spectral uv irradiance on vertical surfaces: a case study

Photochemistry and Photobiology, 1999, 69(4): 464-470

Spectral UV lrradiance on Vertical Surfaces: A Case Study

Ann R. Webb*’, P. Weihs2 and M. Blumthaler3 ’University of Manchester Institute of Science and TechnoloQy, Manchester, England; -. 2Universit& fur Bodenkultur, Vienna, Austria and Wniversity of Innsbruck, Innsbruck, Austria

Received 11 June 1998; accepted 22 December 1998

ABSTRACT

The UV spectral irradiance on horizontal and vertically oriented surfaces was measured throughout a cloudless day (18 July 1995) at Izana station, Tenerife, using a Bentham DTM3OO spectroradiometer scanning from 290 to 500 nm in steps of 5 nm. Results show that irradiance measured on a horizontal surface is not proportional to irradiance on a vertical surface. The relation between the two depends upon orientation of the vertical surface, zenith angle and wavelength. At short UVB wavelengths surfaces directed toward the solar azimuth received their maximum irradiances much closer to solar noon than the maxima for longer wavelengths. Some vertical surfaces also received significantly more irradiance than the horizontal surface at long wavelengths during all but the central hours of the day, while at short wavelengths all vertical irradiances were less than the horizontal except for the measurements at the extreme ends of the day. Erythemally effective radiation followed the diurnal pattern of irradiations for short UVB wavelengths.

INTRODUCTION Solar UV data, whether measured or calculated, are fre- quently used to assess the risk to a biological system from the local UV environment. Standard solar radiation measurements refer to radiation incident on a horizontal surface from the whole sky (the flat-horizon to horizon hemisphere) and are the sum of direct beam and scattered (diffuse) radiation. Other standard but less common measurements are of the diffuse radiation only (on a horizontal surface) and the direct beam radiation only, which can be referred either to a horizontal surface or to a surface normal to the solar beam. Distributions of sky radiance may also be measured (1-3) but this is a research. rather than routine observational technique. The total sum of the radiance from a 4 pi field of view and the direct (normal to the beam) irradiance is the actinic flux and is important, for example, in photochemical reactions. For a target suspended above the surface. radiation from below will depend on backscattering by the atmosphere

beneath the target and the albedo of the surface. If the target for the radiation is at the surface, then radiance from the 2 pi upper hemisphere plus the direct beam irradiance can ap- proximate the actinic flux. Such measurements can be made directly with suitable geometry of input optics (4) but are not common. A common output of the majority of UV model calculations refers to irradiance on a horizontal surface, although models can be used to calculate sky radiances and hence the radiation incident on other surfaces (2 ,3) and actinic flux. Models that derive the irradiance on sloping surfaces from measurements on a horizontal surface have been described for the broadband visible range (5-9) and perform best if both global and direct or diffuse irradiance measurements are available for input, together with some measure of sky “clearness.” However, these models have not been extended into the UV and most available UV data refer to the horizontal surface irradiance.

In contrast, few biological surfaces are horizontal or flat. Surfaces, or small parts of (curved) surfaces, may take any of the orientations available over a whole sphere. Many biological systems can also move and change their orientation, or that of some of their parts, further complicating an assessment of their prominent alignment. In situations where motion and behavior is complex (e .g . for mammals), modeling individual exposure is not possible and a dosimeter must be used if the true irradiance on a given surface is required. Nevertheless, the irradiance on a horizontal surface is often used as a proxy for other surfaces, mobile or not, because no other data are available.

In this work, the relation between spectral irradiance on a horizontal surface, and that on vertical surfaces of different orientation, has been investigated under ideal conditions throughout a single day with a wide range of zenith angles. The vertical surfaces faced north, south, east and west, up- sun (U,t always toward the solar azimuth) and down-sun (D, away from the sun). While they cannot truly represent the exposure of a complex shape, the surfaces could ap- proximate those thin stripes down the vertical surfaces of the upright human body that are normal to the directions of measurement; or the eye, looking straight ahead at the horizon (with no account taken of shadowing by the brows or squinting).

*To whom correspondence should be addressed at: Department of Physics, UMIST. P.O. Box 88, Manchester M60 IQD, UK. Fax: +11 161 300 3941: e-mail: [email protected]

C I YYY Arnericm Socirt? for Phot(4wAopy 003 I -8655/99 SS.00+0.00

tAhhreiiations: D, down-sun; GAW. Global Atmosphere Watch; H, horizontal: NIST, National Institute of Standards and Technology; PTB. Physikalisch-Technische Bundesanstalt; SZA. solar zenith angle: U. up-sun.

