Measurements of total ozone column and effective cloud transmittance have been performed since 1995 at the three Norwegian sites Oslo/Kjeller, Andøya/Tromsø, and in Ny-Ålesund (Svalbard). These sites are a subset of nine stations included in the Norwegian UV monitoring network, which uses ground-based ultraviolet (GUV) multi-filter instruments and is operated by the Norwegian Radiation and Nuclear Safety Authority (DSA) and the Norwegian Institute for Air Research (NILU). The network includes unique data sets of high-time-resolution measurements that can be used for a broad range of atmospheric and biological exposure studies. Comparison of the 25-year records of GUV (global sky) total ozone measurements with Brewer direct sun (DS) measurements shows that the GUV instruments provide valuable supplements to the more standardized ground-based instruments. The GUV instruments can fill in missing data and extend the measuring season at sites with reduced staff and/or characterized by harsh environmental conditions, such as Ny-Ålesund. Also, a harmonized GUV can easily be moved to more remote/unmanned locations and provide independent total ozone column data sets. The GUV instrument in Ny-Ålesund captured well the exceptionally large Arctic ozone depletion in March/April 2020, whereas the GUV instrument in Oslo recorded a mini ozone hole in December 2019 with total ozone values below 200 DU. For all the three Norwegian stations there is a slight increase in total ozone from 1995 until today. Measurements of GUV effective cloud transmittance in Ny-Ålesund indicate that there has been a significant change in albedo during the past 25 years, most likely resulting from increased temperatures and Arctic ice melt in the area surrounding Svalbard.
The amount of stratospheric ozone decreased significantly both globally and
over Norway during the 1980s and 1990s (WMO, 2018; Svendby and Dahlback,
2004). This decrease was mainly caused by the release of ozone-depleting
substances (ODSs). In 1987, the Montreal Protocol was signed with the aim of
phasing out the production of ODSs. Motivated by this treaty, the Norwegian
Environment Agency established the programme “Monitoring of the atmospheric
ozone layer” in 1990. Five years later, in 1995/1996, the network was
expanded and “the Norwegian UV network” was established with funding from the Norwegian Ministry of Health and Care Services and the Norwegian Environment Agency. This network consists of nine ground-based ultraviolet (GUV) radiometers located at sites between 58 and 79
The Norwegian UV network. Grey circles represent stations operated by DSA, whereas red circles represent sites operated by NILU. The large red circle to the south includes the three stations at Østerås (DSA), Blindern, in Oslo, and Kjeller. The instrument in Tromsø (blue circle) was moved to Andøya in 2000.
The spectral distribution of solar UV radiation reaching the ground depends on the optical properties of the atmosphere, the solar zenith angle (SZA), and reflection from the Earth's surface. The transmission of solar radiation in the UVB region (280–315 nm) through the stratosphere is primarily determined by the amount of stratospheric ozone, whereas the attenuation in the troposphere is mainly due to scattering by air molecules (Rayleigh scattering), aerosols, and clouds. Generally, a decrease in total ozone column leads to an increase in UVB radiation, assuming no changes in cloudiness or other UV-affecting parameters.
High-wavelength-resolution spectroradiometers can provide detailed information about the spectral distribution of UV radiation. Stamnes et al. (1991) showed that spectra from such instruments can be used to determine total ozone and cloud transmission accurately. However, simpler and cheaper radiometers with channels in both the UVB and the UVA regions, such as the GUV instruments, have also been demonstrated to be a good alternative to expensive spectroradiometers (Dahlback, 1996; Bernhard et al., 2005; Sztipanov et al., 2020).
In this study, we present a 25-year time series of total ozone column (TOC) from the Norwegian UV Network. We have focused on three stations operated by NILU located in Oslo/Kjeller, at Andøya/Tromsø, and in Ny-Ålesund as shown by red circles in Fig. 1. All stations are equipped with additional total ozone measuring instruments such as Brewer spectrophotometers and a Système d'Analyse par Observation Zénithale (SAOZ) instrument. TOCs derived from the GUV instruments are compared to measurements from other ground-based instruments. In addition, they are compared with satellite retrieved data sets. The current work also presents observed changes in total ozone and effective cloud transmittances.
