A Research Program Aimed at High Altitude Balloon-BorneMeasurements of Radiation Emerging from theEarth's Atmosphere

C. R. Nagaraja Rao and Z. Sekera

The relevance of polarization measurements in the near uv and visible regions of the electromagnetic spec-trum in investigations of the effects of the atmospheric aerosol and the reflective properties of the ground,intermediate layers of the atmosphere, and the atmosphere as a whole on the ambient radiation field is dis-cussed. The salient features of a photoelectric skylight polarimeter characterized by its adaptability toballoon-borne measurements of the degree of linear polarization and relative intensity variations in fournarrow spectral intervals (bandwidth -150 A) centered on XX 3270 A, 3975 , 500 !, and 6120 A are de-scribed. Preliminary results of measurements obtained in a balloon experiment staged at Page, Arizona(36 057'N, 111 027'W, 1300 m above msl) on 30 March 1965 are presented.

1. IntroductionWhen the possibility of performing optical experi-

ments from aboard space vehicles that would eitherorbit around the earth or fly by neighboring planets wasrealized in the Fifties, interest was rejuvenated in studiesof planetary atmospheres, by means of the polarizationof light which was diffusely reflected by the planet, onthe lines of earlier work by Lyot' and Dollfus. 2 Sekeraand collaborators3 computed the polarization para-meters of the radiation emerging from a Rayleigh (mole-cular) atmosphere illuminated at the top by sunlight.Sekera and Vizee4 performed similar computations of theintensity and polarization of the light diffusely reflectedby a planet as functions of the phase angle of the planet.Laboratory studies, directed towards the determinationof surface characteristics of a reflecting layer from thepolarization of the reflected light, were commenced byCoulson', the Cornell group', and Dollfus.7 Rozen-berg8, in his review of the feasibility of optical methodsto probe the Venusian atmosphere, drew attention to aningenious application of the Umow effect, in terms ofwhich the wavelength dependence of the polarizationof the reflected light could be correlated with the spec-tral absorption characteristics of the reflecting medium.

The authors are with the Department of Meteorology, Univer-sity of California, Los Angeles, California.

Received 7 September 1966.The work reported in this paper has been supported by the

Air Force Cambridge Research Laboratories, Office of AerospaceResearch under a contract and by the Atmospheric Sciences Pro-gram of the National Science Foundation under a grant.

Contributions to Meteorology, Department of Meteorology,UCLA No. 130.

Progress achieved in the interim with regard to thesolution of the problem of radiative transfer in inhomo-geneous and turbid atmospheres by Sekera9 and Kano 0

rendered possible, at least in a semiquantitative fashion,the determination of the vertical distribution of inhomo-geneities (aerosols, absorbing constituents) in an other-wise purely scattering planetary atmosphere. Theyinitially assumed that the ground was nonreflecting.These investigations are being extended to the generalcase in which reflective properties of the planetaryboundary can be characterized by a generalized reflec-tion function that has the specular (Fresnel) and diffuse(Lambert) reflection functions as extreme cases.

The advantages of using the earth's atmosphere as aproving ground to test the validity of theoretical pre-dictions and the feasibility of experimental techniques areobvious. Also, the increasing emphasis upon the deter-mination of the contrast transmission properties of theatmosphere warrants a proper understanding of the ex-tent and distribution of the atmospheric aerosol andthe reflecting properties of the ground. Recent inves-tigations by Coulson""l , 2 and Fraser'" must be men-tioned in this context. The meteorological importanceof such investigations, especially of the distribution ofthe atmospheric aerosol in view of the definitive roleplayed by the same in diverse atmospheric processes suchas nucleation, catalysis, and radiation balance, canhardly be exaggerated.

