Atmospheric Research xxx (2010) xxx–xxx

ATMOS-02112; No of Pages 10

Contents lists available at ScienceDirect

Atmospheric Research

j ourna l homepage: www.e lsev ie r.com/ locate /atmos

ARTICLE IN PRESS

Ground-based observations of solar radiation at three Italian sites, duringthe eclipse of 29 March, 2006: Signs of the environment impact on incomingglobal irradiance

Boyan Petkov a,b,⁎, Claudio Tomasi a, Vito Vitale a, Alcide di Sarra c, Paolo Bonasoni a,Christian Lanconelli a, Elena Benedetti e, Damiano Sferlazzo d, Henri Diémoz f,Giovanni Agnesod f, Riccardo Santaguida g

a Institute of Atmospheric Sciences and Climate (ISAC), Consiglio Nazionale delle Ricerche (CNR), Via Gobetti 101, I-40129 Bologna, Italyb International Centre for Theoretical Physics (ICTP), Strada Costiera 11, I-34014 Trieste, Italyc ENEA, ACS-CLIM-OSS, Via Anguillarese 301, I-00123 Santa Maria di Galeria (Rome), Italyd ENEA, ACS-CLIM-OSS, Contrada Grecale, I-92010 Lampedusa, Italye Institute of Acoustic “O.M. Corbino” - ICES Group, Consiglio Nazionale delle Ricerche (CNR), via Fosso del Cavaliere 100, I-00133 Roma Tor Vergata, Italyf Regional Environmental Protection Agency of Aosta Valley (ARPA Valle d'Aosta), Sezione Agenti Fisici, Grande Charriére 44, I-11020 Saint-Cristophe, Aosta, Italyg Italian Air Force Meteorological Service, CAMM - Monte Cimone, Via delle Ville 40, I-41029-Sestola, Modena, Italy

a r t i c l e i n f o

⁎ Corresponding author. Institute of Atmospheric(ISAC), Consiglio Nazionale delle Ricerche (CNR), ViaBologna, Italy. Tel.: +39 051 639 9586; fax: +30 051

E-mail address: b.petkov@isac.cnr.it (B. Petkov).

0169-8095/$ – see front matter © 2009 Elsevier B.V.doi:10.1016/j.atmosres.2009.12.006

Please cite this article as: Petkov, B., et aeclipse of 29 March, 2006: Signs of the e

a b s t r a c t

Article history:Received 10 December 2009Accepted 12 December 2009Available online xxxx

The time-patterns of ground-level solar irradiance during the solar eclipse of 29 March, 2006,were observed at three Italian stations (Lampedusa, Mt. Cimone and Bologna) using differentradiometric techniques. The global irradiance measured at the sites was found to reach theminimum at times not coinciding with those predicted by radiative transfer model evaluations,with ahead or lag times depending on the optical characteristics of the surface–atmospheresystem in the areas surrounding of the stations. This different behaviour has been mainlyattributed to the different influence of the environmental conditions on the diffuse radiancecomponent measured at the observation sites. The present results indicate that the incomingdiffuse radiance recorded at the three stations was appreciably affected by contributionsarising from extended regions of about 30–100 km range from the stations. Such anexplanation agrees very well with the theoretical evaluations obtained in earlier studies. Thesurrounding environmental areas of impact at ultraviolet wavelengths have been found to bewider than those in the visible and near-infrared spectral ranges.

© 2009 Elsevier B.V. All rights reserved.

Keywords:Solar UV irradianceDiffuse UV irradianceSolar eclipse

1. Introduction

Solar eclipse episodes are often favourable opportunitiesfor closely observing the effects of interactions between solarirradiance and the terrestrial atmosphere (Abbott, 1958;Carapiperis and Karapiperis, 1959; Pruitt et al., 1965;Anderson et al., 1972; Zerefos et al., 2000; Koepke et al.,

Sciences and ClimateGobetti 101, I-40129639 9658.

All rights reserved.

l., Ground-based obsernvironmen..., Atmos. Re

2001; Dani and Devara, 2002; Blumthaler et al., 2006).Studies on this topic can be very useful for testing andimproving the radiative transfer models used to simulate thespectrum of ultraviolet (UV) solar irradiance reaching thesurface (Emde andMayer, 2007). Investigations performed sofar have shown that the ground-level solar irradiance duringa period of solar disk obscuration by the Moon begins todecrease at approximately the first eclipse contact time,reaches the minimum in correspondence of the eclipse'smaximum, and then increases again until the fourth contacttime. However, some studies have highlighted appreciabledeviations from the average irradiance time-patterns

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

2 B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

measured during previous eclipse episodes. For instance,measuring the solar direct broadband irradiance during theeclipse of 25 February, 1952, in Greece, Abbott (1958) foundthat the irradiance started to diminish 12 min before the firstcontact time and, subsequently, continued to assume lowervalues than usual throughout a certain period after the lastcontact. A similar irradiance drop prior to the first contact canbe noticed on examining the data reported by Miyake et al.(1949). However, Carapiperis and Karapiperis (1959) did notobserve such occurrences, when they performed solarirradiance measurements in Greece, during the eclipse of 30June, 1954.

Pruitt et al. (1965) carried out solar irradiance measure-ments at Davis (USA) during the eclipse of 20 July, 1963,recording a decrease in solar irradiance reaching the surfaceabout 5–6 min before the first contact time, and the diffuseradiation minimum a only few minutes before the maximumobscuration of the solar disk. Both Zerefos et al. (2000) andBlumthaler et al. (2006) measured the lag of the diffuseirradiance minimum with respect to the eclipse maximumtime, explaining the phenomenon as plausibly due to theappreciable contribution of diffuse radiance from areas farfrom the station and more weakly shadowed by the Moon. Asa result, solar irradiance did not closely follow the normaleclipse evolutionary patterns in such cases, where thediscrepancies were generally attributed to the variable effectsof the eclipse on the solar diffuse radiance field, whosefeatures depend closely on the atmospheric environmentalconditions. In fact, the global solar irradiance reaching theEarth's surface consists of both direct and diffuse radiationcomponents. The former depends only on the parameterscharacterising the optical properties of the atmospheric slantpath described by the incoming solar rays, while the latterarises mainly from the scattering processes due to airmolecules (Rayleigh scattering) and aerosols (Mie scatter-ing), taking place over a large area around the measurementsite (Iqbal, 1983).

