Absolute spectral measurements of direct solar ultraviolet irradiance with a Brewer spectrophotometer
Post on 30-Sep-2016
needed for different purposes. Measurements ofglobal UV irradiance are the most commonly used,
stations. The first reason for this delay was theneed for a pointing system to direct the beam light ofand they are useful in studying the effects of ozone,clouds, and aerosols on solar UV radiation and itschanges with time and for the formation of suffi-ciently long data records that could be used in thefuture to determine long-term UV trends.25 To alesser extent these measurements can also be usedfor biological studies. Actinic flux measurementsare suitable for photochemical as well as for biologicalstudies.6,7 Measurements of the diffuse componentcan be used, together with global and radiance mea-surements, to study the effects of atmospheric aero-sols. Finally direct-Sun measurements are useful todetermine the aerosol optical depth as well as thecolumn abundance of atmospheric species that ab-
the Sun into the spectrometer, which would also beadaptable to existing instruments. This problemwas finally solved by the wide spread use of fibers inUV spectroradiometry. Another reason for the de-lay was the difficulty to establish a reliable method-ology for the absolute calibration of direct solarirradiance. Despite the great improvements madein recent years, some of the uncertainties in themethodology still have not been eliminated.
The Brewer spectrophotometer is among the in-struments capable mechanically, from the firstmoment, of conducting direct-Sun spectral measure-ments. I describe the methodology used for calibrat-ing such measurements taken with the Brewerspectrophotometer MKIII of the Laboratory of Atmo-spheric Physics, University of Thessaloniki. The ab-solutely calibrated direct-Sun spectra were then usedto determine the extraterrestrial solar spectrum fromground-based measurements made at a high-altitudesite through application of the Langley extrapolationmethod. Since it is not within the objectives of thispaper to discuss in detail the accuracy of the Brewer-
The author is with the Department of Physics, Laboratory ofAtmospheric Physics, Aristotle University of Thessaloniki, Cam-pus Box 149, Thessaloniki 54006, Greece.
Received 19 December 1995; revised manuscript received 7 No-vember 1996.
0003-6935y97y215199-06$10.00y0Absolute spectral measuremeultraviolet irradiance with a B
Alkiviadis F. Bais
A methodology for the absolute calibperformed with a Brewer spectrophoglobal and diffuse solar irradiancedirect-Sun component. On the basiof solar zenith angles at a high altitby applying the Langley extrapolatioderived from the airborne tunable lagreement of better than 63%.
Key words: Solar ultraviolet, dir
Measurements of the solar ultraviolet radiationreaching the Earths surface became important dur-ing the last few years, as both observations and pre-dictions suggest the tendency to an increase of UVlevels, following observed ozone decreases.13 Fourdifferent types of solar UV measurement are cur-rently available at various monitoring stations, each 1997 Optical Society of Americants of direct solarrewer spectrophotometer
ration of spectral measurements of direct solar ultraviolet radiation,tometer is presented. The method uses absolute measurements ofobtained practically simultaneously at each wavelength with thes of this calibration, direct-Sun spectra, measured over a wide rangeude site, were used to determine the extraterrestrial solar spectrumn method. Finally this spectrum is compared with a solar spectrumaser absorption spectrometer 3 Space Shuttle mission, showing an1997 Optical Society of Americaect-Sun spectra, extraterrestrial spectrum.
sorb solar radiation in the ultraviolet.8,9 Modelingstudies can benefit greatly from measurements of thedirect-Sun component, since its parameterization ismuch easier than that of global or diffuse compo-nents. Despite their usefulness, spectral measure-ments of the direct component of solar ultravioletradiation have been included only recently in theobservational programs of various UV monitoringderived solar spectra, only a preliminary comparison
20 July 1997 y Vol. 36, No. 21 y APPLIED OPTICS 5199
with an existing extraterrestrial solar spectrum ispresented. Of all such spectra reported in the
ance measurement is within 64%. This estimatedoes not include the systematic error induced by theliterature, one of the most recent spectra wasselected, which was obtained by the airborne tunablelaser absorption spectrometer ~ATLAS 3! mission ofthe Space Shuttle.
