Solar irradiance: total and spectral and its possible variations
Post on 02-Oct-2016
Solar irradiance: total and spectral andits possible variations
M. P. Thekaekara
The present status of our knowledge of the total and spectral irradiance of the sun is briefly reviewed. Thecurrently accepted NASA/ASTM standard values of the solar constant and the extraterrestrial solar spec-tral irradiance are presented. The uncertainties in these values are relatively high. Data on the variabili-ty of the solar constant are conflicting and inconclusive. The variability of solar spectral irradiance. is al-most totally unknown and unexplored. Some alleged sun-weather relationships are cited in support of theneed of knowing more precisely the variations in total and spectral solar irradiance. An overview of thesolar monitoring program of NASA is presented, with special emphasis on the Solar Energy Monitor inSpace (SEMIS) experiment which has been proposed for several of the spacecraft missions. It is a combi-nation of a solar constant detector and a prism monochromator.
1. IntroductionThe solar constant and the solar spectral distribu-
tion have been the topic of extensive investigationsfor a long period of time, and many different valueshave been proposed. At the turn of the century,Hann's Handbook of Meteorology published withoutpreference three values of the solar constant: Pouil-let's 1230 Wm-2, Angstrom's 2717 Wm-2 , and Lan-gley's 2051 Wm- 2 . That was indeed a very widespread of values. The long series of measurementsmade by Abbot resulted in values close to 1350Wm-2 . Subsequent to Abott's work and prior to themore recent measurements made from high altitudeplatforms, two values that were widely accepted wereJohnson's 1396 Wm 2 and Nicolet's 1380 Wm-2 .11. NASA/ASTM Standard of Total and Spectral SolarIrradiance
A listing of the values proposed as a result of mea-surements made from research aircraft, balloons, andspacecraft is given in Fig. 1. They are grouped ac-cording to the three radiation scales to which theyare referred: the International Pyrheliometric Scaleof 1956, the Absolute Electrical Units Scale, and theThermodynamic Kelvin Temperature Scale. Each ofthese values is the result of a long series of measure-ments. The horizontal lines indicate the degree ofuncertainty claimed by each author.
The author is with NASA Goddard Space Flight Center, Green-belt, Maryland 20771.
Received 2 September 1975.
Out of these measurements from high altitudeplatforms resulted a revised value of the solar con-stant, 1353 Wm-2. This has been accepted as thedesign criteria for NASA space vehicles1 and as theengineering standard of the American Society ofTesting and Materials (ASTM).2 The NASA/ASTMstandard includes also a solar spectral irradiancecurve which is shown in Fig. 2. The table of spectralirradiance values is given in Ref. 3. This is basedmainly on measurements made by the GSFC CV 990team who used five large spectral irradiance instru-ments of different types.4 The spectral curve ob-tained from the GSFC data was modified slightly inthe 0.3-0.7-gm range with the aid of the filter radi-ometer data obtained by the Eppley-JPL team andthus was defined the NASA/ASTM spectral irra-diance standard.
A closer look at the currently accepted values ofthe total and spectral irradiance of the sun outsidethe atmosphere shows that these values are by nomeans final, and considerably more work needs to bedone. The value 1353 Wm-2 is significantly lowerthan those of Johnson and Nicolet, but rather closeto those proposed more recently by Labs and Neck-el,5 Makarova and Kharitonov,6 and Stair and Ellis.7And it is also close to the earlier Smithsonian value ofAbbot. But one observes that it is based on ninevalues which range between 1338 Wm-2 and 1368Wm-2 , and the two extreme values are those that, ac-cording to the respective authors, claim least uncer-tainty.
The NASA/ASTM value of the solar constant hasa stated uncertainty of 21 Wm- 2 or 1.5%, which israther large for an important constant of geophysicsand astronomy, when we consider that most other
April 1976 / Vol. 15, No. 4 / APPLIED OPTICS 915
IPS 56 LENINGRAD 135314DENVER 13386EPPLEY-JPL 136023A 6618 134326A 7635 134940
AEUS JPL TCFMGSFC CONEJPL ACR
1353 201358 24
TKTS HY-CAL 1352 22
STANDARD VALUE 135321
1300 1320 1340 1360 1380 1400
1300 1320 1340 1360 1380 1400W. -2
Fig. 1. Values of solar constant derived from high altitudemeasurements.