464

Photochemistry and Photobiology, 1999, 69(4) 465

1.2

@ +' ,@ ,$' *6' ,@ ,,$ @' bpo e5 b6' 4Q Wavelmgth (nm)

Figure 1. Ratio of Innsbruck : UMIST measurements for a horizontal surface. Dotted lines were measurements made close to sunrise/ set (0700 and 1800 UTC). The solid lines refer to measurements at 0900, 1000, 1400 and 1500 UTC.

MATERIALS AND METHODS Measurements were made at the Meteorological Observatory, Izana, on the island of Tenerife, on 18 July 1995. The site is at 28.3N and 16.48W and at an altitude of 2367 in. The observatory is a Global Atmosphere Watch (GAW) station (10). Situated on top of a mountain plateau, the station is usually well above the temperature inversion layer that lies between 1200 and 1800 m. Thus, the air is generally free of local anthropogenic influences, very clean and the site is usually above the convective cloud layer.

The instrument used for the measurements was a Bentham DTM300 spectroradiometer. This is a double monochromator with photomultiplier detector, all enclosed in a temperature-stabilized box. Radiation enters the monochromator viu a cosine response teflon diffuser and a quartz fiber light guide. Instrument specifications are given in Webb ( 1 1). The body of the monochromator was dam- aged in transit, and although it was realigned and calibrated on site with a standard traceable to National Institute of Standards and Technology (NIST, USA), the conditions under which this was done were less than ideal. The absolute calibration of the instrument is therefore questionable, but for the short term the instrument per- formed in a stable manner: thus, the ratios of scans presented in this paper are valid. To tie the ratios to the scale of absolute irradiance the data from a second monochromator, a Bentham DM150. operated by University of Innsbruck have been used ( 1 1 ) . Ozone and turbidity (aerosol optical depth) data were also supplied by Univer- sity of Innsbruck, from direct sun spectral measurements.

The Bentham DM 150 spectroradiometer operated by the Univer- sity of Innsbruck was calibrated with a 1000 W halogen lamp traceable to Physikalisch-Technische Bundesanstalt (PTB, Germany) before and after the campaign, with a relative discrepancy of 3% between the two calibrations. This level of stability was confirmed by continuous comparison with a broadband meter during the campaign. The absolute uncertainty of a calibration is estimated to be 55%. This second instrument was following a different schedule and making scans from 285 to 500 nm in steps of 0.5 nni. There were several points during the day when a scan by the DM150 began at 3 min past the hour and coincided with the start of the second horizontal scan (H2) in the UMIST sequence of measurements. Al- though the two scans were not properly synchronized, they were close enough in time to be compared, given the clear sky conditions. Figure 1 shows the ratios of the two measurements for six times throughout the day. The ratios are essentially independent of wavelength, the structure being a feature of the structure in the solar spectrum combined with the different slit functions of the two spec- troradiometers (0.5 nm for the DTM300 and 0.8 nm for the DM150 (11)). The ratios have values of 0.81 t 0.1 (full range) with the two extremes early in the morning and late in the afternoon when measured signals were small and any discrepancy in the cosine functions of the two instruments would be most marked (zenith angle close to 70" in the afternoon and greater in the morning). Any small error in leveling the UMIST diffuser (which was continually being moved) would also be more apparent at these times. During the

middle part of the day the ratios are consistent to within 0.02. Thus, while the absolute values of irradiance measured by the DTM300 may be on the order of 20% too high, the instrument was stable with both wavelength and time. As the ratios presented below do not depend on absolute calibration but on stability, the original UMIST data are used for all ratios, and a correction applied only when absolute data are presented (Figs. 2, 6 ) .

The teflon diffuser of the DTM300 was mounted in a laboratory stand that held the diffuser casing between two black plates. The flat surface of the diffuser was centered in a hole in the front plate with its surface <1 mm below the front surface of the plate. The quartz fiber light guide protruded through a hole in the back plate. The stand, when held in an optical bench mount. allows tilt of the diffuser from the vertical and rotation about the vertical axis. The freedom to tilt was disabled, with the diffuser in the vertical position. The diffuser could then be directed to any point of the compass or the stand removed from the optical bench mount and laid across two parallel blocks so that the diffuser was horizontal and the fiber fell between the blocks. A compass was used to identify objects on the horizon that were to the north, east, south and west of the site, so that the vertical diffuser could quickly be turned and aligned to these directions. The alignment during measurement sequences was then made by eye, with an estimated uncertainty of Z3". The horizontal position of the diffuser was checked with a small spirit level.

