Water vapor continuum absorption in the 35–40-µm region

Water vapor continuum absorption in the 3.5-4.0-,um region

Kenneth 0. White, Wendell R. Watkins, Charles W. Bruce, Robert E. Meredith, and Frederick G. Smith

Measurements of water vapor continuum absorption in the 3.5-4.0-gm region are presented. The measure-ments were made with both long-path absorption cell and spectrophone systems. A deuterium fluoridegrating tunable laser was the ir source. Measurements were made at 231C and 65 0C with 14.3 Torr and 65

Torr of water vapor, respectively, buffered to 760-Torr total pressure by an 80/20 mixture of N2 /0 2. Both

natural water and a special sample of deuterium depleted water (one-fiftieth the normal concentration) were

used. The 65°C results agree with previous measurements by other workers. The 23°C results indicate acontinuum absorption at this temperature about a factor of 2 larger than expected based on the extrapola-tion scheme and high-temperature data (265°C) of others.

Introduction

An accurate and detailed knowledge of atmospherictransmission in and near the ir spectral windows is es-sential to the design, performance evaluation, andcomparative testing of electrooptical (EO) systems. Forexample, atmospheric transmission is an importantfactor in the selection of optimum spectral bands for irimaging systems. The transmission in and at the edgesof the two major transmission windows (3-5 gm and8-12 gim) is most important since these are the bandsin which most ir devices operate. Also, absorbed ra-diation will reappear as thermal energy, thereby po-tentially leading to nonlinear effects. For example,propagation of laser flux may be limited in a stronglynonlinear fashion for sufficiently high power beams.

The generally low values of extinction in these spec-tral regions necessarily make absorption and scatteringcoefficients difficult to measure and model satisfacto-rily. A typical effect of the uncertainties in water vaporabsorption alone is illustrated by the historical devel-opment of broadband ir imaging systems, which aregenerally designed to operate in either the 3-5-Am or8-12-gm atmospheric windows. Early design studieswere based on atmospheric transmission measurementsby Yates and Taylor,1 which were reduced in such a way

R. E. Meredith and F. G. Smith are with Science Applications, Inc.,Ann Arbor, Michigan 48103; the other authors are with U.S. ArmyElectronics Research & Development Command, Atmospheric Sci-ences Laboratory, White Sands Missile Range, New Mexico 88002.

Received 2 December 1977.0003-6935/78/0901-2711$0.50/0.C 1978 Optical Society of America.

to omit the 10-gm water continuum. Thus, these designstudies predicted optimistically high performance forir imaging systems operating in the long wavelengthwindow, relative to systems operating in the 3-5-gmwindow.

While many constituents contribute to the total at-mospheric attenuation, the absorption by water vaporis of greatest concern for many practical situations,especially for long-path high-visibility conditions.There is hardly a region of the ir spectrum in whichwater vapor does not absorb appreciably. Moreover,of all the important gaseous absorbers, it is the leastunderstood and most difficult to measure quantita-tively. It is now generally accepted that a water vaporcontinuum absorption exists within the two major irwindows (8-12 gm and 3-5 gim). However, the ab-sorption coefficients are not known in either region withsufficient accuracy at present. In the long wavelengthregion, laboratory and field data exist for environmentalconditions, and a bound has been set for the criticalself-to-foreign-broadening ratio.2 In the short wave-length region, no continuum measurements have beenpublished for conditions typical of the natural envi-ronment. In lieu of data, continuum absorption hasbeen extrapolated from high temperature laboratorymeasurements.3

In this paper we are concerned only with water vaporcontinuum absorption in the 3.5-4.0-grm spectral region.Long path (White cell) measurements on high tem-perature samples are reported; and for the first time, wepresent laboratory data obtained on samples repre-sentative of environmental conditions.

The following sections present background and an-alytical procedures, describe the experimental tech-niques, and present the measurement results.

1 September 1978 / Vol. 17, No. 17 / APPLIED OPTICS 2711

Present Understanding of the Water Vapor Continua

In the following discussion, transmittance is taken tobe of the form r = exp(-kL), where k is an absorptioncoefficient expressed in km-' and L is the pathlengthin km.

The available continuum absorption coefficientmeasurements data in both the long and short wave-length regions have been fit to a functional form of thetype2 ' 4

kc(v,T) = n[Cs(vT)p, + Cf(v,T)pf], (1)

where n, is the number of water vapor molecules percm3. C and Cf have units cm3 (km atm molecule)-1and are frequency () and temperature (T) dependentempirical parameters which quantify, respectively, theself- and foreign-broadening contribution. Ps and pfare, respectively, self (i.e., water vapor) and foreign gaspartial pressure in atms. Often, the empirical param-eters are rearranged and expressed as a self-broadeningcoefficient and a self-to-foreign-broadening coeffi-cient:

kc(v,T) = nC 8(v,T) [Ps + B*(t T ' (2)

where B*(v,T)- C(v,T)/Cf(v,T). These forms closelyrelate to the interpretation of the continuum as arisingfrom the wings of distant strong lines, since early line-broadening work5 often was directed toward measuringa ratio of self-to-foreign line width parameters B-Ysbiyf. For Lorentz lines, this interpretation is shownas follows. Far from the center of the ith absorptionline, - >> y and

sins-yi

where S and v' are, respectively, the strength andcenter frequency of the ith absorption line. Since yiis the sum of self- and foreign-broadening contribu-tions,5

' =Y!Ps + YPf. (4)

Then substituting Eq. (4) into Eq. (3),

kwing(v) nS (yP + yPf)* (5)

The continuum parameters may then be related to lineparameters as follows:

Cs () 37._i 2

Cf() si2. (6)

Furthermore, if the ratio -y'/-y is constant across theband(s), as is often assumed for approximate work, aself-to-foreign-broadening efficiency B may be de-fined:

B = Iy/.(7)

In this case, for Lorentz line shapes, the line-broadeningefficiency and the self-to-foreign-broadened continuumratio have identical values:

(Cs/Cf) - B. (8)

In general, however, B is not constant across the mo-lecular band(s); and moreover, either or both the self-and foreign-broadened line shapes are not expected tobe Lorentzian. Therefore, even though the interpre-tation of C/Cf as B is tempting, it can be correct onlyunder very special circumstances (e.g., identical self-and foreign-broadened line shapes). Indeed, significantdeviations of Cs/Cf from the expected B value of a bandimplies (for the line wing interpretation of the contin-uum) that the self-broadened and foreign-broadenedline shapes are not the same.