Meaningful results can be obtained only if the mea-surements of the polarization features of the ambientradiation field are made at various levels in the atmo-sphere from a platform which moves with a low speedrelative to ground so that the assumption of horizontalhomogeneity, implicit in the concept of a plane parallel

February 1967 / Vol. 6, No. 2 / APPLIED OPTICS 221

Fig. 1. Polarization features of upward radiation in the principalplane of the sun.

atmosphere on which most of the theoretical investiga-tions are based, is reasonably justified. Such measure-ments would also aid the determination of the distancefrom the planetary boundary beyond which the surfacefeatures cease to be effective in defining the resultantradiation field. It is desirable that these measurementsbe made above the tropopause and up to stratosphericheights so that the effects of low level dust and haze arefully taken into account. The natural choice for such aplatform would be a stabilized, high altitude, balloon-borne gondola.

In addition to weight-lifting capacity, several otherfactors, such as the slow rate of ascent (as opposed torockets) and the relative simplicity of aerodynamicproblems encountered and the low ground speed (asopposed to an aircraft), favor the choice of a balloon asthe vehicle to carry the stabilized platform. How-ever, allowance must be made for the rather capricioustrajectories and the not-too-infrequent tropopauseshattering suffered by polyethylene balloons.

IL. Theoretical BackgroundThe present discussion is mainly confined to radiation

emerging upward from the earth's atmosphere. Excel-lent reviews of the complementary problem of the down-ward radiation (skylight) may be found elsewhere.'4,"5The dependence of the polarization features on atmo-spheric composition and the reflective properties of theground are discussed as deviations brought about in theambient radiation field of a plane parallel molecularatmosphere.

Coulson's1 8 investigations of the radiation emergingfrom the top of a purely scattering molecular atmo-sphere bounded by a diffusely reflecting (Lambertian)ground have revealed that the over-all polarization dis-tribution curve in the sun's vertical (plane containingdirection of illumination from sun and local nadir) isanalogous to that of skylight with the antisolar pointreplacing the sun. Two neutral points analogous tothe Babinet and Brewster points appear on either sideof the antisolar point (Fig. 1). For lower solar eleva-tions, the Brewster analog disappears and the Aragoanalog appears. The region of maximum polarizationis situated about 90°00' from the antisolar point whenthe diffuse reflectivity of the ground is small (<0.25).The Brewster and Arago analogs are mutually exclusive,

even though they are shown conjunctively in Fig. 1 forpurposes of illustration.

For sufficiently small optical thicknesses (-0.05),i.e., at longer wavelengths, and for large values of dif-fuse reflectivity (0.80), only the light from the hori-zons may show any polarization. This is understand-able since, over most of the sky vault, light that isdiffusely reflected by the ground causes considerabledepolarization. However, because of the long pathlengths involved near the horizons, enhanced scatteringmore than compensates for the depolarization causedby the reflected, unpolarized light, which in turn suffersincreased attenuation. For large optical thicknesses(-1.0), i.e., at shorter wavelengths, this effect is lesspronounced since the atmosphere becomes less trans-parent. But when the diffuse reflectivity changes from0 to 0.80, the location of the point of maximum polar-ization departs considerably from the 90°00' region.

Fraser 7 has recently made extensive computations ofthe polarization parameters of the radiation emergingupward from a planetary atmosphere with a specularlyreflecting (Fresnel) boundary. The results of his com-putations may be summarized as follows-

(1) The degree of maximum polarization of theemergent radiation for moderate optical thicknesses(<1.0) differs widely for the Lambert and Fresnelmodels. The value for the Fresnel model approachesclosely what would prevail if no radiation were reflectedby the ground in the Lambert model.

(2) The point of maximum polarization is locatednear 90°00' from the antisolar point both in the Fresnelmodel and the Lambert model when the ground is non-reflecting. However, when the diffuse reflectivityassumes finite, nonzero values in the Lambert model,the location of the point of maximum polarization de-parts from the 90°00' region.

(3) The Babinet and Brewster analogs approach theantisolar point much more rapidly with decreasingoptical thickness in the Fresnel model than in the Lam-bert model. The neutral points may disappear al-together in the Fresnel model for small optical thick-nesses. This can happen in the Lambert model onlywhen the atmosphere becomes extremely transparent.also observe that, whereas in the Lambert model thepolarization features of radiation emerging both in theupward and downward directions are analogous, theymay differ widely in the Fresnel model.

