solar spectral irradiance and atmospheric transmission at mauna loa observatory

Solar spectral irradiance and atmospheric transmission atMauna Loa Observatory

Glenn E. Shaw

A radiometer was operated at the Mauna Loa Observatory during calendar year 1980 to estimate the spectralirradiance of the sun and its possible fluctuation in time near the peak of solar activity. Data were also ac-quired on seasonal trends of atmospheric transmissivity above the marine mixing layer in the central Pacific.Spectral irradiance remained constant to at least 1/2% at all wavelengths monitored. Furthermore its abso-lute magnitude was in agreement with the Labs and Neckel values to +2% except at blue wavelengths wherethe Mauna Loa values are from 4 to 12% higher and at X = 850 nm where the Mauna Loa value is 9% lower.The residual aerosol optical depth above Mauna Loa Observatory during 1980 averaged ro = 0.020. An in-trusion of dust into the central Pacific from the Gobi Desert (as deduced by the composition of collected par-ticles) invaded the Central Pacific from Mar. to May 1980 and caused a perturbation in optical depth (at X= 500 nm) of ATo - 0.01-0.02. The optical depth increment caused by the Mt. St. Helens volcano was<0.005 in the 2-month period following the eruption.

I. IntroductionThe total radiant output of the sun undergoes slow

and systematic evolutionary alterations over billions ofyears as the solar core alters in composition. Thesechanges are so slight in the main sequence phase thatin the last three billion years there is no evidence thatthe climate of earth has ranged below the ice point orabove the boiling point of water; in other words, thesolar luminosity has remained constant to at least 10 or15% during this time.' On the other hand, there isample evidence indicating that the planet has beenexposed to certain relatively mild climatic changes overthe last 600 million years, which in principle could bedriven or triggered by a fluctuating solar output onrelatively brief time scales.

Solar fluctuations-if they indeed exist-apparentlyhave so far been at or below the limit of detectabilityexcept during a recent measurement conducted fromspace 2 that showed the total solar output to be variableat the 0.1% level for several days at a time. In theweight of all experimental evidence, it seems possibleto conclude that the sun's energy output, the solar

The author is with Geophysical Institute, University of Alaska,Fairbanks, Alaska 99701.

Received 5 October 1981.0003-6935/82/112006-06$01.00/0.© 1982 Optical Society of America.

constant, has been sensibly constant to the extent thatit has varied by no more and probably by much less than1% over several decades.

Although the solar constant seems remarkably con-stant over times of decades, it by no means follows thatthe distribution of solar radiant energy by wavelengthis unalterable. Energy in the extreme short and longwavelength tails of the solar spectrum undergoes largevariation, but it has never been adequately resolvedweather and to what extent spectral variations extend(at reduced levels) into the near-UV-visible-near-IRregion of the spectrum. In principle it is straightfor-ward to test the constancy of the spectral distributionof sunlight by direct experimentation; in practice suchradiometric determinations are notoriously difficult tomake: "the highly accurate seven decimal figures ofmodern science appear elsewhere."3 The situation isso poor that the distribution of energy in the visibleregion of the X > 500-nm solar spectrum is known onlyto -10% accuracy, and at near-UV wavelengths differ-ences of up to 20% are found in standard tabulations.

This paper describes a time series of spectral solarirradiance measurements made during the calendar year1980 at the Mauna Loa Observatory (elevation, 3.5 km)around the time of maximum solar activity. An abso-lute accuracy (SI units) of -2% was achieved for thesemeasurements, but the relative accuracy is about anorder of magnitude higher. Information on the opticaltransmissivity of the atmosphere was of necessity ac-quired during the course of the experiments, and, sincethey are of value and of interest in themselves, they tooare discussed.

2006 APPLIED OPTICS / Vol. 21, No. 11 / 1 June 1982

II. Experimental Technique

A. Radiometer

The radiometer (Fig. 1) is a sequentially driven filterwheel instrument with ten dielectric thin-film inter-ference filters designed to have out-of-band leakage ofless than one part in 107. An EGG model UV-444 sili-con diode served as a stable and convenient radiationdetector, and, to reduce exterraneous effects, the opticalpackage was maintained at constant temperature andhumidity. Active tracking automatically corrected for

atmospheric refraction and pointed the instrument towithin 20 sec of arc; light from the entire disk wasmeasured.

