Journal of Research of the National Bureau ot Standards Vol. 46, No.5, May 1951 Research Paper 2206 Ultraviolet Spectral Distribution of Radiant Energy From the Sun Ralph Stair
This paper gives the results of some preliminary measurements on the ul traviolet spectral energy distribution of direct solar radiation at Washington, D . C. Data are given for wave l e ng~h s extending below 300 millimicrons for air masses approximating M = 1.4. T~ e ne'y equipment and method employed in this work permit the rapid acquirement of a prismatic ultraYlOlet s p ec~ r al energy curve at sea level showing greater Fraunhofer struc t ure than prevIOusly obtamed even at desert and mountain stations. 9ther applications of t?is .equipment, for example, in the study of total ozone and in sky-hght spectral-energy dI stributions are suggested.
1. Introduction II. Instruments and Methods The determination of the ultraviolet spectral radi The apparatus employed in this investigation ant energy of sunlight has long been a subject of consists of a Farrand double quartz prism spectro considerable interest. Nevertheless, but few meas meter, RCA type 935 phototube, 510-cycle light urements for wavelength region s shorter than about modulator, tuned amplifier and auxiliary meters, and 330 or 340 m,u have been r ecorded in the literature recording apparatus (see fig . 1), set up and calibrated [1 to 7).1 This is easily understandable, since double as described elsewhere [19], except that a h eliostat quartz prism spectrometers suitable for the auto was employed for keeping sunlight reflected into the matic and rapid recording of low spectral intensities entrance slit of the instrument. For this purpose have not been generally available until quite recently. th e entrance slit of the instrument was placed toward The b.est radiometric data previously published, the south and the h eliostat arranged so that a beam extendll1g to wavelengths shorter than 300 m,u, are of direct sunlight r efl ected from a plane chrom those of Pettit [1] . Although they were obtaineci at aluminized mirror fell upon th e center of the first high altitudes the low sensitivity of the radiometer lithium-fluoride-quartz collimating lens of the instru n ~cess itat ed the use of wide op tical slits (spectral ment. The spectral energy response characteristic WId th usually 5.0 to 10.0 mIL) and long r esponse times. of the complete instrument, including th e h eliostat The solar spectrum, when photographed with a mirror and photo tube, were determined by using a good grating spectrograph, shows :::.. continuous standard tungsten-filament-in-quartz lamp [19, 20] spec tral energy di stribution on which are superim calibrated for color temperature [11] and evaluated posed thousands of Fraunhofer lines and bands [8, 9] . for spectral emission in the ultraviolet with the These features are mueh too numerous to be detected application of spectral emissivity data for tungsten separately by a prism spectrometer . The prism [12 to 1.5] (see table 1). In this calibration th e ~verages the spectral energy over a wavelength radiant energy from the standard lamp was reflected ll1terval corresponding to th e eff ective slit width of into the spectrometer by th e helio tat mirror. The the instrument. H ence in spectral regions in which lamp distance was arranged so that the illuminated the Fraunhofer lines are numerous and have intense area of the collimating lens approximated that absorption,.depres~ions will oecur in the energy curve illuminated by direct sunlight. observed wIth a prIsm spec trometer. The character No refracting lens or mirror was employed to of the observed curve will be significantly affected produce an image of the sun on the entrance slit of by the slit width employed. In general, the narrower the spectrometer because an inteO'rated solar energy the slit width the more detailed will be the structural spectrum was desired. Although a small area of character of the observed spectrum. the sky surrounding the sun contributed to the total The data on the ultraviolet solar energy curve ~n el'gy respo~se, its eff ect is. necessarily small since recorded in this paper were obtained between 10 :00 It probably dId not at any tIme exceed 1 percent of a. m. and ~2:00 noon during October 1950, when the the total and did not differ radically in spectral sun was slumng through clear atmosphere associated quality from sunlight as observed in ch eck m easure with high-pressure conditions. Since the purpose of ments on the sky only, near th e sun. this investigation was primarily to determine the In the present investigation th e high sensitivity pos ibilities of the available equipment in work of of the detecting and recording apparatus permitted ~his type rather than to acquire extensive data, th e the use of relatively narrow slit widths (spectral ll1struments have not been removed to a lo cation width 2 to 3 m,u), with noise levels less than 1 percent, having day-long clear access to direct solar rays. to about 305 m,u. H ence, with this equipmen t somewhat narrower spectral slit widths could b e
1 Figures in brackets indicate the liternture rcfercnces at the end of this paper practically employed at the altitude of th e Bureau's 353 PH OTOELECTRIC [ DETECTOR
RECORDER
FIGURE 1. Block diagram showing instrumental lay-out.
