spectral solar irradiance and some optical properties for various polluted atmospheres

Solar Energy Vol. 69, No. 3, pp. 215–227, 20002000 Elsevier Science Ltd

Pergamon PII: S0038 – 092X( 00 )00062 – 1 All rights reserved. Printed in Great Britain0038-092X/00/$ - see front matter

www.elsevier.com/ locate / solener

SPECTRAL SOLAR IRRADIANCE AND SOME OPTICAL PROPERTIES FORVARIOUS POLLUTED ATMOSPHERES

,†CONSTANTINOS P. JACOVIDES* , MICHAEL D. STEVEN** andDEMOSTHENIS N. ASIMAKOPOULOS*

*Laboratory of Meteorology, Department of Physics, Division of Applied Physics, University of Athens,Panepistimioupolis, Build. PHYS-V, Athens 157 84, Greece

**Department of Geography, University of Nottingham, University Park, UK

Received 18 November 1998; revised version accepted 7 February 2000

Communicated by RICHARD PEREZ

Abstract—Using ground-based spectroradiometric measurements taken over the Athens atmosphere duringMay 1995, the influence of gaseous pollutants and aerosol on the spectral radiant energy distribution wasinvestigated. It was found that spectral measurements exhibited variations based on various polluted urbanatmospheric conditions as determined via gaseous pollutants record analysis. The relative attenuations causedby gaseous pollutants and aerosol can exceed 27%, 17% and 16% in the global ultraviolet, visible andnear-infrared portions of the solar spectrum respectively, as compared to ‘‘background’’ values. In contrast, anenhancement of the near-infrared diffuse component by 66%, was observed, while in visible and ultravioletbands the relative increases reached 54% and 21% respectively. Experimental total Rayleigh-corrected andspectral aerosol optical depths were retrieved, representing differences in polluted air over the Athensatmosphere. The diffuse component accounts for more than 80% of the total radiation field under high pollutedatmosphere. The observed differences of solar radiation between the Athens center and at a nearby suburbansite are a manifestation of contrasting air properties provided mainly by automotive traffic. 2000 ElsevierScience Ltd. All rights reserved.

1. INTRODUCTION crease in global (direct and diffuse) irradiance thatreaches the ground. There is a need to know the

The increase in terrestrial applications of solarspectral distribution of solar irradiances and the

radiant energy has given impetus to the study ofextent to which changes in environmental factors

solar energy availability in many areas of theaffect this energy distribution. In this article,

world. With the increasing use of spectrallymodification of the spectral composition of solar

selective devices, such as photovoltaic cells forradiation by urban aerosol is investigated.

electrical generation and selective absorbers forFurther, the spectral distribution of solar radia-

thermal collectors, and for practical applicationstion depends on several atmospheric and surface

in environmental and agrometeorological re-properties. In order to improve our understanding

search, current interest is not only in the totalof the factors that affect spectral radiant energy

amount of solar energy reaching the Earth’sdistribution, a variety of radiative transfer models

surface, but also in its spectral composition.have been developed (Bird et al., 1982; Brine and

Atmospheric pollutants and aerosols absorb andIqbal, 1983; Justus and Paris, 1985; Bird and

scatter shortwave solar radiation. The interactionsRiordan, 1986; Gueymard, 1994). Nevertheless,

have resultant impacts on atmospheric radiativevalidation of these models requires accurate and

energy transfer and balance (Charlson et al.,detailed spectral measurements as well as simulta-

1991; Schwartz, 1996). Scattering and absorptionneous observations of environmental factors that

of solar radiation by aerosol particles and gasesaffect spectra at various spectral bands.

results in a remarkable attenuation of the directA joint research experiment between the Uni-

solar beam component and a moderate increase inversity of Athens and the University of Nottin-

the diffuse component. The result is a net de-gham was set up to study the effects of airpollutants on the solar spectral distribution. The

† experiment was deployed in the Athens basinAuthor to whom correspondence should be addressed.(388N, 248E) during May 1995 over 2 weeks,Tel.: 130-1-727-6931; fax: 130-1-729-5281; e-mail:

[email protected] from 16 to 27. Because of the great variability of

215

216 C. P. Jacovides et al.

meteorological conditions it is difficult to project eastern flow, while during spring and early sum-a continuous period of simultaneous observations mer it is a persistent southwestern flow. Localunder clear skies and high pollution levels. To the circulation of sea-breeze may set up a simulta-authors’ knowledge, there has never been estab- neous southwestern flow into the basin, a con-lished a similar experiment in the Athens area. dition which can lead to the ventilation of dow-The present study aims to quantify temporal and ntown Athens (Jacovides and Karalis, 1996).spatial variations in the spectral distribution of Spectral solar irradiance measurements weresolar irradiances in Athens, and to identify obtained using a commercial Licor portable spec-reasons for such variations. Atmospheric optical troradiometer. The instrument takes a sequence ofproperties such as optical depths, the diffuse-to- spectral measurements from 300 to 1100 nm atdirect ratio of solar irradiances and spectral 2 nm interval, the whole operation taking abouttransmittances are retrieved in order to establish two min. The spectroradiometer was controlled bysome ‘‘background’’ characteristics of the atmos- a portable PC and manufacturer’s software andpheric conditions over the Athens atmosphere. conformed to National Institute of Standard and

Technology (NIST) standards. The calibrationresults showed a deviation lower than 5% for

2. EXPERIMENTAL AND SITE DETAILSwavelengths larger than 400 nm. For shorter

The Athens basin is in a region characterized wavelengths this deviation increased to a maxi-by a complex topography. The metropolitan area mum of 23% in the spectral band 300–320 nm.is surrounded by mountains over 1000 m high and These errors are comparable to those reported bythe Saronic Gulf in the southwest (see Fig. 1). Riordan et al. (1989). The measurements taken byThe basin is characterized by high population this instrument may be integrated by a micro-density and serious traffic congestion. The mean processor to give the light incident in a broader

