September 1986 / Vol. 11, No. 9 / OPTICS LETTERS 563

Laboratory measurements of atmospheric temperature andbackscatter ratio using a high-spectral-resolution lidar technique

F. J. Lehmann, S. A. Lee, and C. Y. She

Department of Physics, Colorado State University, Fort Collins, Colorado 80523

Received March 10, 1986; accepted June 20,1986

A narrow-band atomic blocking (bandstop) filter capable of high aerosol rejection has been developed for lidar

applications. Using this filter, and our proposed new high-spectral-resolution lidar technique, laboratory measure-

ments of atmospheric temperature and backscatter ratio have been made, with an accuracy of 1 K and +3%,

respectively.

General lidar (optical radar) techniques can give infor-mation on atmospheric parameters such as tropo-spheric temperature and backscatter ratio only afterthe atmosphere is properly modeled.' Techniquesthat can be used to measure these parameters directlyare therefore of interest. Raman scattering2 and dif-ferential absorption3 have been used to measure tro-pospheric temperature, but when compared with Ray-leigh-Mie scattering, these lidar techniques havemuch lower signal-to-noise ratios.4 However, in orderto extract atmospheric information from Rayleigh-Mie scattering, the aerosol- and molecular-scatteringcomponents, which are contained in a narrow band-width of 1-2 GHz, must be spectrally separated andanalyzed. An obvious solution is to use a scanningFabry-Perot interferometer to analyze the scatteredspectrum induced by a single-frequency laser. In-deed, the early and only atmospheric parameter mea-surement using Rayleigh-Mie scattering was per-formed in this manner.5 In this 1971 experiment,Fiocco and co-workers made a remote temperaturemeasurement at 4-km height; they achieved a mea-surement accuracy of a few degrees by averaging over a2-km path and a 1-h time duration. Since much of thescattered spectrum is undetected at any given instant,a scanning method is time consuming and thereforenot suitable for lidar applications. A desirable high-spectral-resolution lidar (HSRL) technique should beone that uses a single-frequency laser as the lightsource and an adequate narrow-band blocking filterthat permits separation of molecular scattering fromaerosol scattering (or interference) and detection ofthe entire scattering spectrum without scanning.Methods on these lines have been conceived. Using aFabry-Perot interferometer at a fixed setting as theblocking filter, Sroga et al.6 in 1983 made successfulatmospheric backscatter ratio measurements despitealignment difficulties. In 1981 Schwiesow and Lading7

proposed temperature profiling with a HSRL using aninterferometer; no experimental result, laboratory orfield, has yet appeared in the literature. Realizing theproblems in alignment difficulty and small dynamicrange in aerosol rejection associated with interferome-ters, Shimizu et al. proposed the use of narrow-band

atomic blocking (bandstop) filters in a HSRL for at-mospheric parameter measurements.8 In this Letterwe report the development of such an atomic-vaporfilter and our first laboratory atmospheric tempera-ture and backscatter ratio measurements using thisnewly proposed technique. Theoretical calcula-tions4'8 indicate that when this HSRL technique issuccessfully implemented, many existing lidar sys-tems can be modified to perform routine atmospherictemperature profiling with 1-K accuracy, 30-m depthresolution, and a 10-sec measurement time at a rangeof 5 km.

The backscatter ratio, r, is defined as the ratio oftotal backscattering, aerosol and molecular, to thatdue to the molecules alone. As discussed in Ref. 8, rmay be determined from the formula

r = (f - fa)(NJIN2)/(1 - faNlIN2) (1)

Here, N1 1N2 is the measured ratio of the total scatter-ing signal to that attenuated by the bandstop filter.To obtain the attenuation factors fm and fa, the trans-mission spectrum of the atomic filter, F(v), must bedetermined. The aerosol factor, fa = F(vo), corre-sponds to the filter transmission evaluated at the inci-dent laser frequency vo, and the molecular factor, f, isthe convolution of F(v) with the theoretical molecular-scattering spectrum R(v, T, P), which depends on at-mospheric temperature and pressure. To determinethe temperature, two different atomic filters, givingtwo molecular factors, fm and f,,' must be used.8 Bycomparing the theoretically calculated temperature-dependent ratio (at a given atmospheric pressure) fIlft' with the measured ratio of the filtered backscattersignal counts, N2 /N2 ', the atmospheric temperature isdetermined.

