atmospheric scattering coefficients in the visible and infrared regions

JOURNAL OF THE OPTICAL SOCIETY OF AMERICA

Atmospheric Scattering Coefficients in the Visible and Infrared RegionsG. L. KNESTRICK, T. H. COSDEN, AND J. A. CURCIO

U. S. Naval Research Laboratory, Washington, D. C.(Received April 2, 1962)

Atmospheric spectral attenuation coefficients have been measured in ten narrow wavelength bands between0.4 and 2.3 ,u for a variety of weather conditions for two overwater, sea-level paths of 5.5 and 16.3 km. Thewavelength bands were chosen so as to avoid molecular absorption and were isolated by interference filters. A60-in.-diameter high-intensity source and a 24-in.-diameter narrow-field receiver were combined to yieldrelative scattering attenuation coefficients (a) as a function of wavelength X. These were then scaled usingvalues obtained at one wavelength with a visual telephotometer. Logo' vs logX curves show a wide range ofslopes and shapes, with a tendency toward less slope in the infrared (indicating that a- is becoming inde-pendent of X in the infrared). Some correlation with relative humidity was found for relative humiditiesgreater than 70%. An anomalous slope reversal between 1.68 and 2.27 A is discussed, and a possible explana-tion for the reversal is given as selective scattering by the aerosol at these wavelengths.

INTRODUCTION

RADIATION passing through the atmospheresubject to attenuation by molecular and pa]

ticulate scattering and by absorption. The absorptioncaused by permanent gases and water vapor and occuipredominantly at discrete wavelengths. Scattering the other hand is more continuous with wavelength anoccurs to some extent at all wavelengths in the visib]and near-infrared regions of the spectrum. Thereforby isolating narrow wavelength intervals between atsorption bands one can measure the effect of scatterinwith relative freedom from absorption influences.

PROCEDURE

Atmospheric spectral attenuation coefficients of tlhorizontal atmosphere were determined in ten narro,wavelength bands between 0.4 and 2.3 A through thuse of interference filters which isolated the differerbands. The wavelengths were chosen to be as free possible from known molecular absorption. Two diffeient over-water paths were used in the Chesapeake Baarea, a 5.5-km path which was nearly parallel to thshoreline and a 16.3-km path which was entirely ovewater. The period covered by this series of measurements extended from April 1959 to January 1960 anincluded cases of light fog, haze, and clear weatheThe 2% transmittance range' (at 0.67,4) encountereover the 5.5-km path varied from 5.5-72 km, and ovethe 16.3-km path it varied from 37 to 97 km. Relativhumidities varied from 36 to 100%.

The light source was a xenon flash lamp (FT-503) irstalled at the focus of a 60-in.-diameter reflector. Threceiver consisted of an f/3.5 front-surfaced collectcmirror, 24 in. in diameter, and a Newtonian diagonrmirror which was used to deflect the beam to a 2-in.-suncooled, lead sulfide photodetector (Fig. 1). The filteiwere inserted in the focal plane of the mirror, and thdetector was mounted behind the focus. The output the detector was displayed on an oscilloscope screen owhich were read peak deflections from the lamp pulse.

1 This is the range over which the atmospheric transmittancis 2%.

which occurred at 30-sec intervals. Relative transmit-tance values were obtained by comparing the signalsreceived at each wavelength over 5.5 or 16.3 km tothe corresponding signals received on a "zero" (zeroattenuation) run. During a zero run an attenuationdisk with radial slots was placed over the collector tocompensate for the signal increase caused by bringingthe source close to the receiver. A 1000-ft path was usedfor the zero runs, which were made only when themeteorological range was greater than 32 km.

le FIG. 1. Source-receiver system (attenuation diskused for zero runs only).