22wind pattern in the atmospheric boundary layer waveband in W m units.during late summer is a persistent constant north- The spectral global irradiance was measured on

a horizontal plane immediately followed by ameasurement of the diffuse spectral (sky only)irradiance. The technique used to measure diffuseirradiance was the standard one with a shade discdesigned to obscure a 58 radius about the sun.Linke-Feussner’s pyrheliometric readings of thedirect solar beam in several wavebands weretaken, supplementing the Licor’s spectral data. Inthe field, Licor’s calibration was also performedby comparing the measured integrated irradiancewith that of the pyrheliometer in the spectral band300–630 nm. This comparison showed that thetwo instruments reached similar results, display-ing an average difference of less than 2%. Correc-tions of the direct solar beam data due to thepyrheliometer body temperature were taken intoaccount. A broadband correction factor was alsoapplied to pyrheliometric data in order to excludeany parasitic circumsolar contribution (Gueymard,1994).

During the field experiment spectral measure-ments were collected at four sites located (Fig. 1):(a) at the Athens Archeological Museum (MUS),75 m elevation and 0 m above ground, (b) at the

Fig. 1. Map of the Athens basin and its surroundings showing Athens stadium (STA), 108 m elevation, and 0 msites’ locations used in this analysis. Acronyms: MUS- above ground, (c) on the roof of the downtownArcheological Museum of Athens; UNB-University building; University building (UNB), 125 m elevation, 20 mSTA-Athens stadium; UC-University Campus. Open circles

above ground, and (d) at the suburban site ofindicate air pollution monitoring networks, Pat. for PatisionUniversity Campus (UC), 280 m elevation and 0 mand Ng for National garden. Contours have been drawn every

300 m. above ground, 5 km east of the Athens center.

Spectral solar irradiance and some optical properties for various polluted atmospheres 217

The data archive was organized to include two UV spectral band and from Anderson and Mauer-types of study: first, a series of measurements sberger (1992) in the Chappuis band. Ozoneover several days was made at urban MUS, STA column density u was obtained from Dobson’sO

and UNB sites, aiming to assess different con- unit, which is located on the UNB roof. NO2

ditions within the city; and second, a spatial spectral absorption coefficients were taken fromsurvey of sequential measurements at urban Schneider et al. (1987) in the 350 to 500 nm(STA) and suburban (UC) sites (25 min shuttle range and smoothed to 2 nm intervals; whereastrip between sites). NO reduced pathlength was derived from its2

23surface concentrations (mg m ) via simplethermodynamic calculations by assuming that the

3. METHODOLOGY AND DATA ANALYSIS NO scale height is equal to that of aerosol.2

The water vapor optical depth was approxi-The extinction of spectral irradiance is usuallymated as (Bird and Riordan, 1986):modelled via Bouguer’s law,

20.45I 5 I exp 2 m t (1)s d t 5 0.2385a wm (1 1 20.07a wm )bl 0l r l H Ol wl r wl r2

where I is the measured spectral irradiance for (6)bl

wavelength l and I is the extraterrestrial spec-0lwhere a are the spectral water vapor absorptionwltral irradiance corrected for the actual sun–earthcoefficients taken from Bird and Riordan (1986).distance. m is the relative optical air mass and tr lThe atmospheric precipitable water content w wasis the total optical depth comprising optical depthsderived as (de la Casinier et al., 1997):caused by Rayleigh scattering t , aerosol extinc-Rl

tion t , and atmospheric absorptions of gases w 5 0.1 H r (7)al w vsuch as ozone, water vapor, nitrogen dioxide and

w is in cm and the surface water vapor density rvoxygen, t . Apparently, total optical depthsgl 23is in g m . The water vapor scale height H (inwinclude contributions due to: molecular scatteringkm) was derived as (de la Casinier et al., 1997):at all wavelengths, NO absorption band (250–2

700 nm), and ozone absorption in the ultraviolet H 5 0.4976 1 1.5265twband (e.g., strong absorption at Hartley 220–320 nm,

31 exp(13.6897t 2 14.9188t ) (8)and Huggins 300–345 nm spectral bands and weak-

er absorption in the Chappuis band 440–770 nm). where t is the ratio of the ambient temperature (K)Also, several spectral portions are affected by to 273.15.absorption bands of H O and O (680–746, 754–2 2 The relative optical air mass of the average774, 786–844, and 872–1014 nm). The Rayleigh atmosphere was computed from Kasten andscattering component, as it is a significant Young’s (1989) equation, corrected for pressurenonaerosol contributor, can be subtracted from tl variations:the remainder characterizes the aerosol attenua-

21.6364 21m 5 [cos Z 1 0.50572(96.07995 2 Z) ]tion and molecular absorption. The spectral r

aerosol optical depth t was derived as,al (9)

t 5 t 2 t 2 t (2)al l Rl gl where Z is the zenith angle in degrees. Thisexpression can be used to approximate the relativeThe Rayleigh-molecular optical depth is wellatmospheric masses of the different constituents.modelled as (Bird and Riordan, 1986):Appreciable differences, for example, between mrPs24.08 and ozone’s air mass occur only for zenith angles]tRl 5 0.008735l (3)P0 greater than 798.