B. Filter Bandpass CharacteristicsThe filter bandpasses (shown in Fig. 2 and listed in

columns 1 and 2 in Table I) were placed in regions freeof terrestrial atmospheric absorption lines, except forunavoidable broad Chappuis and NO2 bands whichobey Beer's law and which, therefore, can be eliminatedfrom consideration by applying the Langley method ofatmospheric extrapolation. In addition to being inregions free of terrestrial absorption, the passbands werespecifically chosen to lie in relatively smooth regions ofthe solar spectrum so that any slight drifts in the centralfilter passband wavelength would cause the smallestpossible changes in radiometer signal. For choosingappropriate regions of the optical spectrum, high-res-olution solar atlases (e.g., Ref. 4) were smoothed ac-cording to the spectral responses shown in Fig. 2.

C. Extrapolation Through the AtmosphereThe Langley method was used to extrapolate the

surface-based readings through the scattering atmo-sphere. The basic idea behind the extrapolation pro-cess is the known adherence of the atmosphere to theBouguer law:

T(X) = exp -

Fig. 1. Sunray I filter-wheel solar radiometer at Mauna Loa.

(1)

where m (6) is the relative air mass (in terms of thevertical) at solar elevation angle , and ri (X) are theadditive components of optical depth or optical thick-ness consisting of terms from molecular (Rayleigh)scattering, gaseous absorption in weak lines (Chappuis),and absorption and scattering of light by suspendedparticles. A plot of the logarithm of apparent solarintensity against air mass m is, by Eq. (1), linear, in-tercepting the ordinate at the extraterrestrial value of

U.I I 460.0

0.5 433565.7

0.4-415.6

0.3 6.7

02 383.5 6.9 7.86.8

0.1 7.6-

0.0*370 380 390 410 420 450 460 470 480 490 500 550 560 570 580

X (nm)

Fig. 2. Transmission character-istics of filters used in the Sunray

I radiometer.

1 June 1982 / Vol. 21, No. 11 / APPLIED OPTICS 2007

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Table 1. Spectral Solar Irradiance and Estimated Uncertainties

Root mean square Root mean squarevariation" error Ofb Solar

Central of irradiance atmospheric irradiancec Solarwavelength Bandwidth during 1980 extrapolation (lamp calib.) irradiance Solar irradiance 6

(nm) (nm) (%) (%) (gW cm- 2 nm'1) (diode calib.) (,uW cm- 2 nm'1) F/FLN

383.5 7.6 1.0 1.2 97.3 + 3.4 - 87.03 -415.6 6.8 0.5 0.4 177.3 + 4.4 169.1 + 6.2 171.1 1.04460.0 6.7 0.5 0.3 212.6 + 5.3 208.3 + 6.2 198.7 1.07493.3 6.9 0.4 0.3 197.4 + 4.5 190.5 + 4.1 194.8 1.01565.7 7.8 0.5 0.3 182.4 + 4.2 188.2 + 4.0 181.8 1.00616.0 7.6 0.5 0.3 165.4 + 3.6 167.1 + 3.5 167.6 0.99674.2 6.4 0.4 0.3 148.7 + 3.0 149.1 + 3.0 151.5 0.99789.9 6.4 0.4 0.3 113.2 + 2.3 - 116.3 0.97850.4 6.6 0.6 0.8 93.2 4 2.1 - 102.0 0.91

1009.6 7.8 0.5 0.4 73.6 ± 1.8 - 74.0 0.99

a Relative standard deviation (in %) of observed solar irradiance values for 132 days of observation.b Estimated error in percent of the atmospheric extrapolation process.c Uncertainty estimate takes into account uncertainty in standard lamp irradiance, uncertainties introduced by the atmospheric extrapolation

process, and systematic errors involved with transferring standard to instrument.

the sun's spectral irradiance and with a slope corre-sponding to Zri, the total vertical optical extinction ofthe interfering atmosphere. The aerosol optical depthTa, which is a component of interest because it is a cli-matic variable, is recoverable by subtracting knownRayleigh scattering. Langley plots obtained at MaunaLoa were of exquisite quality, the rms residuals of theleast-squares fit being 0.1% or lower.

D. Accuracy of Atmospheric ExtrapolationsThe main systematic error affecting the extrapolation

process is temporal drift in atmospheric extinction overthe day that one is accumulating extrapolation infor-mation. The situation is helped by performing themeasurements from a high-altitude station in the cen-tral Pacific Ocean. The morning hours, in particular,are stable (optically) at Mauna Loa, but around noonthe mixing layer thickens, and upslope winds and con-vection currents carry trace amounts of aerosol materialand water vapor up to the observatory. For this reason,the aerosol count at the observatory was monitoredcontinually; whenever it showed an increase beyond 500particles cm-3 the instrument was turned off.

