1. Introduction
From the total energy emitted by the Sun, only 8% of what arrives at the top of Earth’s atmosphere is ultraviolet radiation, being still reduced to 4% upon arriving at the Earth’s surface due to physical processes in the atmosphere and to additional―geographical, temporal, astronomical, and others―factors (Iqbal, 1983; Huffman, 1992; Guarnieri et al., 2004; Silva et al., 2008; Fountoulakis et al., 2020). This radiation falls mainly into the UV-A range (315-400 nm), which has low atmospheric absorption, and in the UV-B range (280-315 nm) which, even if less intense, is important to several photochemical and biological processes. A third component, UV-C (200-280 nm) is absorbed into the stratosphere during the process of ozone layer formation (Koller, 1965; Robinson, 1966; Guarnieri et al., 2004; Sliney, 2007; Bilbao et al., 2011; Andrade and Tiba, 2016). Monitoring of UV intensity and its fluctuation is important for biological and human health issues, and one of the most used quantifying parameters is the erythemal daily dose (EDD) expressed in J m-2 and defined as the integration over the day of the incoming ultraviolet irradiance on a horizontal surface weighted with the erythema action spectrum (McKinlay and Diffey, 1987; WMO, 2007).
Since the ozone depletion detection (Farman et al., 1985), efforts have been made to quantify EDD magnitude and to estimate long-term variation trends towards higher values in ultraviolet radiance, being this variation continuously estimated from satellite measurements, with a large spatial cover and with data records sufficient to produce trend estimates (Ialongo et al., 2008; Herman, 2010), mostly focused on specific regions. The factors affecting ultraviolet radiation are generally well known, and the importance of each factor varies with latitude, climate, and the amount of atmospheric pollution at each site, giving origin to non-linear interactions involving complex absorption and scattering processes (Koronakis et al., 2002; Kerr and Fioletov, 2008).
The main parameter modulating ultraviolet absorption is stratospheric ozone (90% of total O3), which presents a seasonal pattern due to natural processes of formation, transport, and destruction. Ozone concentrations are at their lowest levels in fall and highest in spring (Wakamatsu et al., 1989; André et al., 2003). However, this variability is influenced by natural phenomena and anthropogenic activities (Fahey and Hegglin, 2011; Bais et al., 2015), the latter being linked to the industrial production of nitrogen dioxide (NO2) formed from the oxidation of nitrous oxide (N2O) coming from the troposphere. NO2 can attain high concentrations in the stratosphere (90% of all NO2), where it destroys O3 through catalytic processes by sequestering active radicals (Seinfeld and Pandis, 1998).
The impact of ultraviolet solar radiation on environmental and human health is well known by the scientific community (Davis and Sims, 1983; Caldwell et al., 2003; Tiegte et al., 2007; Cardoso, 2011; Rodríguez, 2017). However, the knowledge of variability and quantity of spatiotemporal ultraviolet radiation and its relationship to processes in the atmosphere (ozone formation, ozone hole, etc.) has received considerable attention from research projects (Grant and Heisler, 2000; Alados-Arboledas et al., 2003; Efstathiou et al., 2005; Bais et al., 2007; Kerr and Fioletov, 2008; Meleti et al., 2009; Barnard and Wenny 2010; Elsner et al., 2010; Rieder et al., 2010; Antón et al., 2012; Bernhard et al., 2013; Wolfram et al., 2013; Lopo et al., 2014; Liu et al., 2017; Čížková et al., 2018; Jebar et al., 2019; Becerra-Rondón et al., 2021; Gholamnia et al., 2021; Raptis et al., 2021). In this context, the southern part of South America, being closer to the Antarctic ozone hole, is an area exposed to high levels of ultraviolet radiation throughout the year (Corrêa and Pires 2013; Zaratti et al., 2014), making this region important (Díaz, 2006) at a global scale for UV studies.