Atmospheric aerosol optical depth is also de-where P is the atmospheric site pressure and Ps 0scribed as (Angstrom (1961):its sea level value. Ozone and nitrogen dioxide

2aoptical depths follow similar law: t 5 bl (10)al

t 5 a (l)u (4)ONl ON ON where b and a are constants; a is the wavelengthwhere a (l) are the respective spectral absorp- exponent closely related to the aerosol size dis-ON

tion coefficients for O and NO ; whereas u are tribution whereas b is the Angstrom’s turbidity3 2 ON

the respective O and NO column densities in coefficient. Eq. (10) is equivalent to approximat-3 2

atm-cm. Ozone spectral absorption coefficients ing the wavelength dependence of the opticalwere taken from Molina and Molina (1986) in the depth by a straight line on a log–log plot. In such


approximation the slope of the line yields a while meteorological conditions were fair with clear-skyintercept provides b (Michalsky and Stokes, most of the time. The nitrogen dioxide as a strong1983; Cachorro et al., 1989). absorber in the blue spectral region, can color

The direct solar beam spectra were obtained as plumes brown or yellow. A brownish cloud of airthe difference between global and diffuse spectra pollution forms over many industrialized andon a horizontal plane. As shown by various dense urbanized cities, it is seen by the naked eye.authors (Steven, 1984; LeBaron et al., 1990; This cloud denotes both a high level of NO2

Perez et al., 1990; Batlles et al., 1995; de la pollutant and of man-made aerosols. NO is an2

Casinier et al., 1995), the procedure for measur- important indicator of air pollution, because it ising diffuse irradiance by the occultation disc well correlated with polycyclic aromatic hydro-technique needs to be improved especially over carbons and soot (Glasius et al., 1999). Therefore,the wavelength range 300–900 nm. To this re- analysing the 64 spectra in conjunction with airspect, a spectral correction factor has been derived pollution records, 20 spectra were classified asand applied on the diffuse spectral irradiance ‘‘slightly’’ polluted air with low concentrations offollowing de la Casinier et al. (1995). In order to air pollutants, e.g., NO concentration was 47,2

23minimize further possible measurement errors, [NO ],70 mg m , NO concentration was 82,223some additional restrictions were imposed on the [NO],134 mg m , CO concentration was 2.4,

23data: First, restrictions on the optical path length [CO],3.8 g m , and O was 24,[O ],47 mg3 323were imposed by considering an air mass interval m ; 22 as ‘‘moderately’’ polluted with high con-

of 1.39–2.28, i.e. when zenith angle was 44–648. centrations of air pollutants, e.g., 156,[NO ],223 23The choice of this interval results from: (i) the 220 mg m , 288,[NO],394 mg m , 5.5,

23 23significant elevation of the mountains surrounding [CO],8.8 mg m , and 58,[O ],81 mg m ,3

the Athens basin screens larger air masses ac- and 18 as ‘‘highly’’ polluted with much highercounting for relevant observations; (ii) during late concentrations of air particulates, e.g. 279,

23 23mornings, (Z,448) the greater contribution of an [NO ],348 mg m , 408,[NO],519 mg m ,223onshore sea-breeze flow induced more transparent 10.6,[CO],13.9 mg m , and 86,[O ],1083

23conditions thus perturbing atmospheric stratifica- mg m . Only four observations were classified as23tion; and (iii) this interval guarantees the validity ‘‘pollution smog’’ ([NO ] .390 mg m , [NO]2

23 23of the plane parallel atmosphere which in its turn .579 mg m , and [CO] .16 mg m ).decreases the uncertainty in the optical depthdetermination. Second, perturbed data were reject-ed by comparing broad band pyrheliometric data 4. RESULTS AND DISCUSSIONtaken immediately before and after each spectral

4.1. Optical depth versus polluted atmospheresscan, thus ensuring atmospheric stability (Nannand Riordan, 1991). Third, measurements were Stagnating air allowed polluted atmosphererejected if they resulted in unrealistic optical over the city center during the measurementdepth determinations. After this filtering, 64 solar period. The early morning and late afternoonirradiance spectra were available for analysis. observations correspond to periods for which

Further, in Bouguer’s law (Eq. (1)), the ex- there were smaller temperature and relativetraterrestrial constants I were determined by humidity variations. Thus, air temperature varied0l

calibration through the Langley plot method. slightly from 19 to 218C while relative humiditySpectral data taken on two clear and stable days remained almost constant at 62–58%, within thewere used for calibration since the derived ex- same period. After this time interval, air tempera-traterrestrial constants agreed for these days (Lan- ture increased steadily to 288C by midday whilegley plots were straight lines). Calibration was relative humidity fell steadily to about 38% by

21also done using Langley plots taken on the 1500 LST. Light westerly winds of 1.3–2.0 m ssummit of mountain Hymettus before and after were blowing in early morning strengthening later

21the experiment. At this site, the optical depth is to about 3–4 m s , due to an onshore southwes-low and remains almost constant during morning terly sea-breeze flow. Such patterns were ob-making the site an ideal location for calibration served over the whole measurement period.purposes. The uncertainty in the extraterrestrial Fig. 2a,b,c show the experimental total Rayleigh-coefficients for both cases was estimated to be corrected spectral optical depth of the atmosphereless than 2.5% which results in optical depth retrieved from ground-based observations takenuncertainty of about 5%. over the Athens atmosphere during May. In

During the measurement period the addition Fig. 2 illustrate the Angstrom fit and the


Fig. 2. (a) Experimental total Rayleigh-corrected, spectral aerosol (Eq. 2) and Angstrom’s aerosol (Eq. 10) optical depths overslightly polluted Athens atmosphere during May 1995. (b) As in Fig. 2a except over moderately polluted Athens atmosphere. (c)As in Fig. 2a except over highly polluted Athens atmosphere.