The experimental apparatus is schematically shownin Fig. 1. A single-mode cw tunable dye laser (Coher-ent 599-21), with an approximate 1-MHz linewidth, isfocused into a scattering chamber. The backscatteredlight is collected with a mirror set at 450 from the laserbeam, collimated, and then sent through the atomicfilter and detected by a photomultiplier. The narrowbandstop filter consists of a heat-pipe oven9 usingbarium vapor and argon buffer gas. For atmospheric

0146-9592/86/090563-03$2.00/0 © 1986, Optical Society of America

564 OPTICS LETTERS / Vol. 11, No. 9 / September 1986

500 mmLENS

85 mm f/ 1.4

50mm f/1.2

ATOMIC VAPORFILTER

50mm f/1.8

Fig. 1. Schematic of the experimental apparatus.

temperature measurements, the laser is tuned on reso-nance to the 6s2 'So-6s6p IPl transition of barium at553.7 nm. The laser was set to within 10 MHz of thisresonance, within which the filtered counts remain thesame. The filtered photon counts N2 and N2' may beobtained by setting the oven at different temperaturesTo and To'. For the backscatter ratio measurement,the needed photon counts, N2 and N1, are obtainedwith the laser tuned on and 10 GHz off the bariumresonance, respectively, at a single oven setting. Asecond optical pathway (not shown) is used to measurethe filter's transmission spectrum, F(v), as required bythe theory. This path bypasses the scattering cham-ber and goes directly through the filter, and the trans-mitted light is detected by a photodiode. The filter'stransmission spectrum is mapped out by scanning thelaser frequency across the barium absorption peak.In order to prevent erroneous transmission spectradue to optical pumping or saturation, the laser poweris attenuated to a level below which these occur; typi-cal power used was 100 ,4W.

Several sets of experiments, with comparable accu-racies, have been made on different days. In one set ofthese experiments, five oven-temperature settingswere used giving bandstop filters with the followingpeak absorption and FWHM: 92%, 1.4 GHz; 96%, 1.5GHz; 100%, 1.6 GHz; 100%, 1.7 GHz; and 100%, 1.8GHz. The transmission spectra for three of them areshown in Fig. 2. The three filters with 100% peakabsorption, one of which is not shown, may be used toreject aerosol scattering altogether and to determinethe calculated fm/ff' for the atmospheric temperaturemeasurements. Compared with these calculated ra-tios, the three permutated measured ratios N 2 /N 2' ofthe filtered counts provide three determinations of thesame atmospheric temperature. They give values of26.1, 26.3, and 26.60C. Compared with the chamber'stemperature measured by a thermometer, 26.0 +0.50C, these values suggest that the absolute accuracyof our atmospheric temperature measurement is 1 Kindependent of the filter settings used. With the fivefilter transmission spectra, three of which are shown inFig. 2, five determinations of the backscatter ratio, onefor each filter setting, can be made; they are 1.53, 1.45,

1.41, 1.41, and 1.40, corresponding to an average of1.44 + 3%. Tables 1 and 2 summarize the experimen-tal results for easy comparison.

In the above laboratory temperature measurements,the atmospheric pressure of the scattering chamber,857 mbars for example, was measured independentlyby a Baratron and used in the calculation of the molec-ular scattering spectrum R(v, T, P). In a practicallidar system, this is of course not possible, and theatmospheric pressure must be estimated from theheight of the atmosphere that can be range resolvedeasily with a lidar. Since the molecular scattering ismuch less sensitive to pressure variations,8 some un-certainty in the atmospheric pressure does not greatlyaffect the temperature measurement. This point can

0U,0,i0,za:.

-5 GHz 0 GHz

100

U-

z0,co

2UIZSr

0%--5 GHz

7

Ui.

LDzCn00,

zI-

+5 GHz

+5GHz

+5GHz

0 GHz

-5 GHz 0 GHz

FREQUENCY SHIFT

Fig. 2. Transmission spectra, F(v), of the atomic-vaporbandstop filter for different filtration settings.