The diffuser does not have a perfect cosine response, deviating from the ideal response by more than 5% at angles greater than 60". At 70" the uncertainty is 12% and at 80" it is 25% (1 1). The direct and global radiation measurements from the Innsbruck instrument were used to determine the direcudiffuse ratios of radiation at different wavelengths and times and hence indicate the magnitude of error incurred by the imperfect diffuser. For a horizontal surface the error in global radiation was on the order of 3-78 at 300 nm (with the maximum error at zenith angles of 70") and 3-1 1 8 at 400 nm, dependent on solar zenith angle (SZA). The ratios of direcudiffuse radiation for vertical surfaces have a different dependency on SZA for each aspect (N, S, E. W). In addition diffuse radiation comes from part of the sky hemisphere and partially from the ground. Mak- ing a cosine correction for vertical surfaces would involve so many assumptions about the radiance distribution that we have chosen not to make such adjustments. However, the imperfect cosine response would act in the same way for both horizontal and vertical surfaces, serving to underestimate the true irradiance. When there is little direct beam radiation (large zenith angles) the cosine error is similar for both vertical and horizontal surfaces (assuming isotropy from both sky and ground) and there is little resulting error in the vertical/ horizontal ratio. For a large direct beam component (small zenith angle), the error on a vertical surface with a direct beam component will be greater than that on the horizontal surface, resulting in an underestimate of the verticalhorizontal ratio. This worst case would occur for the U and S surfaces at solar noon, when the uncertainty in the ratio is estimated at 15%.

The spectroradiometer was mounted on a ground-level concrete platform to the SSW of the main building of the Izana observatory. The diffuser, in its horizontal position, was 1.5 m above ground level, with the vertical diffuser being about 10 cm higher than this. The unintermpted field of view of the horizontal diffuser was not a complete hemisphere as there were some obstructions on the horizon. The peak of the volcano Teide. at an altitude of 3780 m, is 15 kni to the SW of the observatory, but other than this there is little to obstruct the horizon from W through S to E. The ground drops away from the observatory to the N and E, with a low building about 80 m to the E subtending an angle of 2-3" to the diffuser. The main observatory building, 30 m to the NNE of the instrument. was a two-storey building with a sloping roof covered in a reflective material, subtending an angle of about 10" to the diffuser. From the N to W the land was flat or sloped away from the site with a view of the sea or clouds below. Other instruments were situated on the platform to the N, W and S but were below the level of the horizontal diffuser. The view normal to the vertical diffuser pointing N was clear of the end of the observatory building. but the building would nonetheless have been within the field of view. To the W were the other instruments, red earth and shrubs in the lower part of the field of view, and Teide, at an elevation of 5" above the horizontal, in the distance. Southwards were only earth and shrubs.

466 Ann R. Webb et a/.

and below the horizon the sea and cloud layer. The view to the E was similar, but with the addition of the low building on the slope away from the site in the bottom part of the field of view. The diffuser has a nominal cosine response, so radiation incident normal to the diffuser, i.e. looking straight and level for a vertical diffuser, is of relatively great importance in the overall signal. Radiation, or objects, at the extremes of the field of view (immediately below and far to the sides) are comparatively insignificant. Thus. the view to the N was obstructed the most, with the observatory building filling approximately 8% of the field of view at a distance of 30 m. In other directions the view was essentially ground and sky. The ground surface was a mixture of red earth. low shrubs and some stones and concrete, with an estimated albedo in the UV of about 10% (l2,13).

Spectral scans were made from 290 to 500 nm with S nm steps, which took 25 s. The diffuser was then reoriented and a new scan begun 1 min after the first scan. In a single sequence of scans the diffuser positions were horizontal ( H I ) . U. D, horizontal (H2). S. W. N. E and horizontal (H3). The complete sequence took 9 min and was repeated every half hour, beginning on the hour and half hour, from 0700 to 1900 UTC. For reference. local noon at Izana was 1311 L'TC. The U and D measurements were made with the vertical diffuser pointing to the sun's azimuth (U) and 180" away from the azimuth (D) These alignments were made by eye and have an uncertainty of r S ' .