2. Instrumentation and Measurements
The Brewer spectrophotometer was originally de-signed to perform total ozone measurements in theUV region by using either direct sunlight or diffuseskylight from the zenith.13 With the addition of anUV diffuser the Brewer also became capable of takingspectral measurements of the global solar irradiance.These scans cover the spectral region 290325 nm forconventional single-monochromator instruments~types MKII and MKIV! and 285365 nm for adouble-monochromator ~MKIII! Brewer instrument.Since the instrument was designed to include a Sun-tracking system for total ozone measurements, at-tempts have also been made for the last several yearsto obtain spectra of the direct component of solarultraviolet radiation.
To obtain direct-Sun spectra the pointing system ofthe Brewer instrument is directed toward the Sunsdisk, which is viewed under an angle of approxi-mately 1.5, determined by the aperture of theBrewer input optics. Of course the viewing angle islarger than the Suns angle ~'0.5! and therefore theBrewer instrument senses a fraction of the skylight,which could introduce a small error in the measure-ments. The direct sunlight is then collimated anddirected onto a ground quartz diffuser plate beforeentering the spectrometer. From then on the proce-dure is similar to that followed for gathering globalUV irradiance spectra.1,14,15 The quality of a mea-sured direct-Sun spectrum depends strongly on theeffectiveness of the instruments pointing system intracking the Sun continuously. Failure to achieveaccurate tracking can result in measuring only partof the direct components of solar radiation. Anotherproblem that might be encountered during such mea-surements is the increased signal of the direct beamwhen one samples the long UV wavelengths. Thiscan happen particularly at sites with aerosol-free airand low ozone content as well as at high solar eleva-tions. Under such conditions the signal can becomelarger than the saturation limit of the detector, re-quiring the use of attenuation filters. Although theneutral density filters of the Brewer instrument areconsidered spectrally flat, it is necessary to measuretheir spectral response or at least to take their pres-ence into account during the calibration procedure.
For this study I used a double-monochromatorBrewer MKIII, which has operated at the Laboratoryof Atmospheric Physics since 1993, producing bothglobal and direct-Sun UV spectral measurementsregularly. The absolute calibration of global UVspectra is maintained through comparisons with1000-W reference sources traceable to National Insti-tute of Standards and Technology standards. Theoverall uncertainty associated with the global irradi-
5200 APPLIED OPTICS y Vol. 36, No. 21 y 20 July 1997cosine response of the input optics, which by labora-tory measurements was found to be ;25% for a 60zenith angle, increasing to 230% at 75. These er-rors of course apply only to the direct component thatbecomes less significant at higher solar zenith angles.An internal mercury discharge lamp is used for thewavelength calibration of the instrument, which onecan do by aligning the spectrometer with the302.1-nm line with a precision of 0.007 nm.16 Thiscontrol test is always done before each scan to ensureproper wavelength calibration.
3. Calibration of Direct-Sun Spectra
Calibration of direct-Sun spectra is not as easy ascalibration of global spectra. The former requires apoint source of known characteristics, and a cali-brated lamp at a 50-cm distance could by no means beconsidered as a point source. This could be doneonly if the lamp were moved far from the instrument,e.g., at a distance of 5 m. Although the absoluteirradiance of the lamp at such a distance could becalculated by the inverse square law, a 100-fold de-crease of its signal would occur, which consequentlywould increase appreciably the uncertainty of thecalibration. The sensitivity of the specific instru-ment is such that, when one uses a 1000-W quartz-halogen reference lamp, the uncertainty in themeasured irradiances, induced only by the photon-counting statistics,17 ranges from ;1.5% at 290 nm to;0.5% at 360 nm. If the signal were reduced 100times, these uncertainties would increase to 15% and5%, respectively.
Indirectly, calibration of direct-Sun spectra couldbe done by using simultaneous absolute spectralmeasurements of global and diffuse solar irradiance.The difference between the two measurements D~l!,corrected by the cosine of the solar zenith angle z,gives a good estimate of the direct component of solarirradiance. Then the uncalibrated direct-Sun irra-diance B~l!, also recorded simultaneously, can becompared with that derived indirectly from the globaland diffuse irradiances, thus determining the re-quired calibration factor F~l! as
Although this method seems quite simple, there areseveral restrictions that require considerable atten-tion.
~1! If the method were applied at moderate solarzenith angles, larger than ;30, the cosine error ofthe instruments diffuser could be inherited by thecalibration factor. Therefore one must either correctfor the cosine error or, preferably, perform such mea-surements when the Sun is close to zenith ~i.e., at lowsolar zenith angles!.
