The ultraviolet (UV) band of the electromagnetic spectrum extends from 100 to 400 nm and is divided into three sub-regions: the UV-C (100–280 nm), the UV-B (280–315 nm), and the UV-A (315–400 nm). Only a small fraction (9.3%) of the electromagnetic radiation emitted by the Sun is in the UV spectral region [1
] as most of it is attenuated by the Earth’s atmosphere as it propagates toward the surface.
Despite the small amount of UV radiation that finally reaches the Earth’s surface, its biological significance is exceptional [2
]. UV radiation triggers and/or drives photochemical and photobiological processes, which are necessary for the proper functioning of ecosystems, and has strong direct or indirect effects on human health [3
]. Living organisms have slowly adapted to the current levels of solar UV radiation through the evolution process and fast or abrupt changes impact the health and the diversity of flora and fauna [4
]. Consequently, any change in the ecosystems affects human populations through their interaction with the natural environment.
UV radiation that reaches the surface of the Earth exhibits periodical changes associated with a number of different phenomena: the solar radiation that reaches the Earth’s atmosphere changes by ±3% in the year as a result of periodical changes in the Earth–Sun distance while the angle between the Sun and the zenith of a particular place on Earth also changes periodically leading to corresponding changes in the radiant energy. Periodical changes in solar activity, such as the 11-year solar cycle, the 27-day apparent solar rotation, and dynamical atmospheric processes (e.g., quasi-biennial oscillation (QBO)) also induce changes in the levels of solar UV radiation that reach the Earth’s surface. Living organisms have adapted to the above periodicities, which are predictable and easily modeled.
Non-periodic changes in atmospheric composition and dynamics can also affect the levels of surface UV radiation significantly. Solar UV radiation at wavelengths below 290 nm is absorbed by molecular oxygen and the ozone at the higher atmosphere and does not enter the troposphere. Atmospheric molecules scatter radiation more effectively with decreasing wavelength, thus scatter more UV than visible or infrared light. The ozone effectively absorbs UV-B radiation [22
], allowing to only a very small fraction to reach the Earth’s surface. In addition to the ozone, other gases also absorb part of the UV radiation (e.g., SO2
), though their impact is usually less significant than that of the ozone, either because their absorption efficiency is smaller or because they are less abundant than the ozone.
Clouds and aerosols in the troposphere also scatter (both clouds and aerosols) and absorb (aerosols) solar UV radiation. The spectral attenuation by clouds and aerosols depends on their type and characteristics [26
]. Clouds are the most significant driver of short term variability of UV irradiance at the Earth’s surface [28
]. Although they usually attenuate UV radiation, under particular conditions UV irradiance can be enhanced due to the presence of clouds [29
]. In urban environments changes in aerosols may counterbalance the effect of even extremely high or low total ozone events, and lead to erythemal irradiance above or below the climatological averages respectively [35
]. Surface albedo is an additional important regulatory factor for the levels of UV irradiance, since multiple reflections between the Earth’s surface and the atmospheric constituents enhance UV irradiance over highly reflective (e.g., snow- or ice-covered) terrains [36
]. Over highly reflective terrains and under broken cloud conditions short-term enhancement can be of the order of 50% [38
]. The UV radiation reaching the surface generally increases with altitude, mainly because of the decreasing atmospheric density, thus decreasing attenuation [39
According to the results of many studies, long-term changes in cloudiness, ozone, surface reflectivity, and/or aerosols in the recent past have already altered the levels of UV irradiance significantly in several areas around the globe [41
]. In the 1970s, increased photochemical destruction of ozone due to anthropogenic emissions of the ozone depleting substances (ODS) led to depletion of stratospheric ozone, especially over high latitudes of both hemispheres [42
]. Ozone depletion led to increasing UV-B radiation over the mid and high latitudes [45
], which caused awareness in the scientific community and led to the implementation of the Montreal Protocol, which was signed in 1987. The implementation of the Montreal Protocol in 1989 was successful preventing humanity from extreme exposures to UV radiation [47
]. Currently the first signs of ozone recovery since the mid 1990s, due to the reduced ODS emissions, have been reported over polar latitudes [50
]. Over northern hemisphere mid-latitudes, ozone increases in the upper—but not in the lower—stratosphere have been observed in the same period [52
]. According to Eleftheratos et al. [55
] at northern high latitudes, between 55° and 70° N, the irradiances at 305 nm decreased significantly by 3.9% per decade in the period 1990–2011 mainly because of the reported ozone recovery.