However, it should be noted that, in nature, it isunlikely that a planetary boundary would be eithercompletely specular or diffusive in its reflection proper-ties. The physically more realistic situation when theplanetary boundary is characterized by a reflectionfunction that is a weighted mean of these two extremesis being investigated.

Kano 0 has investigated the sensitive dependence ofthe polarization features of the upward radiation on thelocation of a concentrated turbid layer with a knownsize distribution according to Junge's power law. Hestudied the two extreme but physically feasible caseswhen the turbid layer is located either at the top or atthe bottom of an otherwise purely molecular atmo-

222 APPLIED OPTICS / Vol. 6, No. 2 / February 1967

sphere. For low solar elevations, the presence of ahigh level turbid layer results in pronounced asymmetryin the intensity distribution-the intensity of theemergent radiation being greater on the antisolar sideof the horizon. However, the dispersion of polarizationis less pronounced than when the turbid layer is locatedat the bottom of the atmosphere. The location anddispersion of the neutral points are less sensitivelyaffected by the turbid layer when it is located at thebottom of the atmosphere than when it is at the top.For low solar elevations, the presence of the turbidlayer at the top affects similarly the neutral points ofboth skylight (downward) radiation and the upwardradiation.

Recently, Sekeral has derived the formal solution tothe problem of inversion, in terms of which informationabout atmospheric parameters such as composition,vertical distribution of inhomogeneities, etc., can bederived from measurements of the polarization of thediffusely reflected (upward) radiation made with instru-ments aboard space vehicles or space probes.

111. Experimental Investigations

A. The Photoelectric Skylight PolarimeterOnly the salient features of the photoelectric skylight

polarimeter which has been used in high altitudeballoon-borne measurements are discussed in this sec-tion, since detailed descriptions of design and perform-ance are available elsewhere.19 With the present instru-ment, the degree of linear polarization and relativeintensity variations can be measured in four narrowspectral intervals (bandwidth '150 A) centered on XX3270 A, 3975 A, 5000 A, and 6120 A. The instrumenthas a field of view of 3°00'. Earlier versions of thisinstrument have been used successfully in groundbased observations.

The principal design features are schematicallyshown in Fig. 2. A collimated beam of polarized lightpasses through an optical train consisting of a rotatinghalf-wave retardation plate, a stationary Glan prism,an interference filter with peak transmission in theneighborhood of the characteristic wavelength of theretardation place, and a quartz neutral density wedge.The emergent beam impinges upon the photocathodeof an end-window type photomultiplier tube. Theresulting sinusoidal photo signal passes through auxiliarylogic and detection circuits and is recorded.

It can be shown'4"'5",9 that, under predetermined con-ditions of detection, the amplitude of the sinusoidalsignal is directly proportional to the degree of linearpolarization of the incoming light, provided that the dclevel of the photosignal is maintained constant eventhough the intensity of the incoming beam may vary.This is accomplished with the circular quartz neutraldensity wedge that functions as a light limiter. An acservo system, for which the error signal is derived fromthe photomultiplier output, brings in different portionsof the wedge to intercept the beam emerging from theinterference filter so that variations in intensity of theincident beam are offset. The angular displacements

UGHT LIMITINGCOLLIMATED BEAM STATIONARY QUARTZ WEDGE

OFPOLARIZED GHT GLAN PRISM R O

-tI ~~~~~ TI ~~ELECRONIC

0 Q° ACtr \ jPTU I SIER TION

RTATING \& iX SSEHALFWAVE PLATE FILTER -ENSEMBLE WHEEL l

ILCLOSED SERVO LOOP_

Fig. 2. Schematic of the photoelectric skylight polarimeter.

of the wedge from a fiducial mark in the polarimetercan then be interpreted as relative intensity variationson a logarithmic scale. The predetermined conditionsof detection require that the plane of polarization of theincoming light be oriented at an angle of 45°00' withrespect to the plane of transmission of the analyzingGlan prism. This is achieved by orienting the Glanprism in such a fashion that its transmission plane makesan angle of 45°00' with the plane in which the polari-meter scans and then confining the polarimeter scan tothe principal plane of the sun (sun's vertical), underwhich circumstances the plane of polarization of theincoming signal will always be at 45°00' from the planeof transmission of the Glan prism since in a predom-inantly molecular atmosphere the scattered light ispolarized either parallel or perpendicular to the sun'svertical.