~2! The method for measuring the diffuse compo-nent is to use a shading disk to exclude the direct-Sun
component from the global measurements. The sizeof the disk in relation to its distance from the diffuser
can affect significantly the accuracy of the diffuse-component measurements. A larger than necessarydisk would increase the fraction of the skylight that isobscured, introducing a systematic error in the cali-bration factor. This becomes more important whenthe aerosol optical depth is high, which modifies sig-nificantly the diffuse component. The optimum con-figuration would be such that the solid angledetermined by the size of the disk is equivalent to theviewing angle of the direct-Sun port of the instru-ment.
~3! It was mentioned above that the three types ofmeasurement ~global, diffuse, and direct! must besimultaneous. If one considers using sequentiallyobtained spectra, then the time required to obtain thethree spectra is at least 10 min for the single-monochromator and 20 min for the double-monochromator Brewer instrument.
~4! The shading disk must cover the diffuser platecontinuously in order to obtain consistent measure-ments. However, during a single scan both the ze-nith and the azimuthal angles of the Sun changesignificantly. To compensate for these changes theposition of the disk should be updated regularly tomaintain its relative position to the diffuser.
4. Results and Discussion
Here I describe the methodology followed for the cal-ibration of direct-Sun spectra obtained by the BrewerMKIII of the Laboratory of Atmospheric Physics.The measurements were carried out at the Izana Ob-servatory, Tenerife, during a period of one week inJuly 1995. Izana is a tropical site ~28 N! located atan altitude of 2367 m. Owing to the high altitude,clouds are constantly below the horizon of the site atthis time of the year, except when high-altitudeclouds are present. In addition, the aerosol-free at-mosphere ~e.g., optical depth at 320 nm usually below0.05! makes the site suitable for actinometric mea-surements and calibrations. The methodology fol-lowed to calibrate the Brewer direct-Sun spectralmeasurements is the one described previously thatutilizes global, diffuse, and direct-Sun spectra. Toobtain the diffuse spectrum a disk of 8 cm diameterwas placed 1 m away from the diffuser, toward theSun, supported by a relatively thin, black metal rod.The rod was mounted on the instrument so that itcould rotate about a horizontal axis, located at theplane of the diffuser plate and perpendicular to theSuns meridian. In this way the azimuthal angle ofthe disk was always the same as that of the Sun,since the disks mount was rotated together with theBrewer. Its zenith angle could be set by rotatingmanually the metal rod about its horizontal axis.The experimental setup is shown in Fig. 1.
To fulfill the fourth requirement described in Sec-tion 3 a special routine was made for the Brewer,allowing the instrument to record in sequence and,within only ;30 s, the global, diffuse, and direct ir-radiances for each wavelength. This sequence wasrepeated for the spectral range 290360 nm, in stepsof 2 nm. The calibration factor is expected to be asmooth function of wavelength, and therefore the2-nm step is sufficient to define this function, reduc-ing at the same time the duration of the measure-ments. The alternation between the diffuse andglobal measurements was made by offsetting auto-matically the azimuth of the Brewer by a few degreesso that the shadow of the disk was moving away fromthe diffuser. In this way the obstructions introducedto the radiation field by the metal rod and its mount-ing were the same for both the global and diffusemeasurements, assuming of course that the diffuseskylight within a few degrees of azimuth does notchange significantly. Finally the direct componentwas measured by pointing the Brewer to the Sun andusing the internal diffuser plate as described previ-ously. The signal was checked continuously, andthe appropriate neutral density filter was inserted inthe light path, if the signal exceeded a predefinedcount level. Occasionally and while the Brewer wasperforming a direct-Sun measurement, the disk wasadjusted manually to account for the small changesin the solar zenith angle during the period of mea-surements.
Several spectra of this type were obtained duringthe campaign, covering different solar angles. Notethat the cosine response of the Brewer input opticscan modify significantly the results as the solar ze-nith angle increases. Therefore the calculated di-rect component D~l! was corrected for the cosineerror according to the zenith angle. As mentionedpreviously, the cosine error of the instrument isknown from laboratory measurements.15 Finallythe calibration factor F~l! was determined separatelyfor each wavelength by using Eq. ~1!.
Fig. 1. Experimental setup used for the calibration of the Brewerdirect-Sun spectra.
20 July 1997 y Vol. 36, No. 21 y APPLIED OPTICS 5201
Figure 2 shows the calibration factor as determinedfrom measurements taken at various solar zenith an-gles. It is evident from this figure that measure-ments corrected for the cosine response give similarresults, and therefore the calibrati...