The GUV is a multi-wavelength filter radiometer manufactured by Biospherical Instruments Inc. (BSI), San Diego (Bernhard et al., 2005). The detector unit is environmentally sealed and temperature stabilized, facilitating long-term reliable operation under harsh outdoor conditions. The GUV instruments have five channels in the UV range where each channel has a dedicated filter, a photodetector, and electronics that sample the output at a rate of about 3 Hz. The channels measure simultaneously global (direct and diffuse) solar irradiance at several UV wavelengths, which can be used to reconstruct the solar spectrum in the UV range and to compute biological doses, the UV index, total ozone, and cloud transmittance.
The UV network consists of 12 multiband filter radiometers (model GUV-541
and GUV-511) (Bernhard et al., 2005). Nine of them are continuously operating
at the network locations (Table 1) and three serve calibration purposes and are backups in case of failure at some of the stations. The instrument in Oslo/Kjeller is a GUV-511, whereas the instruments at the other sites are GUV-541. Both instrument types have four channels in the UV region (centre wavelengths 305, 320, 340, and 380 nm). In addition, GUV-541 has a fifth UV channel at 313 nm, whereas GUV-511 has a fifth channel for measuring photosynthetically active radiation (PAR: 400–700 nm). The bandwidths of the UV channels are
Overview of the locations, instrument types, and institutes involved in the Norwegian UV network.
The GUV-511 in Oslo was purchased already in 1993 and was installed at the
University of Oslo (UiO) to test the instrument performance and to develop
appropriate software. In July 2019, this instrument was moved to Kjeller
(
With a few minor exceptions, the GUV instruments have been running continuously since 1995. The GUV instrument at Andøya has been subjected to some problems, most likely caused by an event of water intrusion. In spring 2013 an error with the 380 nm channel was discovered and the instrument was sent to BSI for repair. Two years later, in 2015, the 320 nm channel failed and had to be replaced. During these time periods spare GUV instruments were deployed from the DSA.
As listed in Table 1 there are three Brewer spectrometers in operation in Norway: one in Oslo/Kjeller (B42), one in Tromsø/Andøya (B104), and one in Ny-Ålesund (B50). Generally, the Brewer instruments have been approved by the WMO as reliable high-quality instruments (WMO, 2018; Fioletov et al., 2008). The direct sun (DS) algorithm is the primary measurement mode of the Brewer and is based on measurements of the intensity of direct sunlight at five wavelengths between 306 and 320 nm. The precision of this method can be as high as 0.15 % (Scarnato et al., 2010), but the absolute accuracy relies on an appropriate calibration. Under cloudy conditions, total ozone can be derived by measuring the intensity of scattered radiation from the zenith. As shown by Stamnes et al. (1990) there are some limitations of the zenith sky (ZS) method, but nevertheless this method provides useful information about total ozone content when the DS method cannot be used. Measurements of the Brewer global irradiance (GI) are an alternative to the ZS method and are also based on the principle of measuring scattered UV radiation from the sky.
The Norwegian Brewer instruments have been calibrated by the International
Ozone Service (IOS, Canada) every year since installation in the 1990s,
except from the summer of 2020 when the calibration was prohibited under the
COVID-19 restrictions. These frequent calibrations are done to ensure high-quality Brewer measurements and to make sure that the instruments are well maintained and perform DS measurements within an accuracy of
The GUV data in the present study have been compared to OMI/Aura and
GOME-2/MetOp-A TM3DAM v4.1 total ozone data from Oslo, Andøya, and Ny-Ålesund. The satellite data from OMI and GOME-2 are available from 2004 and 2007, respectively. These data are assimilated products, based on the TM3DAM software developed by Royal Netherlands Meteorological Institute, KNMI (Eskes et al., 2003). The GOME-2 and OMI assimilated TOC values are publicly available and are provided on a daily basis via ESA's TEMIS project (
In Sect. 4.3, trends in effective cloud transmittance from the GUV instruments are discussed, and cloud data from the Norwegian Centre for Climate Services (NCCS;
The procedure for calibrating the GUV instruments is described by Dahlback (1996) and only briefly presented below. When the GUV Teflon diffuser is illuminated by a source, the photodetector transforms the radiation to an electric current which subsequently is converted to a voltage signal. The measured voltage of channel
The shape of the solar UV spectrum at the Earth's surface depends mostly on
the solar zenith angle (SZA) and the TOC. Thus, the spectral distribution of
The calibration procedure described above is normally done during large national or international intercomparison campaigns, where the GUV instruments are operating synchronously with co-located high-resolution reference spectroradiometers. One of these campaigns was arranged in Oslo in 2005, initiated through the national project Factors Affecting UV Radiation in Norway (FARIN) (Johnsen et al., 2008; WMO, 2008). Here the GUV instruments were intercompared with a Bentham spectroradiometer belonging to DSA, which is closely linked to the Quality Assurance of Spectral Ultraviolet Measurements in Europe (QASUME) world travelling reference spectroradiometer. Another large intercomparison campaign, which included the QASUME reference spectroradiometer, was arranged in May/June 2019 (PMOD/WRC, 2019).