While stratospheric ozone is the major atmospheric absorber of ultraviolet radiation, local changes in air pollutants may mask an ultraviolet radiation increase associated with low total ozone episodes, especially in the presence of aerosols (Meleti et al., 2009). This confusing factor may be considered by analyzing the ozone time series, helping to understand UV relationships with pollutants. Besides, considering that for southern Brazil ultraviolet fluxes show significant variability between seasons (Kirchhoff et al., 2000; Guarnieri et al., 2004; Kerr and Fioletov, 2008), quantifying UV variability becomes important on decision-making processes aiming to mitigate effects from exposition to ultraviolet radiation. Based on these perceptions, the objective of this research was: (a) to perform an evaluation, at pixel scale, of the long-term UV variability in Rio Grande do Sul State; (b) to estimate time trends in this variability, and (c) to investigate possible relationships between UV higher values with total O3 and total NO2. To this end, satellite data was our primary source of information. Our research will be presented in the following sections.
2. Material and methods
2.1 Study area
Rio Grande do Sul is the Brazilian southernmost state, having international borders with Argentina to the west and Uruguay to the south (Fig. 1). Its area is 281 707 km², and with more than 11.5 million inhabitants it is the fifth most populated state in the country. The region has a humid subtropical climate with a large seasonal variation with hot summers and well-defined cold winters. Mean temperatures vary from 15 to 18 ºC, with lows as much as -10 ºC (June and July) and highs going up to 40 ºC (December to March) (Livi, 2002). The surface elevation ranges from sea level up to 1200 m with the highest points northeast of the state (Fig. 1).
2.2 Data source
This research was performed from data acquired by the Ozone Monitoring Instrument (OMI) onboard satellite Aura. This sensor is equipped with a spectrometer pointed to the nadir which measures the ultraviolet light (264-504 nm) coming from the Sun and back-scattered by the atmosphere. The differential optical absorption spectroscopy (DOAS) and total ozone mapping spectrometer (TOMS) algorithms were developed to derive several products (Levelt et al., 2006), of which we used nitrogen dioxide (OMNO2d) derived from DOAS and, from TOMS, total ozone (OMTO3d), and EDD (OMUVBd) (Krotkov et al., 2006; Tanskanen et al., 2006). For the product OMNO2d (total column density) data is provided in molecules cm-2, at a spatial resolution of 0.25º × 0.25º (lat/lon), and with a daily frequency. Product OMTO3d (total column density) is also daily but with 1.0º × 1.0º (lat/lon) spatial resolution and in Dobson units (where 1 DU = 2.7 × 1018 molecules O3 cm-2). Product OMUVBd (cumulative daily dose) is in J m-2.
In this study, information of 15 years (2006-2020) including 5452 images with daily measurements (99% of the series) were used to assess and analyze EDD distributions above the study area. This data was acquired from the data provider GES-DISC (NASA, 2021) and processed using free software RStudio. The spatial resolution of products OMUVBd and OMTO3d was resampled from 1.0º × 1.0º to 0.25º using a RStudio functionality to have uniformity with product OMNO2d. For this 0.25º resolution, the study area is covered by 420 cells.
2.3 Analysis
The spatiotemporal analysis of the long-term (2006-2020) EDD consisted of two parts: (1) variation and trend, and (2) relationships of total O3 and total NO2 with higher values of EDD. All the analyses and maps have been processed using statistical functionalities of the software R, version 4.0.5 (R Core Team, 2021). The spatiotemporal variations and the corresponding trends were obtained through the following calculations:
The mean per cell for all 420 cells of the study area for three periods (yearly, monthly and seasonal), using their daily values, was calculated as follows:
where Vari is the mean of the variable in each cell for the respective period; the start and end dates correspond to the first and last date of each period; n is the number of days in the period.
The coefficient of variation (CV), which is a statistical measure of the dispersion of data points in a data series around the mean, is expressed as:
where σ is the standard deviation and μ is the mean.
From annual means of EDD, the spatiotemporal trend was determined using the Mann-Kendall trend test and Theil-Sen’s slope estimator (Mann, 1945; Kendall, 1975), both of them being nonparametric tests. The Mann-Kendall test indicates trend and Theil-Sen’s slope estimator indicates the slope of this trend, as follows:
where EDD i and EDD j indicate the sequential data values of the time series in the years T i and T j , respectively, with j > i. The calculated Q is the estimated magnitude of the trend slope in the time series of the data, where negative values indicate a downward trend and positive values indicate an upward trend.