If the interpretation of the continua as far wings ofLorentz lines is correct, measured absorption valuesshould agree at least quantitatively with line-by-linecalculations, within accuracy of the measured SO and,yi parameters. This is not the case6 in the 8-12-gumregion, where observed absorption is much greater thanpredicted for Lorentz lines. Thus the question persistsas to whether non-Lorentz line wing shapes for eitheror both self- and foreign-broadening cause the contin-ua.

Nevertheless, the magnitude and nature of themeasured continua values do carry definite implica-tions. In the 10-gm region, the ratio Cf/C is measuredto be at most 0.005, with a lower bound of Cf/Cs ap-proaching zero.2 If the lines were even nearly Lorentzthe effective ratio Cf/C8 is at least an order of magnitudetoo small. For this reason, dimer theories7 and clusterexplanationsf have been suggested.

In the 3.5-4.1-gm region, the magnitude of the ab-sorption is compatible with what one would expect ifslightly non-Lorentz shapes caused the discrepancy.Burch3 reports a ratio Cf/C 0.12, a value fairly con-sistent with the value of -ys/yf 5, which is character-istic of water vapor-N2 mixtures.9 In this region, then,an interpretation of distant line wings, for which theself-broadened shape may be slightly more super Lo-rentz than the foreign-broadened shape, appears morereasonable.

At the Ohio State University, some 23°C total watervapor absorption data have been obtained10"' fromwhich one might infer the presence of a room temper-ature continuum. These data were obtained with a DFline tunable source and water vapor on mixtures pre-pared in a White cell.

The most extensive experiments directed towarddetermining a water vapor continuum in this regionprior to the work reported here were elevated temper-ature measurements (65°C) performed by Burch etal. 3 Extrapolations were made from the high-tem-perature data to 230C and to the spectral region be-tween 3.5 ui and 3.7 im. A value of Cf/Cs = 0.12 wassuggested; and, lacking temperature-dependent for-eign-broadened data, the temperature dependence of

2712 APPLIED OPTICS / Vol. 17, No. 17 / 1 September 1978

Cf was assumed to be that of C. This issue is impor-tant since the data show a strong temperature depen-dence.

Analytical Procedure

Attempts to develop from laboratory data empiricalmodels of the 3-5-gm continuum absorption by atmo-spheric water vapor over the extremes of geographicaland seasonal conditions are complicated by the complexnature of the absorption spectrum. In principle, thereis a water vapor continuum contribution for everybroadening agent with which the vapor is mixed; inaddition, a pressure induced nitrogen continuum issuperimposed over the water vapor continuum ab-sorption. Also, HDO and H2 0 line absorptions aresuperimposed over both continua.

In performing quantitative investigations, thisbackground absorption must be subtracted from watervapor spectra to determine the water vapor continuum.Line absorption has a strong spectral dependence,which is usually measured by using high-resolutionscanning spectroscopy and line parameter extractiontechniques. Line absorption for arbitrary conditionsmay then be calculated by line-by-line techniques withan accuracy determined by the accuracy of individualline parameters and by the assumption of far wing lineshape. The techniques for determining continuumabsorption are quite unlike those required for deter-mining line absorption. Continuum absorption doesnot have spectral fine structure; consequently, high-resolution and spectral scanning are not required.High-sensitivity absorption measurements are required,however, since the continuum absorption is weak andcannot be easily increased by enrichment of the sample.Ideally, the wavelengths chosen to measure continuumabsorption should have minimum line contributions tominimize inaccuracy induced by background extraction.In this work, we have eliminated the pressure inducednitrogen continuum experimentally by a ratio techniquewhich is discussed later. In one series of measurementsthe primary source of line background (HDO) waseliminated experimentally by using deuterium depletedwater. The analytical procedure used to subtract linesfrom natural water vapor data are discussed in the fol-lowing paragraphs.

The background line absorption coefficient k(v) isassumed to be a superposition of Lorentz line pro-files:

1 S'n,'ylk (P) =-E 57, i°)2+ 0 i2 *(9)

ir i ( - Pi)S + (-Yi)2

In the above, SO is line strength per water vapor mole-cule, y is the air (80% N2 + 20% 02) broadened Lorentzwidth, and n, is the number per unit volume of thewater vapor component (H2 0 or HDO). The HDOnumber is taken to be 0.03% of the more abundant H20concentration, as is assumed in the Air Force Geo-physics Laboratory (AFGL) data compilation.12"13

The linewidth y is the sum of contributions from self-and foreign-collision partners:

ly= 1YPs + Y'P. + 'YbPb + * * , (10)

where the subscript s refers to self-broadening (H2 0-H2 0 or HDO-HDO collisions), and the subscripts, a,b,... refer to broadening of HDO and H2 0 lines byoxygen and nitrogen and by each other. In the presentwork, line broadening is caused by a mixture of foreigngases having a fixed ratio of a given foreign gas partialpressure to the total foreign gas pressure pf:

(11)paIpf = a; PbIPf = b; etc.

Consequently,

ly= =Pse + pf(ay' + b' + ),

lyi PsYs + Pfi

In analyzing the present data, broadening by N2 and 02individually was not considered. For HDO lines, air-broadened Lorentz widths yf measured by ScienceApplications, Inc. (SAI), were used.14 For H2 0 lines,values listed in the AFGL data compilation were used.13Broadening of HDO lines by H20 and vice versa wasassumed to be self-broadening, even though this as-sumption probably represents an upper (maximum)bound. The large value of the H2 0 dipole momentwould seem to justify this assumption since dipole-dipole broadening varies as the product of the squaresof the dipole moments, whereas resonance effects aregenerally of lesser importance. Accordingly, to accountfor self-broadening, a ratio of B = 5 was used for all H2 0and HDO collisions with air molecules.

