June 10, 2010 / Vol. 8, No. 6 / CHINESE OPTICS LETTERS 533

Measurements for profiles of aerosol extinction coefficient,backscatter coefficient, and lidar ratio over Wuhan in China

with Raman/Mie lidar

Wei Gong (÷÷÷ %%%)1, Jinye Zhang (ÜÜÜ777���)1,2∗, Feiyue Mao (fff������)1, and Jun Li (ooo ddd)1

1State Key Laboratory for Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University,

Wuhan 430079, China2School of Science, Hubei University of Technology, Wuhan 430068, China

∗E-mail: yezi.zh@163.com

Received September 3, 2009

The profiles of aerosol extinction, backscatter coefficient, and lidar ratio in the lower troposphere overWuhan are measured by a multi-channel Raman/Mie lidar. Using the lidar ratio retrieved by Ramanscattering principle, the profiles of aerosol extinction and backscatter coefficients are also retrieved by Miescattering signals, without a prior assumption about their relation in the traditional pure Mie signals dataanalyses. The observations by both Raman and Mie are in good agreement with each other. The highcoherence shows that the system is reliable, and the Mie and Raman channels are in good adjustment andhave the same field of view.

OCIS codes: 010.0280, 010.1100, 280.1100, 280.3640.doi: 10.3788/COL20100806.0533.

Lidar is a powerful active tool for detecting the opticalproperties of aerosols in the atmosphere. Single-channelelastic scattering lidar is often used to retrieve aerosol ex-tinction. But an assumption of the lidar ratio for aerosol(ratio of extinction coefficient to backscatter coefficient)is needed by the retrieval method of Klett[1]. This mayintroduce a considerable amount of uncertainties as thisratio can vary from 20 to 100 sr with the influences of par-ticle type, size, wavelength, range, and time. The meth-ods such as multiple zenith angle measurement[2], highspectral resolution lidar (HSRL)[3], and Raman lidar[4].have been demonstrated to overcome the problem. Mul-tiple zenith angle measurements retrieve the extinctionprofile at two or more different zenith angles simultane-ously or alternately. The aerosol properties are assumedto be the same for both directions to solve the set of twolidar equations and thus the extinction and backscatterprofiles are abtained. Multi-channel elastic (HSRL) orinelastic (Raman) lidar provides more spectral informa-tion that allows the aerosol characteristics to be derivedmore precisely. However, the HSRL is intended for a

specific complex instrument, which involves spectral sep-aration of the molecular and aerosol backscatter returns.Some rather demanding technologies are needed, for ex-ample, a highly stable and narrow band laser system isa must, also a high resolution Fabry-Perot interferom-eter or atomic vapor filters are employed in the HSRLreceiver system to separate the aerosol (Mie) and molec-ular (Rayleigh) scattering.

Raman backscatter signals can be used alone to re-trieve aerosol extinction profiles. With the combinationof Mie data, aerosol backscatter coefficient profiles canbe acquired. This will improve the accuracy of measure-ment compared with that using only the Mie backscattersignals. In this letter, a multi-channel Raman/Mie lidardeveloped by Wuhan University is used to retrieve theoptical properties of aerosols in lower atmosphere. Thelidar ratios acquired by the Raman method work as theprior condition in the Mie Fernald method, and the ob-servation results are compared.

With the detection of Raman backscatter signals, theindependent aerosol extinction coefficient can be deter-mined by[5]

αaer(λL, r) =ddr

[ln nx(r)

r2P (r,λx,λL)

]− αmol (λL, r)− αmol (λx, r)

1 + λLλx

, (1)

where αaer and αmol are the extinction coefficients ofaerosols and molecules, respectively; P (r, λx, λL) is thepower received from the distance r at the Raman-shiftedwavelength λx, λL is the laser output wavelength; nx(r)

is the density number of molecule x.The aerosol backscatter coefficient at the laser wave-

length is solved by

βaer(λL, r) ≈ −βmol(λL, r) + βmol(λL, r0) · P (λL, r) · P (λx, λL, r0) · nx(r)P (λL, r0) · P (λx, λL, r) · nx(r0)