On 18 July the solar declination was approximately ZION, and the SZA at noon in Tenerife was 7". During the day the zenith angle changed at an almost constant rate of 12" per hour, or I" in 5 min. A complete sequence of directional measurements therefore covered a change of almost 2' in SZA. Many of the results are presented as directional data normalized to thc horizontal. and H? was used for this normalization. Avcraging H1, H2 and H3 produced virtually the same results. but several of the HI and H 3 scans were missing. therefore thc single H2 scan, were used for consistency. All directional scans were then also taken within 5 min (or I ' of zenith angle) from the horizontal scan used for the normalization.

RESULTS On 18 July 1995, the sky was cloudless above the Izana site, although clouds developed below the site and beneath the inversion layer. The ozone amount was constant throughout the day at 280 -C 2 DU, surface pressure was 772 hPa, and the maximum ambient temperature reached was 22°C. The turbidity (aerosol optical depth at 320 nm) was more unsta- ble. During the morning, the wind was from the NE and the turbidity was 0.0440.05. At about 1300 UTC the wind direction changed to southerly, and turbidity measurements at 1330 showed an increase to 0.095. Turbidity then decreased slowly through the afternoon to return to the morning values by 1730 UTC. The increase in turbidity can be attributed to Saharan dust advected from the S and visible around the site horizon. The following day, with the wind from the S, the dust and turbidity increased significantly. It was under these relatively stable conditions, with the sun rising close to the zenith at noon, that the measurements were made.

Figure 2 shows the measurements made in all six vertical positions. and the horizontal. throughout the day at wavelengths of 300 and 400 nm. The data have been corrected to the absolute values referenced to Innsbruck and PTB by multiplying by a factor of 0.81 (Fig. 1). The general pattern is the same at both wavelengths (and at all other wavelengths measured). The maximum irradiance of the day occurs on the horizontal surface at solar noon (closest measurement at 1300 UTC), when the sun is almost directly overhead. The E- and W-facing vertical surfaces have maxima in the morning and afternoon. respectively. The N- and S-facing surfaces have similar, low levels of irradiance and show less di-

0700 OBW moD 1000 1100 iz'w i3'w i4:w 15'00 iaoo 1700 18.00 1900

Ttme (UTC)

Figure 2. Measured irradiances (W m-? nm-I) on all surfaces at (a) 300 nm and (b) 400 nm: H (thick line); E, W (solid lines): N, S (dotted lines): U (circles): D (triangles). Absolute values corrected according to Fig. 1.

umal variation than the E and W surfaces, although there is a small maximum in the irradiance of the S-facing surface around solar noon. The U surface parallels the vertical surface of maximum irradiance throughout the day, with two maxima in the morning and afternoon (matching E, W) and a local minimum at noon (matching the maximum in S). The D surface parallels the N surface, with low and comparatively constant irradiance throughout the day. The N-facing vertical surface shows a maximum in the morning for measurements at 400 nm, which does not appear in the 300 nm measurements. One possible explanation is that these early morning measurements are enhanced by reflections from the roof of the meteorological building, which was in the field of view of the N-facing surface. Later in the day the direct angular reflections would not have reached the diffuser. Ex- amination of other wavelengths shows that the N-facing morning maximum is apparent and increases in relative magnitude through the visible from 400 to 500 nm, while it de- creases with decreasing wavelength through the UV. Reflec- tivities for most surfaces are much lower in the UV than in the visible.

Significant differences in the diurnal patterns at each wavelength can be found in the relative magnitudes of the irradiances on the different surfaces. The H surface diurnal peak for a wavelength of 300 nm is steeper and narrower at 300 nm than at 400 nm, as expected because the shorter- wavelength is more sensitive to zenith angle (through path- length and strong absorption by ozone). The short-wavelength radiation is also more susceptible to Rayleigh scattering, which is dependent on the inverse fourth power of the wavelength (P). The overall result of this is that diffuse (scattered) radiation at 300 nm comes mainly from the upper part of the sky hemisphere (small zenith angles) ( I ) , with proportionately less radiance coming from around the hori-


1 4

0700 0800 MIW IOW 1 l W 12W 13W 14W 15W 1600 17W 18W 18W n m (UTC)

07W 0800 MIW 1000 1100 1200 1300 14W 15W 1BW 17W 18W 1900

n m (mci

0700 0800 W W 1000 1100 12W 1300 1400 1500 18W 17W 1000 19W

nm. (rrrc)