The temperature dependences of -y, and yf dependon the details of the collision process and on the popu-lation of the collision partners. If long-range forces areprimarily responsible for the broadening, impact theory(e.g., Anderson theory)15 may be used to predict atemperature dependence for each line. However, thisis often not the case, particularly for foreign broadeningof lines having large rotational energy. Lacking dataon the lines in question and seeking to maintain uni-formity with work performed elsewhere, we have as-sumed the usual inverse square root temperature de-pendence.16

Within the approximations described above, data onnatural water vapor broadened by N2/02 were analyzedby subtracting line absorption by using high-resolutiontransmission codes developed by SAI. Line wing con-tributions beyond 20 cm-' from line center were negli-gible, so all calculations were truncated at this value.No attempt was made to separate self- and foreign-broadening of the continuum. It is noted, however, thatverification of the ratio measured by Burch as well asconfirmation of the value of C8 is of considerable prac-tical importance, since EO systems performance issensitive to these values, except for the extremes ofenvironmental conditions and slant range applicationsfor which molecular absorption is not important.

The validity of this analytical approach was con-firmed early in the measurements program, when 65 0C


(12)

DF transmission measurements near local H20 linesindicated strength values much smaller than were listedin the AFGL data compilation at that time. More re-cently, the computation was updated, independentlyof this work, thereby giving good agreement with lineabsorption measured in the current work. Also, goodagreement with HDO contributions is now obtainedwith the 1976 AFGL compilation,'3 since it has beenupdated to include the latest HDO line parametermeasurements. Measurements of D/H ratio in thewater vapor samples were not performed, but mea-surements performed on deuterium depleted watervapor samples are consistent with the data taken withnatural water vapor.

Experimental Approach and Techniques

The experimental equipment includes absorptioncells, both long-path and spectrophone, and line tunablepulsed lasers. The DF laser sources used for both thelong-path absorption cell and spectrophone measure-ments are grating tunable pulsed Lumonics multigaslasers. The lasers each are operated with an unstableresonator front reflector resulting in a doughnut-shapednear-field TEM-00 output beam.

Fig. 1. Experimental arrangement.

Long-Path Absorption Cell SystemThe experimental setup for the DF laser measure-

ments using the long-path cell is shown in Fig. 1. TheDF laser beam is made coincident with a visible He-Nelaser beam by using several flat mirrors M and onemirror M' with a central hole. Two precision-adjust-able lenses are inserted in the He-Ne beam to match theeffective divergence of the two lasers. The DF outputwavelength is checked by inserting a lens and spectrumanalyzer into the DF laser beam. After the DF laserbeam traverses the length of the laboratory, it is col-lected by mirror CM1. Mirror CM2 is used to collimatethe DF and He-Ne laser beams to 5-mm diam beforeentering the 21-m absorption cell. An optical flat B isused to split a portion of the beam to the cell input de-tector ID. The remaining portion of the beam passesthrough window W. A multipath through the cell isobtained by using White-type reflective optics. Thecell output beam is detected by the cell output detectorOD. Teflon membrane filters F are placed in front ofthe detectors to diffuse the laser beam. Apertures Aare placed in front of the detectors to prevent detectorsaturation. A more detailed account is given else-where.17

For the water vapor absorption measurements, aknown amount of water was introduced into the 21-mcell under vacuum and allowed to evaporate. Theequilibrium pressure was measured by a capacitancemanometer prior to injecting an 80%/20% mixture ofN2/0 2 to buffer the water vapor to 760-Torr totalpressure. After buffering the water vapor partialpressure was monitored by using a dewpoint hygrome-

ter. Measurements were performed at two tempera-tures: 230C and 650C. The reference cell conditionrequired for long-path cell measurements was a 760-Torr 80%/20% mixture of N2/0 2 with a correspondingamount of water vapor pressure replaced by 02 since N2absorbs at several DF frequencies. Additional mea-surement technique details are given elsewhere.18"19

The signal analyzing system for the long-path ab-sorption cell is described in the literature.' 8 The DFlaser pulses are submicrosecond in duration, and thepulse repetition frequency is 0.3 Hz. Both the cell inputand output signals are amplified to a nominal 7-V levelfor digitization. The cell input signal is then appro-priately delayed to match the optical delay experiencedby the beam propagated through the absorption cell.Each signal is fed into a 14-bit analog-to-digital con-verter and is pulse-height analyzed. The two digitizedsignals are then sent to a minicomputer controlled an-alyzing system where they are displayed and analyzedby computer programs. The real-time display of thecell input and cell output signals allows detector-amplifier nonlinearities as well as optical alignment tobe corrected during measurements. In addition a newtechnique, path differencing, is used in data acquisitionto obtain nearly time independent results.' 9 Thetechnique uses rapid changes between two pathlengthsto time average short-term and eliminate long-termsystem drift normally encountered in long-path ab-sorption cell measurements. Path differences of from1008 m to 1512 m were used.


CMI

Spectrophone System

In the spectrophone system as with the long-path cell,a He-Ne laser beam is made coaxial with the DF laserbeam. In this case though, a calorimeter is used fordetermination of the laser pulse energy. The pulsedsource spectrophone used for these water vapor mea-surements consists basically of a semiclosed cylindricalchamber with a dominant longitudinal acousticallyresonant mode near 1 kHz (Fig. 2). This chamber ispositioned within another cylindrical chamber made ofstainless steel which is evacuable and bakeable. Thespectrophone entrance and exit windows are located inthe ends of this outer cylinder and are thus well removedfrom the entrance and exit apertures of the interiorresonant chamber. This arrangement and acousticaldampers serve to avoid contributions to the spectro-phone output signal due to the window absorption. Analuminum coated Mylar diaphragm, forming a completecircular cylinder at the perimeter of the inner chamber,serves as a capacitance microphone sensor for the ab-sorption signal. The time resolved signals from anumber of laser pulses were averaged on a point by pointbasis by using a PAR signal "eductor," and the devel-oping signal was displayed on an oscilloscope andplotted for analysis. The set of relative absorptionvalues was then referenced to a known absorption usingmethane as a trace gas in the same N2 /0 2 buffer gas.Both physical design of the pulsed source spectrophoneand methods of signal treatment have been discussedelsewhere.2 0 ,2'

Several spectrophone measurements were made todetermine whether (1) extraneous sources or (2) im-purities contribute to the water vapor absorption signal.Pure nitrogen was used to verify the absence of extra-neous sources since nitrogen absorbs in a continuumwhich is very weak at the shorter wavelength end of theDF laser spectrum (<10-4 km-1 at 2800 cm-'). Noabsorption was detected; therefore, we conclude that noextraneous sources exist. The approach with regard tothe impurities was less direct. The same singly distilledwater (metallic still) was used for both spectrophoneand White cell measurements. In addition, several setsof spectrophone measurements were made for each oftwo other water samples which were prepared in a spe-cially designed, all-glass still. This still forms a closedsystem and is purged with argon during the distillationprocess. The two samples prepared in this still weresingly and multiply distilled. The results were notsignificantly different for the three conditions in spiteof the differential distillation of the components.