experimental spectral aerosol optical depth, as series centered at the air mass of interest. Airdetermined via Eq. (2). The observations corre- pollution records were taken from two nearbyspond to the early morning data when zenith stations: Patision, being within a distance of aboutangle was 46.8–588. Fig. 2a,b,c correspond to 0.5 km from MUS and 1.5 km from UNB site,‘‘slightly’’, ‘‘moderately’’ and ‘‘highly’’ polluted and Athens Stadium (mobile station of Athensatmospheres respectively. Note that each curve is district), a few meters from STA site.in fact the average optical depth of the morning Optical depth values observed over the Athens


atmosphere compare well with those reported by concentration. Nevertheless, the larger standardVasilyev et al. (1995) for the highly polluted error of the polluted cases is indicative of theMexico City atmosphere, representing visibilities greater variation in aerosol characteristics arisingof about 5 km. In the graph 2c, the optical depth from differences in aerosol composition and sizeobserved over the Mexico City atmosphere is distribution. It is worth noting that the errors inshown for comparison purposes. It is noted that the ‘‘slightly’’ polluted case were less than halfVasilyev et al. used the concept of experimental those corresponding to the other cases. It istotal Rayleigh-corrected optical depth as in this apparent that the early morning weak insolationanalysis. Cachorro et al. (1996) also proposed a causes a weak variation of the meteorologicalsimilar procedure in obtaining spectral optical parameters and consequently also of the aerosoldepths while Wenny et al. (1998) reported optical extinction. Thus, in spite of the short period ofdepth values of the same magnitude. The spectral observation, it seems that the relative humiditydependence of the curves shown in Fig. 2 is also effect on the aerosol optical depth is ratherevident. The relatively strong wavelength depen- negligible (Nilsson, 1994). Overall, the errordence of the polluted optical depth is possibly analysis supports the assumption that the trans-associated with small particle scattering and with formation of spectral optical depths cannot betrace gases absorption at shorter wavelengths. attributed to experimental errors but can be in-Strong absorption due to NO and O is revealed ferred to gaseous absorbers which dominate over2 3

by the UV spectral signature resulting in steeper the Athens atmosphere. Nonetheless, the assump-slopes of the optical depth at this waveband. An tion made concerning the plane parallel atmos-additional factor results from the multiple scatter- phere in the early morning, is justified.ing which increases with increasing particle con- Table 1 also includes Angstrom’s parameters b

centration, thus changing the slope of the t vs. l and a, as determined from spectral extinctionl

(Vasilyev et al., 1995). The differences between plots of the aerosol optical depth measurements,the experimental total Rayleigh-corrected and the defined by (2), e.g., log(t ) versus log(l), in theal

remaining experimental aerosol optical depth spectral range 350–862 nm, having omitted themight stem from absorptions over UV and visible spectral portions due to water vapor and oxygenwavebands. As it is clear a great sensitivity of the absorptions. This spectral restriction was decidedozone and water vapor absorption in the visible in order to minimize the influence of light scatter-region is observed. Especially the experimental ing and humidity. In general, turbidity valuesaerosol spectrum (Eq. (2)) exhibits significant relate well to climatological values of the areaerror noise mainly due to imprecise absorption (Jacovides and Karalis, 1996). It is mentionedbands of oxygen and water vapor. Removal of this here that early morning optical depth value wasbias would require simultaneous measurements of high always, dropping to half of this maximum bythe vertical atmospheric water content. midday and increasing to almost the morning

Table 1 gives morning average aerosol optical value in the evening; this results from the induceddepth values including standard errors, which higher transparency conditions due to an onshorecorrespond to the above curves. It appears that the sea-breeze flow. Mean morning aerosol opticalhighly polluted case exhibited the largest average depth values on these twelve days, are given invalues for the three operational wavelengths. The Table 2. There is a distinct contrast in opticaldifferent mean aerosol optical depth for the depths between the highly polluted cases consid-polluted cases suggests that the atmospheric opti- ered. The optical depths for May 18 and 19 arecal properties vary in relation to air particulates higher than on May 20 and 24. Nevertheless, a

Table 1. Mean aerosol optical depths at three specified wavelengths for the polluted cases considered. Also, the Angstromparameters, b and a, and NO optical depths, are given2

Wavelength Highly polluted Moderately polluted Slightly polluted

Mean SE Mean SE Mean SE

380 nm 0.829 0.022 0.578 0.020 0.237 0.011500 nm 0.603 0.022 0.394 0.018 0.174 0.009676 nm 0.425 0.019 0.262 0.022 0.123 0.008Angstrom b 0.336 0.032 0.169 0.025 0.078 0.019Exponent a 1.125 0.056 1.162 0.069 1.298 0.085NO Optical Depth2

at 400 nm 0.096 0.012 0.048 0.009 0.0074 0.0002(t )NO2


Table 2. Morning averaged aerosol optical depth observations over the Athens atmosphere. Also, the ozone amount and theAngstrom exponent (a), are given. Letters in parentheses indicate slightly (S), moderately (M) and highly (H) pollutedatmospheres

Date May’95 Ozone 380 nm 500 nm 676 nm Angstrom (a)

(Classification) (DU) Mean SE Mean SE Mean SE Mean SE

16/5(S) 352 0.389 0.023 0.186 0.022 0.129 0.010 1.470 0.10717/5(M) 338 0.592 0.025 0.402 0.023 0.286 0.022 1.156 0.09218/5(H) 334 0.989 0.034 0.655 0.027 0.448 0.023 1.128 0.08819/5(H) 332 0.968 0.038 0.633 0.025 0.436 0.023 1.123 0.08220/5(H) 335 0.876 0.032 0.545 0.028 0.402 0.022 1.139 0.07821/5(S) 334 0.376 0.027 0.164 0.021 0.122 0.014 1.322 0.10222/5(S) 338 0.398 0.022 0.159 0.019 0.116 0.012 1.328 0.104

a23/5(S) 342 0.402 0.025 0.176 0.022 0.131 0.016 1.491 0.10924/5(H) 332 0.912 0.032 0.560 0.025 0.416 0.020 1.129 0.08225/5(S) 339 0.378 0.018 0.168 0.019 0.124 0.012 1.286 0.10126/5(M) 329 0.628 0.028 0.419 0.024 0.238 0.019 1.236 0.09827/5(M) 331 0.668 0.026 0.425 0.022 0.284 0.022 1.248 0.101

a Rural atmosphere.