Figure 3. Irradiances for the vertical surfaces normalized by irradiance on the horizontal surface at wavelengths of (a) 300 nm, (b) 350 nm and (c) 500 nm U (thick lines). E, W (solid lines), N, S (dotted lines), D (triangles)

zon, while the total amount of scattered radiation is >50% of the global radiation. At 400 nm, less radiation is scattered from the direct beam, and the radiance pattern is different, with more radiance coming from larger zenith angles. Thus, at 400 nm the E/W surfaces have their maxima at 0900 and 1700 UTC, respectively, and at times beforelafter the morning/evening peaks, respectively, their irradiance is greater than that for the H surface. When the solar elevation is small and a vertical surface is pointed in the general direction of the sun it will receive more radiation than a horizontal surface, if scattering from the direct beam is not too great. In contrast, at 300 nm the E/W maxima occur at 1100 and 1500 UTC and are less than the H irradiance at all times except before 0900 and after 1800 UTC. At 300 nm the consider- able scattering of radiation from the direct beam, and its return from around the hemisphere, especially small zenith angle directions, reduces the irradiance of vertical surfaces even when facing toward the sun, because direct beam irradiance is small and the surface only sees part of the ce- lestial hemisphere. At the same time, the irradiance of the H surface, which receives radiation from the whole sky and is near normal to small zenith angles, is relatively enhanced.

Figure 3 shows the irradiance on each of the vertical sur-

faces, normalized by the irradiance on the horizontal surface (H2), at wavelengths of 300, 350 and 500 nm. Similar fig- ures (not shown) have been calculated for other wavelengths. The general diurnal patterns for each surface are similar at all wavelengths and match the general patterns illustrated in Fig. 2. The U surface has maxima in the morning and afternoon, E/W have a maximum during morning/aftemoon, respectively, while the N, S and D surfaces are more constant during the day. However, within these broad general- izations there are interesting and important spectral differences.

First, note the different vertical scales on the graphs in Fig. 3, which indicate the maximum ratios of vertical : horizontal irradiance and range from 1.4 at 300 nm to 7.0 at 500 nm. At all wavelengths the ratios are less than one for all surfaces throughout the middle part of the day, but at the beginning and end of the day the irradiances on the vertical surfaces facing the sun increase. These increases are more dramatic and occur earliedlater in the day, as wavelength increases. Thus, at 300 nm the only ratio > 1 .O is on the E- facing slope at 0900 UTC, while at 500 nm ratios >1.0 are found on surfaces facing the sun at 1000 UTC and before, and at 1700 UTC and after, the ratios increasing rapidly to reach a maximum of 6.3 at the extremes of the measurement day. At wavelengths less than 400 nm there is a distinct peak in the ratios during morning and afternoon, while at longer wavelengths the ratios simply increase to maxima at the lim- its of the measurement period. Note that the negative ratios at 300 nm are due to instrumental noise. On the D vertical surfaces there was essentially no radiation at this wavelength by 1900 UTC.

At the long UVA and visible wavelengths a significant proportion of the radiation remains in the direct beam even at large zenith angles. Thus, as the sun approaches the horizon and its direct beam is close to normal to a vertical- facing slope, that slope will receive more radiation than a horizontal surface (whose irradiation by the direct component = beam X cos SZA). By contrast, at the short UVB wavelengths virtually all the radiation is scattered or ab- sorbed from the direct beam at large zenith angles, so that a surface normal to the direct beam does not gain as much advantage over a horizontal surface as it does at long wavelengths.

It is clear that there is not a simple relation between irradiances on horizontal and vertical surfaces. Nonetheless, there may be ways of assessing vertical irradiance from horizontal irradiance in specified situations, and the data gath- ered in Izana were used to investigate these possibilities for clear skies. Initially, the ratios of vertical to horizontal irradiances, adjusted for incidence angle, for the S, E, W and U surfaces, were plotted as a function of angle to the solar beam. The angle between the sun and normal to the plane of the surface (the zenith angle for a horizontal surface) was calculated for each measurement, and its cosine was used to normalize the irradiance on each surface. Figure 4b shows that for incident angles between 10 and 75", knowledge of horizontal irradiance and surface orientation with respect to the sun allows assessment of vertical irradiances to within 20%, for a wavelength of 500 nm. At large incident angles, using horizontal irradiances to estimate vertical irradiances will lead to large errors (up to 300%). However, in terms of

468 Ann R. Webb eta/.

0016

0014

7 0012 -

0 0 1 ~

5 o m

E o w 6

3 o w 4

-

0 W2

0

2 5

1 5

- -

0 5

0 0 10 x1 30 40 50 60 70 80 90

Anph b.Mn nwnvl LO M p m m .nd Mnc( brm

Figure 4. Ratios of vertical : horimntal irradiances. when normalized by the anglcs of the sun to the normal plane of the surface at (a) 300 nm and ( b ) SO0 nm. S (square): E (cross): W (triangle): U (diamond).

absolute irradiance. these exposures would be small as the solar beam is almost parallel to the surface, and their con- tribution to a daily total exposure would be minimal. This treatment essentially considers all radiation to be in the direct beam, which is not too unreasonable at long wavelengths. The same treatment at 300 nm is not valid (Fig. 4a). Shorter wavelengths require a more rigorous treatment and additional information.