The accuracy of the extrapolation process was judgedby analyzing data acquired between two different airmass limits. On a given day, if the extinctionthroughout the atmosphere were constant, the slope andintercept values of the Langley plot would be identical,conversely any differences would indicate time-changing optical transmissivity of the atmospherecaused (most probably) by changes in the overlyingaerosol. Thus, for example, the annual mean solar ir-radiance acquired at a 493.3-nm wavelength for an airmass range of 1 < m < 3.5 agreed with that calculatedover an air mass range of 3.5 < m < 7 to 0.4%. Theuncertainties so introduced are in Table I, column 4from which it can be seen that they were very small.

E. Radiometric Scales

The radiometer was calibrated two ways: (1) byreferencing it to an absolute electrical cavity radiometerusing the technique of Geist et al.,5 and (2) by refer-

encing it to a 1000-W quartz-iodide standard lamp(NBS lamp F-56), which had been previously referencedagainst a gold point blackbody source. These methodstie the scale to the electrical and thermodynamic systemof SI units.

The electric method of calibration was performed atthe World Radiation Center, Davos, Switzerland, inSept. 1977. Monochromatic light from dye and argonlasers were used to calibrate a transfer standard silicondiode (United Detector Technology model PIN-10 DP),the reference standard being the WRC electrical cavityradiometer. Since the intercalibrations were performedtwo years before the Mauna Loa experiment, it is pos-sible that the substandard might have drifted. Al-though the drift rate is unknown, periodic intercali-brations with four other substands over two years pro-vided relative sensitivity ratios constant to +0.2%. (Itis of course conceivable, but unlikely, that all the diodesmay have drifted the same way.) The excellent agree-ment between independent detectors attests to theexcellent long-term stability of the PIN-doped siliconjunction photodiode.

Ill. Observed Solar Spectral IrradianceThe Mauna Loa spectral irradiance experiment gave

the results summarized in Table I. Column 3 is therelative standard deviation (in percent) for all measuredsolar irradiance values. The fourth column (Table I)lists the estimated uncertainty of the atmospheric ex-trapolation process. It is interesting to note that theelements of column 3 and 4 are quite comparable inmagnitude suggesting that the greatest part of thevariability is due to uncertainties in the atmosphericextrapolation.

Columns 5 and 6 (Table I) are tabulated values ofannual mean solar spectral irradiance in (SI units) asdetermined from the NBS standard lamp (column 5)and from the diode intercomparison traceable to theWorld Radiation Center's active cavity radiometer(column 6). The error limits incorporate uncertaintiesin the calibration transfer plus the estimated uncer-tainty with which the reference Standard (NBS lamp


for column 5 and active cavity radiometer for column6) is known on the absolute scale of SI electrical units.In all cases, when two calibrations are available, theerror bars for the two determinations overlap.

Column 7 is a tabulation of the Labs and Neckel6 7

values of solar irradiance shown for comparison pur-poses; the last column is the ratio of the NBS lamp-derived Mauna Loa determinations of spectral irra-diance to the Labs and Neckel values.

The time series of the daily acquired values of ex-traterrestrial spectral irradiance determinations (re-ferred to 1 a.u.) are shown for all wavelengths in Fig. 3

L 616

4k5 L 674

460

493

566

A 790

I 1009

Fig. 4. Frequency distribution of extrapolated values of solarspectral irradiance for the Jan. 1980-Jan. 1981 period.

415.6

460.0e

493. 3

565.7

616.0

-4V~~~-H4-±A-4-~~~ 4 4,, t.

. r 674.2

7S9.9

- 1--1" K-u- tt--Xi

4513.4

109 .6

.I I I I I I I I I I I I IJ IF ' M A ' M ' J | J A'

4980S 0'N DIJ

and the corresponding frequency distribution in Fig. 4(see also column 3, Table I). A comparison of estimateduncertainties (column 4, Table I) with observed variance(column 3, Table I) suggests that the majority of thevariance is caused by temporal changes in the opticaldepth of the atmosphere during the acquisition of data.It appears that the sun's spectral irradiance at allwavelengths monitored varied by no more than a fewtenths of a percent during 1980.

The zero time delay correlation between the timeseries of total solar irradiance measured by satellite2 andthe time series of spectral irradiance monitored from thesurface was not significant. Given the accuracy of theatmospheric extrapolation (e.g., -0.5%) and the size ofthe reported variations in the solar constant by Willsonet al. (0.1%), it is not particularly surprising that nocorrelation is found.