Discussion of Experimental Results

65CC Water Vapor Absorption

As stated above, measurements were made at tem-peratures of 650 C and 230C. The 650 C measurementswere made to allow a comparison with the water vaporcontinuum results of Burch et al.3 near their experi-mental conditions. A direct comparison of the totalwater absorption as measured here by using the White

ISOLATEDMICROPHONE CAVITY

Fig. 2. Schematic drawing of pulsed source spectrophone.

E

zwILILw0

0F

0C')

2800FREQUENCY (cm-1)2700 2600 2500

1.0 _

.9 0 0 MEASUREMENTS THIS WORK

.8 + 4 PREDICTIONS

.7

.6

.5

3 + 0~~~~~~~

.2 _4.* 8

0 LASER LINES

3 4 5 6 7 8 9 10Pueo" ' ' I ' ' I I

3 4 5 6 7 8 9 10 1IP2

1 ' ' ' ' '

4 5 6 7P3-+2 ' ' I .

8 9 10 11

Fig. 3. Total water vapor absorption (line plus continuum) mea-surements and predictions at 650C with 72-Torr water vapor buffered

to 760 Torr.

cell system with the latest calculation13 is given in Fig.3. The over-all agreement between the measurementand the prediction is good. The disagreements betweenthe two results occur mainly in the region near 2800cm-I, where the HDO line absorption is large.

This set of measurements has been used to infer thewater continuum. The inferred values were obtainedby subtracting calculated HDO and H20 line contri-butions from the measured total water absorptionvalues. The results are shown in Fig. 4 and Table Ialong with the continuum predicted from Burch'shigh-temperature measurements. The present resultsand the Burch prediction are in substantial agreementbetween 2450 cm-' and 2700 cm-'; however, the in-ferred continuum values have considerable scatter be-yond 2700 cm-'. The large apparent scatter in the in-ferred continuum does not reflect measurement preci-sion; the precision is indicated by the error bar in Fig.4. The scatter is caused by uncertainty in the line ab-sorption coefficients which are subtracted from themeasured data to infer the continuum. At somefrequencies, the line absorption dominates the total(line plus continuum) absorption as indicated in TableI. Burch did not directly measure the self-broadenedcontinuum at this temperature beyond 2650 cm-1. Byusing his measured results between 2400 cm-' and 2650cm-1 and extrapolation scheme (discussed later), the


. . . .

FREQUENCY (cm- 1)E

I-zIW

0UL.w0

FL

0a: 0CD,

2800 2700 2600 2500.30 I II -

25 - . 0 0 THIS WORK

20 _ BURCH RESULTS -

.15 s . o, ___ 0 0

.10 0 - 00 00 0 0

.05 _00L

LASER LINES

3 4 6 8 9 102 'I 3 7 9 101,

3 4 45 6 7 8 9 10 11P3 2t , , , , I I I

Fig. 4. Inferred continuum this work at 650C with 72-Torr watervapor buffered to 760 Torr compared to Burch results. Burch resultsare as follows: Solid curve based on 65°C self-broadened data andhigher temperature foreign-broadened extrapolation; dashed curvebased on extrapolation of higher temperature self- and foreign-

broadened data.

Table I. Water Vapor Absorption Coefficients (km 1) X 103 with 72-TorrWater Vapor at 65°C Buffered to 760 Torr

.~~~~~~

SL4 - - 1L'CU -' '

Ql ' 4JW; 0 OCO ) 4 E

a L 3 o ' 3 CL 3 L0 :aj CD I IlUM 0 0 4

'-e 'a C) OW D- 0.0 c'. '.C T S- aJ J ~~~~~~~~~~~~~~~~~3 : F :3

P1(3) 2839.796 564 198 6 360 138P1(4) 2816.385 410 222 13 175 131P,(5) 2792.434 264 165 2 97 123P1(6) 2767.968 313 228 1 84 115P2(3) 2750.096P,(7) 2742.988 152 25* 0 127 107P2(4) 2727.312 218 114 2 102 101P1(8) 2717.543 883 662* 1 220 98P2(5) 2703.993 185 46 0 139 93P1 (9) 2691.605 145 62* 5 78 89P2 (6) 26800173 227 195 3 29 86P1(10) 2665.218 221 96 0 125 82P2(7) 2655.861 424 367* 0 57 80P3(4) 2640.075 236 164 0 72 77P2(8) 2631.067 124 17 44 62 75P3(5) 2617.389 94 23 0 71 72P2(9) 26050808 267 187* 0 80 70P3(6) 2594.201 190 83 4 103 69P2(10) 2580.102 120 29 0 91 69P3(7) 2570.523 111 64 1 46 70P2(11) 2553.954 83 8 0 75 72P3(8) 2546.375 92 20 1 71 74P3(9) 2521.769 95 3 1 91 81P3(10) 2496.722 96 1 8 87 90P3(11) 2471.245 137 0 33 104 100

-ThnoZudes updated HDOvalues from references 14 and 22.

2716 APPLIED OPTICS Vol. 17, No. 17 I 1 September 1978

Table II. Natural Water Vapor Absorption Coefficients (km-1 ) X 103 with14.3-Torr Water Vapor at 23°C Buffered to 760 Torr

C~~~~~~jLa 4. O - L 4.3

-O a L O CS n WEE 3 ) 0

= _ < 3 I *_ C 3 = <~~~C 1

InC > Ea 4 - *- 0CC 4)C*_- 1.4' *I)We'a.' *_ C. OW. 0 1 NC' aO. o 'aeo o D )-s-i - s-z.- ~~~~..icS S- S-c .