Calculations of the expected irradiance on the vertical surfaces were also made using a series of sets of direct (normal to the beam) and global (cosine-corrected) radiation measurements from the Innsbruck instrument to partition radiation at each wavelength into direct and diffuse components at a specified time. The diffuse irradiance was assumed to be isotropic, both from the sky and reflected from the ground. This defined the diffuse irradiance, which was the same on all vertical surfaces. The additional direct beam component was calculated using the angle between the solar beam and normal to the plane of each surface. Thus for a vertical surface

diffuse(V) = [diffuse(H)/21 + [global(H) X 0.1/2]

where 0.1 is albedo and

dircct(V) = direct(H) X sin(SZA) X cos(a1 - az)

where 0') and (H) refer to vertical and horizontal surfaces, respectively, a1 is azimuth of the surface and az is azimuth of the sun.

Qualitatively. diurnal patterns of irradiance agreed well with the measured irradiances on the vertical surfaces made by the UMIST instrument: compare Fig. 5 with Fig. 2. At longer wavelengths ( e .g . 400 nm, Figs. 2h and Sb) there is also reasonable quantitative agreement for all surfaces, while at 300 nm (Figs. 2a and 5a) only the horizontal surface shows concurrcnce between measured and modeled values.

1 8 I

01 7w B W 9w T O W 1 i . w IZM 13-w i 4 w 1500 16w i 7 . w ia:w ww

n m (UTC)

Figure 5. Calculated irradiances in W m- nm I at (a) 300 nm and (b) 400 nrn on all surfaces based on the measured global irradiance (horizontal) and direct beam irradiance. H (thick line); E. W (solid lines): N, S (dotted lines): U (circles): D (triangles).

For the vertical surfaces, the modeled data consistently over- estimate the measured data. Contributions to this discrepancy may come from uncertainties in defining the diffuse horizontal irradiance at 300 nm (the starting point of the calculation), erroneous assumptions about the isotropic nature of diffuse radiation or the spectral albedo of the surface and the fact that it was not flat to the horizon.

When considering the effects of UV radiation, it is the biologically weighted irradiance that is of interest rather than the irradiance at a single wavelength. Figure 6 shows the diurnal pattern of irradiances on all surfaces for the erythemally effective radiation (spectra weighted with the CIE action spectrum (14) and corrected for absolute values according to Fig. 1). The calculation of erythemally effective radiation from spectra at 5 nm intervals suffers an error of less than 5% when compared to the same calculation from spectra at 1 nm intervals. The all encompassing curve of the horizontal surface and the time of morning and afternoon maxima on thc vertical surfaces are much closer to the 300 nm case (Fig.

0 4

p @@ @@ ,6$ ,,rp ,.? ,$rp ,&@ p ,@ *@ ,B" @rp Time (UTC)

Figure 6. Erythema1 irradiance (W m 2)eq on all surfaces (absolute values corrected according to Fig. 1) . H (thick line): E. W (solid lines); N, S (dotted lines): U (circles): D (triangles).


2a) than the 400 nm case (Fig. 2b). This is expected as erythema1 radiation has diurnal and annual variations most like those of wavelengths of approximately 306 nm.

DISCUSSION

The measurements presented here are for a single day at a particular and special site. Conditions were almost perfect for radiation measurements with a cloudless, clean and relatively stable atmosphere and a horizon with few significant obstructions. The diurnal range of zenith angles was large, with the sun almost reaching the zenith at solar noon. In addition, the site was at high altitude and aerosol was very low, resulting in a diffuse-to-global ratio that is smaller than expected at locations with high aerosol load where the radiation passes through more scattering medium. The results should not be applied quantitatively to other locations and conditions, but by studying what happens in these ideal conditions a more general qualitative pattern of irradiances can be inferred and applied to other situations.