Nevertheless, recent studies reporting results from mid-latitude stations of the northern hemisphere show that UV irradiance has increased in the last two decades mainly as a result of the global brightening effect, while total ozone has either remained stable or increased in the same period [56
]. Increasing aerosols over South-East Asia have had the opposite effect resulting to decreased levels of solar UV radiation [41
]. Changes in surface albedo and cloudiness have also affected the levels of UV radiation in the past decades, especially over sub-polar and polar regions [60
]. For example, statistically significant negative trends of the monthly average noon erythemal and 345 nm irradiances for October, of the order of −11% to −14%, have been found for Barrow, Alaska for the period 1991–2011, which have been attributed to a statistically significant decrease in the number of days during which the land surface around the station was covered by snow. Over such high latitudes decreased ice and snow cover (i.e., decreased surface albedo) and increased cloudiness are projected to drive significant changes of UV irradiance in the future [63
Due to the very complex interactions of solar UV radiation with atmospheric constituents and the features of the Earth’s surface [65
], modeling and predicting its changes is in many cases very uncertain [41
]. Large uncertainties in the future projections (by climate models) of the main factors controlling the levels of the solar UV irradiance at the Earth’s surface (especially clouds and aerosols) further increase the uncertainty in estimates of future UV changes due to the corresponding changes in climate and air quality [66
]. The spectral characteristics of the interactions between UV radiation and aerosols are not yet completely understood, and are usually poorly described in radiative transfer models [41
]. Uncertainties about the future rate of ozone recovery arise from the recent unexpected emissions of anthropogenic ODS already controlled by the Montreal protocol, as well as emissions of ODS, which have not yet been forbidden [68
]. Climate change—induced alterations of stratospheric temperatures and circulation patterns are also expected to have a significant impact on the future levels and spatial distribution of stratospheric ozone [70
], which in turn would affect solar UV radiation [74
Continuous estimates of the levels of UV irradiance at the Earth’s surface on a global scale are available for the last four decades from satellite measurements [80
]. Although in the last years there has been significant progress in the algorithms used for the retrieval of surface UV irradiance from satellites [84
], the retrievals are still not sufficiently accurate over mountainous sites [86
], highly reflective terrains [88
], as well as over highly polluted environments [89
], mainly because of the use of climatological data (for e.g., surface albedo and aerosol absorption) and simplifications in the algorithms. Furthermore, satellite retrievals represent the average of a finite area covered by the satellite pixel and are not necessarily representative for each point of the pixel, especially over complex, inhomogeneous terrains [92
]. Thus, under the conditions described above, comparisons between ground-based UV measurements and satellite retrievals may yield differences of 20% for clear skies and up to 50% for cloudy skies.
The accuracy of ground-based measurements is limited mainly by the characteristics of each particular instrument [93
]. The standard uncertainty in good quality spectral UV measurements from well maintained and calibrated sensors is of the order of 5% for wavelengths above 305 nm [95
], although it can be much lower, less than 2%, for properly designed and accurately characterized instruments [97
]. Performing continuous and high quality global spectral UV measurements is a difficult task, which demands expensive equipment and properly trained personnel. Although monitoring of the UV irradiance can be achieved using much cheaper broad-band instruments, spectral measurements have important advantages. Analyzing accurate long-term spectral UV measurements with respect to measurements of atmospheric constituents (e.g., total ozone column, aerosols, etc.) and the Earth’s surface properties (e.g., surface albedo) allows the detection of trends [98
], but also the identification of the main factors and mechanisms that drive them [26
]. Furthermore, co-located measurements of the solar spectral UV irradiance and other atmospheric components improve our understanding regarding the complex interactions between UV radiation, atmospheric constituents, and the Earth’s surface properties. In this way, analyzing spectral UV measurements contributes to the improvement of the accuracy in UV modeling and the satellite retrievals.
Spectral UV measurements can be weighted with well defined action spectra and then used for the calculation of biologically effective quantities in order to directly quantify the effect of UV radiation on biological processes [101
] and human health [102
]. A quantity that is commonly used for public information and in human health-related studies is the UV index [104
]. The UV index is a metric of effectiveness of UV irradiance on causing erythema in human skin, and is calculated by dividing the erythemal irradiance (in mW/m2
) by 25. The erythemal irradiance is the spectral UV irradiance weighted with the erythemal action spectrum (i.e., the relative contribution of the irradiance at each wavelength to the induction of erythema in the human skin) [103
]. The integral of erythemal irradiance over a certain time period is usually referred as the erythemal dose. Highly accurate spectral measurements can be also used for the calibration and/or the validation of broad-band and narrow-band instruments [90
] used to directly measure UV index.