In addition, the solar radiation partly reflected upward bythe surface or by clouds can be scattered by air molecules andaerosols, sometimes enhancing considerably the incomingdiffuse radiance flux measured at the observation site. SinceRayleigh scattering closely depends on wavelength, thediffuse radiance is comparable to, or may become evenmore intense than the direct component at UV wavelengths,with significantly lower values at the longer wavelengths inthe visible and near-infrared spectral ranges. Dependingstrongly on the atmospheric turbidity conditions and surfacereflectivity characteristics, UV diffuse radiance can varygreatly as a function of the environmental parameters.These dependence features have been the subject of numer-ous studies.

Huber et al. (2004) investigated the influence of nonhomogeneous surface albedo on the diffuse radiation, findingthat large heterogeneities can characterise the distribution ofthe UV component over a not uniform observation area, withdiscrepancies of up to 40% between the radiation fieldsmeasured over a snow-free terrain and a snow-coveredsurface. Kylling and Mayer (2001) compared field measure-ments of UV irradiance over an area containing high, snow-covered mountains and ice-free water surfaces with 3-Dmodel simulations. Considering various snow-cover charac-

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. Re

teristics, they gave evidence of a radiation enhancement ofabout 23–27% for clear-sky conditions and 40–60% in overcastcases, due to the high reflectance features of the snow surface.Similar results were obtained by McKenzie et al. (1998) oversome snow-covered areas, showing an increase in the globalsolar irradiance, equal to 22% and 28% in the UV-A and UV-Bbands, respectively. In Antarctica, Smolskaia et al. (1999)measured an increase in the UV global irradiance of about 10%over snow surfaces, with respect to analogous measurementstaken over open sea-water surfaces. The results weresubsequently confirmed by Mayer and Degünther (2000)using model simulations.

Also employingmodel evaluations, Lenoble (2000) studiedthe impact of surface reflectance characterising the surround-ing area on the zenith radiance reaching the ground forcloudless sky conditions. She found that (i) approximately 12–15% of the reflected photons came from the region within1 km around the observation site, (ii) about 25–30% from thesurrounding areawithin the 1–5 kmradius range, (iii) 43–47%from the 5–30 km radius range, and (iv) 10–15% due tomultiple reflections from more distant areas.

With regard to cloud effects, Degünther and Meerkötter(2000) studied the impact of remote clouds on ground-levelsolar UV radiation, by analysing the different spatial distri-bution of cloud layers around the observation site. Using athree-dimensional radiative transfer model, they found thatclouds about 15–20 km from the site can strengthen theincoming UV radiation at the surface by 15% and more, orreduce it by more than 50%, while more distant clouds wereestimated to exert a less marked influence, contributing tothe ground-level UV irradiance by a few percent only.

Thus, the environmental characteristics around an obser-vation site are expected to impact significantly on the solarradiation density flux measured at the ground, especially onits UV spectral component. However, while the influence offactors located in proximity of an observation site has beeninvestigated with both theoretical simulations and experi-mental methods, the influence of diffuse radiation originatingfrom distant regions has generally been analysed andestimated by means of physical models only, without usingexperimental data. Given the scales of the Moon penumbra,where the distances between sites presenting appreciablediscrepancies of the shadow intensity reach 100 km or more,it is evident that solar eclipse events can present favourableconditions for monitoring the effects produced by remoteenvironmental factors on ground-level solar radiation. Whenthe Moon's umbra moves on the Earth surface, it shadowsconsecutively the diverse parts of the areas around themeasurement station, to various extents. By associating suchoccurrences with the surface radiance time-patterns, thepresent study attempts to evaluate the extension of the areaaround the station that contributes to cause a significantimpact on the solar radiance field measured at the ground.

2. Instruments used at the three Italian sites anddescription of local eclipse circumstances andmeteorological conditions

The incoming solar irradiance was measured withindifferent spectral bands at three Italian sites, with latitudesranging from 35 °N to 45 °N, during the eclipse of 29 March,

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

3B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

2006: (i) the “R. Sarao” ENEA station for Climate Observationson Lampedusa Island, not far from the African coast, (ii) theMt. Cimone GAW station, on the roof of the ISAC-CNR(Institute of Atmospheric Sciences and Climate - ItalianNational Research Council) “O. Vittori” Laboratory, locatedat the top of the highest peak of the Northern Apennines, and(iii) the Bologna ISAC-CNR station, on the roof of the ISACbuilding, in Northern Italy, at the foot of the Apennines. Thegeographical coordinates of the three stations and their mainsolar eclipse circumstances are given in Table 1. Fig. 1presents the map prepared by Espenak and Anderson(2004), in which the positions of the three stations areindicated, with respect to the most significant eclipsepatterns and magnitude isopleths.

Different instruments were employed at the stations, withall their clocks adjusted to the Universal Time Clock (UTC), torecord homogeneous measurement times:

(1) Solar irradiance measurements were performed at theLampedusa ENEA station (Pace et al., 2006), using theMulti Filter Rotating Shadowband Radiometer(MFRSR) (Harrison et al., 1994) to record a pair ofspectral series of signals, the first relative to globalradiation and the latter to diffuse component, with afrequency of 1 overall measurement every 15 s; eachspectral series consisted of signals taken within sixchannels of half-bandwidth HBW=10 nm, centred at415.6 (hereinafter referred to as 416 nm), 495.7, 614.6,672.6, 868.7 (hereinafter referred to as 869 nm) and939.6 nm, respectively.