On the other hand, solar greenhouse effects8 oper-ating in the photosphere could alter the spectral dis-tribution of radiant energy, while not necessarily re-sulting in any substantial change in the wavelength-integrated energy (the solar constant). The most sig-nificant variability would be expected to occur in regionsof large line blanketing.9 Such selective spectral am-plification mechanisms, however, do not show up in theMauna Loa data to any detectable extent at the tenwavelengths monitored. The results can be summa-rized by stating that visible band spectral irradiancechanges in the sun during 1980 were less than a half ofa percent at all wavelengths monitored, while simulta-neously the solar constant varied by 0.1%.2

IV. Atmospheric Transmissivity Above Mauna LoaDuring 1980

The wavelength dependence of aerosol extinction(i.e., the wavelength dependence of aerosol opticaldepth) is relevant to terrestrial atmospheric aerosolphysics and theories of climatic change.10 '1'

At Mauna Loa the aerosol extinction spectrum variedin magnitude and shape on a day-to-day basis and alsosystematically over the year indicating that the aerosolabove the mixing layer of the Central Pacific Ocean isnot homogeneous even on short time scales. The time


Fig. 3. Deviation of daily extrapolated values of spectral solar irra-diance about the mean at ten wavelengths. Tic marks are 1%

intervals.

383

383.6Y-. , 11 � r I , � _,, t, _ � I ��, � -1

series of day-averaged aerosol extinction (shown in Fig.5) indicates the following:

There was little if any evidence of the Mt. St. Helensstratospheric dust cloud from the May 1980 eruptionat Mauna Loa. The cloud's optical thickness pertur-bation was at most = 0.005 at X = 493.5 nm in thetwo-month period following the eruption.

An episode of what is tentatively identified as GobiDesert sand affected the central Pacific Ocean duringMar., Apr., and May 1980 (see Fig. 5). Similar Gobidust intrusions were observed from MLO in the spring

383

445

460

493

566

I I I I 0 T To .05

3

674

790

850

a 4009

I_ _ I I0 .. 05

'A+ LO

3133.5

., I III II I . I I I.. / _ _ _ 1_I I I I I I I

415.6

46.

493.3

1 1-+-I-- --- - Ace-+ I I I I -I565.7

616.0

. 674.2

. I + I e- I I - + - I I I -

789.9

.~~ -A-2 ual j+w i~ D T 'f m

8513.4

1009.6

.J F }M |A; M J tJ A S 0 N 0 J4980

Fig. 5. Time series of aerosol optical depth at different wavelengthsabove Mauna Loa. Gobi dust was detected at the observatory in

Mar.-Apr. 1980. Distance between tic marks, AT = 0.01.

I Fig. 6. Frequency distribution of aerosol-plus-ozone optical depthsfor calendar year 1980 above Mauna Loa.

of 1976 and 1979.12 The observed anomaly can be saidto be almost certainly due to imported crustal materialat an altitude above the observatory since studies of thecomposition of particles collected at Mauna Loa andanalyzed by x-ray spectrometry (particles collected byDittenhoefer)13 were rich in silicon and aluminum.

The frequency distribution of the aerosol extinctionduring 1980 at Mauna Loa is shown in Fig. 6 for each ofthe ten wavelengths; aerosol extinction was smaller andless variable in the red than blue. However, the relativevariability T5/T was about constant at all wave-lengths.

V. ConclusionsThe following conclusions can be stated on the basis

of the 1980 experiments at Mauna Loa:(1) Solar spectral irradiance (at ten-wavelength

bands) in the visible-near-IR region varied by <0.5%during the Feb. 1980-Feb. 1981 period, and most (80%)of the variance was apparent rather than real, occurringfrom uncertainties in the atmospheric extrapolationprocess, even though Mauna Loa is certainly one of thecleanest observatories on earth.

(2) The absolute values (on the thermodynamic scaleof SI units) of solar spectral irradiance were found to bein agreement to within 2% of the Labs and Neckelstabulations at all wavelengths except X = 415.6 nm and460.0, where the Mauna Loa values were 4 and 7%higher, and at X = 850.5 mm where the Mauna Loavalues were 9% lower.

(3) The background atmospheric optical deptharising from absorption and scattering from aerosolabove Mauna Loa averaged To = 0.020 0.005 at X =500 nm, but an enhancement in the extinction occurredfrom March to May, attributable to desert soil particlestransported across the Pacific Ocean from the region ofthe Gobi desert. The stratospheric plume from the Mt.St. Helens volcano was not apparent at Mauna Loa; itresulted in an increased stratospheric extinction of atmost T = 0.005 in the two-month period following theeruption.


