sky daily sums could be also interpolated from the observed data corresponding to average typical conditions by selecting the cloudless days proceeding from the recorded daily AOD values. We have used the latter version to avoid systematic differences in scales.The precipitable water vapour variations influence the clear-sky values of *Gclear* and *I'clear* as well but the range of variations of AOD influence is about ten times larger than that of water vapour (Russak et al., 2005).