A key factor for the maintenance of a homogenous and stable calibration
scale for the network instruments is a system for quality control which
accounts for long-term changes in the absolute response factors
The GUV data products described in this work consist of measurements used in
combination with a radiative transfer model (RTM) based on the discrete
ordinate method (Stamnes et al., 1988; Dahlback and Stamnes, 1991). When solar radiation passes through the atmosphere, a portion of the UVB radiation will be absorbed by ozone. Other fractions of the radiation will be multiple scattered or absorbed by air molecules, aerosols, and clouds (Stamnes et al., 2017). The total ozone column (TOC) is determined from the GUV instruments by comparing a measured and calculated
The
To quantify the effects of clouds, aerosols, and changing surface albedo, a cloud transmission factor is introduced. It is defined as the measured
irradiance at wavelength channel
Ratios of GUV
As described in Sect. 2.3, each GUV instrument has a unique set of
Figure 2 shows the GUV
As seen from Fig. 2 there is a clear seasonality in the TOC ratio. This can both be attributed to an instrumental SZA dependence and/or a seasonal variability related to the atmospheric profile in the RTM and
GUV TOC from Ny-Ålesund measured throughout two selected periods: April 2018
When all measurements and seasons are considered as a whole, we have chosen
an SZA correction of GUV TOC data to harmonize with other ground-based
instruments at the stations. All available GUV
Results from statistical fit of GUV
The harmonization method described above is applied to the three GUV instruments operated by NILU, which are co-located with other ground-based ozone monitoring instruments. Total ozone is also derived for the other stations in the UV network (presented in Table 1 and Fig. 1), but for these instruments a different approach is used. A description of the method and results will be presented in a separate paper.
Under heavy cloud conditions the ozone retrievals are usually less accurate.
An extreme example is discussed by Mayer et al. (1998) for a thunderstorm
case. They found that multiple scattering caused errors as large as 300 DU.
A less extreme situation, which is more representative for Norway, is
exemplified in Fig. 4. The figure shows eCLT (black line) and total ozone column (red line) derived from GUV measurements in Oslo between 11:00 and 17:00 UTC on 9 September 2018. Figure 4 indicates a gradual ozone decrease throughout the day, but what is most interesting is the occurrence of ozone
peaks when eCLT is very low. The uncertainty in total ozone increases as the
cloud optical depth becomes very large, and normally we use a cut-off at
eCLT
Total ozone and eCLT during 1 d (9 September 2018) with heavy clouds at Blindern, University of Oslo. Black arrow indicates a time where eCLT drops below 20 %.
The example in Fig. 4 shows that total ozone increases by 15 DU (
Ozone difference between GUV and Brewer DS (and SAOZ) as a function of eCLT: Oslo
Based on this analysis we have introduced a linear ozone correction
Ozone cloud correction for eCLT
The full GUV TOC time series from 1995 and onwards have been harmonized with
respect to the SZA and eCLT corrections described above. Specifically, TOCs
have been divided by the fit function
Correlation, bias, and SD in total ozone from GUV and Brewer (and SAOZ) instruments. The left columns are for uncorrected GUV data, whereas the right columns are for SZA- and CLT-corrected GUV total ozone data. Bias and SD are both expressed in DU and percent (in parenthesis).