(2) The SKU 430 High Output Sensor and the SKS 1110Single Channel Light Sensor, bothmanufactured by SkyInstruments LTD, were employed at Mt. Cimone tomeasure the incoming global solar radiation during theeclipse day. The first sensor was equipped with opticalfiltering components, giving an overall spectralresponsivity curve peaked at 297 nm and coveringthe UV-B band (280–315 nm). Simultaneously, the SKS1110 pyranometer performed continuous measure-ments of the solar global broadband irradiance withinthe 400–1100 nm wavelength range.

(3) At Bologna, the measurements were performed usingthe Ultraviolet Radiometer, model UV-RAD, and theprecision pyranometer CM22, which are routinelyemployed to carry out calibration tests, scheduled as

Table 1Local circumstances of the solar eclipse of 29March, 2006, at the three Italianstations.

Station Lampedusa(ENEA)

Bologna(ISAC-CNR)

Mt. Cimone(ISAC-CNR)

Instruments MFRSR UV-RADCM22

SKU 430SKS 1110

Latitude 35° 31′ N 44° 31′ N 44° 12′ NLongitude 12° 38′ E 11° 20′ E 10° 42′ EAltitude (m a.m.s.l.) 50 34 2165First contact time (UT) 09:13:50 09:33:01 09:31:59Maximum eclipse timetem (UT)

10:26:51 10:38:03 10:36:47

Fourth contact time (UT) 11:41:53 11:44:11 11:42:46Eclipse magnitude 0.707 0.532 0.526

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. Re

part of the BSRN (Baseline Surface Radiation Network)programme (see ARM/BSRN web site). Manufacturedat the ISAC-CNR Institute, the UV-RAD radiometermeasures the UV global solar irradiance within sevenspectral channels centred at peak-wavelengths of 300,306, 310, 314, 325, 338 and 364 nm, with values ofHBW varying between 0.6 and 0.9 nm (Petkov et al.,2006). During the three time-periods from around5 min before to 5 min after the first contact time,eclipse maximum time and fourth contact time,respectively, the UV-RAD measured continuously theincoming solar global irradiance within the channelcentred at 325 nm, taking measurements with anaverage frequency of 3 signals per minute. The Kipp &Zonen CM22 pyranometer was used to perform precisemeasurements of the broadband solar irradiance, overthe wide wavelength range from 290 to 3600 nm,simultaneously to the UV-RAD.

The local eclipse evolutionary times at Bologna have beencalculated by Espenak and Anderson (2004) and are takeninto account in the present study. Since this report does notprovide detailed information on the times at the other twostations, their local circumstances were defined using theEclipse Software EmapWin, Version 1.21, created by Takesako(2001). The average discrepancies, calculated following thetwo approaches applied to several Italian sites, were foundequal to (i) 4.5±0.6 s for the first contact time, (ii) 5.2±0.6 sfor the eclipse maximum time tem, and (iii) 5.9±0.4 s for thefourth contact time. Such discrepancies of a few seconds canall be considered relatively small for the present applicationpurposes. For this reason, both Espenak and Anderson (2004)and Takesako (2001) evaluations have been considered to bereliable for use in the present study.

Meteorological conditions during the eclipse day can bebriefly summarised as follows:

- at Lampedusa, the cloud cover consisted of extended cirros-tratus, andwas (i) 5 octas or higher before∼09:00UTC, (ii) 1–2 octas between 09:00 and 10:10 UTC, and (iii) less than 1octa subsequently,when the cloud bandmoved on and clear-sky conditions prevailed, as shown by a sky imager used totake regular pictures of the skyevery5 minduring theeclipse.

- At Mt. Cimone, the sky was cloudless during the morning,until several minutes after the last eclipse contact, whennumerous cumulus congestus clouds appeared, stronglyand irregularly attenuating the incoming solar radiation.

- At Bologna, the sky was generally cloudless, except in thesouth-western quadrant, where slight cumulus congestusclouds appeared, incidentally covering the solar disk onlyafter the fourth eclipse contact time.

A layer of desert aerosol particles extending up to about5 km altitude was observed by the ENEA Lidar (Di Iorio et al.,2003) at Lampedusa, on 29 March, 2006. Correspondingly,the daily average aerosol optical depth (AOD) was evaluatedas 0.12 at the visible MFRSR wavelengths, this value beingrather low for such a remote Mediterranean site on a daycharacterised by a Saharan dust transport episode. A value ofthe Ångström (1964) exponent α equal to 0.04 wasdetermined using the spectral values of AOD obtained fromthe MFRSR global and diffuse irradiance measurements,

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

Fig. 1. Map of the Moon's umbra and penumbra over the North African and Mediterranean region, as defined by Espenak and Anderson (2004). The maximumeclipse time isopleths are given in steps of 30 min (see labels), while those of constant eclipse magnitude are given from 1.0 to 0.0, in steps of 0.20. The positions othe three observation stations are indicated by triangles labeled with L (Lampedusa ENEA station), MC (Mt. Cimone ISAC-CNR station) and B (Bologna ISAC-CNRstation).

4 B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

indicating the presence of a significant columnar load ofcoarse particles, capable of causing almost neutral extinctionfeatures. Such values of AOD and α clearly suggest thepresence of an optically dense layer, consisting of relativelylarge size particles of desert origin. In general, Saharan dustparticles exhibit relatively low values of the single scatteringalbedo (Meloni et al., 2006), associated with moderateabsorption of the short-wave radiation, thus enhancingatmospheric opacity at UV wavelengths, and appreciablyattenuating the diffuse radiation reaching the measurementsite.

The Alps, situated around 120–150 km North of Bologna,were clearly visible to the naked eye on the eclipse day.Applying the Horvath (1971) correction to visibility observa-tions of a target covered by snow and, hence, behaving quitedifferently froman ideal black body, it can be roughly evaluatedthat the visual range was 65–80 km during the eclipse. Theseground-level visibility conditions correspond to values of theaerosol volume extinction coefficientβa(550 nm) at the surfacevarying between less than 0.04 and about 0.05 km−1. For thisrange of βa(550 nm), values of AOD ranging between 0.04 and0.07 are obtained at visiblewavelengths, under the assumptionof a scaleheight close to1.25 km, chosen in accordancewith thePenndorf (1954) evaluations and the average estimates of 1.0–1.5 km determined by Tomasi (1982) in the Po Valley regionduring the early spring.