To evaluate the relationships between higher values of EDDs with the variables total O3 and total NO2, we compiled the spatial and temporal distribution of EDDs at each month for the period 2006-2020 (Fig. 2a). We were then able to establish the 90th percentile (EDD90+) by pixel in each month (Fig. 2b). Once the EDD90+ days were identified and the values of total O3 and total NO2 were selected, corresponding to these higher UV values for the concerned pixels, we performed a partial correlation (rpartial) and we valued only those results with a level of statistical significance α ≤ 0.05. This method allowed us to evaluate the correlation between two variables considering the effect (variance) of a third one (Fig. 2c), and is typically applied when there is reasonable evidence of correlations between the considered variables, which is the case involving UV, NO2, and O3. Therefore, we applied this method successively for total O3 and total NO2, obtaining the respective correlations with EDD90+, while the other variable remained constant. To better understand and describe the results, classification criteria according to the type of correlation and the respective level of significance were established.
Additionally, having generated monthly EDD90+ thresholds, these results were grouped by year to estimate (a) the number of times (frequency) that a pixel is repeated under this condition each year, and (b) the number of days per year that registered this condition.
3. Results
3.1 Spatial and temporal distribution over Rio Grande do Sul State
The average daily ultraviolet erythemal irradiance for Rio Grande do Sul State, derived from OMI observations for the 15-year studied period, is presented in Figure 3a. Values vary from 3300 to 3700 J m-2, with a steady irradiance gradient in latitude of about 66.7 J m-2 degree-1, decreasing southwards. The long-term spatiotemporal variations are presented in Figure 3b, with values from 1.8 to 3.6%, a fluctuation that approximately follows the gradient shown in Figure 3a. Figure 3c shows that 29.76% of the area had a long-term decreasing trend, while 6.19% had a long-term increasing trend; the remaining state’s area didn’t record significant (α>0.05) trends. The largest increasing trends coincide with north-east areas, which have the highest elevations.
Monthly averages had values as high as 6028 J m-2 in December and as low as 1123 J m-2 in June (Fig. 4 and Table I). Besides, it was observed that April and August were the months with the largest changes in ultraviolet erythemal irradiance, dropping from 2811 J m-2 in April to 1633 J m-2 in May and the inverse between August (1746 J m-2) to September (2666 J m-2) (Fig. 4, Table I). Regarding seasonal variations, these are more intense in the spring-summer period (4007 J m-2 to 5411 J m-2) rather than in the autumn-winter period (1378 J m-2 to 2899 J m-2) (Table I).
Average age | |||||||||||||||||
2006 | 2007 | 2008 | 2009 | 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | |||