22 21‘‘pollution smog’’ event was observed on these irradiance reduction to 0.672 W m nm .days resulting in higher optical depth values. Such Scattering by aerosol and absorption by both tracedifferences illustrate the wide range of possible gases and aerosol are the resultant processesaerosol optical depth that the Athens atmosphere altering the radiation field under cloudless skies.periodically experiences. Additionally, Table 2 The differences in the air gaseous and aerosolincludes the Angstrom exponent a and the ozone loadings for the three polluted cases becomeamount (in DU unit); this amount of ozone was evident from the previous discussion of theused in deriving its optical depth. aerosol optical depth. In the surface atmospheric

The optical depths of NO are shown in Figs. 2 boundary layer, the absorption by trace gases is2

for comparison purposes. The resultant NO more effective in the UV spectral range under2

optical depth values at 400 nm were found to be high aerosol concentrations (Webb, 1992;(see Table 1), 0.096, 0.048, and 0.0074 under Jacovides et al., 1998).highly, moderately and slightly polluted atmos- Table 3 summarizes integration values of thesepheres respectively. The scale height used for curves in several narrowbands. It is clear that theNO optical depth retrievals ranged between 1.62 greatest attenuation caused by gaseous pollutants2

km for low turbidity (,0.1) and 0.940 km for was observed in the UV spectral band, in agree-large turbidity (.0.3), which are in line with ment with most observations in different placesKing and Buckius (1979) values. over the globe (Bird et al., 1982; Lorente et al.,

1994; Wenny et al., 1998). The highly polluted4.2. Spectral irradiances versus polluted case exhibited, on average, the highest UV rela-atmospheres tive attenuation, 27%, compared to the slightly

In view of the above, the contrasting sen- polluted case. In the visible (VIS) and nearsitivities of the two radiative fluxes (i.e. global infrared (NIR) parts of the solar spectrum theand diffuse), to changes in air conditions, are highly polluted case provided relative attenuationsshown in Fig. 3a,b. From the available data of the order of 17% and 16% respectively, inarchive, coincidentally measured spectra of global comparison to the slightly polluted case. It isand scattered irradiances, are compared. It must evident that the highly polluted atmosphere whichbe noted that spectral curves are interpreted in exhibited the highest optical depth, also exhibitedview of the previous classification. Examples of the greatest attenuation of global irradiance, main-the measured spectral energy distribution of glob- ly in the UVB spectral band (33%).al irradiance, taken at an air mass interval 1.62– The corresponding curves for the scattered1.81 when zenith angle was 52–56.48, are shown irradiance are shown in Fig. 3b. It is clear that,in Fig. 3a. The graph illustrates the influence of apart from the significant increase in diffusegaseous pollutants and aerosol on the spectral irradiance for all wavelengths, the differenceirradiance. As it is seen, for the ‘‘slightly’’ between the two relative maxima in the VIS band,polluted case (curve 1) the maximum spectral i.e. around 414 and 450 nm, increases. However,irradiance occurs at a wavelength of 480 nm, under slightly polluted conditions, the scattered

22 21reaching 0.820 W m nm ; whereas the ‘‘high- irradiance results mainly from air molecules,ly’’ polluted case (curve 3) leads to a spectral while under high polluted atmosphere the increase


(UV) in the slightly polluted case; whereas therelative contribution of longer wavelengths in-creases with increasing concentration of air par-ticulates. Specifically, the relative increase ofscattered irradiance produced by gaseous pollu-tants and aerosol was greater in the NIR spectralband. An enhancement of about 66% was ob-served comparing the highly to slightly pollutedcases. VIS and UV spectral bands follow withpercentages of the order of 54 and 21%, respec-tively.

Further, in order to isolate the influence of apolluted air on the spectral characteristics of solarradiation, the ratio of diffuse-to-direct solar ir-radiances is considered. This ratio is a usefulparameter in estimating atmospheric optical prop-erties, such as optical depths and light absorptionindex (King, 1979) and in monitoring absorptionby light cirrus clouds. In addition, through thisratio any propagation error resulting from possibleinstrument calibrations on the measured spectra,can be eliminated. Correlations between diffuse-to-direct ratios and selected aerosol optical depths,for two polluted cases presented above, are shownin Fig. 4. The aerosol optical depth values usedcorrespond to the recommended wavelengths byWMO (Adeyefa and Holmgren, 1996), e.g., 380,400, 450, 500, 550, 676, 778, and 862 nm. Foranalyses, averaged optical depth values taken overurban MUS site only, were considered. Thecorrelations are well described by exponentialfitting:

Diffuse /Direct 5 a exp(b.t ) (11)al

with a50.04460.005 and b54.3260.03 forhighly polluted air or a50.02660.004 and b5

10.9560.08 for slightly polluted air. It is obviousthat the decrease of ratios dependence to opticaldepth is much smoother under slightly pulluted airthan under highly polluted air, reflecting the lackof gaseous pollutants and aerosol. As the aerosoloptical depth increases the ratio increases con-siderably especially at short wavelengths. Asshown earlier (Jacovides et al., 1998) the ratio isparticularly sensitive to both changes in solarFig. 3. (a) Spectral distribution of global solar irradianceszenith angle and gaseous pollutants and aerosol,under cloudless skies and for three polluted cases considered.