JJ ~ ~~~ ~~~ 3 - O nO 30a) InD 5

P1 (3) 2839.796 145 53 92 39

Pj(4) 2816.385 143 48 95 36

P1 (5) 2792.434 95 35 60 33

Pj(6) 2767.968 111 51 60 31

P2 (3) 2750.096 63 20 43 29

P1(7) 2742.988 69 5* 64 28

P2 (4) 2727.312 69 24 45 27

P,(8) 2717.543 167 120* 47 26

P2 (5) 2703.993 61 7 54 24

P1 (9) 2691.605 66 14* 52 23

P2(6) 2680.173 93 43 50 22 75

P(10) 2665.218 78 20 58 21

P2(7) 2665.861 129 78* 51 20 103

P3(4) 26400075 78 36 42 19

P2 (8) 2631.067 44 8 36 19 38

P3 (5) 2617.389 44 4 40 18

P2 (9) 2605.808 67 32* 35 17

P3 (6) 25940201 64 13 51 17 59

P2(10) 2580.102 43 4 39 17

P3 (7) 2570.523 63 8 55 18 35

P2 (11) 2553,954 52 1 51 18

P3 (8) 2546.375 41 3 38 19 29

P3 (9) 2521.769 41 0 41 21

P3 (10) 2496.722 44 1 43 23

P3 (11) 2471.245 38 4 34 26

*IncZudes updated HDO vaZues of SAI from references 14 and 22.

continuum between 2650 cm-1 and 2820 cm-1 can bemodeled.3 The error limit on the Burch predictions isJ50%.

230C Water Vapor Absorption

Natural water vapor absorption coefficients havebeen measured in both the White cell and in the spec-trophone, while deuterium-depleted water vapor valueshave been measured in the White cell. The naturalwater vapor absorption coefficients measured in theWhite cell for 25 DF laser lines are given in column 3 ofTable II. These results were obtained at 23°C with awater vapor pressure of 14.3 Torr and buffered to a totalpressure of 760 Torr by an 80%/20% mixture of N2/0 2.Since the absorption was referenced to that of an80%/20% mixture of N2/.02, the N2 pressure inducedcontinuum absorption was subtracted out experimen-tally. The above corresponds to the midlatitude sum-mer (MLS) atmospheric model, which has become thestandard for such measurements. Table II also showscalculated values for the various component absorptioncontributions. The theoretical line absorptions andcalculations are based on AFGL absorption line pa-rameter compilations' 3 (column 4 of Table II). HDOcalculations (indicated by asterisks) have been modified

based on recent SAI spectral measurements.'4 ' 22 Thecontinuum predictions (column 6 of Table II) are basedon Burch's measurements.3 Since his measurementswere made between 65°C and 155°C, he obtained thecontinuum values by extrapolating the higher temper-ature data to 23°C.

The extrapolation procedure used by Burch et al.3 toobtain the 23°C air-broadened water vapor continuumis described completely in ref. 3. Since this referenceis not readily available the procedure will be discussedbriefly. Several wavelengths were chosen in the 3.5-4.1-gm region where normal water absorption was aminimum. Absorption measurements were made atelevated temperatures of 65°C, 1110C, and 1550C be-cause sufficient water vapor resulting in measurableabsorption could be introduced into their absorptioncell. Pure water vapor absorption (i.e., with no buf-fering by other atmospheric gases) was measured as afunction of water vapor pressure. The results weregiven in terms of the self-broadened water continuum(assumed to be far wing) empirical parameter CS whichis related to the absorption coefficient k as follows [seeEq. (1)]:

k = nCsp, (13)

where p is the water vapor pressure. k/n, was foundto vary linearly with water vapor pressure, and Cs wasfound to be proportional to exp(constant/0), where isthe temperature in Kelvin. Next, the foreign-broad-ened water continuum absorption coefficient CN2 wasmeasured. Measurements were made at 1550C with 2atm of water vapor buffered by nitrogen to 4.5 atm, 5.7atm, 7.5 atm, and 10 atm. From these measurements,a measure of CN2 was obtained which relates to the self-and foreign-broadened water vapor continuum ab-sorption coefficient as shown in Eq. (1). Assuming thenitrogen broadening to be the same as oxygen broad-ening, the water vapor continuum absorption due toatmospheric water vapor can be obtained by extrapo-lation of the above measurements. The results ofBurch's measurements and his extrapolation scheme(i.e., the extrapolated value of Cs at 230 C with the as-sumption that the ratio of CN2to C8 at 1550C is the sameas that at 230 C for 14.3 Torr of water vapor in a 760-Torr total pressure atmosphere) are shown by thedashed curve in Fig. 5 with an estimated :65% errorlimit. A similar approach, with regard to the foreignbroadening, was followed for Burch's 65°C water vaporabsorption data presented earlier.

Inferred continuum absorption coefficients from thepresent measurements are calculated by subtractingtheoretical HDO and H20 line absorption coefficientsfrom measured total water absorption coefficients.These are given in column 5 of Table II.

A more direct measure of the water vapor continuumwas made by using HDO depleted water.2 3 '2 4 Watervapor absorption in this spectral region is comprised ofthe water vapor continuum, substantial HDO line ab-sorption, and weak H20 line absorption. By using anHDO depleted water sample the need to subtract ab-sorption line contributions was eliminated for about

- .11

~ .10b9

-~ .09z .08c) .07

06

8 .05

0g JD4I, .03

6 .02

M .01

2800FREQUENCY (cm')2700 2600 2500

LASER LINES

3 4 5 6 7 8 9 10

3 4 5 6 7 8 9 10 11P2-#1 '''

4 5 6 7 8 9 10 11P3' ' '''''Fig. 5. Comparison of measured water continuum at 230 C with

14.3-Torr water vapor buffered to 760 Torr.

Table I. HDO Depleted Water Vapor and Water Line AbsorptionCoefficients (km-1 ) X 103 with 14.3-Torr Water Vapor at 230C Buffered to

760 Torr

HDO + H20

HDO Depleted Water Line' 3

Laser Frequency Vapor Absorption AbsorptionLine (cm'1) Coefficients* Coefficients

P1(3) 2839.791 117 ± 2 -

P1(4) 2816.380 128 ± 4 3

P1(5) 2792.434 91 ± 15 1

Pj(6) 2767.968 82 ± 10 1

P2 (3) 2750.093 76 ± 7 -

P1(7) 27420997 65 ± 12 -

P2 (4) 2727.308 90 ± 6 1

P 1 (8) 2717.538 81 ± 6 2

P2 (5) 2703.998 70 ± 5 -

P,(9) 2691.608 50 ± 2 1

P 2 (6) 2680.178 53 ± 9 1

P1 (10) 2665.218 46 ± 18 -

P2 (7) 2655.863 54 ± 7 2

P,(4) 2640.075 30 ± 3 1

P2 (8) 2631 .066 33 ± 10 5

P3(5) 2617,386 47 ± 4 -P2 (9) 2605.806 35 ± 6 1

P3 (6) 25940197 49 ± 11 1

P2(10) 2580.095 25 ± 13 -

P,(7) 25700522 43 ± 10 -

P2(11) 2553.951 42 ± 7 -

P3 (8) 2546.373 33 ± 10 -

P3 (9) 2521.769 43 ± 11 -

P3 (11} 2471o243 74 ± 6 4

*The uncertainties shoun are RMS measurement uncertainties, i.e., theprecision of the measurement.