September 1986 / Vol. 11, No. 9 / OPTICS LETTERS 565

Table 1. Backscatter Ratio Measurements

FilterA B C D E

1.53 1.45 1.41 1.41 1.40

Table 2. Local Laboratory Temperature (in DegreesCelsius) Measured by Three Pairs of Bandstop Filters

Measured Filter Count RatioN2(D)/N 2 (C) N2 (E)/N 2(C) N2 (E)/N 2 (D)

26.1 26.3 26.6

0.886

0.885

E 0.884

0.883

294

, 10 mbars

Atmospheric

Pressure = 857 mbars

299 304

TEMPERATURE (K)

Fig. 3. Temperature dependence of the ratio of molecularattenuation factors, fm/fm', at two filter settings (E to C).The error bars indicate a +0.5-K temperature uncertaintyfor a +10-mbar pressure variation.

be seen in Fig. 3, where the calculated molecular atten-uation ratio, fmt/fm', for a constant pressure (857mbars) and a pair of atomic filter settings is plotted asa function of atmospheric temperature. At a giventemperature, e.g., 299 K, a pressure variation of ±10mbars gives rise to an uncertainty in fm/fm' as indicat-ed by the vertical error bar. If the pressure is notindependently monitored, this uncertainty translatesinto +0.5-K temperature variation as indicated by the

horizontal error bar. This +10-mbar pressure varia-tion corresponds to the maximum deviations that theatmosphere undergoes. Under normal conditions thepressure variations are much less, and hence the re-sulting temperature uncertainty will still be within the1-K accuracy quoted.

As pointed out,' 0 the S6 model for nitrogen mole-cules" is more accurate than the simpler SM model formonatomic molecules.'2 We therefore used the S6model here for calculating the scattering spectrumR(v, T, P) of air molecules. The SM model has alsobeen used for comparison; the atmospheric tempera-tures so determined show a comparable relative accu-racy but with a systematic error of roughly 5 K toohigh, confirming that the S6 model is more accurate.

In order to test the dynamic range of the bandstopfilter, an elastic scattering target was used to mimicthe effect of aerosol scattering. When the total scat-tered light is increased this way by 3200 times, thefilter, at a setting corresponding to that used in Fig.2(b), can still reject the elastically scattered light,keeping the filtered total counts the same. The high-rejection dynamic range of the atomic bandstop filterdeveloped here should make our HSRL techniquesuitable for measurements in normal as well as aero-sol-loaded atmospheres.

We acknowledge the contributions of G. W.Kattawar, K. Kobashi, and G. Tenti, who helped us inestablishing a working computer program for calculat-ing the molecular scattering spectrum. We also ac-knowledge helpful discussions with Bob Drullinger onheat-pipe oven construction. This research is fundedby the U.S. Army Research Office under grant no.DAAG29-83-K-0095.

References

1. M. J. Kavaya and R. T. Menzies, Appl. Opt. 24, 3444(1985).

2. J. Cooney, J. Appl. Meteorol. 11, 108 (1972).3. J. E. Kalshoven Jr., C. L. Korb, G. K. Schwemmer, and

M. Dombrowski, Appl. Opt. 20,1967 (1981).4. H. Shimizu, K. Noguchi, and C. Y. She, Appl. Opt. 25,

1460 (1986).5. G. Fiocco, G. Benedetti-Michelangeli, K. Maischberger,

and E. Madonna, Nature 229,78 (1971).6. J. T. Sroga, E. W. Eloranta, S. T. Shipley, F. L. Roesler,

and P. J. Tryon, Appl. Opt. 22, 3725 (1983).7. R. L. Schwiesow and L. Lading, Appl. Opt. 20, 1972

(1981).8. H. Shimizu, S. A. Lee, and C. Y. She, Appl. Opt. 22,1372

(1983).9. C. R. Vidal and F. B. Haller, Rev. Sci. Instrum. 42,1779

(1971).10. A. T. Young and G. W. Kattawar, Appl. Opt. 22, 3668

(1983).11. G. Tenti, C. D. Boley, and R. C. Desai, Can. J. Phys. 52,

285 (1974).12. S. Yip and M. Nelkin, Phys. Rev. A 135,1241 (1964).

I

.