Please cite this article as: Petkov, B., et al., Ground-based observations of solar radiation at three Italian sites, during theeclipse of 29 March, 2006: Signs of the environmen..., Atmos. Res. (2010), doi:10.1016/j.atmosres.2009.12.006

f

3. Field measurements and main results

Fig. 2 shows the solar irradiance time-patterns measuredat Lampedusa on 29March, 2006. The upper part of the graphreports the MFRSR measurements recorded within thechannel centred at 415.6 nm, on the eclipse day, which giveevidence of the strong reduction effects caused by the eclipseon the global, direct and diffuse irradiance. In addition, thetime-patterns of the irradiance measurements were alsosubject during the morning to some marked and suddendecreases, caused by irregular passages of clouds in front ofthe Sun. The lower part of Fig. 2 shows a comparison amongthe global solar irradiance and its diffuse and directcomponents measured within the 868.7 nm channel, withsimilar features to those observed for the 415.6 nm channel.

Fig. 3 shows the comparison between the solar irradiancetime-patterns measured at Mt. Cimone and Bologna. Theresults shown in Figs. 2 and 3 indicate that the solarirradiance reaching the Earth's surface on 29 March followedthe obscuration phases of the solar disk: it begins to drop afterthe first contact time, reaches the minimum at around themaximum time tem, and subsequently increases graduallyuntil assuming normal values during the final phase. Therelative irradiance drop at time tem for each of the threestations approximately corresponds to the eclipse magnitudevalue presented in Table 1. Fig. 3 shows that the solar

Fig. 2. Comparison of the time-patterns of solar global irradiance and itscomponents, measured by the MFRSR at Lampedusa on 29 March, 2006,within the channels centred at the 416 nm (a) and 869 nm (b) wavelengths.Solid, dashed and dash–dotted curves refer to the global, direct and diffuseirradiance measurements, respectively. The dotted vertical lines mark thetimes relative to first contact, maximum eclipse and fourth contact.

5B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

irradiance time-patterns measured at both stations weresubject to quite similar variations, with a sudden decrease inthe broadband irradiance after the first contact time and ashort lagging of the UV irradiance decrease measured by boththe UV-RAD and SKU 430 radiometers. During the periodincluding the eclipse final phase, the instruments used atBologna and Mt. Cimone also yielded very similar irradiancetime-patterns for both UV and broadband components. TheUV irradiance reached its maximum about 6 min before theeclipse end, and varied very slowly after the fourth contact,presenting almost constant values during the 5–10 minfollowing the final time of eclipse.

An important result highlighted by Figs. 2 and 3 is thatboth UV and broadband components of global solar irradiancereached their minima at times tmin, which do not coincideexactly with the eclipsemaximum time. The reasons for thesetime discrepancies are examined below, through a detailedanalysis of the field data.

4. Determination of irradiance minimum time at thestations and model simulations

In order to define the precise value of tmin, it was decidedto fit the measured irradiance values with a third degreepolynomial curve expressed in the following form,

P3ðtÞ = a0 + a1t + a2t2 + a3t

3; ð1Þ

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. Re

where t is the time. This approximation represents very wellthe time-patterns of the solar irradiance within ±5–10 minaround the eclipse minimum, the corresponding best-fitsolutions being obtained at the three sites for rather highvalues of the regression coefficient r, all close to 0.98.Parameter tmin can be determined analytically as theminimum of the best-fit polynomial function P3(t) inEq. (1). Fig. 4 shows a pair of application examples of theprocedure, relating to the global irradiance measurementstaken with the UV-RAD and MFRSR at the 325 and 415.6 nmwavelengths, respectively, and presents the correspondingbest-fit curves together with the pairs of curves yielding theso-called standard errors of estimate (Spiegel, 1972). Thestandard error curves define the time-intervals within whichthe best-fit polynomial can vary, giving the measure of theerror affecting parameter tmin. The corresponding values oftmin determined for all the instruments employed at the threestations are presented in Table 2.

It may be obvious to assume that the surface solarirradiance should reach its minimum at the eclipse maximumtime. By contrast, the model evaluations showed that adiscrepancy exists between these two times, the differencebeing attributed to interferences between two oppositetemporal trends: (i) the natural daily change in the surfacesolar irradiance, as a result of the solar zenith angle variations,and (ii) the evolutionary trend of the irradiance caused by theeclipse. Evaluations of solar irradiance minimum time weremade through simulations of irradiance time-patterns by theTUV solar radiative transfer model (Madronich, 1993), fromwhich the corresponding minimum time tTUVmin was determinedfollowing the above described procedure. Emde and Mayer(2007) demonstrated that a 1-D model realistically evaluatesthe solar irradiance at the ground within the penumbra; thus,it can be assumed that the chosen model is suitable forestimating the corresponding times tTUVmin . The spectral solarirradiance trend at the top of the atmosphere during theeclipse was estimated following the approach developed byKoepke et al. (2001).

Fig. 5 shows the spectral distribution curve of irradianceminimum time evaluated by the model in the case of the Mt.Cimone station, which can be assumed to represent also thewavelength-dependent behaviour of this parameter at theother stations. The irradiance measured at wavelengthsshorter than 330 nm reached the minimum significantlybefore that relative to longer wavelengths, due to the limbdarkening effect (Koepke et al., 2001). The main part of UV-A(λN330 nm), visible and near-IR diapasons shows a slightvariability of irradiance minimum time, with discrepancies of15 s between 700 and 330 nm, 6 s between 700 and 400 nm,and 4 s between 700 and 600 nm.