SOLAR CONSTANT 1353 Wm-2
1.0 1.2 1.4 1.6WAVELENGTH (,m)
Fig. 2. The NASA/ASTM standard curve of extraterrestrial solarspectral irradiance.
constants like the velocity of light, Planck's constant,or electron charge are quoted to an accuracy of a fewparts in a million. The uncertainty in the spectral ir-radiance is considerably greater than in the solar con-stant itself and varies with the wavelength range. Inthe visible and near ir the GSFC experimenters useddifferent instruments which did not yield identicalspectral curves. There were greater differences be-tween the GSFC data and the Eppley-JPL filterdata. Equally disturbing, if not more so, are the dif-ferences between the NASA/ASTM CURVE curveand those published earlier, those of F. S. Johnson,Labs, Neckel, and others.Ill. Solar Variability and its Effects: InconclusiveEvidence
In many applications of the solar irradiance values,both total and spectral, a question of major concernis the variability of these values. The changes in thesolar irradiance in the uv below 0.3 m and in the mi-crowave range between 2 cm and 10 m have been wellestablished. As for the solar constant itself the vari-ability is over a smaller range, probably 1% or 2% orperhaps less; but the data available in literature donot permit any firm conclusions. A precision consid-erably better than what is possible for absolute accu-racy is required to determine the variability of thetotal and spectral irradiance of the sun. Measure-ments made from sea level and even from mountaintops and research aircraft do not have the requiredaccuracy and precision because of the highly variableatmospheric attenuation and the errors inherent inextrapolation to zero air mass.
The most extensive data on the solar constant arethose collected by Abbot and his co-workers at theSmithsonian Institution. They cover a period of over50 yr. There have been numerous attempts at an
Table I. Major Contributions to Analysis of Smithsonian Data for CorrelationBetween Solar Constant E and Sunspot Number N
Refer-Data period Year ence Main conclusions of the authors
Angstr6m 1915-17 1922 8 E is max. for 100 < N < 160.E increases by 2.4% for N 0-80.
Abbot, et al. 1920-39 1942 9 E does not vary with N.E varies by 1% with faculae area.
Aldrich 1923-44 1945 10 E is maximum for N near 20.Greatest variation of E was 0.4% in 1923-33.Aldrich and Hoover 1922-52 1952 11 E has small irregular variations, less than 2%.Largest increase during 1946-50 when N was high.
Aldrich and Hoover 1944-52 1954 12 E increases by 0.8% as N increases from 0 to 175.Allen 1915-52 1958 13 Variations in E do not exceed 0.1%.Sterne and Dieter 1923-58 1958 14 There are no periodicities in N common to data from two
stations: Table Mountain and Montezuema.Abbot 1923-58 1958 15 N has many periodicities with submultiples of 273 months.
N changes up to 4% with large sunspots and magnetic storms.Angstr6m 1923-55 1970 16 Variations in N are less than 0.2%.Kondratyev and 1928-32 1970 17 There is indubitable correlation between N and faculae area.
Nikolsky 1943-52 There are synchronous variations up to 0.8% at three stations.
916 APPLIED OPTICS / Vol. 15, No. 4 / April 1976
I I I
analysis of these data to determine whether the solarconstant varies with the sunspot number, the mag-netic field, or any other observable feature of the sun.The sunspots have the well known 11-yr cycle. Inaddition there are several other cycles, those of solarrotation, the solar magnetic field, the magnetic sectorboundaries, the longer periodicities of 90 yr andmore. There are also the sporadic and unpredictableevents that last for periods from a few minutes toseveral days.
A few major contributions to the analysis of theSmithsonian data for correlation between the sun-spot number N and the solar constant E are given inTable I. The Table lists the authors, the period ofthe Smithsonian data that they considered, the yearof publication, the serial number of the publication inthe list of references at the end of this paper, and themain conclusions that the authors have derived.This summary statement of the main conclusionsdoes not do full justice to the wealth of informationcontained in the respective publications. It is ob-vious that different authors examining the same datacome to entirely different conclusions.
Three studies that are independent of the Smith-sonian data should also be cited. Bossolasco et al.'8made an analysis of ground based measurements atfour stations, Uccle (Belgium), Jerusalem (Israel),Krippenstein, and Sonnblick (Austria), over the1956-1960 period and concluded that E has a pro-nounced maximum for No 160 and that it decreasesby 15-17% as N increases to 250 or 300. This is anupper extreme for suggested variations in the solarconstant. Kondratyev attributes this finding to in-creased atmospheric absorption;'7 there is a closecorrelation between the nine dates of observations ofBossolasco et al. with N > 250 and the dates of theU. S. nuclear tests of the 1956-1958 period. Fromthe Leningrad balloon data Kondratyev and Nikol-sky17 conclude that E is maximum for 80 < N < 100and that there is a decrease of N of 2-2.5% for higherand lower N values. They also find that Angstrom'sanalysis made in 1922 had yielded a curve of E vs Nof much the same shape, and Thekaekara's value ofthe solar constant based on measurements made inAugust 1967 when N was 96 is in general agreementwith the Leningrad data. The Temperature ControlFlux Monitor (TCFM) of JPL was flown on two MarsMariner Missions in 1969.19 Data were available ona period of 200 days, but the signals normalized toone astronomical unit showed a steady downwardslope, decreasing from 1352.5 Wm-2 at launch to1288 Wm-2 at encounter with Mars. The probablecause was instrument drift which could be correctedby making a pitch turn of 100 after encounter anddetermining the shift in the zero of the instrument.Major conclusions from the TCFM were that thereare variations in E of the order of tenths of 1% thatare random, that the largest observed variation was0.4%, and that long term cyclic variations of severalpercent were not observed.