one-half of the laser lines used and was only a few per-cent of the total absorption for the remaining lines. Ananalysis of the deuterium depleted water sample indi-cated 2% the natural HDO concentration.25 The resultsof these measurements with twenty-four DF laser linesare given in Table III. HDO absorption has beenmeasured2 6 with a spectrophone, and these measure-ments show general agreement with the predictions.


0\ ~~MEASUREMENTS

\ o~~~ PRESENTo\ ~+ OSU\ ~MODELS

- o >s~ - PRESENT0 0 -- BURCH EXTRAPOLATION -

_ -_ _

o\V | , a

Table IV. Water Vapor Continuum Absorption Coefficients (km- 1) X 103for 14.3-Torr Water Vapor at 230 C Buffered to 760 Torr

Natural Natural HDO DepletedLaser Water Vapor Water Vapor Water VaporLine Spectrophone* White Cell* White Cell* Average*

P,(3) 92 + 4 117 ± 2 105 ± 18

P,(4) 95 ± 10 125 ± 4 110 ± 21

P1 (5) 96 ± 5 60 ± 7 90 ±15 82 ±19

P1(6) 68 ± 5 60 ± 4 81 ± 10 70 ± 11

P2 (3) 43 ± 10 76 ± 7 60 ± 23

P,(7) 78 ± 4 64 ± 6 65 12 69 ± 8

P2 (4) 62 ± 6 45 ± 5 89 ± 6 65 ± 22

P1(8) 75 ± 7 47 ± 11 79 ± 6 67 ± 17

P2 (5) 57 ± 3 54 ± 7 70 ± 5 60 ± 9

P1 (9) 48 ± 2 52 ± 8 49 ± 2 50 ± 2

P2(6) 49 ± 3 50 ± 8 52 ± 9 50 ± 2

P5(10) 28 ± 2 58 ± 8 46 ± 18 44 ± 15

P2 (7) 48 ± 5 51 ± 6 52 ± 7 50 ± 2

P3(4) 19 ± 2 42 ± 10 29 ± 3 30 ±12

P2(8) 38 ± 4 36 ± 7 28 ±10 34 ± 5

P3(5) 42 ± 3 40 ± 3 47 ± 4 43 ± 4

P2 (9) 37 ± 4 35 ± 9 34 ± 6 35 ± 2

P3(6) 46 ± 2 51 ± 11 48 ± 11 48 ± 3

P2 (10) 39 ± 4 39 ± 20 25 ±13 34 ± 8

P3 (7) 39 ± 2 55 ± 10 43 ±10 46 ± 8

P2 (l1) 51 ± 8 42 ± 7 47 ± 6

P3 (8) 36 ± 3 38 ± 10 33 ±10 36 ± 3

P3 (9) 40 ± 3 41 ± 7 43 ±11 41 ± 2

P3 (10) 43 12 43 ±12

P3 (1) 34 ±15 70 ± 6 52 ±25

*The uncertainties shown are RMS measurement uncertainties, i.e., theprecision of the measurement.

Thus, for an HDO depleted water vapor sample, thewater vapor continuum level is not significantly alteredby the decrease in the HDO molecules.

The total natural water vapor absorption was alsomeasured with a spectrophone. From these results, aninferred continuum was obtained in the same manneras that used for the White cell measurements. Thespectrophone system was calibrated by using knownconcentrations of methane and measured methaneabsorption coefficients. 21 27

The inferred and directly measured (HDO-depleted)water vapor continuum values at 23°C are given inTable IV with the rms uncertainties. The uncertaintiesshown for the average values listed in column 5 of TableIV represent typical measurement set rms reproduci-bilities. The uncertainties of the average values forlaser lines at the beginning and ending of series arelarger than others probably due to weaker laser emis-sion. The average of the three sets of data is shown ascircles in Fig. 5 along with the continuum extrapolatedfrom the Burch measurement and published measure-ments by the Ohio State University (OSU) group11' 28

(see column 7 of Table II). The solid curve in Fig. 5 isa least squares fit with an estimated 25% uncertainty.The present results indicate a water continuum roughlytwice as large as that of Burch near 3.8 gm and about afactor of 3 greater near 3.5 gim. The inferred OSUcontinuum values indicate a continuum level slightlyless than present values.

The Burch measurements of water vapor continuumdid not include an analysis of how the expected errorbounds on the self- and foreign-induced coefficients (C8and C) influence the error in the total water continuumabsorption coefficient kc. A Monte Carlo techniquewas used for evaluating the combination of the variouserrors, including the effects of temperature extrapola-tion.2 9

Briefly, the Monte Carlo technique employed selectsstatistically, from appropriate distribution functions,values for all the variables which have an associateduncertainty. The values for these variables are thencombined, according to the governing equation, toevaluate a statistical sample for k0 This procedure isrepeated several hundred times (each time using newrandomly generated numbers) until sufficient statisticsare accumulated to evaluate accurately the total un-certainty in kc.

Results of this analysis on Burch's data and a com-parison with present results are shown in Fig. 6. Thepresent values are well above the closest error bound.It is concluded that the high-temperature measure-ments agree quite well with the Burch measurements,but the 230C extrapolation of Burch underestimates thecontinuum by at least a factor of 2.

Impact of Results on Modeling

An understanding of the water vapor continua in boththe 3.0-5.0-gm and the 8.0-12.0-gm regions is impor-tant for predicting performance of high-energy laser(HEL) devices and for performance evaluations ofbroadband systems.

The significance of determining accurately the water

T .10E

I-z .08

U.)8 D

4 .02

0

2800FREQUENCY (cm-")2700 2600 2500

LASER LINES

3 4 5 6 7 8 9 10

3 4 5 6 7 8 9 10 IP2.I ' - ' I I I '

4 5 6 7 8 9 10 IP3--2 ' ' ' - I 1 I

Fig. 6. Comparison of water continuum absorption measurements:Burch's uncertainty bounds and this work measurements. Tem-perature = 231C, pressure H20 = 14.3 Torr, total pressure = 760

Torr.