Min | 3268.80 | 3150.66 | 3147.99 | 3047.56 | 3062.55 | 2923.35 | 3144.24 | 3127.65 | 3062.54 | 3038.59 | 3107.56 | 3098.80 | 3103.57 | 3066.23 | 3258.64 | 3523.18 | |
Max | 4048.69 | 3816.12 | 3890.14 | 3834.19 | 3938.40 | 3817.10 | 3948.04 | 3980.54 | 3766.24 | 3817.86 | 3938.17 | 3898.87 | 3950.04 | 3832.75 | 4047.60 | ||
Avg | 3682.38 | 3448.17 | 3557.41 | 3455.17 | 3519.18 | 3423.95 | 3579.51 | 3592.31 | 3445.45 | 3388.14 | 3539.02 | 3526.36 | 3538.92 | 3478.90 | 3672.87 | ||
SD | 169.21 | 140.71 | 155.85 | 167.83 | 191.46 | 180.91 | 164.58 | 186.94 | 152.03 | 169.42 | 178.87 | 168.47 | 185.78 | 155.05 | 181.05 | ||
January | Min | 5488.48 | 5939.72 | 5617.14 | 5504.68 | 4961.95 | 4994.64 | 6129.48 | 5905.89 | 5352.54 | 5365.28 | 5844.16 | 5643.63 | 5064.32 | 4578.50 | 5716.81 | 5961.76 |
Max | 6429.91 | 6797.85 | 6359.55 | 6183.63 | 6557.13 | 6215.91 | 6800.09 | 6674.06 | 6503.99 | 6242.56 | 6679.80 | 6385.24 | 6298.20 | 6154.74 | 6395.92 | ||
Avg | 5894.54 | 6279.19 | 6080.60 | 5856.46 | 5843.93 | 5718.32 | 6452.78 | 6283.06 | 6005.87 | 5741.27 | 6237.72 | 5908.81 | 5681.96 | 5431.39 | 6010.57 | ||
February | Min | 5131.20 | 5251.41 | 4581.02 | 5082.78 | 4674.11 | 4193.23 | 4538.52 | 4412.04 | 4767.73 | 4971.58 | 4879.41 | 4871.09 | 5434.91 | 4965.13 | 5298.93 | 5430.43 |
Max | 5907.30 | 6044.07 | 5697.91 | 6044.57 | 6082.00 | 5348.34 | 6213.93 | 5603.45 | 5808.65 | 6007.45 | 5841.39 | 5684.70 | 6162.54 | 5623.77 | 5875.65 | ||
Avg | 5608.42 | 5558.54 | 5372.52 | 5590.69 | 5531.27 | 5005.12 | 5423.73 | 5091.80 | 5335.45 | 5317.14 | 5479.22 | 5292.31 | 5883.53 | 5365.86 | 5600.85 | ||
March | Min | 3916.93 | 3546.14 | 3833.16 | 3834.76 | 4105.68 | 3793.35 | 3690.78 | 3638.05 | 3864.86 | 4158.70 | 3551.23 | 3636.17 | 3938.33 | 3722.94 | 4022.40 | 4252.25 |
Max | 4759.97 | 4738.58 | 4565.17 | 4866.91 | 4760.19 | 4438.87 | 4823.19 | 4177.45 | 4381.57 | 4527.89 | 4273.16 | 4464.31 | 4847.67 | 4322.19 | 5116.54 | ||
Avg | 4410.86 | 4025.67 | 4255.23 | 4373.04 | 4456.69 | 4157.43 | 4495.57 | 3954.00 | 4129.94 | 4346.50 | 3868.19 | 4232.09 | 4390.95 | 4046.54 | 4641.13 | ||
April | Min | 2326.92 | 2455.26 | 2462.56 | 2552.55 | 2464.89 | 2279.12 | 2342.64 | 2412.18 | 2336.02 | 2585.23 | 1657.28 | 2164.06 | 2279.33 | 2382.82 | 2447.82 | 2811.85 |
Max | 3510.89 | 3172.53 | 3097.19 | 3580.86 | 3049.51 | 3114.04 | 3120.64 | 3500.83 | 3072.16 | 3155.04 | 3092.69 | 3185.53 | 3400.84 | 3191.08 | 3471.86 | ||
Avg | 2975.73 | 2800.14 | 2846.04 | 3093.56 | 2685.01 | 2756.59 | 2762.93 | 2936.22 | 2719.92 | 2925.25 | 2336.12 | 2615.73 | 2918.77 | 2727.55 | 3078.20 | ||
May | Min | 1405.48 | 1261.47 | 1302.16 | 1417.23 | 1256.34 | 1287.07 | 1447.21 | 1319.54 | 1249.94 | 1356.93 | 1186.56 | 1256.72 | 1328.20 | 1337.50 | 1530.71 | 1633.16 |
Max | 2358.26 | 1764.32 | 2045.70 | 1937.72 | 1646.95 | 2037.09 | 2150.80 | 2002.00 | 1903.50 | 1931.15 | 1733.36 | 1687.89 | 2001.26 | 1627.08 | 2025.22 | ||