Ratios of GUV
The ratios between GUV and Brewer DS (and SAOZ) TOC are visualized in Fig. 6 for the three stations: Oslo (top), Andøya (centre), and Ny-Ålesund (bottom). Compared to Fig. 2 no systematic seasonality can be seen in the ratios. Ny-Ålesund is possibly an exception, where low GUV TOC values are seen in late fall most of the years. These measurements are performed at very high SZA (84–89
Total ozone differences (in %) between GUV and GOME-2
Corrected GUV TOCs have been compared to GOME-2A and OMI TM3DAM v4.1 (Eskes
et al., 2003) data for Oslo, Andøya, and Ny-Ålesund. It should be
emphasized that GUV data are homogenized with respect to Brewer DS (and
SAOZ) data and that any offset between Brewer and satellite data most likely
will be reflected by offset in GUV–GOME-2 and GUV–OMI ozone data. Figure 7 shows the difference (in %) of daily noontime GUV and GOME-2 total ozone for the period 2007–2019 (left column) and GUV vs. OMI for the period 2004–2019 (right column). Results for Oslo are shown in the top row, Andøya in the centre row, and Ny-Ålesund in the bottom row. The correlations, biases, and SDs are listed in Table 5. At Oslo, the noontime total ozone is never
calculated at SZA
Correlation, bias, and SD in daily noontime total ozone from
Figure 7 and Table 5 show that GOME-2 gives slightly better agreement with GUV TOC compared to OMI. For all stations, the SD is higher for GUV–OMI than for GUV–GOME-2, both when the entire GUV time series and data with SZA
For total ozone assessment and trends studies, the established Brewer instruments would normally be used. However, as demonstrated in previous sections, GUV measurements can provide realistic and stable time series and are suitable for separate studies of long-term changes of the ozone layer. GUV instruments that are co-located with a Brewer or another standard TOC instrument for 2–3 years (until harmonization parameters are established) can afterwards be moved to a new location for independent TOC measurements. The harmonization procedure is used to minimize small systematic errors in GUV TOC data and assumes that Brewer data are without error. However, it should be noted that TOC retrievals at large SZAs can be uncertain if the new site has a very different ozone climatology compared to the original site, as explained in Sect. 2.3. Data from the GUV instruments are also very useful to extend the measuring season at sites with reduced staff and/or characterized by harsh environmental conditions. The case of Ny-Ålesund, where Brewer data are very sparse due to a rough climate that requires a high attendance, is a clear example of GUV usefulness. In Ny-Ålesund as much as 52 % of TOC daily means have solely been based on GUV measurements during the last 5 years.
Even at sites like Oslo and Andøya, where good attendance and less harsh conditions allow more robust Brewer operations, GUV TOC can fill in missing data and extend the measuring season. Brewer zenith sky (ZS) or global irradiance (GI) measurements (WOUDC, 2019) are normally performed under cloudy conditions. However, these measurements can also be impacted by high SZA, heavy clouds, or technical problems. The last 5 years, 14 % of the daily mean TOC values at Andøya are retrieved from GUV to fill in for missing Brewer DS/ZS/GI measurements.
The overall GUV data coverage at the Norwegian stations is very good. If we
disregard the two calibration campaigns in 2005 and 2019, the GUV-511 in
Oslo has been in operation
The GUV network was established during a period where a significant downward
trend in total ozone had been observed for most places on Earth. Statistical
analysis of the Dobson (D56) time series from Oslo 1978–1998 revealed an
annual average total ozone decrease of
Linear trends in the annual average total ozone at the three stations have been calculated, and the results are shown in Fig. 8: Oslo in the top panel, Andøya in the centre panel, and Ny-Ålesund in the bottom panel. For the Oslo station we have a full year of data in 1995, whereas the measurements in Tromsø (Andøya) and Ny-Ålesund started in mid-1995, and a full year of data is not available until 1996. Thus 1995 is omitted from the time series at these two stations. Results from the linear regression analyses are presented in Table 6. In addition to changes in annual mean total ozone, the table includes also linear trends for winter (December–February), spring (March–May), summer (June–August), and fall (September–November).
