Accepted Manuscript Evaluation of the Cloudy Sky Solar Uva Radiation Exposures A.V. Parisi, N. Downs, J. Turner PII: DOI: Reference:
S1011-1344(14)00173-0 http://dx.doi.org/10.1016/j.jphotobiol.2014.05.012 JPB 9752
To appear in:
Journal of Photochemistry and Photobiology B: Biology
Received Date: Revised Date: Accepted Date:
20 December 2013 6 May 2014 8 May 2014
Please cite this article as: A.V. Parisi, N. Downs, J. Turner, Evaluation of the Cloudy Sky Solar Uva Radiation Exposures, Journal of Photochemistry and Photobiology B: Biology (2014), doi: http://dx.doi.org/10.1016/ j.jphotobiol.2014.05.012
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
EVALUATION OF THE CLOUDY SKY SOLAR UVA RADIATION EXPOSURES A.V. Parisi,1,* N. Downs1 and J. Turner1 1
Faculty of Health, Engineering and Sciences, University of Southern Queensland,
Toowoomba, Australia.
*To whom correspondence should be addressed: Ph: 61 7 46312226; Email: [email protected]
1
Abstract The influence of cloud on the solar UVA (320-400 nm) exposures over five minute periods on a horizontal plane has been investigated. The first approach used cloud modification factors that were evaluated using the influence of clouds on the global solar exposures (310-2,800 nm) and a model developed to apply these to the clear sky UVA exposures to allow calculation of the five minute UVA exposures for any cloud conditions. The second approach established a relationship between the UVA and the global solar exposures. The models were developed using the first six months of data in 2012 for SZA less than or equal to 70 o and were applied and evaluated for the exposures in the second half of 2012. This comparison of the modelled exposures for all cloud conditions to the measured data provided an R2 of 0.8 for the cloud modification model, compared to an R2 of 0.7 for the UVA/global model. The cloud modification model provided 73% of the five minute exposures within 20% of the measured UVA exposures. This was improved to 89% of the exposures within 20% of the measured UVA exposures for the cases of cloud with the sun not obscured.
Keywords: Cloud; UVA; Solar radiation; Exposure; Cloud Modification Factor; Ultraviolet
INTRODUCTION For a given solar zenith angle (SZA), cloud cover plays the most significant contribution compared to other influencing factors on the solar UV exposures on the surface of the earth [1]. The influence of cloud can vary from enhancements in the UV over that on a corresponding clear day [2] to a reduction to almost zero. The influence of clouds on the UV is determined by the amount of cloud, cloud height, cloud optical depth and cloud type [3][4]. Additionally, the influence by cloud varies with wavelength [5][6][3]. It has been reported that for the UVA waveband, the attenuation by clouds increases with wavelength [3]. 2
Previous research has considered the influence of cloud on the daily exposures to UVA (320 – 400 nm) and UVB (280 – 320 nm) for a higher latitude station at 58o15’ N [3]. During the period when the SZA range is higher, the average cloud modification factor (CMF) for the UVA daily exposures was 0.36. However, there was a wide range of CMF values ranging from 0.15 to 0.76. Other research has employed two years of hourly data [7] and employed the cloud modification factor [8] to determine a non-linear relationship with the UV irradiances in the 295 to 385 nm waveband. The hourly UVA and UVB exposures and their relationship to the broadband solar radiation have been reported [9]. Other research has incorporated two cloud height categories of low/medium and high and half hourly exposures for the evaluation of the erythemal UV [1].
The CMFs have also been employed by [10] to investigate linear, quadratic and potential function regressions of the influence of cloud as measured at one point of the day at 13:00 EST on the erythemal UV measured with a radiometer with a spectral response that approximates the erythemal action spectrum. The erythemal UV exposures over a day have been evaluated using measured ozone and cloud modification factors [11]. Other research has employed daily and hourly exposures to establish relationships between global solar radiation (295-2,800 nm) in Brazil and the ultraviolet (290-400 nm), photosynthetically active (380700 nm) and near infrared (695-2800 nm) exposures [12]. Further research has investigated other non-linear models to improve the relationship between hourly data on the CMF with the integrated irradiances over the waveband (295-385 nm) to the global radiation [13]. FoyoMoreno et al. [7] found that cloud above five oktas has an increasing influence on the reduction of irradiances in the band 295 to 385 nm. These researchers developed a relationship between the cloud modification factor for global radiation (CMFG) and the cloud
3
modification factor for total UV in the waveband 295 to 385 nm. The photosynthetically active radiation compared to clear sky values has been employed to determine the UVB under variable cloud [14]. Another approach has employed a statistical analysis of the known parameters of cloud to predict the resulting UV [15]. A sky/cloud formula incorporating various parameters of the cloud has been developed [16]. Other research has employed the influence of cloud on the CMFG to determine a relationship between this and the CMF for eye damaging UV [17] and the vitamin D effective UV [18].