Avg | 1804.05 | 1564.14 | 1717.16 | 1707.28 | 1417.34 | 1698.00 | 1857.55 | 1660.58 | 1547.57 | 1649.24 | 1530.44 | 1462.03 | 1583.99 | 1534.18 | 1763.86 | ||
June | Min | 913.73 | 804.79 | 885.78 | 824.74 | 865.82 | 879.28 | 826.25 | 975.40 | 791.18 | 940.61 | 782.16 | 998.74 | 852.47 | 802.04 | 857.91 | 1123.32 |
Max | 1479.06 | 1185.81 | 1289.37 | 1322.81 | 1427.01 | 1219.37 | 1328.42 | 1381.97 | 1177.11 | 1529.54 | 1357.12 | 1356.21 | 1223.61 | 1595.85 | 1286.04 | ||
Avg | 1206.58 | 996.13 | 1099.70 | 1161.73 | 1094.31 | 1101.48 | 1074.86 | 1222.13 | 987.52 | 1182.22 | 1131.44 | 1153.66 | 1069.88 | 1287.23 | 1081.01 | ||
July | Min | 992.79 | 892.97 | 1073.82 | 1017.59 | 894.56 | 902.76 | 1064.08 | 1007.08 | 836.25 | 1002.54 | 892.31 | 928.12 | 743.58 | 933.30 | 996.78 | 1266.24 |
Max | 1724.80 | 1306.05 | 1868.98 | 1376.06 | 1493.15 | 1444.95 | 1391.53 | 1772.42 | 1420.21 | 1417.55 | 1682.35 | 1773.31 | 1521.86 | 1512.09 | 1556.63 | ||
Avg | 1364.39 | 1139.92 | 1358.41 | 1254.90 | 1233.95 | 1122.99 | 1278.12 | 1425.48 | 1205.34 | 1159.73 | 1347.39 | 1495.27 | 1072.57 | 1215.55 | 1319.58 | ||
August | Min | 1393.45 | 1142.04 | 1570.57 | 1410.61 | 1446.86 | 1240.69 | 1380.80 | 1449.65 | 1611.09 | 1257.35 | 1467.05 | 1294.07 | 1396.50 | 1467.94 | 1519.99 | 1746.56 |
Max | 2025.51 | 1820.27 | 1919.62 | 1888.74 | 2029.95 | 1788.92 | 2154.31 | 1991.12 | 2177.76 | 2252.10 | 1976.43 | 2094.30 | 1932.47 | 2248.12 | 2216.56 | ||
Avg | 1796.03 | 1448.47 | 1752.28 | 1615.48 | 1705.71 | 1564.51 | 1789.24 | 1715.55 | 1924.75 | 1793.68 | 1788.36 | 1829.37 | 1645.72 | 1861.23 | 1967.97 | ||
September | Min | 2666.08 | 2157.96 | 2172.60 | 2117.35 | 2226.36 | 2372.50 | 2319.42 | 2138.77 | 2369.27 | 2299.69 | 2441.69 | 2295.98 | 2178.74 | 2521.37 | 2228.43 | 2666.57 |
Max | 3089.37 | 3049.60 | 2924.10 | 2606.62 | 3162.04 | 3094.03 | 3078.70 | 3008.47 | 2747.02 | 3205.11 | 3392.04 | 3236.68 | 2814.52 | 3297.93 | 3217.45 | ||
Avg | 2881.63 | 2528.25 | 2579.68 | 2398.34 | 2582.10 | 2633.46 | 2668.19 | 2647.56 | 2527.15 | 2729.82 | 2984.77 | 2694.28 | 2565.38 | 2947.72 | 2630.15 | ||
October | Min | 3932.25 | 3218.74 | 3429.20 | 3606.20 | 3916.43 | 3174.74 | 3424.19 | 3811.28 | 3553.57 | 3442.41 | 3638.36 | 3267.41 | 3352.05 | 3139.08 | 3586.36 | 3945.37 |
Max | 4534.77 | 3796.06 | 3944.03 | 4604.51 | 4756.87 | 4794.01 | 3964.24 | 4667.41 | 4425.30 | 4133.68 | 4664.19 | 4060.87 | 4321.33 | 4004.53 | 5037.91 | ||
Avg | 4342.24 | 3401.74 | 3687.50 | 4130.93 | 4332.84 | 4083.44 | 3615.90 | 4249.34 | 4001.07 | 3783.76 | 4073.70 | 3640.47 | 3917.30 | 3595.26 | 4325.11 | ||
November | Min | 4811.50 | 5289.39 | 4943.70 | 3938.55 | 4630.76 | 4984.41 | 5322.94 | 4970.05 | 4824.56 | 4360.12 | 5352.40 | 5117.16 | 5193.20 | 4940.17 | 4884.87 | 5411.89 |
Max | 5904.39 | 5866.41 | 6287.10 | 5099.73 | 5817.45 | 5901.41 | 5989.90 | 6207.80 | 5637.73 | 5246.93 | 6067.33 | 6268.46 | 6094.22 | 5744.24 | 5839.23 | ||