(b) As in Fig. 3a but for scattered irradiances. in the shorter wavelength region. The importantrole of scattering is verified at these wavelengths;the direct irradiance that is depleted by scattering

of aerosol scattering, causes the combined scat- processes is regained as increased diffuse ir-tered irradiance at 450 nm to be higher than that radiance leading to large ratio’s values. It iscorresponding to 414 nm. Table 4 summarizes notable that at longer wavelengths the ratio forintegrating scattered irradiance values over sever- both atmospheres appears to be of the same orderal narrowbands. The scattered component has a of magnitude. This may be attributed to waterpredominantly short wavelength composition vapor effect, which is the regulating process at


22Table 3. Global irradiance (W m ) for various narrow bands (nm) and different polluted classes considered. Air mass interval1.62–1.81. Percentages in parentheses indicate relative differences with slightly polluted case. The total ozone amount (in DU)was: 342, 336 and 330 for the slightly, moderately and highly polluted atmospheres respectively

Polluted UVB UVA UV VIS NIRConditions 300–320 nm 320–400 nm 300–400 nm 400–700 nm 700–1100 nm

22 22 22 22 22(W m ) (W m ) (W m ) (W m ) (W m )

Slightly 1.198 23.2 24.4 214.4 129.9Moderately 0.879 19.6 20.5 193.9 116.7

(226%) (216%) (216%) (210%) (210%)Highly 0.803 16.9 17.7 178.1 109.2

(233%) (227%) (227%) (217%) (216%)

22Table 4. Diffuse irradiance (W m ) for various narrow bands (nm) and different polluted classes considered. Air mass interval1.62–1.81. Percentages in parentheses indicate relative differences with slightly polluted case. The total ozone amount (in DU)was: 342, 336 and 330 for the slightly, moderately and highly polluted atmospheres respectively

Polluted UVB UVA UV VIS NIRconditions 300–320 nm 320–400 nm 300–400 nm 400–700 nm 700–1100 nm

22 22 22 22 22(W m ) (W m ) (W m ) (W m ) (W m )

Slightly 0.832 12.0 12.9 46.3 13.2Moderately 0.814 13.3 14.2 55.5 15.6

(22%) (111%) (110%) (120%) (118%)Highly 0.801 14.8 15.6 71.1 21.9

(24%) (123%) (121%) (154%) (166%)

these wavelengths. It is also apparent that for differences in transmittance. The effect of gaseousshorter wavelengths the diffuse component ac- pollutants and aerosol on the transmittance degra-counts for more than 80% of the radiation field dation is more important in the UV and VISunder polluted air. spectral bands. Employing narrowband transmitt-

The spectral transmittances over the Athens ances for the cases considered, it was found thatatmosphere for the two distinct polluted cases the highly polluted atmosphere provided lessconsidered are compared in Fig. 5. The difference transmission of approximately 62% in UVB, 59%in attenuating properties between highly and in UVA, 37% in VIS and 19% in NIR, asslightly polluted cases are evident by noting compared to good transparency conditions values.

Fig. 4. Spectral dependence of the diffuse-to-direct ratioversus aerosol optical depth for two distinct polluted cases Fig. 5. Comparison of total spectral transmittances over theconsidered. Athens atmosphere for two polluted cases considered.


These findings are in agreement with the spectral 4.3. Spatial variability of spectralcharacteristicsattenuation dependence observed for varying tur-

bidity levels reported by Lorente et al. (1994) and The incident solar irradiance at two close sitesWenny et al. (1998). may differ due to different atmospheric properties,

Further, the total spectral transmittance of a e.g., O , NO , water vapor, aerosol and automo-3 2‘‘pollution smog’’ event together with Rayleigh- tive traffic. Thus, spatial variations of gaseousmolecular scattering and aerosol extinction is absorbers and aerosol can be regarded as the maingiven in Fig. 6. Although the sample is limited, causes of the radiation difference at nearby urbanthe results have their peculiarities. Gaseous pollu- and suburban cloud free sites. The present datatants resulted in significant transmittance degra- archive includes spectral scans taken for twodations. Comparing with the Rayleigh scattering consecutive days, May 26 and 27, at urban STAtransmittance, the ‘‘pollution smog’’ transmittance and suburban UC sites. Stagnating air allowed awas low, resulting in significant degradation of moderately polluted atmosphere over the cityvisibilities. Some deviations are observed in the center mainly on May 27. Early morning tempera-NIR region, which probably resulted from singu- tures in downtown area remained around 228C onlar values of the Angstrom parameters employed both days while relative humidity fell steadilyfor the whole spectral range, i.e. 350–1050 nm. from 62% at 0800 to 42% by 1500 LST. Westerly

21As shown by Adeyefa and Holmgren (1996) such light winds of 2 m s blew during morning,anomalies may be attributed to the deviation of while visibility was limited to 8–9 km. At the UCthe Angstrom formula at longer wavelengths. This site temperatures remained nearly constant atverifies the previously mentioned spectral restric- 218C and relative humidity dropped from 66 totion (350–862 nm) in obtaining Angstrom param- 46% between 0800 and 1600 LST; however, localeters. Thus, following Bird and Riordan (1986) katabatic wind blowing in the early morning fromand Adeyefa and Holmgren (1996) a multi-term Mt. Hymettus over the UC site resulted in pollu-Angstrom law was applied with b and a chang- tants concentration to be kept low.ing their values at 550 and 800 nm, but the Fig. 7 compares the experimental total Rayleigh-improvement was marginal. corrected optical depths together with the Ang-