Due to their high cost, long-term, continuous, and accurate spectral UV measurements are available only from a limited number of stations. An investigation into the databases of the larger networks and data centers providing ground-based remote sensing measurements (the Network for the Detection of Atmospheric Composition Change (NDACC), the World Ozone and Ultraviolet Radiation Data Centre (WOUDC), and the European UV Database (EUVDB)) reveals that there are less than seventy stations around the world that have provided spectral UV measurements for relatively long periods, of a few years or more, in the last three decades. Less than half of them have provided measurements continuously for more than ten years. In Europe, measurements of the UV index and other integrated quantities are widely available, mainly from broad-band instruments [107
]. There are also studies discussing the changes of the UV irradiance, even before the 1990s, based on reconstructed UV time-series from ground based or satellite retrievals of other parameters. Bilbao, et al. [108
], for example, analyzed a reconstructed dataset of the erythemal irradiance for central Spain and reported a significant increase of 3.5% and 4.1% per decade for summer and autumn between 1991 and 2010. Analysis of satellite measurements also shows increasing trends of the UV-A radiation over Europe in the period 1979–2011 [109
]. Systematic, continuous spectral measurements of global solar UV irradiance, which can provide more reliable information regarding the trends of UV irradiance, are available from a few European stations, in some since the early 1990s [58
The present study provides a brief review of the results presented in recent studies reporting changes of the spectral UV irradiance and effective UV doses over Europe, with respect to changes in total ozone, surface albedo, clouds, and aerosols. The review is provided in Section 3
, where long term spectral UV records from four different European stations at latitudes between 40 and 67° N are also being analyzed to provide further context. Interesting findings from the analysis of spectral UV measurements around Italy are discussed in Section 4
. Thereafter, in the same section, the necessity of having continuous, accurate long-term spectral UV measurements is highlighted through the climatological analysis of the spectral datasets of Rome and Aosta, as well as comparison of the two datasets with UV estimates from the Deutscher Wetterdienst (DWD) forecast model and the ozone monitoring instrument (OMI). The methodology followed in order to analyze the data presented in Section 3
and Section 4
is discussed in Section 2
5. Summary and Conclusions
Since the 1990s, the levels of the spectral UV irradiance have changed over Europe, mainly as a result of changes in aerosols, clouds, and surface albedo. All studies referring to stations in South, East, and Central Europe report positive trends of the UV irradiance, mainly resulting from reduced attenuation by aerosols and clouds. A number of studies for different stations in the UK report decreasing UV irradiance in the last three decades. Significant trends of −7 to −8% per decade (in the erythemal UV dose and the irradiance at 307.5 nm) have been detected for the period between the mid-1990s until the mid-2010s in Reading and Chilton, UK. However, further analysis for the latter station showed that UV irradiance was increasing up to 2004 and decreasing thereafter. For the same period, significant positive trends of the same magnitude have been detected for Uccle, Belgium, which is within a distance less than 400 km from Reading and Chilton. Analysis of the UV datasets for the two stations in the context of the present study confirmed these findings and showed that UV-B and UV-A irradiances at the particular stations change with the same rate (possibly for different reasons) during the period 1996–2017. The big difference between the results for Uccle and Reading is a typical example of the large variability of changes in aerosols and cloudiness, which subsequently result in large spatial variability in the trends of UV irradiance. A large significant increase of 8% has been found for the irradiance at 307.5 nm at the station of Thessaloniki for 1996–2017, despite the absence of significant trends in total ozone or cloudiness at the particular station [56
]. This trend has been mainly attributed to changes in aerosols.
Studies referring to higher latitude stations report decreasing UV irradiance. A positive, not significant trend of the average daily erythemal dose has been reported for the arctic station of Hornsund, Svalbard for the period 1996–2016. Analysis of the UV time-series for different stations in Norway, as well as for Sodankylä, Finland, reveals negative trends in the erythemal irradiance and the irradiance at different UV-B and UV-A wavelengths, ranging between 2% and 5%. However, these trends are not statistically significant. Analysis in the context of the present study shows that in the period 1996–2017 the UV irradiance at 307.5 and 324 nm decreases in Sodankylä, with the trends being non-significant and of similar magnitude as those reported from previous studies. Lakkala et al. [140
] reported larger but insignificant trends, of the order of 10% for the period 1990–2014 in the monthly average levels of the irradiance in the UV-B region in spring and the early summer, mainly resulting from increasing ozone. At higher latitudes, part of the changes in UV can be attributed to changes in surface albedo and cloudiness.
In Italy several studies report significant changes in cloudiness during the last decades [172
], which are expected to have an impact on the levels of UV irradiance. Although, continuous spectral UV measurements of good quality are available from different stations, the long-term changes of UV irradiance have not been investigated yet. The inhomogeneous terrain throughout the Italian territory and the large altitudinal gradients in particular regions, lead to large differences between the factors controlling the variability of UV irradiance even within nearby sites. Differences between the daily minimum SZA were also significant since Italy is extended in a long latitudinal zone. In order to discuss these effects, the annual variability of the irradiance at 307.5 and 322.5 nm and the erythemal irradiance were investigated for the urban, low altitude site of Rome in central Italy, and the alpine site of Aosta at the northern borders of the country.