Avg | 5381.80 | 5573.39 | 5749.70 | 4365.35 | 5251.96 | 5473.05 | 5742.53 | 5486.37 | 5391.29 | 4770.12 | 5674.64 | 5852.64 | 5767.54 | 5361.81 | 5336.20 | ||
December | Min | 6246.74 | 5848.03 | 5904.21 | 5263.71 | 5306.82 | 4978.41 | 5244.60 | 5491.89 | 5193.51 | 4722.65 | 5598.16 | 5712.44 | 5481.18 | 6004.03 | 6012.68 | 6028.75 |
Max | 6860.07 | 6251.92 | 6682.90 | 6498.12 | 6478.56 | 6408.23 | 6360.69 | 6779.48 | 5939.91 | 6165.36 | 6498.18 | 6588.90 | 6782.01 | 6671.44 | 6532.23 | ||
Avg | 6522.29 | 6062.50 | 6190.04 | 5914.31 | 6095.00 | 5772.99 | 5792.72 | 6435.61 | 5569.49 | 5258.93 | 6016.23 | 6139.60 | 5969.40 | 6372.44 | 6319.77 | ||
Summer | Min | 5622.14 | 5679.72 | 5367.46 | 5283.72 | 4980.96 | 4722.10 | 5304.20 | 5269.94 | 5104.59 | 5019.84 | 5440.58 | 5409.05 | 5326.80 | 5182.55 | 5676.14 | 5806.98 |
Max | 6399.09 | 6364.61 | 6246.79 | 6242.11 | 6372.56 | 5990.83 | 6458.24 | 6352.33 | 6084.18 | 6138.46 | 6339.79 | 6219.61 | 6414.25 | 6149.98 | 6267.93 | ||
Avg | 6008.42 | 5966.74 | 5881.05 | 5787.15 | 5823.40 | 5498.81 | 5889.74 | 5936.82 | 5636.94 | 5439.11 | 5911.05 | 5780.24 | 5844.96 | 5723.23 | 5977.07 | ||
Min | 2549.78 | 2420.96 | 2532.63 | 2601.51 | 2608.97 | 2453.18 | 2493.54 | 2456.59 | 2483.60 | 2700.29 | 2131.69 | 2352.32 | 2515.29 | 2481.09 | 2666.98 |
*Data obtained from the OMI dataset.
Considering that the ultraviolet erythemal irradiance in Rio Grande do Sul has a defined spatial and temporal pattern (Figs. 3, 4), and based on the 90th percentile for each pixel in each month (see Table SI in the supplementary material), we derived that the highest incidence of high values of ultraviolet erythemal irradiance took place in 2006 (a maximum of 67 occurrences per pixel and a total of 157 days), and the lowest incidence was recorded in 2011 (a maximum of 20 occurrences per pixel and a total of 49 days) (Fig. 5).
The temporal relationships of total O3 and total NO2 with the higher values of ultraviolet erythemal irradiance are presented in Figure 6. It is observed that ultraviolet erythemal irradiance and total NO2 present a similar pattern throughout the year with a minimum between the months that correspond to winter and maximum between the months that correspond to summer (Fig. 6a, c). Meanwhile, total O3 exhibits minimum in summer and maximum in spring (Fig. 6b).
The partial (r partial ) pixel-to-pixel relationships between EDD90+ and total O3 and total NO2 vary by month (see Tables SII-SXIII in the supplementary material). In particular, we observed that occurrences of EDD90+ can be grouped in five situations (Table II), as follows: (a) pixels where there is a dominant negative correlation of EDD90+, both with total O3 and total NO2; (b) pixels where there is a negative correlation with total O3 and a positive one with total NO2; (c) negative correlation predominates with total NO2; (d) negative correlation predominates with total O3, and (e) cases where EDD90+ occurrence is not explained by its correlation neither with total O3 nor total NO2.