Fig. 7. Comparison of experimental total Rayleigh-correctedFig. 6. Spectral transmittance of the ‘‘pollution smog’’ event. optical depths and Angstrom fits observed over the AthensMolecular-Rayleigh scattering, aerosol extinction and the atmosphere: 1 – over the Athens center, and 2 – over thecombined effects of scattering and absorption (total), are also University Campus. Important NO , O , O and H O absorp-2 3 2 2

shown. tion bands, are indicated.


was over the west and central basin during ‘‘rushhour’’ in the morning has been advected up thesouthwest facing mountain slope and over the UCsite (Lalas et al., 1983). Fig. 8 gives an exampleof the direct solar beam spectra taken over urbanand suburban atmospheres at air mass m 51.86.r

The observed difference in solar irradiances overthe sites is a manifestation of contrasting airproperties, e.g. the present of various radiativelyactive gases. Integrated values of the direct solarbeam in several narrowbands are given in Table 5.A 50% increase in attenuation of direct UVBirradiance was observed between the sites whilethe respective percentages in UVA, VIS and NIRspectral irradiances were found to be 35, 23 and17%.

5. CONCLUSIONS

Results obtained in the present analysis of thespectral composition of global and scattered ir-

Fig. 8. Comparison of the spectral direct solar beam dis- radiances showed that various atmospheric param-tribution: (1) over the Athens center and (2) over the

eters cause considerable changes to the spectralUniversity Campus.distribution of radiant energy reaching the ground.Our conclusions are summarized as follows:

Spectral measurements showed that the urbanstrom fits observed over the UC site and over the atmosphere produces a greater relative effect ondowntown Athens area. In spite of sites’ differ- the UV than on VIS and NIR wavebands. Pollutedences, the variation structure of the optical depths conditions cause a variation in atmospheric com-exhibited similar signatures. This may be due to position and consequently wide changes in thesmall particles scattering and trace gases absorp- radiative attenuation. The greater attenuations intion that are conceivably present over the active the UV band were about 27% while in VIS andAthens atmosphere. Nonetheless, differences in NIR bands attenuations summed up to 17% andpollution level between the Athens center and its 16% respectively during morning hours when thesuburbs might stem from differences in local urban atmosphere is loaded with higher levels ofemissions, e.g., automotive exhausts. In dow- gaseous pollutants and aerosol. On the other handntown Athens the optical depth values were found the relative increase in scattered irradiance pro-to be 55% higher in the UV spectral band than at duced by gaseous pollutants and aerosol wasthe UC site and up to 25% higher in the NIR greater in the NIR band. In the case of highband. It is worth noting that by midday both sites aerosol concentrations its value was found 66%exhibited aerosol optical depth values of the same higher compared to that corresponding to goodorder of magnitude. This can be explained by transparency conditions. VIS and UV bands pro-considering prevailing wind flow patterns over the vided lower increases by 54% and 21% respec-Athens basin. A southwesterly sea breeze flow is tively.established toward noon; the polluted air which Experimental total Rayleigh-corrected and

22Table 5. Direct irradiance (W m ) in various narrow bands (nm) over urban and suburban atmospheres. Air mass 1.86.Percentages in parentheses indicate sites’ difference

Aerosol depth UVB UVA UV VIS NIR(at 500 nm) 300–320 nm 320–400 nm 300–400 nm 400–700 nm 700–1100 nm

22 22 22 22 22(W m ) (W m ) (W m ) (W m ) (W m )

UC 0.26460.008 0.218 13.9 14.2 275.8 201.3STA 0.41260.014 0.110 9.1 9.3 213.4 167.9

(250%) (235%) (235%) (223%) (217%)


Charlson R. J., Langner J., Rodhe H., Leovy C. B. and Warrenspectral aerosol optical depths were retrieved,S. G. (1991) Perturbation of the northern hemispheric

representing differences in polluted air over the radiative balance by backscattering from antropogenic sur-Athens atmosphere. Urban polluted air was found face aerosols. Tellus 4AB, 152–163.

Glasius M., Carlsen M. F., Hansen T. S. and Lohse C. (1999)to significantly increase aerosol optical depthMeasurements of nitrogen dioxide on Funen using diffusionvalues. The Angstrom parameters were derived tubes. Atmos. Environ. 33, 1177–1185.

through spectral extinction curves determined Gueymard C. (1994) Updated transmittance functions for usein fast spectral direct beam irradiance models. Proc. of thefrom a log–log plot fit.23rd Am. Solar Energy Soc. Ann. Confer., San Jose, CA,The diffuse-to-direct ratios of solar irradiancesJune 1994.

were found to depend strongly on the polluted Jacovides C. P. and Karalis J. D. (1996) Broad band turbidityparameters and spectral band resolution of solar radiationatmospheric conditions. The depletion in directfor the period 1954–1991, in Athens, Greece. Int. J.irradiance is significant, particularly in the short-Climatol. 16, 229–242.

est wavelength domain, e.g. UV spectral band but Jacovides C. P., Giannourakos G. P., Asimakopoulos D. N. andis compensated to some extent by an increase in Steven M. D. (1998) Measured spectra of solar ultraviolet

irradiances at Athens basin, Greece. Theor. Appl. Climatol.the diffuse component. It was established that the59, 107–119.diffuse component accounts for more than 80% of Justus C. G. and Paris M. V. (1985) A model for solar spectral

the radiation field, at least for shorter wavelengths irradiance and radiance at the bottom and top of a cloudlessatmosphere. J. Appl. Meteor. 24, 193–205.under highly gaseous pollutants and aerosol load-

Kasten F. and Young A. T. (1989) Revised optical air massings. tables and approximation formula. Applied Optics 28,The high urban polluted air has significant 4735–4738.