Research employing the CMF has generally employed these to provide the erythemal UV, UVB or total UV, with minimal research [9] undertaken to provide the CMF for the evaluation of the UVA. The UVA waveband has been implicated in initiation of skin cancers [19], along with premature skin ageing and wrinkling [20][19]. The research that has considered the influence of cloud on the UVA has employed manual observation of cloud by trained observers. However, the influence of cloud can change over short time frames. This paper will address this by considering the influence of cloud on the UVA over short time frames by the automatic measurement of the cloud cover with a sky camera and the concurrent measurement of five minute UVA exposures. The research in this paper investigates the influence of cloud on the UVA five minute exposures and the use of the cloud modification of the global solar irradiances to enable evaluation of the UVA five minute exposures for all sky conditions.
MATERIALS AND METHODS UVA and Solar Radiation Data The UVA (320-400 nm) exposures to a horizontal plane for every five minute period were measured by a calibrated UVA meter (model 501A, Solar Light Inc, PA, USA) recording 4
UVA exposures for each five minutes on an unobstructed University building roof under all sky conditions at the sub-tropical site of Toowoomba, (27.6o S, 151.9o E, 693 m) Australia. This instrument was calibrated in April 2012 to a spectroradiometer (DTM300, Bentham Instruments Inc., UK) with calibration traceable to the National Physical Laboratory, UK. The global solar exposures for each five minutes were recorded by a nearby solar pyranometer (model CMP3-L, Kipp and Zonen, supplied by Campbell Scientific, Australia Pty Ltd) recording data in the waveband 310 nm to 2,800 nm. This instrument was calibrated by the manufacturer at the time of purchase and recorded data on a Campbell data logger. The fraction of cloud cover was provided by a sky camera (model TSI440, Yankee Environmental Systems, USA) located on the same building roof. This instrument recorded at each five minute point the fraction of sky covered in cloud and whether or not the solar disc was cloud obscured.
The first six months of data from 2012 for SZA less than or equal to 70o with the solar noon SZA over this period ranging from 5 o to 51 o was employed using two approaches to establish models for the evaluation of UVA exposures. The first approach establishes the relationship between the cloud modification factor for global solar radiation and the cloud modification factor for the UVA. The second approach establishes the relationship between the UVA and the global solar exposures. The reason for using the first six months is that the data covers the range of SZA encountered. This then left the second six months of data from 2012 to be employed to provide an evaluation of this developed model.
Cloud Modification Factors
5
The cloud modification factors were defined as the exposures under all sky conditions compared to the corresponding clear sky condition exposures [7]. For the UVA waveband (CMFUVA), these were calculated as:
ܨܯܥ =
(1)
ೌೝ
where UVA are the exposures over each five minute period and UVAclear are the corresponding exposures for a cloud free period as determined from the cloud free envelope for the respective SZA.
The UVAclear was calculated using an empirical model fitted to the clear sky five minute exposure data for the first six months of 2012. The empirical model was employed due to the more straight forward approach of this compared to using a transfer model. The data for the times when the sky was clear of cloud were selected by determining the cases where the cloud cover was not more than 2% as measured by the TSI cloud camera. The data were not classified according to ozone levels as the influence of ozone absorption in the UVA waveband is minimal. The influence of aerosols and albedo were not taken into account as the measurement site does not have significant changes in the ground cover albedo and the atmospheric aerosols. There were 3,475 five minute exposures that satisfied the clear sky requirement or 10.6% of the total number of 32,709 five minute exposures during 2012. These clear sky data for the first six months were plotted as a function of cos-1(SZA) [7] and a power function fitted to provide:
ܷܸܣ = 14,838ି ݔଵ.ସଽ
ܬ. ݉ ିଶ
(2)
6
with an R2 = 0.97, where x = cos-1(SZA) and SZA is the solar zenith angle in radians.
Similarly, the cloud modification factors for the global solar irradiances (CMFsolar) were calculated as:
ܨܯܥ௦ =
ீ
(3)
ீೌೝ
where G are the five minute exposures and Gclear are the corresponding exposures for a cloud free period as calculated from the cloud free envelope for the respective SZA. In a similar manner to the UVAclear, this was calculated by fitting an empirical model to the cloud free data for the first six months of 2012 when the cloud cover was not more than 2%. These were plotted as a function of cos(SZA) and a power function fitted to give:
ܩ = 68,107ݖଵ.ଶଷଽ
ܬ. ݉ିଶ
(4)
where z = cos(SZA), with an R2 = 0.96.