The values of parameters tmin and tTUVmin are given inTable 2: in the case of Lampedusa, only the 416 nm irradianceminimum is given, assuming that this wavelength representsthe behaviour of all the spectral bands measured by MFRSR(see Fig. 5). The results in Table 2 show that the measuredtimes of irradiance minimum significantly differ from thecorresponding values determined by the model. The dis-crepancies Δtm= tmin − tTUVmin showed a marked minimumanticipation in the case of Bologna, and noticeable and lessmarked lags at Lampedusa and Mt. Cimone, respectively.Fig. 6 illustrates the differences between evaluated and

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

Fig. 3. Upper part: time-patterns of the solar irradiance measured on 29 March, 2006, at Bologna with (a) the UV-RAD radiometer and (b) the CM22 pyranometer.Lower part: as in the upper part for themeasurements performed atMt. Cimone with (c) the SKU 430 sensor, and (d) the SKS 1110 sensor. The dotted vertical linesmark the times relative to the first contact, maximum eclipse and fourth contact, respectively.

6 B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

measured minimum times in the three cases of (i) 416 nmMFRSR channel at Lampedusa, (ii) UV irradiance measure-ments at Mt. Cimone, and (iii) UV-RAD measurements atBologna, respectively. The corresponding results for thebroadband irradiance are presented in Fig. 7, where theirradiance minimum times were calculated at the 500 nmwavelength, since it represents in practice the time when theirradiance signals measured within large spectral intervals byCM22 and SKS 1110, respectively, reached their minima (seeFig. 5). Such measurements indicate that the surface solarirradiance observed at Bologna and Mt. Cimone reached theirminima nearly simultaneously, with discrepancies of 11 s forUV and 16 s for the broadband irradiance observations. Inboth cases, the differences are smaller than the error made inestimating the minimum time. At the same time, theseminima should occur with differences of 1 min and 47 s forUV and 1 min and 18 s for broadband irradiances,respectively.

At Mt. Cimone, the solar irradiance minimum shouldanticipate the maximum eclipse by 1 min and 54 s for UV-Band 1 min and 5 s for the broadband irradiance. At Bologna, thedifferences were evaluated as 1 min and 23 s at 325 nm and1 min and 3 s for the broadband irradiance, respectively, thesevalues being very close to the corresponding ones found at Mt.Cimone. The discrepancy of 31 s between the anticipationsobserved for theUVmeasurements atMt. Cimone and Bologna,respectively, is presumably due to the spectral differencescharacterising the UV irradiance measurements performed atthe two stations. Thus, it can be concluded that the irradiance

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. Re

anticipations remained constant at both stations, when thepenumbra was passing over the region around these twostations. In this case, the existence of two isopleths requiresinvestigation to explain the differences between observationsand simulations, the first relative to the eclipsemaximum time,and the second to the irradiance minimum time, the twoisopleths being parallel and moving at the same forwarddisplacement velocity.

5. Discussion of the results

As pointed out in the Introduction, the diffuse componentof the incoming irradiance at an observation site originatesfrom the surrounding environmental area, through interac-tion processes mainly due to Rayleigh and aerosol scatteringand surface reflectance effects. Assuming that the stationenvironment contributions to the diffuse irradiance do notvary as a function of the incoming direction, and, hence, thediffuse irradiance presents a uniform spatial distributionaround the site, the irradiance minimum should be observedat the time when the incoming solar irradiance reaches itsminimum above the station. At this time, the areas situatedon either side of the isopleth containing the obscurationmaximum points and passing over the station should beshadowed to the same extent. Thus, the solar irradianceminima observed at the three Italian stations during the solareclipse of 29 March, 2006, should occur at the correspondingtimes tTUVmin predicted by the model evaluations. Instead, thefield measurements clearly show that this expectation was

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

Fig. 4. Application examples of the method followed in the present study todetermine the solar irradiance minimum time. The values of global solarirradiance measured at Bologna by the UV-RAD radiometer at the 325 nmwavelength are represented with circles in the upper part (a), while thosetaken at Lampedusa with the MFRSR instrument at the 416 nm wavelengthare represented with triangles in the lower part (b). The field data arecompared with the corresponding polynomial best-fit (solid) curves, giventogether with the pairs of dashed curves defining the standard errors ofestimate. The irradiance minimum time tmin is indicated by the solid verticalline. For comparison, the maximum eclipse time tem (see Table 1) is markedby the dotted vertical line.

Fig. 5. Model simulation of the spectral distribution (solid curve) ofirradiance minimum time in the case of Mt. Cimone. The dotted horizontalline gives the eclipse maximum time.

7B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

not confirmed by observations, and the correspondingdiscrepancies Δtm assumed not negligible values, withpositive signs at Lampedusa and Mt. Cimone and negativesign at Bologna. The different circumstances lead us toconclude that the spatial distributions of diffuse irradiancearound the three stations were not homogeneous. Thus, thesurrounding environmental regions are expected to givecontributions of different intensity to the diffuse solarirradiance component, coming from different directions,due to the diverse directional enhancement of horizontalphoton transport (Emde and Mayer, 2007). When the Moon'spenumbra moves forward, shadowing the surrounding areasthat mainly contribute to the overall diffuse radiance

Table 2Occurrence times of the global solar irradiance minimum during the eclipse of 29 M

Instrument (station) Spectral band of the irradiance measurement

FMRS (Lampedusa) 416 nmUV-RAD (Bologna) 325 nmCM22 (Bologna) BroadbandSKU 430 (Mt. Cimone) UV-BSKS 1110 (Mt. Cimone) Broadband

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. Re

measured at the station, the global irradiance recorded atthe observation site reaches its minimum. To illustrate such ahypothesis, the specific environmental conditions of theeclipse day should be analysed more closely. However, themain environment characteristics can be observed in Fig. 8,which shows a satellite image extracted from the data-setavailable at the Satellite Receiving Station web site (seeReferences) of Dundee University (UK), made in the visiblespectral band at 10:22 UTC of 29 March, 2006.