Seasonal and annual changes in total ozone in Oslo, at Andøya,
and in Ny-Ålesund for the period
The annual means in Oslo are based on data from January to December, for Andøya the means are calculated for the months from February to mid-November, and data from Ny-Ålesund are based on data from March to October. For the two northernmost stations the winter averages cannot be retrieved because of the polar night. Note also that the fall trend results for Ny-Ålesund, presented in Table 6, do not include November.
Due to different months included in the Oslo, Andøya, and Ny-Ålesund annual means, the absolute values are not comparable. Still, there are many similarities in the three data sets. Even though Oslo and Ny-Ålesund are separated by more than 2000 km, the years with low annual average TOC often coincide. Annual variations in the ozone transport from its source region in the tropics toward the polar regions during the winter will often have similar impacts at all our stations, and variations in the Quasi-Biennial Oscillation (QBO), El Niño–Southern Oscillation (ENSO), the solar cycle, and stratospheric aerosols will give significant interannual variability in total ozone (WMO, 2018; Svendby and Dahlback, 2004). The explanatory variables mentioned above are often used in TOC trend studies to eliminate variability caused by natural sources and to get a more precise picture of trends related to emissions of anthropogenic sources such as ODSs.
Annual average total ozone in Oslo, at Andøya/Tromsø, and in Ny-Ålesund. Linear trends for the whole period 1995/1996–2019 are marked with orange lines; ozone changes for 1999–2019 are in blue.
Total ozone column measured in Ny-Ålesund in spring 2020 with the SAOZ instrument (black triangles), GUV (black line), OMI satellite (orange line), and GOME-2 (blue line).
In Fig. 8, linear observational trends for the entire period (from 1995/1996 to 2019) are marked in orange, whereas changes for the last 20 years are marked in blue. The latter trend estimate is done to eliminate the years in the mid-1990s with very low ozone, partly influenced by the Mt. Pinatubo eruption and the cold Arctic winters in 1996 and 1997 (Solomon, 1999). The analysis reveals a total ozone increase for the period 1995/1996–2019 at all stations and for all seasons. However, only half of the positive trend results are statistically significant to a 95 % confidence level (
Despite a general increase in TOC during the last decades, Lawrence et al. (2020) reported that the TOC over the northern polar region was exceptionally low in late winter and early spring 2020. The average total ozone for February to April was the lowest value registered since the start of satellite measurements in 1979. The low TOC was partly caused by an exceptionally cold and persistent stratospheric polar vortex, which provided ideal conditions for chemical ozone destruction (Grooß and Müller, 2021; Manney et al., 2020; Wohltmann et al., 2020). These low ozone values resulted in enhanced UV radiation, and the average UV index measured by the GUV instrument in Ny-Ålesund in April 2020 was elevated by 34 % relative to the average 1979–2019 level (Bernhard et al., 2020).
Figure 9 shows GUV total ozone in Ny-Ålesund from mid-February to May 2020, and the low ozone levels from the end of March to mid-April are clearly seen. Total ozone values from SAOZ, GOME-2, and OMI (TM3DAM v4.1) are included in the figure for comparison. The study from Wohltmann et al. (2020) showed that the Arctic ozone at 18 km altitude was depleted by up to
Total ozone column values from Oslo/Kjeller in 2019 measured with the GUV instrument (black line), OMI satellite (orange line), and GOME-2 (blue line). The red circle indicates the mini ozone hole over Scandinavia on 4 December 2019.