d CALIBRATION

rd Method,r The transmittance of the atmosphere at 0.67 /u was'e measured using a visual telephotometer for the Kos-

chmieder method of contrast measurement, as described- by Curcio and Durbin.2 The telephotometer consistede of a standard red-filtered Leeds and Northrup opticalir brightness pyrometer fitted with a telescopic lens. Itl was used to measure the brightness of a distant darkq object, such as pine woods, and of the sky horizon. Byrs converting the brightness temperature readings of bothe objects to spectral radiant emittances and substitutingif in the Koschmieder relationship, one obtains the trans-n mission at 0.67 A to the woods.2 The transmission value3, at 0.67 g was used to scale the photoelectric relative

:e 2 J. A. Curcio and K. A. Durbin, "Atmospheric Transmissionin the Visible Region," NRL Rept. 5368 (October 1959).

1010

VOLUME 52, NUMBER 9 SEPTEMBER 1962

ATMOSPHERIC SCATTERING COEFFICIENTS

transmission curve to one of absolute transmission. Theabsolute transmission values were then converted toattenuation coefficients for a 1-km path. At this stage,the coefficients were plotted versus wavelength andwere then replotted after a subtraction of the computedcontribution due to Rayleigh scattering from airmolecules.

Filters

A composite curve of transmittance vs wavelengthfor all the filters used is shown in Fig. 2. In the visibleregion the bandwidths at half-peak transmittanceaverage about 140 A (4 m4) for nearly parallel light.The filters for 1.6 and 2.24 have half-bandwidths ofthe order of 100 m/i. Since the filters were used in con-verging light, the bandwidths in practice were greaterthan the figure indicates. From measurements ofspectral transmittance versus angle of incidence it wasfound that the transmittance curve shifted 6 for one

I

I

I 4

z

2

t 0.43 r-04 6

r0 67

0. 1. 040 1 .5 .

to - 0 0.78

0.3 0.4 0.5 0.6 0.7 0.8 0.!oF 2.2?

toF-S01- 3.04

1~~~~~~~~36

0.9 3.0 3.5 2.0WAVELENGTH (MICRONS)

2.5

FIG. 2. Composite transmittance curve for filtersused to measure attenuation coefficients.

of the visible filters and 3 mg for the 1.6 filter for anincidence or convergence angle of 8. The total averagebandwidth at half peak transmittance then becomes20 mA for the visible region and still of the order of100 m/ for the infrared. The peak wavelengths of the

TABLE 1. Scattering coefficients at 10 wavelengths, including contributions due to Rayleigh scattering by air molecules(values in the last two columns include corrections for water vapor and methane absorption).