TABLE 1. Black-body data and emissivity factoTs fOT tungsten be changed at intervals of 10 to 30 m,u. For at operation temperalure of standard tungsten-filament-in-quartz example, below about 304 m,u for an air mass M=1.4 lamp; l'e/ative spectral energy emission of standard lamp, based on tabulaled values of tungsten emissivity; Telative an increase of 1 m,u in wavelength results in nearly spectral transm~ssiv e factor (response factor) of Farrand 100-percent change in the intensity of solar radiant spectrometer, including heliostat mirror, phototube and all energy reaching the earth's surface. This rapid other components of the instrument; transmission coefficients increase of intensity with wavelength made difficult of 0.18 cm of ozone per unit air mass through an air mass of M-1.43; and the calculated Rayleigh scattering spectral trans the accurate interpretation of recorder records below mission coefficients for ail' mass, M = 1. 43, at Washington 320 or 330 m,u. Hence, within this spectral region, (approximately sea level, 355 ft) supplementary readings were usually made by manual operation of the wavelength drive of the instrument and by recording the energy values as Relative Relative Transmis- energy Relative spectral sion coem- Rayleigh indicated on a Ballantine alternating-current VTVM Emis- energy response scatter- Wave- black sivityof tungsten factor of cient of 01' by the Leeds & Northrup recorder. length body. lamp, spectrom- 0.18-em coefIicient, 2945° K tungsten ozone, ntp M=1.43 c'z=1.43S' 2,915° I( eter,0.25-mm slits M=1.43 ------III. Ultraviolet Solar Energy Curve mil 300 100 0.422 422 48 0. 060 0.197 305 ----- . 425 509 59 .217 .219 310 143 .4275 611 82 . 457 . 242 The data on which the solar energy curves dis 315 ----- . 430 728 96 .653 .265 played in figures 2 and 3 and in table 2 are based 320 199 .433 862 139 .814 . 287 were obtained in the Radiometry Laboratory at this 325 .--.- . 436 1, 007 169 . 903 .310 330 269 . 4385 1,180 199 . 953 .332 Bureau, which is located in the northwest section 335 ----- . 441 1,367 241 .980 .354 of Washington away from a large part of the city 340 358 . 443 1, 586 271 . 997 .376 345 ----- . 4445 1,825 307 .998 .399 smoke. The air mass penetrated by the direct 350 469 . 446 2,092 335 . 999 .421 solar rays ranged from approximately M = 1.35 to 355 -- --- . 447 2, 369 405 .999 . 444 M=1.50 during the course of the mesasurements . 360 fi98 .448 2,679 407 1. 000 .466 365 ----- . 449 3, 017 444 ------.--.--- Representative data for specific air masses are given 370 752 . 450 3,384 458 ------.------for a few days during the early part of October. 375 .4505 3,776 479 ------On some of the days on which 10 to 20 spectral 380 --931 .451 4,199 507 ------385 ----- . 451 4,605 549 ------... _---- determinations were rapidly made within the spectral 390 1, 137 .4515 5,134 579 ---_.--.------400 1,369 .452 6,188 630 range from 299 to 315 (or 320) mJL considerable ------variation was noted in the shorter wavelength intensities relative to that at 315 or 320 mJL. As the Rayleigh scattering factor is not greatly different Radiometry Laboratory (355 ft) and much narrower (see table 1) between the limits of this wavelength ones at mountain elevations. The rapid spectral interval, it is to be concluded that the greater scanning rate (3 00 to 400 m,u in about 3 min) permits fluctuations at the shorter wavelengths must be the completion of an ultraviolet spectrogram before a primarily associated with rapid changes in the significant change occurs in the air mass 2 even for amount of ozone penetrated. The data in table 2, air masses M = 2.0 to M = 3.0. and for the shorter wavelengths in figures 2 and 3, are Within the wavelength range extending from the averages of three to seven sets of measurements about 299 to 330 m ,u the spectral intensities increase at air masses approximating the values applied to so rapidly with wavelength that in order to evaluate the curves. the data properly the instrumental sensitivity must The solar energy curves displayed contain numer ous absorption bands varying in magnitude both in 2 In this paper air mass, represented by j\£, refers to the amount of air between the observer and the sun; air mass, .M = 1, being assigned to a single thickness spectral width and in intensity of absorption. If of atmosphere above the particular station. For an y solar position air mass is nearly proportional to the cosecant of the angle the sun's rays make with the reference is made to the Rowland maps and wave earth's surface. For sunlight outside the earth's atmosphere, air mass, j\1=O. length data [8] and to the Utrecht photometric atlas Tho reader must not co nfuse th;s usc of the term, air mass, with the same term n sed to designate air of a particular origin, temperature, or other property. of the solar spectrum [9], it is found that each band 354 T ABLE 2. Observed relative spectTaI-energy d'istribution of direct sunlight at W ashington, D. C. on 3 days in Oclober 1950 WASHINGTON, D.C. 195 20 October 3 October 6 October 6 October 11 Wavelength M = l.37 M= J. 43 .JI1=1.4l .111 = 1.51 18 m l' 299.2 5 3 4 300.2 10 4 7 8 16 301. 