Analysis of satellite data shows that in Rome the attenuation of UV radiation by aerosols was more important than in Aosta, throughout the whole year. Attenuation by clouds in months October–March was stronger at Rome, while in the summer (June–August) attenuation by clouds was stronger at Aosta. UV irradiance at Aosta in months October–April was enhanced by the high surface albedo. The total ozone was higher by 5–10 DU in Aosta in spring and by 5–10 DU in Rome in autumn. When the differences between the irradiance over the two sites were studied for a particular SZA (=65°) the differences in surface albedo and cloudiness were clearly depicted in the results, while the effect of differences in total ozone was less significant. The ratio between the irradiances at Aosta and Rome (for both, 307.5 and 322.5 nm) ranged from 0.8 in summer to 1.25 in winter. Theoretical analysis shows that at SZA = 65° the ratio would be 1.05 for 322.5 nm and 1.08 for 307.5 nm due to the higher altitude of Aosta, assuming that all other factors were constant.
When the irradiances measured at local noon at the two sites were compared to each other, the effect of different SZA became important and more pronounced for the 307.5 nm. The ratio for 307.5 nm ranged from a minimum of 0.7 in December to a maximum of 1.05 in May. Minimum and maximum values of the ratio were found for the same months for 322.5 nm and the UV index. Nevertheless, the variability is smaller, (i.e., 0.85–1.1 for the 322.5 nm irradiance and 0.8–1.05 for the UV index). The climatological noon UV index in Rome in July was 7.3 and was larger than the UV index in June (7.0), although the noon SZAs were smaller in June. The reason was possibly that clouds attenuate more UV radiation in June than in July. At Aosta, increased cloud cover in June relative to July counterbalanced the effect of SZA less effectively than in Rome, leading to nearly the same value of the UV index equal to 6.6 for both months.
Comparison of the noon UV index measured at the sites of Rome and Aosta for a four year time-period with the corresponding UV index from OMI and DWD forecast model estimates, further highlighted the significance of having accurate ground-based UV measurements. Large differences up to 6 units were found between ground-based measurements and both the DWD model and OMI for particular days. These differences resulted from an inaccurate description of the attenuation by clouds by both model and satellite estimates. Predictions of the current DWD model algorithm refer to a pixel of 13 × 13 km2 instead of a particular point like in are the ground-based measurements. The OMI pixel refers to an even larger area of 13 × 100 km2, wherein cloudiness is again not necessarily representative for Rome or Aosta. The use of climatological values for aerosol absorption properties, in both the OMI algorithm and the DWD model, which in many cases underestimated absorption by aerosols in the UV, possibly leads to the average overestimation of the UV index by 0.5 in winter to 1 in summer over Rome. In Aosta, the UV index was mainly overestimated in the second half of the year, possibly as a result of the inaccurate description of the aerosol absorption in the model and the satellite algorithm, and the underestimation of total ozone by OMI. Underestimation of the UV index in the first half of the year, mainly from OMI, could be due to the systematic overestimation of the attenuation by clouds and underestimation of the effect of surface albedo.
The comparison between the levels of the UV irradiance at the stations of Rome and Aosta revealed the significant role of aerosols, clouds, and surface albedo in the formulation of the levels of the spectral UV irradiance at each site. Changes in the particular parameters in the future might induce large changes in the levels of the UV irradiance that reaches the Earth’s surface. Study of the trends of spectral UV irradiance in the past, with respect to the changes of the latter parameters and total ozone might provide useful information on how the spectrum of UV irradiance changed in the past decades. Investigation of the interactions between spectral UV irradiance and the factors discussed above might also provide very useful information, which would contribute to the improvement of the modeling of the UV irradiance.
The present study focused on pointing out the significance of having long-term, accurate UV measurements in Italy, in the context of changes in air quality and climate. Thus, scientific questions arising from the present study were in many cases not fully investigated and remained unanswered. Extension of the present study could include validation of different satellite and forecast UV products for a longer time-period with respect to total and tropospheric ozone, clouds, snow cover, and aerosol optical properties, since ground-based measurements of all the above parameters are available for Rome and Aosta. Development and application of new algorithms in the European SkyNet Radiometers Network (EUROSKYRAD) network is expected to give new products in the near future, which will contribute significantly in the understanding of the complex interactions between aerosols and solar UV radiation. Total solar radiation measurements and sky camera products, which are available for both stations could also provide useful information for the role of clouds. Analysis of the results from more UV monitoring stations in Italy, in addition to those of Rome and Aosta, is necessary in order to investigate the spatial distribution of changes throughout the country.
In any case, the results of the present study clearly demonstrated the importance of maintaining and analyzing long-term, continuous, high quality ground-based measurements.