The results illustrate the changing nature of irradiance on a vertical surface compared to that of a horizontal surface. Not only does the ratio of vertical : horizontal irradiance change throughout the day but also the spectral character of the vertical irradiance compared to the horizontal. This is particularly evident in the current experimental example for the E- and W-facing surfaces. Inspection of Fig. 3 shows that the maxima on these surfaces occur closer to noon for short wavelengths than for long wavelengths, thus the irradiance is UVB enriched approximately 2 h either side of noon and blue enriched 4 h either side of noon. For biologically effective radiation, the timing of maximum effective irradiance will depend upon the action spectrum in question, the more emphasis that is given to short wavelengths, the closer to noon the maxima become.

The S-facing slope receives comparatively little radiation at the Izana site. This is because when the sun is to the S it is also very close to the zenith (in July), thus a vertical surface is far from normal to the direct beam. At times or places when the noontime zenith angle is greater (the sun is lower in the sky), the S-facing vertical surface should receive a comparatively greater irradiance. Blumthaler et al. (15) made measurements of erythemally effective radiation on a horizontal and a vertical S-facing surface at the high-altitude Jungfraujoch site from 1983 to 1992. The surface at this site is high albedo snow, in contrast to the low albedo earth at Izana, and with the noon solar elevation at summer solstice of 65", the ratio of vertical to horizontal irradiance was 0.67 compared to 0.24 in Izana. Schauberger further details the influence of albedo (snow vs asphalt) on irradiance of inclined surfaces (16), while Weihs et al. (17) show that the impact of albedo on irradiance of a horizontal surface is on average limited to 4.54% in clear-sky conditions.

The other vertical slopes will also have irradiance patterns that peak at different times and different heights throughout the year, thus the ratio between variously oriented surfaces at a given time will also change. The intuitive thought that a S-facing slope (in the Northern Hemisphere) receives most radiation is not necessarily always true (see Figs. 1 and 6). However, at a time or location where the noon zenith angle is larger and therefore subtends a smaller angle to a vertical

surface, the S-facing surface will have a more pronounced maximum at noon, and the U line in Figs. 1 and 6 should take on a more bell-shaped curve. If the energy incident on a flat surface is to be maximized (e.g. for solar energy pur- poses), then the surface is inclined so that it is perpendicular to the direct solar beam at local noon, rather than vertical. The D and (in the Northern Hemisphere) N-facing slopes will always have low irradiances compared to the other surfaces. Any direct beam radiation that reaches the N-facing slope must do so when the sun is low in the sky, i.e. early morning or late evening at appropriate locations and times of year.

Because most biological surfaces are not horizontal, information such as that in Fig. 6 may help to provide a more realistic assessment of the radiation received by a surface. Taking the upright human body at Izana in July as an example, it is clear that the hours close to solar noon are the most dangerous for horizontal areas of flesh (top of head, shoulders, bridge of nose), but the rest of the body (with more vertical orientation) may be more at risk during the mid-morning and mid-afternoon periods. If considering ocular damage and making the gross assumption that the eye- ball is essentially vertical and looking straight ahead, the same mid-morning/afternoon periods are most dangerous if facing in the general direction of the sun. In reality the average direction of gaze is - 15" (18), and facing the sun may induce squinting so the risk assessment becomes more complex in this case. A moving target will have a more complex pattern of exposure than one that is restrained to a single position, and only measurement by a dosimeter system on the surface can really identify individual exposures in such cases.

The data were collected under cloudless skies, and the results cannot be applied to alternative conditions. Mc- Kenzie, investigating horizontal and normal incidence irradiances in New Zealand (19). found similar condition-dependent results. In overcast conditions, when there is no direct be& radiation, the relative irradiance of the different surfaces would depend on the degree of isotropy of the diffuse radiation. This in turn would be dependent on the nature and stability of the cloud and on wavelength. A vertical surface only "sees" half of the sky, so in near-isotropic conditions vertical surfaces would therefore receive approximately equal radiation and less radiation than a horizontal surface. In broken cloud conditions, with intermittent direct beam solar radiation, the changing irradiances on different surfaces become very complex and are not easy to general- ize.

CONCLUSION

The measurements illustrate the changing nature of the solar spectrum and levels of irradiance on variously oriented surfaces throughout the day. They show that irradiance measured on a horizontal surface is not a good indicator of irradiance on vertical surfaces, either for intensity or diurnal pattern. The relation between horizontal and vertical irradiances is both wavelength dependent and zenith angle dependent, the latter being determined by latitude and time. How- ever, given sufficient information about the radiation (the direct as well as global component) and the solar angle rel-

470 Ann R. Webb eta/.

ative to the surface, it is possible to get a good estimation of the vertical irradiance from the horizontal measurements for clear sky conditions with low aerosol load.