2 18 9 13 15 302.2 35 20 29 30 303. 2 59 37 '15 51 14 304.2 89 60 7 8 314.2 502 502 510 478 315.2 498 507 519 497 316.2 534 534 317.2 572 6 318.2 605 319.2 629 320.2 646 4 Nu mber of settings . . 2 Fraunhofer lin es responsible for the observed ab O ~~~-L~~~~-L~~~L-~-L~ __L-J sorption spectrum is given in table 3. The waye-' 290 330 350 370 390 4 10 4 30 length position of the ab orption bands is found to' WAVEL ENGTH, MILLIMICRONS be in close agreemen t 'with that of the main concen FIGUR E 2. Spectral dis17'ibulion oj Tadiant ene1'gy at W ashing trations of the winged lines [10] in the solar spectrum. I... ton, D. C., from the sun for 2 days in October 1950; ordinates The large drop in solar radiant energy intensity at' arbitTary. about 390 mIL, previously noted by Pettit [1], is very interes ting. results from one or more groups of intense Fraunhofer When consideration is given to th e winged lines lines. Some of these resul t mainly from one or two [10] of the solar spectrum and their eff ect upon th e elemen ts in the solar atmosphere--others from sev observed intensity as vividly displayed on the Utrech t eral elemen ts. For example, in the spectrall'egion of 309 mIL the principal absorbing elemen ts are the neutral metals, N i, Mn, Al, and F e, accompanied by WASH INGTON,D.C. OH, while at 315 to 316 mIL, Cr, F e, Ti and others 20 are presen t. Again at 322 to 323 mIL F e and several S UN other metals have strong absorption lines, while at 18 MoO 335 to 33 7 mIL H shows additional absorption. At (C ALCU L ATE D I (OCT 3,5,6,11, 2 6) 345 to 348 mIL F e is the principal absorbing element, 16 1950 although many lines of Mn, Ni, and. other elemen ts are present. At 358 to 359 mIL and at 375 to 376 mIL >- 14 0:"' the absorption bands result mainly from many W z 12 strong F e lines. The strong CN band (at 388 mIL) w w accompanied by many strong F e lines (around 385 > ~ 10 mIL), together with Mg absorption results in the .J W strong band centered a t about 386 mIL . The band 0: at 395 mIL is a combination of the intense H and Ie 8 Fraunhofer lines lo cated at a bout 393.5 and 396.8 6 mIL and results primarily from absorption by ionized Ca and H atoms in the solar atmosphere. At 408 4 m IL and again at 432 mIL Fe and H are primarily responsible in producing two of the important ab 2 sorp tion bands of the visible spectrum of the sun. T wo r elatively weak bands occurring at 465 and 489 m IL are caused principally by F e and H , respec tively, while the strong band between 520 and 53 0 F I CURE 3. Spectral distribution of radiant energy at W ashing mIL results mainly from Mg absorption in the solar ton, D. C., f 01' 1 day in October 1950; and calwlated mean atmosphere. A more detailed listing of the principal outside the atmosphere for several days; ordinates arbitrary. 355 TABLE 3. Relationship between the location of winged and ~he I?-ethod employed in their acquirement, especially other strong Fra1tnhofer lines and the observed minima in the spectral solar energy curve 111 VIew of the fact that at sea level solar radiant energy intensity values are extremely low at the Very strong lines result from a bsorption b y the italicized elements shorter ultraviolet wavelengths. The quality of these data indicate the pTacticability of using much Wavelength Observed !larrower spectral slit widths if the apparatus were Rowland Principal absorbing elements absorption 111stalled at mountain or desert locations. At suit and Utrecht band maps able stations much valuable solar spectral data should be obtainable. mp' m IL 308 to 310 Fe, N i, Mn, AI, Oll ...... 308 to 3n9 Data of the type obtained with this instrument 316 to 317 Fe T and n, Or II , Ni, Ti II, OH... ______315 to 316 have long been urgently needed for use in ozone 323 Fe I and n, rl'i II, NL. __ . ______. ______. 322 to 323 331. 5 Fe, Zr II , Na It Ti n , Fe IL . ______determinations by means of the phototube and filter 334 Mg, rJ' i I , Ti n, Fe, OrlL ______335 to 337 336 N H , Ti If, N i , Or IL ______method [16 to 18] . Previously a smooth curve best 344 Fe, i\1n H, 'J' i Il, N i, Co ______345 to 348 representing the filter measurements has been em 358 Fe,Ni, v II , Be Ti , Co, Ti II, Or IL ______358 \0 359 374 Fe, N i, rri, rfi ]L ______375 to 376 ployed. The limited data recorded herein should 385 Fe, M g, CN...... 386 394 Ca ll, AI, H ...... 395 serve to permit the use, in future phototube-filter ~10 ll6...... 408 ozo~e w 92 lOfJ - ;:il - - :? 357