ppt, H20 RH T/kmDate (cm) (%) (at 0.67,u) 0.40 0.43,u 0.46 0.47 0.56 A 0.67 A 0.78 A 1.04,p 1.68 2.27 AApril 8 6.3 40 0.88 0.24 0.225 0.190 0.175 0.170 0.130 0.115 0.122 0.065 0.0709 7.1 42 0.92 0.134 0.112 0.105 0.105 0.075 0.084 0.067 0.084 0.019 0.02713 2.9 70 0.90 0.24 0.210 0.180 0.185 0.145 0.105 0.109 0.096 0.045 0.03216 4.4 36 0.95 0.140 0.125 0.180 0.093 0.073 0.052 0.034 0.048 0.025 0.00021 4.6 79 0.88 ... ... 0.210 0.200 0.185 0.122 0.112 0.100 0.045 0.03424 4.5 55 0.85-0.89 0.25 0.23 0.200 0.195 0.170 0.140 0.130 0.125 0.081 0.07027 7.0 94 0.49a 1.36 1.26 1.17 1.10 0.98 0.71 0.68 0.53 0.38 0.3428 4.8 97 0.78a 0.405 0.385 0.345 0.34 0.30 0.25 0.245 0.225 0.151 0.14429 (1100) 5.8 93 0.70a 0.62 0.59 0.55 0.54 0.46 0.35 0.32 0.225 0.102 0.07429 (1300) 6.1 88 0.78a 0.46 0.43 0.40 0.38 0.31 0.25 0.23 0.169 0.081 0.06030 4.0 36 0.60 0.69 0.67 0.63 0.64 0.58 0.505 0.48 0.385 0.28 0.26Sept. 18 3.7 44 0.89 0.210 0.190 0.167 -0.170 0.140 0.120 0.110 0.088 0.055 0.06222 9.8 81 0.79 0.40 0.37 0.32 0.30 0.24 0.23 0.194 0.150 0.093 0.11229 (0810) 11.0 ... 0.78-0.70 0.38 0.345 0.325 0.325 0.305 0.29 0.26 0.245 0.183 0.18229 (0825) 11.0 ... 0.70-0.67 0.49 0.46 0.445 0.44 0.405 0.38 0.34 0.315 0.255 0.2429 (0850) 11.0 ... 0.78-0.73 0.365 0.345 0.33 0.315 0.295 0.28 0.25 0.24 0.181 0.17629 (0910) 11.0 ... 0.65-0.75 0.425 0.40 0.38 0.385 0.36 0.345 0.31 0.30 0.235 0.2629 (0945) 11.0 ... 0.79-0.77 0.335 0.305 0.29 0.285 0.25 0.25 0.22 0.21 0.158 0.17629 (1030) 11.0 95 0.77 0.33 0.31 0.29 0.285 0.265 0.265 0.235 0.235 0.173 0.204Oct. 2 7.6 76 0.88 0.22 0.186 0.162 0.171 0.133 0.122 0.105 0.075 0.0405 0.05226 11.0 56 0.96 0.091 0.066 0.064 0.061 0.039 0.039 0.040 0.040 0.043 0.04527 11.5 58 0.92 0.165 0.153 0.136 0.133 0.095 0.087 0.073 0.066 0.053 0.04828 8.0 47 0.95 0.125 0.115 0.100 0.087 0.070 0.056 0.050 0.041 0.023 0.03529 9.0 59 0.94 0.157 0.137 0.115 0.105 0.072 0.064 0.061 0.051 0.030 0.04130 15.5 77 0.91 0.25 0.22 0.197 0.184 0.138 0.096 0.091 0.042 0.017 0.024Nov. 2 7.0 45 0.96 0.078 0.064 0.059 0.052 0.035 0.039 0.027 0.033 0.014 0.0233 8.0 57 0.95 0.112 0.090 0.093 0.080 0.056 0.047 0.049 0.042 0.040 0.0284 13.0 49 0.92 0.159 0.127 0.122 0.107 0.097 0.087 0.070 0.064 0.060 0.06313 14.0 75 0.75 0.29 0.26 0.188 0.132 0.12520 7.0 58 0.91 0.196 0.174 0.146 0.149 0.118 0.090 0.080 0.066 0.044 0.045530 4.2 38 0.95 0.099 0.101 0.085 0.072 0.055 0.056 0.046 0.041 0.031 0.033

Jan. 5 1.8 44 0.86 0.196 0.180 0.171 0.171 0.137 0.145 0.137 0.137 0.101 0.10513 (1110) 3.7 92 0.60 1.14 1.04 0.97 0.88 0.74 0.51 0.49 0.32 0.182 0.14213 (1135) 3.7 ... 0.58- 0.53a * 1.24 1.16 1.08 0.89 0.58 0.58 0.34 0.200 0.2613 (1345) 3.7 ... 0.58a 1.18 1.10 1.03 0.95 0.80 0.54 0.53 0.32 0.18 0.11713 (1440) 3.7 92 0.67a 0.94 0.86 0.79 0.75 0.60 0.40 0.37 0.205 0.092 0.06113 (1520) 3.7 92 0.67a 0.82 0.75 0.70 0.66 0.54 0.40 0.37 0.24 0.148 0.121

Rayleigh 0.045 0.034 0.025 0.023 0.011 0.005 0.003 0.001 ...

a Obtained from visual estimate of meteorological range.