aTotal NO2 (TNO2); btotal O3 (TO3). | |
Clasification criteria | Class |
rpartiala ≤ 0 & rpartialb ≤ 0; pvaluea ≤ 0.05 & pvalueb ≤ 0.05 | TO3-TNO2 |
rpartiala ≥ 0 & rpartialb ≤ 0; pvaluea ≤ 0.05 & pvalueb ≤ 0.05 | TO3-TNO2 |
rpartialb ≤ 0; pvaluea ≥ 0.05 & pvalueb ≤ 0.05 | TO3 |
rpartiala ≤ 0; pvaluea ≤ 0.05 & pvalueb ≥ 0.05 | TNO2 |
pvaluea ≥ 0.05 & pvalueb ≥ 0.05 | Unexplained |
The spatiotemporal distribution of these five situations is shown in Figure 7a, b, in terms of the percentages which explain the EDD90+ observations. From the analysis of results presented in Figure 7, it is suggested that total O3 is the main factor explaining EDD90+ observations in all months, having a weight ranging from 25% (April) to 93% (November), with the notable exception of May, where its contribution (and of all other considered factors) is not significant. Synergies, either positive or negative, of total O3 and NO2 did not seem to play significant roles in EDD90+ observations, while total NO2 explains EDD90+ observations at most for 4% of all pixels. On the other hand, between 5 to 96% of pixels with EDD90+ seem not to be linked neither to O3 nor to NO2, with higher frequencies in January (40% of the area) and May (96% of the area). EDD90+ data for these months remains mostly unexplained by our approach.
4. Discussion
4.1 Spatial and temporal distribution over the Rio Grande do Sul State
The latitude dependence of ultraviolet erythemal irradiance within Rio Grande do Sul shown in Figure 3a is part of an ampler dependence already reported (Corrêa, 2015). Even if the overall ultraviolet irradiance in southern Brazil, and especially in Rio Grande do Sul, presents lower values if compared with northern regions, detailed knowledge of its magnitude and distribution is of interest by demographical reasons: the state presents the country’s highest rates of melanoma cancer, a fact that is attributed to a higher percentage of people of Caucasian or European origin that present a greater predisposition to develop this pathology when exposed to UV radiation (Lee-Taylor et al., 2010; Corrêa, 2015; Imanichi et al., 2017; INCA, 2020). From Figure 3a it can be seen that UV erythemal irradiance varies significantly across the state, a perception that is reinforced by the results presented in Figure 3b and especially in Figure 3c, which suggest that irradiance had a positive time variation in a non-negligible fraction of the state’s area during the studied period. Presently, information on skin cancer incidence in different regions of the state is still lacking, and this study provides useful information for the understanding of data that eventually will be available and the corresponding management. It is also important to note that the erythemal ultraviolet irradiance received in the fall of 2020 (with 3161 J m-2) was higher than in the previous 14 autumns, in agreement with what was reported by Becerra-Rondón et al. (2021) on increases in the ultraviolet radiation index for that season in previous years. Additionally, we also note that the highest EDD values, which have been measured in 2006, coincide with a year of very low solar activity, as measured by the number of solar spots, while the lowest EDDs have been measured in 2011, that is, about a half solar cycle after 2006 (Rampelotto et al., 2009; Valachovic and Zurbenko, 2014; BOM, 2021).
Presently, we observe that the northeast part of the study area has the lowest latitudes and the highest altitudes, associated with the highest radiation values and with changes in intensity according to the period (for the monthly and seasonal case) (Fig. 4). This behavior is due, at least in part, to the fact that at higher sites the path through the atmosphere is shorter and scattering processes (Rayleigh and Mie) are not as important compared to lower sites, where radiation is reduced by increasing density and atmospheric components, which promotes greater attenuation (Ziemke et al., 2000). This effect is intensified when combined with the variation of the solar zenith angles (geographical and astronomical effects) (Schmucki and Philipona, 2002). Although these two effects establish a spatial and temporal distribution pattern in our study area, changes in cloud cover, temperature, secondary effects of the Antarctic ozone hole, and other variables (e.g., ozone, aerosols, and air pollutants) also greatly influence short- and long-term variations (Kirchhoff et al., 1996; Koronakis et al., 2002; Schmucki and Philipona, 2002; Guarnieri et al., 2004; Kerr and Fioletov, 2008; Salgado et al., 2010; Schmalfuss et al., 2014; Nunes et al., 2020).
4.2. Relationships of total O 3 and total NO 2 with high values of erythemal daily dose
Even if the temporal behavior of total O3 and total NO2 (which affect ultraviolet radiation [Fig. 6b, c]) is well defined, the simultaneous presence of more than one of them can lead to non-linear interactions involving complex absorption and scattering processes (Kirchhoff et al., 2000; Krotkov et al., 2001; Koronakis et al., 2002). Moreover, the magnitude of the ultraviolet intensity varies in function of the chosen period (daily, monthly, and annual mean), and from geographical characteristics (Kerr, 2005; Ialongo et al. 2011; Čížková et al., 2018). From these perceptions and considering the atmospheric processes and associated health hazards characteristic of our study area, we highlighted where higher values of ultraviolet radiation occurred, spatially and temporally (EDD90+) (Fig. 5). From these results, the relationships of the two modulating factors (total O3 and total NO2) with respect to EDD90+ were investigated.