Evaluation of UVA Exposures The data for the first six months of 2012 comprised of 14,720 values when all of the data from the TSI, UVA meter and the global solar meter were being collected. Various models were tested for the relationship between the CMFsolar and CMFUVA data for the whole set of data from 1 January 2012 to 30 June 2012 for SZA less than or equal to 70 o. The model fitted was a power function of the CMFsolar which is a modified version of a model previously employed [7] to provide the UVA exposures as follows: 7
ఉ
ܷܸܣܸܷ = ܣ × ߙ × ܨܯܥ௦
ܬ. ݉ିଶ
(5)
where α and β are coefficients determined from the first six months of data. These were evaluated by plotting CMFUVA versus CMFsolar and the fitting of the model to determine α and
β.
The second approach established a model between UVA and G as follows: ܷܸܩ × ߛ = ܣ
ܬ. ݉ିଶ
(6)
where ߛ is a coefficient to be determined from the first six months of data.
The models were tested on the data from 1 July 2012 to 31 December 2012 to calculate the UVA exposures for all sky conditions which were compared to the measured UVA exposures. The data for the second six months of 2012 comprised of 19,061 values when all of the data from the TSI, UVA meter and the global solar meter were being collected. The cases of the solar disc obscured and the solar disc not obscured were also considered for the approach using the cloud modification factors.
RESULTS AND DISCUSSION The measured UVA exposures over the five minute intervals were averaged for each of the cloud cover ranges and for each of the SZA ranges of SZA ≤ 20o, 20o < SZA ≤ 40o, 40 o < SZA ≤ 60 o and 60o < SZA and are provided in Figure 1. These four SZA ranges have been employed in a similar manner to previous research [7]. The error bars are plotted for the SZA range of ≤ 20 o only, to show the range of the data and represent one standard deviation. There
8
is a decrease in the UVA exposures with increasing SZA. Even though for a fixed SZA range the means of the UVA exposures for the different cloud amounts from 2 oktas to 7 oktas are within the error bars, there is a general downward trend in the exposures for increasing cloud cover. The size of the error bars is due to the range of cloud types and cloud optical depths encountered.
The CMFUVA values calculated for each of the cloud amounts are shown in 1.4
CMFUVA
SZA
S1011-1344(14)00173-0 http://dx.doi.org/10.1016/j.jphotobiol.2014.05.012 JPB 9752
To appear in:
Journal of Photochemistry and Photobiology B: Biology
Received Date: Revised Date: Accepted Date:
20 December 2013 6 May 2014 8 May 2014
Please cite this article as: A.V. Parisi, N. Downs, J. Turner, Evaluation of the Cloudy Sky Solar Uva Radiation Exposures, Journal of Photochemistry and Photobiology B: Biology (2014), doi: http://dx.doi.org/10.1016/ j.jphotobiol.2014.05.012
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
EVALUATION OF THE CLOUDY SKY SOLAR UVA RADIATION EXPOSURES A.V. Parisi,1,* N. Downs1 and J. Turner1 1
Faculty of Health, Engineering and Sciences, University of Southern Queensland,
Toowoomba, Australia.
*To whom correspondence should be addressed: Ph: 61 7 46312226; Email: [email protected]
1
Abstract The influence of cloud on the solar UVA (320-400 nm) exposures over five minute periods on a horizontal plane has been investigated. The first approach used cloud modification factors that were evaluated using the influence of clouds on the global solar exposures (310-2,800 nm) and a model developed to apply these to the clear sky UVA exposures to allow calculation of the five minute UVA exposures for any cloud conditions. The second approach established a relationship between the UVA and the global solar exposures. The models were developed using the first six months of data in 2012 for SZA less than or equal to 70 o and were applied and evaluated for the exposures in the second half of 2012. This comparison of the modelled exposures for all cloud conditions to the measured data provided an R2 of 0.8 for the cloud modification model, compared to an R2 of 0.7 for the UVA/global model. The cloud modification model provided 73% of the five minute exposures within 20% of the measured UVA exposures. This was improved to 89% of the exposures within 20% of the measured UVA exposures for the cases of cloud with the sun not obscured.