September 1962 1011

KNESTRICK, COSDEN, AND CURCIO

two infrared filters are not quite centered between thewater-vapor absorption bands but are located slightlyon the long wavelength side of center, resulting in aneffective shift of this side of the filter passband de-pendent upon water content of the atmosphere. Thedata at these points were corrected for this effect usingthe Passman and Larmore3 values for transmissivity asa function of water content.

Methane

When the attenuation values at 2.27, seemed to beconsistently high we investigated the possibility ofabsorption by other gases in the atmosphere and foundthat Migeotte4 lists several methane bands in thisregion: namely, at 2.20, 2.32, 2.37, and 2.43 y. Accord-ingly, the transmissivity of methane was measured inthe laboratory using the PbS cell and filter combination.It was found that 1 atm-cm of methane, which wasequivalent to that which would frequently be en-countered in a 5.5-km path, had an effective trans-mittance of 97% and the transmittance for a 16.3-kmequivalent path was 91%. The amounts of methanein the two paths were calculated using the ARDC modelatmosphere figure of 1.4 parts per million of methaneby volume.5 This would be a minimum value for thepaths used here because the Chesapeake Bay locationhas appreciable surrounding swamp land, a suggestedsource of methane. 6

RESULTS

There are available a total of 37 measurement runs,which were obtained on 27 different days. The datafrom these runs are given in Table I which includes in-formation on the precipitable (ppt) water content(column 2) and the transmissivity (kilometer) (T/km)at 0.67 A (column 4) as measured with the telephotom-eter. In some cases of haze.or light fog the transmissionchanged during a run. In these cases, the transmittancesbefore and after the run were known and are given incolumn 4 of Table I. The median value was used fornormalizing the data. The values of the coefficientslisted in Table I for 1.68 and 2.27 A contain correctionsfor water vapor and methane absorption within thepassband of the filter. The contributions due to Ray-leigh scattering are included in the tabulated data andare listed separately at the end in case one desires toeliminate this component from the data. A majorityof the runs were made with the 5.5-km path, the longpath being used from Oct. 26 through Nov. 30 only.The data corrected for water vapor and methane absorp-tion, and including the Rayleigh contribution, are shown

I S. Passman and L. Larmore (Unclassified), IRIS Proc. 1, No. 2,15 (Dec. 1956), Classified Symposium.

IM. V. Migeotte The Atnospheres of the Earth and Planets,edited by G. P. Kuiper (University of Chicago Press, Chicago,1949) 2nd ed., Chap. 10, p. 284.

s L. E. Miller Handbook of Geophysics, U. S. Air Force (1957),1st ed., Chap. 8, p. 1.

I J. Strong, Phys. Today 4, 14 (1951).

I-

U.0

0

L)

I-n

0.8

0.6

0.4

0.21

0.I

0.08

0.06

0.041

0.02

3 0.4 0.6 0.8 1.0 2.0 4.0WAVELENGTH (MICRONS)

6.0

FIG. 3. Scattering coefficient vs wavelength including Rayleighscattering. Corrections for water vapor and methane absorptionhave been made.

plotted as scattering coefficient versus wavelength inFigs. 3 through 8. For convenience, the curves weregrouped as much as possible according to slope. Forexample, all curves in Fig. 3 are rather flat in the in-frared. The lower three curves have increasingly steeperslopes in the visible region. The slope of the Oct. 26curve approaches the slope of a Rayleigh scattering

0.8

0.6

O.E

0.4

8Y 0.2b

zI-0 0.1

0 0.0809 0.06

CZzl_ 0.04I-

1-

,,SEPT. 22

-+APR. 8

/ APR. 9

l l l

0.02-

0.01LO.:

l 111110.4 0.6 0.8 1.0 2.0 4.0 6.0

WAVELENGTH (MICRONS)

FIG. 4. Scattering coefficient vs wavelength including Rayleighscattering. Corrections for water vapor and methane absorptionhave been.made.

1012 Vol. 52

SEPT. 29 (0910)

SEPT. 29 (0850)

- JAN. 5

NOV. 4

OCT. 26

NOV. 3

_ I I I I I I I I I

lI.".