616

I I I I I I I I I I r I

The knowledge of something as elementary andfundamental as the spectral distribution of light emittedby the sun has still not been achieved to any reasonableaccuracy, although the problem is not nearly so muchthe atmosphere we look through as it is the onerous andunrelenting difficulties in establishing reliable radio-metric reference scales, especially in the near UV. Thisis where research should be conducted in the immediatefuture, and only after an order of magnitude improve-ment is made will it be justifiable to make spectral ir-radiance measurements outside the atmosphere.

The author wishes to acknowledge meticulous designwork done by R. Domke, J. Knox, and P. Groves.Furthermore, I am indebted to V. Ferrell for operatingthe spectral radiometer. To the staff and director ofthe Mauna Loa Observatory I extend thanks for friendlyand helpful cooperation. This research was supportedby National Science Foundation grant ATM 78-12185.

References1. M. Mitchell, Jr., "Paleoclimatic Evidence for the Solar Cycle and

its Variation," in The Solar Output and its Variations, 0. R.White, Ed. (Colorado Associated U. P., Boulder, 1977).

2. R. C. Willson, S. Gulkis, M. Janssen, H. S. Hudson, and G. A.Chapman, Science 211, 700 (1981).

3. A. K. Pierce and R. G. Allen, "The Solar Spectrum Between 0.3and 10 Am," in The Solar Output and its Variations, 0. R. White,Ed. (Colorado Associated U. P., Boulder, 1977), pp. 169-192.

4. J. M. Beckers, C. A. Bridges, and L. B. Gilliam, "A High Resolu-tion Spectral Atlas of the Solar Irradiance from 380 to 700 nm,"Air Force Geophysics Laboratory AFGL-TR-76-0126, Vols. 1 and2 (1976).

5. J. Geist, B. Steiner, R. Schaefer, E. Zalewski, and A. Corrons,Appl. Phys. Lett. 26, 309 (1975).

6. D. Labs and H. Nickel, Zs. Astrophys. 69, 1 (1968).7. D. Labs and H. Nickel, Sol. Phys. 15, 79 (1970).8. W. C. Livingston, "Solar Input to the Terrestrial System," in

Solar-Terrestrial Influences on Weather and Climate, B.McCormac and T. A. Seliga, Eds. (Reidel, Boston, 1979), pp.45-57.

9. D. F. Heath and M. P. Thekaekara, "The Solar Spectrum Be-tween 1200 and 3000 A," in The Solar Output and its Variations,0. R. White, Ed. (Colorado Associated U. P., Boulder, 1977).

10. W. Bach, Rev. Geophys. Space Phys. 14, 429 (1976).11. S. F. Singer, The Changing Global Environment (Reidel, Dor-

drecht, 1975).12. G. E. Shaw, J. Appl. Meteorol. 19, 1254 (1980).13. A. C. Dittenhoefer, "An Investigation of the Effects of Sulfate and

Nonsulfate Particles on Light Scattering at the Mauna Loa Ob-servatory," submitted to Water Air Soil Pollut. (1981).

Meetings Calendar continued from page 1988

1983 March20-25 185th ACS Natl. Mtg., Seattle A. T. Winstead, 1155 16th

St. N. W., Wash., D.C. 20036

April

19-21 Atmospheric Spectroscopy Int. Workshop, OxfordshireR. W. Saunders (I. W.A.S.), Rutherford Appleton Lab.,Chilton, Didcot, Oxon, OX11 OQX, UK

May

17-19 OSA Conf. on Lasers & Electro-Optics, BaltimoreOSA, Mtgs. Dept., 1816 Jefferson Pl., Wash., D.C.20036

June

? Coherence and Quantum Optics, 5th Rochester Mtg. L.Mandel, U. Rochester, Dept. Phys. & Astronomy,Rochester, N.Y. 14627

27-30 4th Int. Conf. Integrated Optics & Optical Fiber Com-munication, Tokyo Sec., IOOC'83, c/o N. Kuroyanagi,Res. Div., Musashino Elec. Comm. Lab., NipponTelephone & Telegraph Public Corp., Musashino-shi,Tokyo 180, Japan

29-7 July CIE 20th Session, Warsaw U.S. Nat. Comm., CIE, coNBS, Wash., D.C. 20034

July

4-5 4th Top. Mtg. on Gradient Index Optical Systems, Kobe,Japan I. Kitano, Nippon Sheet Glass Co., Ltd., 1,Kaido Shita Konoiki, Itami, Hyogo 664, Japan

August

28-2 Sept. 186th ACS Natl. Mtg., Wash., D.C. A. T. Winstead, 115516th St. N.W., Wash., D.C. 20036 continued on page 2021


solar spectral irradiance and atmospheric transmission at mauna loa observatory

Documents