Keywords: Cloud; UVA; Solar radiation; Exposure; Cloud Modification Factor; Ultraviolet
INTRODUCTION For a given solar zenith angle (SZA), cloud cover plays the most significant contribution compared to other influencing factors on the solar UV exposures on the surface of the earth [1]. The influence of cloud can vary from enhancements in the UV over that on a corresponding clear day [2] to a reduction to almost zero. The influence of clouds on the UV is determined by the amount of cloud, cloud height, cloud optical depth and cloud type [3][4]. Additionally, the influence by cloud varies with wavelength [5][6][3]. It has been reported that for the UVA waveband, the attenuation by clouds increases with wavelength [3]. 2
Previous research has considered the influence of cloud on the daily exposures to UVA (320 – 400 nm) and UVB (280 – 320 nm) for a higher latitude station at 58o15’ N [3]. During the period when the SZA range is higher, the average cloud modification factor (CMF) for the UVA daily exposures was 0.36. However, there was a wide range of CMF values ranging from 0.15 to 0.76. Other research has employed two years of hourly data [7] and employed the cloud modification factor [8] to determine a non-linear relationship with the UV irradiances in the 295 to 385 nm waveband. The hourly UVA and UVB exposures and their relationship to the broadband solar radiation have been reported [9]. Other research has incorporated two cloud height categories of low/medium and high and half hourly exposures for the evaluation of the erythemal UV [1].
The CMFs have also been employed by [10] to investigate linear, quadratic and potential function regressions of the influence of cloud as measured at one point of the day at 13:00 EST on the erythemal UV measured with a radiometer with a spectral response that approximates the erythemal action spectrum. The erythemal UV exposures over a day have been evaluated using measured ozone and cloud modification factors [11]. Other research has employed daily and hourly exposures to establish relationships between global solar radiation (295-2,800 nm) in Brazil and the ultraviolet (290-400 nm), photosynthetically active (380700 nm) and near infrared (695-2800 nm) exposures [12]. Further research has investigated other non-linear models to improve the relationship between hourly data on the CMF with the integrated irradiances over the waveband (295-385 nm) to the global radiation [13]. FoyoMoreno et al. [7] found that cloud above five oktas has an increasing influence on the reduction of irradiances in the band 295 to 385 nm. These researchers developed a relationship between the cloud modification factor for global radiation (CMFG) and the cloud
3
modification factor for total UV in the waveband 295 to 385 nm. The photosynthetically active radiation compared to clear sky values has been employed to determine the UVB under variable cloud [14]. Another approach has employed a statistical analysis of the known parameters of cloud to predict the resulting UV [15]. A sky/cloud formula incorporating various parameters of the cloud has been developed [16]. Other research has employed the influence of cloud on the CMFG to determine a relationship between this and the CMF for eye damaging UV [17] and the vitamin D effective UV [18].
Research employing the CMF has generally employed these to provide the erythemal UV, UVB or total UV, with minimal research [9] undertaken to provide the CMF for the evaluation of the UVA. The UVA waveband has been implicated in initiation of skin cancers [19], along with premature skin ageing and wrinkling [20][19]. The research that has considered the influence of cloud on the UVA has employed manual observation of cloud by trained observers. However, the influence of cloud can change over short time frames. This paper will address this by considering the influence of cloud on the UVA over short time frames by the automatic measurement of the cloud cover with a sky camera and the concurrent measurement of five minute UVA exposures. The research in this paper investigates the influence of cloud on the UVA five minute exposures and the use of the cloud modification of the global solar irradiances to enable evaluation of the UVA five minute exposures for all sky conditions.
MATERIALS AND METHODS UVA and Solar Radiation Data The UVA (320-400 nm) exposures to a horizontal plane for every five minute period were measured by a calibrated UVA meter (model 501A, Solar Light Inc, PA, USA) recording 4
UVA exposures for each five minutes on an unobstructed University building roof under all sky conditions at the sub-tropical site of Toowoomba, (27.6o S, 151.9o E, 693 m) Australia. This instrument was calibrated in April 2012 to a spectroradiometer (DTM300, Bentham Instruments Inc., UK) with calibration traceable to the National Physical Laboratory, UK. The global solar exposures for each five minutes were recorded by a nearby solar pyranometer (model CMP3-L, Kipp and Zonen, supplied by Campbell Scientific, Australia Pty Ltd) recording data in the waveband 310 nm to 2,800 nm. This instrument was calibrated by the manufacturer at the time of purchase and recorded data on a Campbell data logger. The fraction of cloud cover was provided by a sky camera (model TSI440, Yankee Environmental Systems, USA) located on the same building roof. This instrument recorded at each five minute point the fraction of sky covered in cloud and whether or not the solar disc was cloud obscured.