A~k17owledgetnents-We thank the Instituto Nacional de Meteorol- ogia (INM) for allowing this work at the Izana GAW station. Par- ticular thanks go to Emilio Cuevas. Albert0 Redondas and Carmen Rus Jimenez for their support and help during the project. A.W. and P.W. were working at Reading University, UK, at the time of the study. Dr. Weihs was supported by the European Community, and the equipment was funded by the UK Department of Environment. The work of M.B. was supported by the Austrian Science Founda- tion.

REFERENCES I . Blumthaler, M.. J . Grobner. M. Huber and W. Ambach (1996)

Measuring spectral and spatial variations of UVA and UVB sky radiance. Geophys. Rex Lett. 23, 547-550.

2. Grant, R. H., G. M. Heisler and W. Gao (1997) Ultraviolet sky radiance distributions of translucent overcast skies. Theor. Appl. Clim. 58. 129-140.

3. Grant. R. H., G. M. Heisler and W. Gao (1997) Clear sky radiance distribution in ultraviolet wavelength bands. Theor. Appl. C h i . 56, 113-135.

4. Miiller, M., A. Kraus and A. Hofzumahaus ( 1995) 03->0( ID) photolysis frequencies determined from spectroradiometric measurements of solar actinic UV-radiation: comparison with chem- ical actinometer measurements. Ceophys. Res. Lett. 22, 679- 682.

5. Behr. H. D. ( 1997) Solar radiation on tilted south oriented surfaces: validation of transfer models. Solnr Energy 61, 339413 .

6. Perez, R., R. Stewart, C . Arbopast, R. Seals and J . Scott (1986) An anisotropic hourly diffuse radiation model for sloping surfaces. Description. perfonance validation, site dependency evaluation. Solar Energy 36, 48 1 4 9 8 .

7. Perez, R., P. lneichen. R. D. Seals and A. Zelenka ( 1990) Mak-

ing full use of the clearness index for parameterizing insolation conditions. Solar Energy 45, 11 1-1 14.

8. Perez, R., P. Ineichen, R. D. Seals, J. Michalsky and R. Steward ( 1990) Modeling daylight availability and irradiance from direct and global irradiance. Solar Energy 44, 271-289.

9. Ma. C. C. Y. and M. Iqbal (1983) Statistical comparison of models for estimating solar radiation on inclined surfaces. Solar Energy 31. 313-317.

10. World Meteorological Organisation (1997) The strategic plan of the Global Atmosphere Watch (GAW). WMO CAW publication No. 11 3, WMO, Geneva.

1 1 . Webb, A. R. (Ed.) (1997) Advances in solar ultraviolet spectro- radiometry. CEC Air Pollution Research Report 63.

12. Blumthaler, M. and W. Ambach (1988) Solar UVB-albedo of various surfaces. Photochem. Photobiol. 48. 85-88.

13. Feister, U. and R. Grewe (1995) Spectral albedo measurements in the UV and visible region over different types of surface. Photochem. Photobiol. 62, 136-744.

14. McKinlay. A. F. and B. L. Diffey (1987) A reference action spectrum for ultraviolet induced erythema in human skin. CIE J. 6, 17.

15. Blumthaler, M., W. Ambach and R. Ellinger (1996) UV-Bes- trahlung von horizontalen und vertikalen Flachen im Hochge- birge. Werter wid Leben 1, 25-31.

16. Schauberger. G. (1987) Verteilung von solarer UV-Strahlung auf nicht-horizontale Flachen. Med. Phys. 1987, 18 1-1 86.

17. Weihs, P., S. Simic, W. Laube, W. Mikielewicz, G. Rengarajan and M. Mandl (1999) Albedo influences on surfaces UV irradiance at the Sonnblick High Mountain Observatory (3106 m altitude). J. Appl. Mereorol. (In press)

18. Sliney, D. H. (1996) Ultraviolet radiation ocular exposure radiation dosimetry of the eye: a report to the World Health Or- ganization and United nations Environment Programmc. WHO/ EHG195.18, WHO. Geneva.

19. McKenzie, R. L., K. J. Paulin and M. Kotkamp (1997) Erythe- ma1 UV irradiance at Lauder, New Zealand: relationship between horizontal and normal incidence. Photochem. Photobiol. 66, 683489 .

spectral uv irradiance on vertical surfaces: a case study

Documents