Episodes of very low total ozone content are not limited to early spring and periods of several weeks. They can also occur for a few days because of unusual meteorological or atmospheric conditions, as observed at Kjeller in late 2019. In Fig. 10, GUV noontime total ozone from Oslo and Kjeller in 2019 is compared to GOME-2 and OMI data from Oslo (12:00 UTC values). The black line shows GUV TOC data, whereas blue and orange lines represent GOME-2 and OMI measurements, respectively. The lack of GUV data from mid-May and June is caused by the calibration campaign at DSA (see Sect. 2.2). GUV data prior to mid-May 2019 are from Oslo, whereas measurements after July 2019 were performed at Kjeller outside Oslo. The GUV comparison to GOME-2 and OMI overpass data from Oslo indicates that the agreement between ground-based measurements and satellite data is as good at Kjeller as in Oslo. A very interesting episode is the extremely low total ozone values measured on 4 December 2019 (red circle in Fig. 10). On this day, the noontime GUV ozone value at Kjeller was only 193 DU. This is the lowest value measured by the GUV instrument in Oslo/Kjeller the last 20 years. GOME-2 and OMI from Oslo also measured very low total ozone at 12:00 UTC on this day, 201 and 203 DU, respectively. At 18:00 UTC the previous day the total ozone value from OMI was as low as 193.5 DU.
In the fall/winter of 2019 the Arctic polar vortex formed earlier than usual (Manney et al., 2020; Lawrence et al., 2020). Temperatures were low enough for PSC formation by mid-November 2019, earlier than in any previous year since at least 2004. PSCs were visible over Norway during a large part of winter 2019/20. However, in early December, chorine activation and associated chemical ozone loss were still limited. Dameris et al. (2021) indicate that a mini ozone hole over southern Norway on 4 December 2019 was caused by advection of lower-latitude air masses and increased tropopause height. Figure 11 shows total ozone from the GOME-2 satellite at 12:00 UTC on this day. As seen in the figure, the TOC was below 200 DU in the middle parts of Norway, northern Sweden, and southwestern Finland.
Total ozone column on 4 December 2019 at 12:00 UTC from the GOME-2A satellite (data downloaded from
As described in Sect. 2, the effective cloud transmittance (eCLT) expresses the effect of clouds, aerosols, and surface albedo on the UV radiation reaching the ground. In the present study an eCLT of 100 % represents a clear sky with no surface reflection. An eCLT value above 100 % can occur in case of scattered clouds and/or enhanced surface reflection, e.g. snow.
Figure 12 shows annual average noontime eCLT values and trends at the three stations: Oslo (orange line), Andøya/Tromsø (grey/black line), and Ny-Ålesund (blue line). Linear regression analyses indicate that there are no changes in eCLT at Oslo or Andøya. However at Ny-Ålesund, eCLT has decreased over the last 25 years, and a negative trend of 5 %–6 % is evident from Fig. 12. The change in eCLT is even more pronounced if we only consider the months from late spring and early summer (April–June), as shown in Fig. 13. For these 3 months the overall decreases in eCLT are
Annual average noontime eCLT measured in Oslo, Tromsø/Andøya, and in Ny-Ålesund from 1995/1996 to 2019. Trends in eCLT are indicated as dotted lines.
Monthly mean eCLT in Ny-Ålesund for April, May, and June 1995/1996 to 2019. Trends in eCLT are indicated as dotted lines.
To examine possible monthly differences and changes in the cloud cover in
Ny-Ålesund for the period 1995–2019, cloud data from the Norwegian
Centre for Climate Services (NCCS) have been utilized (see Sect. 2.1). NCCS
cloud data at 12:00 UTC have been selected to reflect the period where GUV eCLT noontime values are measured. Figure 14 shows the number of clear days for April (blue), May (orange), and June (black) for the years 1995–2019. The average number is
Number of monthly clear-sky days observed in Ny-Ålesund in April, May, and June 1995–2019. Trends are indicated as dotted lines. Data are from the NCCS database.
Effective cloud transmittance (eCLT) in Ny-Ålesund for 1996–2019. “All data” represent monthly noontime average eCLT where all days are included. “Clear-sky data” represent monthly eCLT noontime average for days with eCLT
The cloud data from NCCS will partly explain why the overall eCLT in Fig. 13 is highest for April and lowest for June. However, the data will not necessarily give the full explanation of the decreasing GUV eCLT trend from 1996–2019. To examine whether the decrease in eCLT also is affected by albedo change, clear-sky data (defined as noontime eCLT
Monthly mean clear-sky eCLT in Ny-Ålesund for April, May, and June 1996 to 2019. Trends in clear-sky eCLT are indicated as dotted lines.