Any of the four matched sets of half-wave plate andinterference filter can be brought into the light path bya switch closure either manually or by a telemeteredcommand. This renders possible the use of a singledetector, resulting in considerable reduction of un-certainty in calibration procedures. The polarimetercan scan in any predetermined plane through a variableangular range.

The system which weighs about 10 kg and measures25 cm X 25 cm X 25 cm is designed to operate on 28V dc to facilitate operation either from an aircraft or aballoon. The analog signal voltages corresponding tovarious polarimeter functions, such as the degree oflinear polarization, relative intensity variations, scan

Fig. 3. The photoelectric skylight polarimeter. A, collimator;B, fastening plate; C, lid; D, ac servo system; E, waveplateensemble; F, filter wheel; G amplifiers; H, ASCOP model541A photomultiplier tube; I, high voltage power supply; J,

quartz neutral density wedge.

February 1967 / Vol. 6, No. 2 / APPLIED OPTICS 223

-- 0

ALTITUDE 370 .000 60'

/ SOLAR ELEVATION 44-0 \ //OOA/i \ ~~~~~SCALE |' 20% POLON X

_ / < A ~LONO RADI US V ECTO R . TO\\\ 70

7y f - 08

ZETIITON.~ ~~~~~~~ 90

_0 '

50' 40' 30' 00' 10' 0' 10' 00' 30' 40' 50!

SCR REVERS Xs ;/SCAR REVERSE

60' \ \ ALTITADE 40,000 36,000 \ /\/ \/\ / SCAN 0959-1004 MST A / \ /~~~~~.111-1

SOLAR ELEVATION 40-30

/ \/ \XSCALE I . 20% POLXN

TO' ALN RADIAN VETO

:~~~~~~~

I0' 40- O- 20 10 0 10- 20- N- - SO

O.

COLOR RED X 12; \ SCAN REERSE

60' \ \ ALTITADE 36.0 37.000 \ / .\/ \A / ~~~~~SCAN 1004-lOON lST7 \ \/6

/\ /\ \ / ~~SOLAR ELEVATION 43'O00 \ /.\ -/ \/ \ X SCALE I'. 20% POLON

TO' / \ALONO RADIUS VECTOR T O'

6 0 ' 'C N I T V / ' .-l D V '

' i ~~-. ! X A\ /./\; ,- . -~~ \

o~~~~~~ -7- -

-7--

, ~ ~ , I

-7

-I1/ --I //

/1 /

50- C- O 20- 10- 0' I- 0' 33- 0'

Fig. 4. Specimen results. Degree of linear polarization. - - - Intensity (arbitrary units).

angle, d level and optics identification, can be recordeddirectly or may be interfaced with either frequencymodulation (FM) or pulse code modulation (PCIM)telemetry systems. The polarimeter is calibrated witha simple polarization calibrator based upon the polar-ization of specularly reflected light.20 Details of thepolarimeter are shown in Fig. 3.

B. The Balloon ExperimentAfter a careful survey of the frequency of occurrence

of favorable sky conditions (less than 0.1 cloud cover),Page, Arizona (36 057'N, 111 027'W, 1300 m above msl)was chosen as the balloon launch site. The principalpayload on the truncated, pyramidal gondola, charac-terized by a low center of gravity, consisted of thepolarimeter, a time lapse camera, a seven-channeloscillograph recorder as a backup system, a Barocoder,and telemetering equipment to receive and executecommands sent from the ground station and to transmiton the PCM telemetry system scientific and auxiliarydata to the ground station. On-board banks of silvercadmium and solar cells furnished the required power.