Previous studies, such as those of McKenzie et al. (1998),Smolskaia et al. (1999) and Huber et al. (2004), demonstratedthat surface reflectivity around an observation station is themost important parameter among those impacting on the skydiffuse radiance observed at the ground. Reflected back to thespace, solar irradiance turns out to be scattered againdownward by the atmosphere, thus enhancing the diffuseradiation field forming at the terrestrial surface during thefirst passage of the solar beams. Considering the crucial role ofsurface albedo, it was decided to analyse more closely thesurface reflectance characteristics of the surrounding areas atthe three sites.

North of Bologna and Mt. Cimone, from west to east, thePo Valley region (not covered by snow) and the Adriatic seapresented low surface reflectivity, while the Apennines chainsouth of Mt. Cimone and Bologna showed high surface albedoduring the observation period, due to the presence of a thicksnow cover. The Mediterranean sea area around Lampedusawas characterised by low reflectivity and only the cloud bandsituated south-east of Lampedusa presented high reflectivityfeatures.

The thick white line in Fig. 8 indicates schematically theeclipse maximum points, and the arrow shows the forward

arch, 2006, and corresponding time discrepancies Δtm.

tmin, UTC tTUVmin UTC Δtm, min

10:27:26±00:00:15 10:26:21 01:05±00:1510:35:30±00:00:15 10:36:40 −01:10±00:1510:36:36±00:00:15 10:37:00 −00:24±00:1510:35:19±00:00:15 10:34:53 00:26±00:1510:36:20±00:00:15 10:35:42 00:38±00:15

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

Fig. 7. Time-patterns of the broadband irradiance determined in terms ofbest-fit polynomials at Mt. Cimone and Bologna. As in Fig. 6, the dash–dot–dot vertical lines represent the irradiance minimum times defined by themodel estimation curves (see explanation in text), the dash–dotted verticallines indicate the corresponding times for the field measurements (errorsrepresented by horizontal solid bars) and, the vertical dotted lines give theeclipse maximum times at the three stations, respectively.

Fig. 6. Time-patterns of modelled (dashed curves) and measured (solidcurves), presented by the best-fit polynomial solutions of solar irradiancerelative to the cases observed at (a) Lampedusa, (b) Mt. Cimone and (c)Bologna stations. The dash–dot–dot vertical lines represent the irradianceminimum times defined by the model estimation curves, the dash–dottedvertical lines indicate the corresponding times for the field measurements(errors represented by short horizontal solid bars) and, the vertical dottedlines give the maximum eclipse times at the three stations, respectively.

8 B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

moving direction of the Moon's penumbra. As pointed out atthe end of the previous section, the isopleths relative to theminimum points of the incoming solar irradiance forLampedusa station and Mt. Cimone–Bologna area, respec-tively, can be assumed to be parallel to the eclipse maximumisopleth, as shown in Fig. 8, where they are represented by apair of thick white dashed lines pertaining to different timesof the eclipse event.

Bearing in mind the above assumptions, the followingscenario can be posited. Firstly, the penumbra shadowedLampedusa to the maximum extent at 10:26:21 UTC, but theirradiance measured at this station reached its minimumabout 1 min later, when the eastern part of the station

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. Re

environment was shadowed to the greatest extent, asindicated by the white dashed line in Fig. 8. Thus, on thebasis of the remarks made at the beginning of this section, itcan be assumed that the eastern sector contributed to thediffuse irradiance measured at Lampedusa more stronglythan the western one, so that the global irradiance reachedthe minimum when the eastern part of the station environ-ment was most shadowed. Espenak and Anderson (2004)evaluated that the penumbra velocity during the eclipse overItaly was around 0.8 km/s; therefore, at time tmin relative toLampedusa, the minimum of incoming solar irradiance waspositioned close to the place marked by the dashed line, i.e.about 50 km north-east of the station. In this case, the mostoptically dense part of the cloud band in Fig. 8, which alsopresents the highest reflectance characteristics (Kylling et al.,2005), was shadowed to the maximum extent. Since it isestimated that the cloud band was about 100 km south-eastof Lampedusa, it can be concluded that this sector of the cloudcover around Lampedusa contributed predominantly to themeasured diffuse irradiance and, consequently, to theincoming global irradiance.

Degünther andMeerkötter (2000) adopted a 3-D radiativetransfer model to investigate the impact of remote clouds onsurface UV radiation. They evaluated the irradiance in themiddle of a square sector, which was surrounded by a cloudlayer, by varying the sector dimensions and assuming anaerosol optical depth of 0.33 at the 330 nmwavelength. Theirestimates indicate that the clouds situated about 50 km from

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

Fig. 8. Satellite image of the Mediterranean area, taken by the NOAA 17satellite on 29 March, 2006, at 10:22 UTC, from the web site of the SatelliteReceiving Station at Dundee University (UK) (see http://www.sat.dundee.ac.uk/). The locations of the three observation sites are indicated by whitecircles marked with labels L (Lampedusa ENEA station), MC (Mt. CimoneISAC-CNR station) and B (Bologna ISAC-CNR station). The thick white linedefines the eclipse maximum isopleth, the arrow giving its forward movingdirection. The dashed white lines define the isopleths of the surface pointsshadowed to the maximum extent at time tmin for the Lampedusa station onthe left side of the eclipse maximum isopleth, and for Mt. Cimone andBologna stations, on the right side. The white and black rank given in abscissadefines the scale of the image in steps of 100 km.

9B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

the station should contribute to enhance UV-B (300 nm) andUV-A (330 nm) irradiances by about 1% and 2%, respectively.The results of the present analysis indicate that the distanceof the remote cloud bands capable of impacting mainly thesurface solar irradiance at the 416 nm wavelength (i.e. closeto the limit separating the UV and visible components) couldextend up to 100 km. It should be pointed out that the aerosoloptical depth measured at Lampedusa on the eclipse day was0.12, which is appreciably lower than that assumed byDegünther and Meerkötter (2000) in their calculations, andthis difference implies that less intense attenuation processesoccurred at Lampedusa than in the model simulation.