0 50

0 0 THIS WORK- BURCH FIT-- BURCH UNCERTAINTY

MEASUREMENT TEMPERATURE0 O A' 65'C

. O o ~~C1 15S°CC~~~~~~BII

0 0 a 55eO

C% 0 0~~~~~~~~~~~~~~~b

0 0 0 00

'0AidC C~~ 00

17 - . . . . . . . s

FREQUENCY (CM-I)2800. 2700. 2600. 2500. 2400. 2300. 2200. 2100.

1.00

0.80 X

0.60 II I I J I.0

3.7 3.9 4.1 4.3 4.5 4.7WAVELENGTH (MICROMETERS)

Fig. 7. Calculated spectral atmospheric transmission for variouswater continuum models: tropical model atmosphere, 8.5-km visi-bility, 0.75-km altitude, 5-km range: (a) zero continuum; (b) con-tinuum extrapolated from measurements of Burch; (c) continuum

measured this work.

vapor continuum absorption in the 3.0-5.0-gm regionis indicated in analysis of performance of thermalimaging systems. Comparisons have been made ofrelative performance between forward-looking ir (FLIR)devices designed to operate in the 3.0-5.0-gm and8.0-12.0-gm windows.30 Transmission predictions, andtherefore FLIR performance, are impacted by themagnitude of the self- and foreign-broadening coeffi-cients, their relative magnitude, and their temperaturedependence. The effect at the shorter wavelengthwindow is shown in Fig. 7 where various continuamodels are used in the propagation model LOWTRAN3A.31 The results show the transmission without con-tinuum absorption, with the Burch extrapolation, andwith the present results. As can be seen there is sig-nificant effect on the transmission depending on whatvalue is used for the continuum absorption. For theconditions considered in Ref. 29, the difference betweenthe Burch continuum value and the present values hasa factor of 5 effect on the relative performance of a 3-5-gm FLIR as compared to an 8-12-gm device.

SummaryThis paper has presented measurements of water

vapor continuum absorption in the 3.5-4.0-gm regionat 650 C and 230 C. The 230 C measurements are thefirst reported at a temperature below 650C, to ourknowledge. Measurements made at the higher tem-perature agree with those of Burch et al. The mostsignificant result of the present investigation is that theair-broadened water vapor continuum at 230C is sometwo to three times greater than expected based on thehigh temperature data and extrapolation procedure ofBurch et al. This is significant not only because thegreater absorption we measure has impact on practicalapplications, but it also indicates at this time that thetemperature dependence of the continuum at the lowertemperature above and below ambient is in question.

The greater absorption at 230 C than expected fromthe Burch extrapolation also implies a wide bound ofuncertainty in the room temperature values of C8 andCf and for the ratio B* for air-broadened water vaporin this spectral region. This is a question which hasgreat practical significance for electrooptical imagingsystems's performance, since transmission as a functionof range and humidity is sensitive to the magnitude ofthese parameters. In the absence of self- and air-broadened water vapor data required to determine Cf,C8, or even B*, it is not possible to identify the mecha-nism by which the water vapor (continuum) absorbs inthis spectral region. Measurements on self- and for-eign-broadened water vapor directed to answering thisquestion are in progress in our laboratory and will bereported at a later date.

The authors gratefully acknowledge the help of BrianZ. Sojka in the data-taking portion of this work andDonald E. Snider and David G. Snyder for their criticalreviews of the manuscript.

References1. H. W. Yates and J. H. Taylor, "Infrared Transmission of the

Atmosphere," NRL Report 5453 (1960), AD 240180.2. R. E. Roberts, J. E. A. Selby, and L. M. Biberman, Appl. Opt. 15,

2085 (1976).3. D. E. Burch, D. A. Gryvnak, and J. D. Pembrook, "Investigation

of the Absorption of Infrared Radiation by Atmospheric Gases:Water, Nitrogen, Nitrous Oxide," AFCRL-71-0124 (1971).

4. D. E. Burch, D. A. Gryvnak, and J. D. Pembrook, "Investigationof the Absorption of Infrared Radiation by Atmospheric Gases,"AFCRL-70-0373 (1970).

5. D. E. Burch, E. B. Singleton, and D. Williams, Appl. Opt. 1, 359(1962).

6. C. B. Farmer, "Extinction Coefficients and Computed Spectrafor the Rotational Band of Water Vapor," EMI Electronics, LTD,Report DMP 2780 (1967).

7. P. Varanasi, S. Chou, and S. S. Penner, J. Quant. Spectrosc. Ra-diat. Transfer 8, 1537 (1968).

8. H. R. Carlon, Appl. Opt. 10, 2297 (1971).9. W. S. Benedict and L. D. Kaplan, J. Chem. Phys. 30, 388

(1959).10. E. K. Damon, J. C. Peterson, F. S. Mills, and R. K. Long, "Spec-

trophone Measurement of the Water Vapor Continuum at DFLaser Frequencies," RADC-TR-75-203 (1975), OSU Report ESL4045-1.

11. F. S. Mills, R. K. Long, and E. K. Damon, "Laser AbsorptionStudies," RADC-TR-75-289 (1975), OSU Report ESL 4054-2.

12. Field data obtained after completion of this work by the NavalResearch Laboratory using the gas filter correlation techniqueindicate that a value of 0.02% of the H2 0 concentration is moreappropriate. The reason for this is not clear, nor is it clear thatthis lower value is more appropriate for the sample used in thiswork. This will be investigated during the course of our ongoingmeasurements.

13. R. A. McClatchey, W. S. Benedict, S. A. Clough, D. E. Burch, R.E. Calfee, K. Fox, L. S. Rothman, and J. S. Garing, "AFCRL At-mospheric Absorption Line Parameter Compilation," AFCRL-TR-73-0096 (1973). Analyses were performed using the data tapepublished January 1976.

14. D. R. Woods, W. Flowers, R. E. Meredith, T. W. Tuer, and J. P.Walker, "DF Laser Propagation Analysis," Science Applications,Inc., Report SAI-77-001-AA (1977).