Theoretical calculations (Degünther et al., 1998; Degünther and Meerkötter, 2000; Lenoble, 2000) show that surface ultraviolet irradiance measurements may be influenced by albedo variations more than 10–20 km away. Kylling and Mayer (2001) showed that for Tromsø, Norway, a declining snowline in mountainous areas may have about a 25 % (50 %) effect on cloudless (cloudy) surface irradiance measurements. These findings support the suggestion that the clear-sky eCLT trends in Ny-Ålesund are due to albedo changes. The changes can be attributed to local snow/ice conditions but also to ice/snow changes several kilometres away from the measuring site.
As seen from Fig. 15, there can be large eCLT variations from one year to another. In April 2006 there was a minimum eCLT value of only 103 %. As indicated in Fig. 14, there was only one clear day in this month (20 April), a day which was classified as category 2 from the NCCS data (a quarter of the sky had clouds). The GUV eCLT minute values indicate that a thin cloud or haze occasionally covered the sun and resulted in relatively low noontime average eCLT on this day. April 2009 is an opposite example where the noontime eCLT was very high. On this day a large fraction of the NCSS cloud data were classified as category 0, meaning that the sky was cloud-free for several days. According to snow data from NCCS, the snow depth in Ny-Ålesund was high in April 2009. In addition, the ice extent in the Barents Sea in spring 2009 was large compared to previous years (Norwegian Polar Institute, 2020). The combined effect of these three factors resulted in a peak eCLT in April 2009.
Clear-sky eCLT mean values and trends from the GUV are summarized in Table 7. The average clear-sky eCLT for April 1996–2000 is 125.7 %, whereas the April average for 2015–2019 is 115.2 %, a decline of
The eCLT results from Ny-Ålesund imply that there has been a significant change in albedo with reduction of snow/ice in the Svalbard area throughout the last 25 years, especially for the spring months. Related results were found by Bernhard (2011), who showed that the onset of snowfall at Barrow, Alaska, advanced by almost 2 weeks per decade for the period 1991–2011. Also, albedo studies from Möller and Möller (2017) have demonstrated a significant negative albedo trend of the glaciers of Svalbard over the period 1979–2015, and data from the Norwegian Polar Institute show that the sea-ice extent in April in the Barents Sea has considerably declined the last decades (Norwegian Polar institute, 2020). These findings on Arctic albedo change and ice melt clearly support existing reports and publications on ongoing climate change (Wunderling et al., 2020; IPCC, 2018).
The Norwegian UV network has been in operation for 25 years, and the unique GUV data can be used to derive a broad range of atmospheric and biological exposure parameters, including total ozone column (TOC), UV index, and cloud transmittance. The instruments are relatively simple to operate and maintain and measure continuously throughout the day with 1 min time resolution.
The 25-year-long records of GUV TOC measurements in Norway have been
re-evaluated and harmonized. For the three stations located in Oslo, at
Andøya, and in Ny-Ålesund there are annual TOC increases of
GUV measurements of effective cloud transmittance (eCLT) in Ny-Ålesund, Svalbard, reveal a negative eCLT trend for the spring, indicating that the albedo at this site has decreased over the past 25 years. This is most likely a consequence of an ongoing ice melt caused by increased temperatures in the Svalbard area.
Harmonized GUV TOC and eCLT data can be accessed at
TMS designed the study and performed the analyses. BJ, AD, AK, and GHB performed supporting simulations and analyses. BP and VV provided Brewer no. 50 data. GHH was responsible for SAOZ data. TMS wrote the paper, and all authors provided input on the paper for revision before submission.
The authors declare that they have no conflict of interest.
We thank the Norwegian Environment Agency for funding total ozone and UV measurements in Oslo/Kjeller, Andøya, and Ny-Ålesund. The authors would like to thank Reidar Lyngra at Alomar (Andøya) and staff at the Norwegian Polar Institute for keeping the instruments running. The contribution from LATMOS, especially Andrea Pazmino, on SAOZ data processing is highly acknowledged. We would also like to thank the two referees for helpful comments and suggestions.
This paper was edited by Stelios Kazadzis and reviewed by two anonymous referees.