The first six months of data from 2012 for SZA less than or equal to 70o with the solar noon SZA over this period ranging from 5 o to 51 o was employed using two approaches to establish models for the evaluation of UVA exposures. The first approach establishes the relationship between the cloud modification factor for global solar radiation and the cloud modification factor for the UVA. The second approach establishes the relationship between the UVA and the global solar exposures. The reason for using the first six months is that the data covers the range of SZA encountered. This then left the second six months of data from 2012 to be employed to provide an evaluation of this developed model.
Cloud Modification Factors
5
The cloud modification factors were defined as the exposures under all sky conditions compared to the corresponding clear sky condition exposures [7]. For the UVA waveband (CMFUVA), these were calculated as:
ܨܯܥ =
(1)
ೌೝ
where UVA are the exposures over each five minute period and UVAclear are the corresponding exposures for a cloud free period as determined from the cloud free envelope for the respective SZA.
The UVAclear was calculated using an empirical model fitted to the clear sky five minute exposure data for the first six months of 2012. The empirical model was employed due to the more straight forward approach of this compared to using a transfer model. The data for the times when the sky was clear of cloud were selected by determining the cases where the cloud cover was not more than 2% as measured by the TSI cloud camera. The data were not classified according to ozone levels as the influence of ozone absorption in the UVA waveband is minimal. The influence of aerosols and albedo were not taken into account as the measurement site does not have significant changes in the ground cover albedo and the atmospheric aerosols. There were 3,475 five minute exposures that satisfied the clear sky requirement or 10.6% of the total number of 32,709 five minute exposures during 2012. These clear sky data for the first six months were plotted as a function of cos-1(SZA) [7] and a power function fitted to provide:
ܷܸܣ = 14,838ି ݔଵ.ସଽ
ܬ. ݉ ିଶ
(2)
6
with an R2 = 0.97, where x = cos-1(SZA) and SZA is the solar zenith angle in radians.
Similarly, the cloud modification factors for the global solar irradiances (CMFsolar) were calculated as:
ܨܯܥ௦ =
ீ
(3)
ீೌೝ
where G are the five minute exposures and Gclear are the corresponding exposures for a cloud free period as calculated from the cloud free envelope for the respective SZA. In a similar manner to the UVAclear, this was calculated by fitting an empirical model to the cloud free data for the first six months of 2012 when the cloud cover was not more than 2%. These were plotted as a function of cos(SZA) and a power function fitted to give:
ܩ = 68,107ݖଵ.ଶଷଽ
ܬ. ݉ିଶ
(4)
where z = cos(SZA), with an R2 = 0.96.
Evaluation of UVA Exposures The data for the first six months of 2012 comprised of 14,720 values when all of the data from the TSI, UVA meter and the global solar meter were being collected. Various models were tested for the relationship between the CMFsolar and CMFUVA data for the whole set of data from 1 January 2012 to 30 June 2012 for SZA less than or equal to 70 o. The model fitted was a power function of the CMFsolar which is a modified version of a model previously employed [7] to provide the UVA exposures as follows: 7
ఉ
ܷܸܣܸܷ = ܣ × ߙ × ܨܯܥ௦
ܬ. ݉ିଶ
(5)
where α and β are coefficients determined from the first six months of data. These were evaluated by plotting CMFUVA versus CMFsolar and the fitting of the model to determine α and
β.
The second approach established a model between UVA and G as follows: ܷܸܩ × ߛ = ܣ
ܬ. ݉ିଶ
(6)
where ߛ is a coefficient to be determined from the first six months of data.
The models were tested on the data from 1 July 2012 to 31 December 2012 to calculate the UVA exposures for all sky conditions which were compared to the measured UVA exposures. The data for the second six months of 2012 comprised of 19,061 values when all of the data from the TSI, UVA meter and the global solar meter were being collected. The cases of the solar disc obscured and the solar disc not obscured were also considered for the approach using the cloud modification factors.
RESULTS AND DISCUSSION The measured UVA exposures over the five minute intervals were averaged for each of the cloud cover ranges and for each of the SZA ranges of SZA ≤ 20o, 20o < SZA ≤ 40o, 40 o < SZA ≤ 60 o and 60o < SZA and are provided in Figure 1. These four SZA ranges have been employed in a similar manner to previous research [7]. The error bars are plotted for the SZA range of ≤ 20 o only, to show the range of the data and represent one standard deviation. There
8
is a decrease in the UVA exposures with increasing SZA. Even though for a fixed SZA range the means of the UVA exposures for the different cloud amounts from 2 oktas to 7 oktas are within the error bars, there is a general downward trend in the exposures for increasing cloud cover. The size of the error bars is due to the range of cloud types and cloud optical depths encountered.
The CMFUVA values calculated for each of the cloud amounts are shown in 1.4
CMFUVA
SZA