15. P. W. Anderson, Phys. Rev. 76, 647 (1949).


16. S. S. Penner, Quantitative Molecular Spectroscopy and GasEmissivities (Addison-Wesley, Reading, Mass., 1959), p. 28.

17. K. 0. White and W. R. Watkins, "Absorption of DF Laser Ra-diation by Propane and Butane," U.S. Army Electronics Com-mand Report ECOM-5563 (1975), AD A012169.

18. K. 0. White, G. T. Wade, and S. A. Schleusener, Proc. IEEE 61,1596 (1973).

19. W. R. Watkins, Appl. Opt. 15, 16 (1976).20. C. W. Bruce, "Development of Spectrophones for CW and Pulsed

Radiation Sources," U.S. Army Electronics Command ReportECOM-5802 (1976).

21. C. W. Bruce, B. Z. Sojka, B. G. Hurd, W. R. Watkins, K. 0. White,and Z. Derzko, Appl. Opt. 15, 2970 (1976).

22. D. R. Woods, R. E. Meredith, F. G. Smith, and T. W. Tuer, "HighResolution Spectral Survey of Molecular Absorption in the DFLaser Region: Measurements and Calculations," Rome AirDevelopment Center, Griffiss AFB, N.Y., RADC-TR-75-180(1975).

23. The HDO depleted water sample was obtained from ScienceApplications, Inc., Ann Arbor, Mich.

24. W. R. Watkins and K. 0. White, Opt. Lett. 1, 31 (1977).25. W. M. Thurston, Atomic Energy of Canada, General Chemistry

Branch, Chalk River, Ontario; private communication (1976).26. C. W. Bruce and K. 0. White, J. Opt. Soc. Am. 66, A1088

(1976).27. D. J. Spencer, G. C. Denault, and H. H. Takimoto, Appl. Opt. 13,

2855 (1975).28. F. S. Mills, "Absorption of Deuterium Fluoride Laser Radiation

by the Atmosphere," Ohio State U. PhD Dissertation, Electros-cience Laboratory 4054-3 (1975).

29. T. Tuer, SAI, Ann Arbor, Mich.; private communication(1976).

30. T. W. Tuer, "Thermal Imaging Systems Relative Performance:3-5 tm Versus 8-12 Mm," AFAL-TR-76-217 (1977).

31. J. E. A. Selby and R. M. McClatchey, "Atmospheric Transmit-tance from 0.25 to 28.5 Mm: Computer Code Lowtran 3,"

AFCRL-TR-75-0255 (1975).

Meetings Calendar continued from page 2689

17-20 2nd Joint MMM-Intermag Conference, New York R.M. Josephs, Sperry Univac Computer Systems, P.O.Box 500, Blue Bell, Pa. 19422

18-21 USNC/URSI/IEEE, met., Seattle A. Ishimaru, Dept.of Electrical Eng., FT-10, U. of Washington, Seattle,Wash. 98195

18-22 Laser '79 Exhibition and Congress, West Germany C.Werner, DFVLR, inst. fur Physik der Atmosphare, 801Oberpfaffenhofen, Post Wessling/OBB, GermanFederal Republic

July1-6 21st Colloquium Spectroscopium International/8th In-

ternational Conference on Atomic Spectroscopy,Cambridge Secretariat, P.O. Box 109, Cambridge CB12HY, U.K.

2-6 9th Internat. Laser Atmopheric Studies Conf., MunichC. Werner, DFVLR, Inst. for Atmospheric Phys., 8031Oberpfaffenhofen, Post Wessling/OBB, W. Germa-ny

AugustNational Heat Transfer Conference, San Diego R. Vis-

kanta, School of Mechanical Eng., Purdue U., WestLafayette, Ind. 47907

27-31 Amorphous and Liquid Semiconductors, 8th internat.conf. Harvard U. Sci. Ctr. Conf. Secretariat, 20 Gar-den St., Cambridge, Mass 02138

September10-14 Symp. on Atomic Spectroscopy, Tucson J. 0. Stoner,

Jr., Phys. Dept., U. of Ariz., Tucson, Ariz. 85721

17-19 Optical Communication Conf., Assoc. of 5th EuropeanConf. on Optical Communication and 2nd Internat.Conf. on Integrated Optics and Optical Communica-tion, Amsterdam J. H. C. van Heuven, Philips Labs.,Eindhoven, Netherlands

October7-12 OSA Annual Mtg., Rochester J. W. Quinn, OSA 2000

L St. N. W., Washington, D.C. 20036

November25-30 ASME Winter Mtg., New York ASME, 345 E. 47th St.

New York, N. Y. 10017

December10-15 Submillimeter Waves and Their Application, 4th inter-

nat. conf., San Juan K. J. Button, MIT, NationalMagnet Laboratory, Cambridge, Mass. 02139

1979April

28-3 May 81st Annual Meeting and Exposition of the AmericanCeramic Society, Cincinnati F. P. reid, ACS, 65 Ce-ramic Drive Columbus, Ohio 43214

May27-1 June

JuneI?

7th Canadian Congress of Applied Mechanics, U. ofSherbrooke N. Galanis Faculty of Appl. Sci., U. ofSherbrooke, Sherbrooke, P.Z. JiK 2R1

AAS Meeting, Wellesley, Mass. L. W. Frederick, P.O.Box 3818, Univ. Station, Charlottesville, Va 22903

13-15 Applications of Ferroelectrics, IEEE internat. symp.,Sheraton Ritz Hotel, Minneapolis S. T. Liu, Honey-well Corp. Res. Ctr., 10701 Lindale Ave., S., Bloom-ington, Minn. 55420

19806th Internat. Conf. on Vacuum Ultraviolet radiation

Physics, USA R. P. Madden, NBS, Washington, D.C.20234

March5-9 Pittsburgh Conference on Analytical Chemistry and

Applied Spectroscopy E. S. Hodge, 440 Fifth Ave.,Pittsburgh, Pa. 15213

1980JuneI? AAS Meeting, Cambridge, Mass. L. W. Fredrick, P.O.

Box 3818, Univ. Station, Charlottesville, Va. 22903

2-6 USNC/URSI/Canadian National Committee URSI,Quebec J. A. Cummins, Universite Laval, Dept. deGenie Electriquie, Faculte des Sciences, Cite Univ-ersitaire, Quebec GIK 7P4

continued on page 2729


Documents

Water vapor continuum absorption in the 35–40-µm region