Later, the penumbra moved towards the region includingthe Mt. Cimone and Bologna stations. The minima of both UVand broadband components of solar irradiance occurred at thefirst station about 30 s after the passage of the correspondingirradiance minimum isopleth (indicated by the white dashedline in Fig. 8) over Mt. Cimone. However, the correspondingirradianceminima observed nearly simultaneously at Bolognapresented an appreciable anticipation of 1 min and 10 s for UVradiation, and a shorter anticipation of 24 s only for thebroadband spectral band, respectively. Thus, theUV irradianceminima at Mt. Cimone and Bologna were recorded when the

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. Re

penumbra shadowed to the maximum extent the area of theApennines with the higher surface reflectivity characteristics.As can be seen in Fig. 8, the surface reflectance distributionaround the two stations presented asymmetrical features inthe Apennine area, the southern snow-covered region beingthe one with higher surface albedo. Thus, the southern regionof Apennines was the main source of the diffuse irradiancecoming from the environment around the two stations.Whenthe Moon penumbra shadowed this mountain chain to thegreatest extent, the diffuse component, and, consequently, theglobal irradiance as well, reached their minima at both Mt.Cimone and Bologna. These remarks suggest that the regionsat distances of about 30–70 km from the stations could impactappreciably on the surfacemeasurements of diffuse and globalirradiance.

In addition, it is clearly seen that the anticipationmeasuredat Bologna for UV irradiance is appreciably higher than that ofthe broadband radiation, which includes more importantcontributions of visible and near-IR radiation. The time-difference between the two minima indicates that the rangeimpacting the UV irradiance in this case was about 30 kmgreater than that affecting the visible and near-IR bands. Thisoccurrence could be explained by the extremely low aerosolcontent suspended in the atmosphere on the eclipse day,which produced a lower flux of visible and near-IR diffuseradiation, and, at the same time, extinguished the incomingUV diffuse irradiance from the sky to a more limited extentthan on days characterised by higher turbidity conditions.

The above-mentioned model simulations showed that (i)environment areas at distances greater than 30 km cancontribute to the measured UV irradiance by up to 15%(Lenoble, 2000), and (ii) neglecting the surface albedodistribution within the range of 40 km from the observationsite can cause an error of up to 3% in the UV irradianceestimation (Degünther et al., 1998). Thus, the results of thepresent study, which indicate that neighbouring areas in therange of 100 km can influence the surface irradiancemeasurements at some stations, are in good agreementwith theoretical evaluations.

6. Conclusions

The global solar irradiance was measured by variousinstruments during the solar eclipse of 29 March, 2006, atthree Italian stations,within different spectral intervals. It wasfound that the times when the measured irradiances reachedtheir minimawere appreciably different from those predictedby the radiative transfer model. The discrepancies wereattributed to the contributions arising from the diffuse solarradiation produced by scattering processes mainly due toenvironmental factors, such as clouds and surface reflectance.The analysis of the results highlights that the local globalirradiance measurements were impacted by areas within a30–100 km range, varying from one station to another. Theasymmetrical distribution of the surface reflectance charac-teristics around an observation site can create an asymmetryin the horizontal photon fluxes (Emde and Mayer, 2007)coming from different sectors of the surrounding environ-ment. Thepresent results describing the evolutionary patternsof the eclipse episode indicate that such an occurrence cancause appreciable variations in the surface irradiance time-

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006

10 B. Petkov et al. / Atmospheric Research xxx (2010) xxx–xxx

ARTICLE IN PRESS

patterns observed during an eclipse episode, as compared tothose defined on the basis of astronomical parameters only.

Acknowledgments

This research was supported by the Programma Nazionaledi Ricerche in Antartide (PNRA) and developed as a part ofSub-project 2004/2.4 “Accurate surface-based radiationmeasurements: implementation of a BSRN station at DomeConcordia”. Contributions by S. Piacentino (ENEA, ACS-CLIM-OSS, Santa Maria di Galeria, Italy), T. Di Iorio (Dept. of Physics,University “La Sapienza”, Rome, Italy), and R. Werner (STIL,Stara Zagora, Bulgaria) are also gratefully acknowledged.

References

Abbott, W.N., 1958. On certain radiometric effects during the partial eclipseof February 25, 1952. Geofis. Pura Appl. (Milan) 39, 186–193.

Anderson, R.C., Keefer, D.R., Myers, O.E., 1972. Atmospheric pressure andtemperature changes during the 7March 1970 solar eclipse. J. Atmos. Sci.29, 583–587.

Ångström, A., 1964. The parameters of atmospheric turbidity. Tellus 16, 64–75.ARM/BSRN web site, see http://www.arm.gov/instruments/instrument.php?

id=bsrn.Blumthaler, M., Bais, A., Webb, A., Kazadzis, S., Kift, R., Kouremeti, N.,

Schallhart, B., Kazantzidis, A., 2006. Variations of solar radiation at theEarth's surface during the total solar eclipse of 29 March 2006. Proc. ofthe SPIE Conference “Remote Sensing 2006”, 6362-14, Session 3, p. 37.SPIE, Stockholm, Sweden, 11–14 September. doi:10.1117/12.689630.

Carapiperis, L.N., Karapiperis, Ph.P., 1959. On the variation of solar radiationand blueness of the sky during the eclipse of June 30, 1954 at Athens.Pure Appl. Geophys. 43, 323–328.

Dani, K.K., Devara, P.C.S., 2002. Aerosol optical depth and ozone variationsduring the total solar eclipse of 24 October 1995. Atmos. Res. 65, 1–15.

Degünther, M., Meerkötter, R., 2000. Effect of remote clouds on surface UVirradiance. Ann. Geophys. 18, 679–686.

Degünther, M., Meerkötter, R., Albold, A., Seckmeyer, G., 1998. Case study onthe influence of inhomogeneous surface albedo on UV irradiance.Geophys. Res. Lett. 25, 3587–3590.

Di Iorio, T., di Sarra, A., Junkermann, W., Cacciani, M., Fiocco, G., Fuà, D., 2003.Tropospheric aerosols in theMediterranean: 1. Microphysical and opticalproperties. J. Geophys. Res. 108 (D10), 4316. doi:10.1029/2002JD002815.