In 1952 Aldrich and Hooverl concluded theirpaper "The Solar Constant" with the following re-mark: "There is currently much interest in travelbeyond the stratosphere. When this is accom-plished, direct measurements of the solar constant,unhampered by an ever-changing and complex atmo-sphere will follow." Travel beyond the stratosphereis a reality. Apart from the TCFM and the more re-cent Earth Radiation Budget(ERB) experiment noattempt has been made from spacecraft to determinewhether the solar irradiance, total and spectral, ischanging along with all the other observable featuresof the sun, which are known to be changing. Thequestion is not new. The opening paragraph of apaper published in 1966 by Abbot, "Solar Variation,a Weather Element,"20 is well worth quoting: "Theeminent astronomer Dr. Samuel Pierpoint Langley,third secretary of the Smithsonian Institution, at Dr.George E. Hale's invitation, sent me to Mt. WilsonObservatory in 1905 to observe the radiation of theSun. Langley suspected that this might be variable,and that its variation might be a cause of weatherchanges. If the suspected solar variations proved pe-riodic, they might lead to long-range weather fore-casts of great value to agriculture and water supply."
While there a great deal in literature aboutchanges in the solar constant and their effects onweather, a great deal that is conflicting and inconclu-sive, there is hardly any mention of changes in thespectral distribution in the visible and near ir wherethe energy output of the sun is the greatest. Thereason is not that changes do not exist, but that theyare totally unknown and unexplored. Almost. allsolar energy effects on the atmosphere and the earthare wavelength dependent. Localized radiation bal-ance and the transport of large masses of air, increaseof atmospheric pollution and its sink mechanisms,the making of weather and climate and the numericalmodeling required for predicting weather, changes inozone and their erythemal effects on humans, all de-pend on some limited portions of the solar spectrummore than on others. The atmosphere is far frombeing a neutral density filter, or is the land and oceansurface of the earth a gray absorber. The earth albe-do spectrum is very different from the solar spec-trum. Ozone production is due to solar uv. Chloro-phyll photosynthesis essential for all life support isdue to- wavelength bands centered around 450 nmand 650 nm, with half-intensity bandwidths of about30 nm. Other resonance phenomena are photomor-phogenic responses like seed and flower develop-ment, shape and size of leaves, plant height, leafmovements as in mimosa, phototropism; the associ-ated wavelengths are nearly the same, but withslightly greater bandwidths. Absorption by watervapor with all its major effects on the making ofweather is in narrow-wavelengths bands, all beyond700 nm, which is the more poorly known part of thespectrum. If the solar constant is shown to bechanging, we would want to know what spectralranges are causing this change. Nor is there any rea-son to assume that the changes in solar spectral irra-
April 1976 / Vol. 15, No. 4 / APPLIED OPTICS 917
diance over limited wavelength bands are as small asin the solar constant itself.
If it is known with sufficiently high precision whatare the changes in solar radiant flux, in what wave-length ranges and with what periodicities, thatwould provide a solution to many intriguing correla-tions which have been observed between solar activityand terrestrial phenomena. Among such phenomenaare wintriness index of northern hemisphere sea levelpressure, the annual march of temperature of differ-ent cities of Europe, the changes in meridional sealevel pressure, the annual frequency of Etesian windsover Greece. Thus, for example, the number of daysper year when Etesian winds, winds from the NE andNNE as opposed to sea breeze, blew over Athens inthe 1893-1961 period followed the same trends ofmaxima and minima as the annual mean sunspotnumbers. The geomagnetic disturbances have a pe-riodicity of 27 days superposed on an 11-yr period.Long period correlations have been observed in suchweather related phenomena as the water level in riv-ers and lakes, annual growth rings of petrified andliving trees, the advance and retreat of glaciers, thefrequency of lightning strikes on the electrical powergrid, the annual rainfall in different locations. In1953 Brooks and Carruthers observed "There is acorrelation coefficient of +0.8...