We found that for nine months of an average year (from June to February) between 53 to 93% of the study area have EDD90+ primarily associated with total O3 in the form of a negative correlation (Fig. 7a, b). This behavior has also been reported in other works under all weather conditions (Kerr and McElroy, 1993; Fioletov et al., 1997; Guarnieri et al. 2004; Rieder et al., 2010; Bernhard et a., 2013; Lopo et al., 2014; Čížková et al., 2018; Raptis et al., 2021). In the remaining months (March to May) from 70 to 96% of the area have EDD90+ unexplained by total O3 and/or total NO2, being therefore attributed to variables not considered in this study. A large inversion of the total O3 weight between May and June was observed. While it is true that total O3 is the main driver of long-term variations in ultraviolet radiation intensities, it can also be argued that the solar zenith angle (SZA) and cloud cover are non-negligible factors in the short term, gaining more importance during autumn and summer, respectively (Lopo et al., 2014; de Bock et al., 2014; Čížková et al., 2018). In this context and considering the complexity of interactions with ultraviolet radiation, during autumn and winter Rio Grande do Sul State is also under the effect of atmospheric circulation patterns (equatorial Atlantic mass, tropical Atlantic, continental tropical, and polar) that promote the transport of flows of ozone-poor air from the tropics towards high latitudes, accompanied by higher air pressures that lead to clear or partially cloudy skies, leading to the incidences of EDD90+ (Stick et al., 2006; Rossato, 2011).
Meanwhile, although in a smaller proportion of the studied area (between 0.2-6.4%) throughout the year, the relationship of total NO2 with EDD90+ values exists in three situations (Table II), depending of the pixels (regions) considered, which shows how the complexity of the interaction of more than one variable in time and space generates the same effect.
Associating our results and their derived perceptions to what has been reported elsewhere (Rieder et al., 2010; Cížková et al., 2018), high values of ultraviolet radiation tend to be associated to low total O3 levels and with the absence of clouds or with partly cloudy skies. Furthermore, the influence of air pollutants (e.g., NO2) over ultraviolet radiation is significant, acting in complex ways, even on cloudy days when the prevailing effect of clouds was expected. It was reported that two of these outcomes are that both high and low concentrations of NO2 lead to higher ultraviolet radiation (Dickerson et al., 1997; WMO, 2020; Musiolková et al., 2021), which strongly suggests that although there are specific patterns for each variable depending on the season and cloud cover, the latter is possibly the most important, since it plays a crucial role in the influence of incident solar radiation during the day (Krotkov et al., 2001; Alados-Arboledas et al., 2003; Musiolková et al., 2021)
5. Conclusions
Human exposure to solar ultraviolet radiation has important public health implications, and evidences of hazards associated with UV overexposure have been reported in many investigations. Starting from these perceptions and taking them as motivation, this study used remote sensing data to analyze the spatiotemporal characteristics of EDDs in Rio Grande do Sul, a region where this important information is still lacking. Our results showed that (1) the annual average of the daily accumulated EDDs was 3523.18 J m-2, with a long-term decreasing trend in 29.76% of the state’s area; (2) places located in lower latitudes and higher altitudes had higher incidences of EDD; (3) there is a well-defined temporal pattern, with a higher average value in summer (5806.98 J m-2) and a lower average value in winter (1378.71 J m-2); (4) during the 15 years of the study (2006-2020), the highest number of days with high EDD was in 2006, a year of lower solar activity, and (5) the incidence of EDD90+ (the more intense radiation for the whole studied period) showed a negative correlation with total O3, and in a few cases either positive or negative correlations with total NO2, however, the lack of correlations of EDD90+ with both total O3 and total NO2 in many instances (places and dates) is an indication that there are variables not considered in this study. Even if part of our generated EDD90+ data remains unexplained, our results on the spatiotemporal distribution of the highest incidence of this radiation can be relevant for official policies concerning the regional prevention of skin neoplasms, which are already among the highest in Brazil.