.3

0.0110.

September 1962 ATMOSPHERIC SCATTERING, COEFFICIENTS

0.6 0.8 1.0 2.0WAVELENGTH (MICRONS)

that the particles are larger than the wavelength of theradiation.

DISCUSSION

Scattering Coefficient Curves

Points on the a- vs X curves were connected by straightlines rather than by drawing a smooth curve of bestfit because particle scattering does not necessarily varymonotonically with wavelength. However, most of thedeviations of the points from a smooth curve are prob-ably due to accumulated errors in the system, includingtwinkle, fluctuation of source output, and errors inreading the scope trace. The effect of twinkle is espe-cially noticeable in the data for clear atmospheres,namely, on April 9 (Fig. 4) and November 2 (Fig. 5),

1.5

1.0

0.8

0.6

0.4


distribution below 0.56 ji and has a flat portion whichextends into the green. This behavior indicates anaverage particle size for which the scattering is almostindependent of wavelength or, in other words, indicates

,,SEPT. 29(1030)

\JAN. 13 (1440)/ OCT. 2

,,+OCT. 297OCT. 28

I I I I0.6 0.8 1.0 2.0 4.0 6.0



.I-b

0U.U-

0.2 -

0.1

0.08CDza: 0.06

o 0.04U)

0.02-

i I ll I0.3 0.4 0.6 0.8 1.0 2



even though the average of several readings was taken.The effect was measured on a clear day and found toinvolve fluctuations of ±46% based on the average of5 readings. This is believed to be the maximum varia-tion due to twinkle.

The relative variation in the point at 2.27 p does notseem to be due entirely to error. In half the cases thevalue of the coefficient at 2.27 A was greater than thatat 1.68,g. Such a slope reversal does not occur this oftenat any other wavelength. In addition, the data of Yatesand Taylor7 show the same behavior of this point,although with less regularity. The reason for this be-havior may involve weak absorption bands in the 2.2 ,t

7 H. W. Yates and J. H. Taylor, "Infrared Transmission of theAtmosphere," NRL Rept. 5453 (June 1960).

1013

I.

0Q-

13 (1345)

-- CT. 27

\APR. 13

l l l l

1.0

0.8

0.6

0.4

E

b0.2z

C.)

L

o 0.1C.)

CD 0.08Z' 0.06- .-

Lu) 0.04

2 _

4.0 6.0

0.021-

0.3 0.4

5.

1.5.

n no L 1 |

U . u l * 1 | . . .V.VI -

I l l l l l

KNESTRICK, COSDEN, AND CURCIO

'E JAN. 13 (1135)

b 0.21-

U_ O.OEX +XXJAN. 13 (1520)

w 0.10 00

o 0 .08 APR. 29 (1100)zz 0.06 APR. 29 (1300)

0.04W ~~~~~~~~~~APR. 21

OCT. 300.02

APR. 16

0.3 0.4 0.6 0.8 1.0 2.0 4.0 6.0WAVELENGTH (MICRONS)


region. These appear in the spectral curves of reference7, one of which is reproduced as Fig. 9. At least a smallamount of absorption is always present even for a pathas short as 1000 ft. This absorption as well as the highmeasured value of scattering coefficient at 2.27 A arenot related to the absolute or the relative humidity,which would seem to rule out water vapor as thecausative agent. The only other known absorbing gasin this region is methane (for which an approximatecorrection has been made). Therefore, if the phenom-enon is due to absorption, it is believed that it is notcaused by either of the gases for which corrections have

'100

90

80

70

60

50

I 40

w30

20

10

0A

1.9 2.0 2.1 2.2 2.3 2.4 2.6WAVELENGTH, NICRONS

FIG. 9. Atmospheric transmittance for a 16.25-km path as afunction of wavelength (from reference 7) showing weak-absorp-tion structures in the 2 t region.

already been made. Another possible source of thehigh scattering coefficient may be some sort of selectivescattering.