Dundee Satellite Receiving Station web site, http://www.sat.dundee.ac.uk/.Emde, C., Mayer, B., 2007. Simulation of solar radiation during a total eclipse:

a challenge for radiative transfer. Atmos. Chem. Phys. 7, 2259–2270.Espenak, F., Anderson, J., 2004. Total Solar Eclipse of 2006March 29, NASA/TP –

2004 – 212762. see http://sunearth.gsfc.nasa.gov/eclipse/eclipse.html.Harrison, L., Michalsky, J., Berndt, J., 1994. Automated multifilter rotating

shadow-band radiometer: an instrument for optical depth and radiationmeasurements. Appl. Optics 33, 5118–5125.

Horvath, H., 1971. On the applicability of the Koschmieder formula. Atmos.Environ. 5, 177–184.

Huber, M., Blumthaler, M., Schreder, J., Schallhart, B., Lenoble, J., 2004. Effectof inhomogeneous surface albedo on diffuse UV sky radiance at a high-altitude site. J. Geophys. Res. 109, D08107. doi:10.1029/2003JD004013.

Please cite this article as: Petkov, B., et al., Ground-based obsereclipse of 29 March, 2006: Signs of the environmen..., Atmos. ReAll in-text references underlined in blue are linked to publications o

Iqbal, M., 1983. An introduction to solar radiation. Academic Press Canada,Don Mills, Ontario, pp. 107–168.

Koepke, P., Reuder, J., Schween, J., 2001. Spectral variation of the solarradiation during an eclipse. Meteorol. Zeitschrift 10, 179–186.

Kylling, A., Mayer, B., 2001. Ultraviolet radiation in partly snow coveredterrain: observations and three-dimensional simulations. Geophys. Res.Lett. 28, 3665–3668.

Kylling, A., Webb, A.R., Kift, R., Gobbi, G.P., Ammannato, L., Barnaba, F., Bais,A., Kazadzis, S., Wendisch, M., Jäkel, E., Schmidt, S., Kniffka, A., Thiel, S.,Junkermann, W., Blumthaler, M., Silbernagl, R., Schallhart, B., Schmitt, R.,Kjeldstad, B., Thorseth, T.M., Scheirer, R., Mayer, B., 2005. Spectral actinicflux in the lower troposphere: measurement and 1-D simulations forcloudless, broken cloud and overcast situations. Atmos. Chem. Phys. 5(1975–1997).

Lenoble, J., 2000. Influence of the environment reflectance on the ultravioletzenith radiance for cloudless sky. Appl. Opt. 39, 4247–4254.

Madronich, S., 1993. UV radiation in the natural and perturbed atmosphere.In: Tevini, M. (Ed.), Environmental Effects of UV (Ultraviolet) Radiation.Lewis, Boca Raton, pp. 17–69.

Mayer, B., Degünther, M., 2000. Comment on “Measurements of erythemalirradiance near Davis Station Antarctica: effect of inhomogeneoussurface albedo”. Geophys. Res. Lett. 27, 3489–3490.

McKenzie, R., Paulin, K., Madronich, S., 1998. Effects of snow cover on UVirradiance and surface albedo: a case study. J. Geophys. Res. 103,28785–28792.

Meloni, D., di Sarra, A., Pace, G., Monteleone, F., 2006. Optical properties ofaerosols over the central Mediterranean. 2. Determination of singlescattering albedo at two wavelengths for different aerosol types. Atmos.Chem. Phys. 6, 715–727.

Miyake, Y., Sekihara, K., Kawamura, K., 1949. Results of the observation ofintensity and distribution of ultraviolet sky radiation at Rebun Island.Geophys. Mag. 19, 72–77.

Pace, G., di Sarra, A., Meloni, D., Piacentino, S., Chamard, P., 2006. Aerosoloptical properties at Lampedusa (Central Mediterranean). 1. Influence oftransport and identification of different aerosol types. Atmos. Chem.Phys. 6, 697–713.

Penndorf, R., 1954. The vertical distribution of Mie particles in thetroposphere. J. Meteorol. 11, 245–247.

Petkov, B., Vitale, V., Tomasi, C., Bonafè, U., Scaglione, S., Flori, D., Santaguida,R., Gausa, M., Hansen, G., Colombo, T., 2006. Narrow-band filterradiometer for ground-basedmeasurements of global UV solar irradianceand total ozone. Appl. Optics 45, 4383–4395.

Pruitt, W.O., Lourence, F., Crawford, T.V., 1965. Radiation and energy balancechanges during the eclipse of 20 July 1963. J. Appl. Meteor. 4, 272–278.

Smolskaia, I., Nuncz, M., Michael, K., 1999. Measurements of erythemalirradiance near Davis station, Antarctica: effect of inhomogeneoussurface albedo. Geophys. Res. Lett. 26, 1381–1384.

Spiegel, M.R., 1972. In: Boxer, R.W. (Ed.), Schaums's Outline of Theory andProblems of Statistics in SI Units. McGraw-Hill Internat. Book Co, NewYork, pp. 250–252. SI Edition.

Takesako, S., 2001, http://www.kotenmon.com/index_e.htm.Tomasi, C., 1982. Features of the scale height for particulate extinction in

hazy atmospheres. J. Appl. Meteor. 21, 931–944.Zerefos, C., Balis, D.S., Meleti, C., Bais, A.F., Tourpali, K., Kourtidis, K., Vanicek,

K., Capelani, F., Kaminski, U., Colombo, T., Stubi, R., Manea, L., Formenti,P., Andreae, M.O., 2000. Changes in surface solar UV irradiances and totalozone during the solar eclipse of August 11, 1999. J. Geophys. Res. 105,26463–26473.

vations of solar radiation at three Italian sites, during thes. (2010), doi:10.1016/j.atmosres.2009.12.006n ResearchGate, letting you access and read them immediately.