King M. D. (1979) Determination of the ground albedo andinfluence on the atmospheric spectral transmit-the index of absorption of atmospheric particulates bytance. Transmittance degradations caused by aremote sensing. Part II: Application. J. Atmos. Sci. 36,

highly polluted atmosphere are more important in 1072–1083.King R. and Buckius R. O. (1979) Direct solar transmittancethe UV and VIS spectral bands, where the trans-

for a clear sky. Solar Energy 22, 297–301.mittance is less wavelength dependent in com-de la Casinier A., Cabot T. and Benmansour S. (1995)

parison to that of slightly polluted atmospheres. Measuring spectral diffuse solar irradiance with non-cosineflat-plate diffusers. Solar Energy 54, 173–182.

de la Casinier A., Bokoye A. I. and Cabot T. (1997) DirectAcknowledgements—This work was supported in part by a solar spectral irradiance measurements and uptated simplegrant from the Greek General Secretariat of Research and the transmittance models. J. Appl. Meteor. 36, 509–520.British Council of Athens. The Associate Editor of the journal, Lalas D. P., Asimakopoulos D. N., Deligiorgi D. and HelmisDr. R. Perez, and the anonymous reviewers are highly C. G. (1983) Sea breeze circulation and photochemicalacknowledged for their comments and suggestions in improv- pollution in Athens. Atmos. Environ 17, 1621–1632.ing the revised manuscript. LeBaron B. A., Michalsky J. J. and Perez R. (1990) A simple

procedure for correcting shadowband data for all skyconditions. Solar Energy 44, 249–256.

Lorente J., Redan A. and de Cabo X. (1994) Influence ofREFERENCESurban aerosol on spectral solar irradiance. J. Appl. Meteor.

Adeyefa Z. D. and Holmgren B. (1996) Spectral solar 33, 406–415.irradiance before and during a Harmattan dust spell. Solar Michalsky J. J. and Stokes G. M. (1983) Mt. St. Helen’sEnergy 57, 195–203. aerosols: some tropospheric and stratospheric effects. J.

Anderson S. M. and Mauersberger K. (1992) Laser measure- Appl. Meteor. 22, 640–648.ments of ozone absorption cross sections in the Chappuis Molina L. T. and Molina M. J. (1986) Absolute absorptionband. Geophys. Res. Lett. 19, 933–936. cross section of ozone in the 185 to 350 nm wavelength

Angstrom A. (1961) Technique of determining the turbidity of range. J. Geophys, Res. 95, 14501–14508.the atmosphere. Tellus 13, 214–223. Nann S. and Riordan C. (1991) Solar spectral irradiances

Batlles F. J., Olmo F. J. and Alados-Arboledas L. (1995) On under clear and cloudy skies: measurements and a semiem-shadowband correction methods for diffuse irradiance mea- pirical model. J. Appl. Meteor. 30, 447–462.surements. Solar Energy 54, 105–114. Nilsson B. A. (1994) Model of the relation between aerosol

Bird R. E., Hulstrom R. L., Kliman A. W. and Eldering H. G. extinction and meteorological parameters. Atmos. Environ.(1982) Solar spectral measurements in the terrestrial en- 28, 815–825.vironment. Applied Optics 21, 1430–1436. Perez R., Ineichen P., Seals R., Michalsky J. J. and Stewart R.

Bird R. E. and Riordan C. J. (1986) Simple solar spectral (1990) Modelling daylight availability and irradiance com-model for direct and diffuse irradiance on horizontal and ponents from direct and global irradiance. Solar Energy 44,tilted planes at the Earth’s surface for cloudless atmos- 271–289.pheres. J. Climate Appl. Meteor. 25, 87–97. Riordan C., Myers D., Rymes M., Hulstrom R., Marion W.,

Brine D. T. and Iqbal M. (1983) Diffuse and global spectral Jennings C. and Whitaker C. (1989) Spectral solar radiationirradiance under cloudless skies. Solar Energy 30, 447–453. data base at SERI. Solar Energy 42, 67–79.

Cachorro V. E., Gonzalez M. J., Frutos A. M. and Gazanova J. Schneider W., Moortgat G. K., Tyndall G. S. and Burrows J. P.L. (1989) Fitting Angstrom’s formula in spectrally resolved (1987) Absorption cross section of NO in the UV and2

optical thickness. Atmos. Environ. 23, 265–270. visible (200–700 nm) at 298 K. J. Phot. and Photobiol., A:Cachorro V. E., Duran P. and de Frutos A. M. (1996) Retrieval Chemistry 40, 195–217.

of vertical ozone content using the Chappuis band with high Schwartz S. E. (1996) The Whitehouse Effect-shortwavespectral resolution solar radiation measurements. Geophys. radiative forcing of climate by anthropogenic aerosols: anRes. Let. 23, 3325–3328. overview. J. Aerosol Sci. 27, 359–382.


Steven M. D. (1984) The anisotropy of diffuse radiation Webb A. R. (1992) Spectral measurements of solar ultraviolet-determined from shade ring measurements. Quart. J. R. B radiation in southeast England. J. Appl. Meteor. 31,Met. Soc. 110, 261–270. 212–216.

Vasilyev O. B., Conteras L. A., Velazquez A. L., Fabi P. R., Wenny B. N., Schafer J. S., DeLuisi J. J., Saxena V. K.,Ivlev L. S., Kovalenko A. P.,Vasilyev A.V., Jukov V. M. and Barnard W. F., Petropavlovskikh I. V. and Vergamini A. J.Welch R. M. (1995) Spectral optical properties of the (1998) A study of regional aerosol radiative properties andpolluted atmosphere of Mexico City. J. Geophys. Res. 100, effects on ultraviolet-B radiation. J. Geophys. Res. 103,26027–26044. 17083–17097.

spectral solar irradiance and some optical properties for various polluted atmospheres

Documents