The amount of precipitable water contained in thepath does not correlate with the scattering, but there issome correlation of scattering with relative humidity(except at the wavelength mentioned above). In Fig. 10the scattering coefficient at 5600 A (0.56 /.L) is plottedversus relative humidity (RH). For humidities of 70%or greater the coefficient depends to some extent on therelative humidity. Below 70% RH, however, no de-pendency exists, which is in agreement with thefindings of Junge,8 Wright,9 and others, and bears outMiddleton's statements that there is comparativelylittle change in the radii of the nuclei with changes ofrelative humidity below about 70%. Recently, Boileau"and Bumal2 have reported similar findings and in addi-

I I I I I I I 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9SCATTERING COEFFICIENT o(Kmri) AT 0.56.

FIG. 10. Relative humidity vs scattering coefficient at 0.56,u.

tion have found an apparent discontinuity for relativehumidities between 65 and 85%. Such a discontinuityis suggested by the present data (Fig. 10), but theevidence is not strong enough to justify it.

Coefficient Ratios

One would like to be able to predict the attenuationat a certain wavelength or several wavelengths bymeans of one or two simple measurements. The meteoro-logical parameters seem to be of little help, but there issome limited correlation between coefficients at variouswavelengths. For example, Curcio and Durbin2 in-vestigated the relationship of the coefficients at threewavelengths in the visible region and obtained averagevalues for the ratios of the coefficients at one wave-length to those at another wavelength. A similar in-vestigation was made using the present data.

8 C. Junge, Gerlands Beitr. Geophys. 46, 108 (1935).9 H. L. Wright, Quart. J. Roy. Meteorol. Soc. 66, 66 (1940).'SW. E. K. Middleton, Vision Through the Atnosphere (The

University of Toronto Press, Toronto, 1952), Chap. 3, p. 24.i1 A. R. Boileau, "Correlation between Measured Path Function

and Relative Humidity," SIO Reference 59-5, Scripps Institutionof Oceanography, University of California (Feb. 1, 1959).

12 T. J. Buma, Bull. Am. Meteorol. Soc. 41, 357 (1960).

1014 Vol. 52

w ho

September 1962

1.56

I.0

0.8

0.6ZL

t 0.40

E 0.2bI-zL 0.1

W 0.080

, 0.06

I.- 0.04.4

0.02

i.

ATMOSPHERIC SCATTERING COEFFICIENTS

=~~~~~~~~~~ /+.t +t . +

_4 ++/

= +/

- /

_///_/ 0-o47= 15 -0.67

I I I I II l I I I I I II0.01 0.02 0.04 0.06 0.08 0.1 0.2 0.4 0.6 0.8

SCATTERING COEFFICIENT 0-(Km1

') AT 0.67k

FIG. 11. Scattering coefficient at 0.47 .u vs scatteringcoefficient at 0.67 A (data from Table I).

In Fig. 11 the attenuation coefficient at 0.47 isplotted against the coefficient at 0.67 M. The line ofbest fit would have a slope of +0.97 but a line of slope+ 1 has been fitted to the data for convenience. Thiscurve gives an average relationship of 0-0.47=1.5o-0.67.

This compares favorably with the value of 1.54 ob-tained by Curcio and Durbin.

In Fig. 12 the attenuation coefficient at 0.56 isplotted vs the attenuation coefficient at 0.67 , and byfitting the data with a line of + 1 slope, which is within4% of the slope of the best-fit line, the relation derived

.0

0.6

loI-. 0.2

0b

2 0.1

I 0.04W

Z

I-4 0.04

0

0.02t

O.C

SCATTERING COEFFICIENT 0-(Km1) AT 0.679

FIG. 12. Scattering coefficient at 0.56 vs scatteringcoefficient at 0.67 (data from Table I).

1.1

0.1

I-O.