The polarimeter, encapsulated in 3.2-cm Styrofoam,was mounted onto the gondola in such a fashion thatit could scan through 270°00' symmetrically aboutthe local nadir. The objective of the Wollensak ModelTL 35 time lapse camera was pointed toward the nadir,

and the direction of the film transport was chosen to beparallel to the plane in which the polarimeter wouldscan. The color photographs, which were taken onceevery minute, have been utilized in two ways; namely,to determine the type and nature of the underlyingterrain and surrounding sky and to fix the azimuthalorientation of the scan plane with the aid of the knowntrajectory of the balloon and prominent topographicalfeatures. This was later used to check the perform-ance of the azimuth stabilization control.

The azimuth stabilization control, which would posi-tion the gondola so that the polarimeter would alwaysscan in the principal plane through the sun, was built byHi-Altitude Instruments of Denver, Colorado, for theNational Center for Atmospheric Research (NCAR),Boulder, Colorado. It was driven by a torque motorthat would position the relatively large inertial load.The error signal for azimuth stabilization was derivedfrom a bank of eight coarse eyes (photosensors) locatedat equal intervals on a ring frame. When the azimuthstabilization had been achieved, a pair of fine eyes,driven by a programmed cam, locked on to the sun.The limits on achievable tracking accuracies both inelevation and azimuth were set at 100'.

The 6750 m3 polyethylene balloon with a gross load of300 kg was launched at 0845 MST (Mountain StandardTime) on 30 A'arch 1965. The static launch technique

224 APPLIED OPTICS / Vol. 6, No. 2 / February 1967

SCAN REVERSE \ / COLOR RLAE X 36731 N 0 SCAN REVERSE

60-' \ \ ALTITUDE 07000 42000/

N\ / SOLAR ELEVlATION 3R'3I / \/ i X ~~SCALE I'* 20% POLT O

To' N \ X < ALON RA DIUS VECTOR -\

90- < -btfy I (~~~~~~~0

.7-

p /-- K � � // /

7 - 'I /

,! \�\2 / -� -PR--

--- �..T0bIT00.30.20' 10' 0' IA' 00' 30' 40' TO' .

. .

fO- fO-

.o.

80-

60-

.o.

I,

W.

.O.

TO-

.80 80

TO-

60-

jS BO-

"I I

60-60' I.0-

fO-

was adopted. The ceiling of 13 km was attained at arelatively slow ascent rate of 200 m/min. The balloonfloated at altitude for some time and then began todescend slowly. It suffered tropopause shatteringat 1019 MST.

IV. ResultsA comparison of the photographs obtained with the

time lapse camera and the trajectory of the balloonmapped by visual tracking and the chase plane indi-cated that the azimuth stabilization control had notfunctioned properly when it was intermittently switchedon while the balloon was ascending and while it was atceiling. Hence, the polarimeter scan was not confinedto the principal plane through the sun. This resulted inthe measured values of polarization and relative in-tensity variations in the four colors being not trulyrepresentative of the actual parameters to be associatedwith the incoming beam but only of the lower limitsthat can be set on these parameters. This ambiguitywas caused by the fact that the transfer functions ofsome phase sensitive networks in the circuitry had asensitive dependence on the angular separation betweenthe plane of polarization of the incoming beam and theplane of transmission of the Glan prism. Therefore,a rigorous comparison with available numerical com-putations would not be meaningful. Some of the re-sults are shown in Fig. 4.

The question as to how such measurements of thepolarization features of the upward radiation, whenavailable, could be used to derive information aboutthe composition of the atmosphere remains to be an-swered. It will now be possible, in the light of therecent solution by Sekera9'l 9 of the problems of inversion-and of radiative transfer in inhomogeneous media,to compute the polarization parameters of radiationemerging at different levels in the atmosphere. And, asin all geophysical phenomena, the model for which thecomputations yield the best fit with experimental resultswill be deemed as representative of the prevalent situa-tion.

The authors wish to take this opportunity to record

their sincere appreciation for Al Shipley of NCAR,Boulder, who directed the balloon operations. DanielDibble of the Department of Meteorology, UCLA andPaul Furukawa of NCAR, Boulder actively participatedin the flight program. Judith Phillips and CaroleSadler prepared the illustrations. Sherry Lovell isresponsible for the execution of the manuscript.

References1. B. Lyot, Ann. Obs. (Paris) 8, 1 (1929).2. A. Dollfus, thesis, University of Paris (1955).3. K. L. Coulson, J. V. Dave, and Z. Sekera, in Tables Related

to Radiation Emerging from a Planetary Atmosphere withRayleigh Scattering (University of California Press, Berkeley,1960).

4. Z. Sekera and W. Vizee, U. S. Air Force Project Rand ReportR-389-PR, The Rand Corporation, Santa Monica, 1961.

5. K. L. Coulson, private communication.6. Bruce Hapke and H. von Horn, J. Geophys. Res. 68,

4545 (1963).7. A. Dollfus, Physics and Astronomy of the Moon, Z. Kopal,

Ed. (Academic Press, Inc., New York, 1962), pp. 131-161.8. G. V. Rozenberg, Soviet Phys.-Doklady 8, 1 (1963).9. Z. Sekera, U. S. Air Force Project Rand Report R-413-PR,

The Rand Corporation, Santa Monica, 1963.10. Muneyasu Kano, Ph.D. Dissertation, Dept. of Meteorology,

Univ. of California (unpublished), 1964.11. K. L. Coulson, E. L. Gray, and G. M. Bouricius, Icarus 5,

139 (1966).12. K. L. Coulson, Appl. Opt. 5, 905 (1966).13. R. S. Fraser, J. Opt. Soc. Am. 54, 289 (1964).14. Z. Sekera, Advances in Geophysics (Academic Press, Inc.,

New York, 1956), Vol. III, pp. 43-104.15. Z. Sekera, in Handbuch der Physik, S. Flugge, Ed. (Springer

Verlag, Berlin, 1957), Vol. XLVIII, pp. 288-328.16. K. L. Coulson, J. Plan. Space Sci. 1,265 (1959).17. R. S. Fraser, Final Report, Contract NA 55-3891, TRW

Space Technology Laboratories, Redondo Beach, Calif.(1965).

18. Z. Sekera, Icarus 6 (1967).19. Z. Sekera, C. R. Nagaraja Rao, and D. Dibble, Rev. Sci.

Instr. 34, 764 (1963).20. C. R. Nagaraja Rao, Appl. Opt. 5, 1187 (1966).

Books continued from page 208

For this purpose, he limited the text to fundamental outlooksand the most important systems, and accepted simplifications insome cases where deeper knowledge is available. He succeeded,nevertheless, in creating a feeling for these advanced areas bypursuing them in some instances.

A novel and apparently rewarding approach is the method ofstarting (Chap. 2) and ending (Chap. 4) with the consideration ofstructure, making use in this resumption of the theme of theknowledge presented in the sandwiched chapters on properties.The discussion of the properties is concise, up to date, andsystematic. Viscosity, expansivity, density, refraction, me-chanical properties, strength, strain, hardness, electrical prop-erties, dielectric constant, dielectric loss, transmittance, sur-face tension, chemical resistivity, and thermal properties aretaken up in this order, with the general plan of consideringprinciples, measuring techniques, dependence upon composition,temperature, and heat treatment.

The limitation to 370 pages presents a formidable problem, butmakes the book readable and immediately useful for the purpose.In the case of single properties, this limitation made it necessaryto abandon some depth; and certain recent subjects like crystalfield theory are only briefly mentioned (but are well referenced),others like defect centers and radiation effects are sacrificed (butdefect theory is used in the discussion of melts and glass forma-tion).

However, considerable depth and a perfect awareness of pres-ent concepts is achieved in the essential concluding section on thenature of glasses. The book is the only comprehensive textgiving its due to the kinetic concepts which now dominate theunderstanding of glass; yet clear descriptions of the older geo-metrical and force rules are found. Recent ideas and data onphase separation are introduced briefly but adequately in thisconnection. The most outstanding progress in our knowledge ofthe fine structure of glasses has been caught in time as this and itsassociates on borate systems are referenced in the proper context.

The book is considered helpful for glass technologists engaged inteaching or studying the subject as a specialty on almost all levels,

February 1967 / Vol. 6, No. 2 / APPLIED OPTICS 225