0E

b

z-)0.W 0.00

O 0.0Z

0.0

0.0

0.01 _02 0.04 0.06 0.08 0.1 0.2 0.4 0.6 0.8SCATTERING COEFFICIENT 0(KmK) AT 0.56

FIG. 13. Scattering coefficient at 0.47ju vs scatteringcoefficient at 0.56 (data from Table I).

is 0-0.66= 1.2a0.67. Curcio and Durbin used 0.535 A in-stead of 0.56 1t, so no direct comparison can be madewith their value of 1.32.

A similar plot, Fig. 13, was made of the coefficientsin the blue against those in the green resulting in a ratioof 00.47= 1.25ao.66. The degree of correlation of coeffi-cients at two different wavelengths depends on thecloseness of the wavelengths. For example, when anattempt was made to plot coefficients in the red versusthose in the infrared, a rather large spreading of thepoints resulted, as shown in Figs. 14 and 15 (comparewith Figs. 11 to 13). A best-fit line was not applied to

0.8

0.6

oI-IR 0.2

0

E,0.06

b

z

2 ojo

W 0.040

0.021

O.SCATTERING COEFFICIENT O-(Km'I) AT 1.04j.

FIG. 14. Scattering coefficient at 0.67 A vs scatteringcoefficient at 1.04,u (data from Table I).

1015

8 7

_ /+

4. - /+

2 /+4+ ~ ~ /

I. _ +/ +4 ++Di /4

4 /

2- / 0*047 1250o.5e

. I I I I I I no n II

+

= +r/_ /+~~/4_ ~~~/+

__~ 4+

= /g

- /

_ /+

- / /0.6 ' I20'0.67/

I /II I I I I1 0.02 0.04 0.06 0.08 0.1 0.2 0.4 0.6 0.8

_ /~~~~~~/

_ + +

- A/+/

/ I 0, 4 111111 0 0 0 .01 -0.02 0.04 0.06 008 O 0.2 0.4 Q6

o. ..

I

2

COSDEN, AND CURCIO

0.

0.6

~L0.40

t-8r 0.2

b

L 0.08

0C.J 0.06

e 0.04

0.02[

0.01 0.02 0.04 0.06 0.08 0.1 0.2SCATTERING COEFFICIENT 0(Kml) AT 1.68±

0.4

change occurs only part of the time. On the basis ofexisting data there seems to be little hope of reliablypredicting infrared scattering attenuation from a knowl-edge of the attenuation in the visible. The difficulty isillustrated very pointedly in Fig. 6 by the two curves ofSeptember 29 and January 13 which cross at about 1 .

A discussion of these results in relation to estimatedparticle-size distributions has been published pre-viouslyl3'14 and will not be discussed in detail here.Suffice it to say that, in general, the aerosol on anyparticular day behaves, insofar as the variation of with is concerned, like a two-component compositedistribution, the main component being described by aJunge distribution of the form dN/d logr=-Crs andthe other either by a distribution similar to that ofaerosols found in maritime air, or by a relatively mono-disperse distribution contained in a narrow-radiusinterval.

0.6

FIr. 15. Scattering coefficient at 0.67 JU vs scatteringcoefficient at 1.68 A (data from Table I).

Fig. 15 due to the wide spread in the data. This spreadis no doubt caused by the change in slope betweenvisible and infrared regions and the fact that this

ACKNOWLEDGMENT

The authors wish to thank Dr. L. F. Drummeter forhis guidance and encouragement.

13 J. 'A. Curcio, G. L. Knestrick, and T. H. Cosden, "Atmo-spheric Scattering in the Visible and Infrared," NRL Rept. 5567(January 1961).

14J. A. Curcio, J. Opt. Soc. Am. 51, 548 (1961).

-t +

+ +

.i.+ +I I~ + + +I

_ + + + +~~~~+

_ ~ ~~~~~ + +

_~~ ~ ~~~~ ++

++ {11 14 +

I -

1016 Vol. 52KNESTRICK,

I

atmospheric scattering coefficients in the visible and infrared regions

Documents