·exp[−

r∫r0

αaer(λx, r) + αmol(λx, r)dr]

exp[−r∫

r0

αaer(λL, r) + αmol(λL, r)dr], (2)

1671-7694/2010/060533-04 c© 2010 Chinese Optics Letters

534 CHINESE OPTICS LETTERS / Vol. 8, No. 6 / June 10, 2010

where P (λL, r) is the Mie scattering return power, r0 isthe reference height, βaer and βmol are the backscattercoefficients of aerosols and molecules, which can be es-timated according to Rayleigh scattering. In Eqs. (1)and (2), Angstrom’s law has been used. It describes theextinction coefficients for different wavelengths and as-sumes that they can be expressed as αaer(λ, r) ∝ λ−k.For aerosol particles, the wavelength dependence param-eter k = 1 is appropriate and it has been shown that theuncertainty caused by k is negligible[6]. So αaer(λx, r)in Eq. (2) can be computed if αaer(λL, r) is known. To

evaluate this equation, r0 needs to be set. When thebackscatter coefficient of aerosols is much smaller thanthat of molecules, the aerosol backscatter coefficient canbe ignored.

The method of Fernald[7,8] is used to analyze the Miebackscatter signals to retrieve the optical property ofaerosols. In this method, the atmosphere is seen as beingcomposed of gas molecules and aerosol particles. Aerosolbackscatter and extinction coefficients can be expressedby these recursive formulae correspondingly:

βaer(I − 1) + βmol(I − 1) =Q(I + 1) exp[−A(I, I + 1)]

Q(I)βaer(I)+βmol(I) + S1{Q(I) + Q(I − 1) exp[A(I − 1, I)]}∆r

, (3)

αaer(I − 1) +S1

S2αmol(I − 1) =

Q(I − 1) exp[A(I − 1, I)]Q(I)

αaer(I)+S1/S2αmol(I) + {Q(I) + Q(I − 1) exp[A(I − 1, I)]}∆r, (4)

where Q is the range-corrected (R-corrected) lidar dataQ(r) = P (r)·r2, ∆r is the range resolution, A is given by

A(I, I + 1) = (S1−S2)[βray(I) + βray(I + 1)]∆r.

S1 and S2 are the extinction-to-backscatter ratios foraerosols and molecules, respectively; βray is the rayleighscattering coefficient. For molecules, the ratio remainsconstant when S2 = 8π/3. S1 is always assumed to bea constant with range in pure Mie lidar analysis. Butin our work, S1 is a real measured value by the Ramanmethod presented above.

The lidar system consisted of three parts: a laser,a telescope optical receiver, and a signal acquisitionrecorder. Detailed specifications are shown in Table 1.The system was based on a double frequency Nd:YAGlaser at 532 nm. The Mie backscatters at 532 nm and thenitrogen Raman backscatters at 607 nm were collectedby a Schmidt Cassegrain telescope, and were separatedby dichroic mirrors. To reduce sky background and blockstrong Mie signals return into the Raman channel, inter-ference filters were employed. A semi-custom bandpassfilter and a standard bandpass filter were used in Ra-man and Mie channels, respectively. In the Mie channel,the center wavelength (CWL) was 532 nm and the full-width at half-maximum (FWHM) was 3±0.5 nm. In theRaman channel, the center wavelength was 607 nm andthe FWHM was 5.0±1 nm. The out-of-band rejection(blocking) of the filter was <10−4. In the Raman chan-nel, two of the same bandpass filters were used, and theout-of-band rejection of dichroic mirror before the filterswas about 10−2. So the total out-of-band rejection in theRaman channel was about 10−8−10−9. The accumulatedminimum transmission of Raman filters was 25%. TheHamamatsu photomultipliers (PMTs) 7400 series wereselected to detect Mie and Raman signals. These are sub-miniature PMT tubes with 16-mm diameter and 12-mmseated length, and the electron multiplier was composedof metal channel dynodes. The peak wavelengths are500 and 630 nm for the Mie channel and Raman chan-nel, respectively. The data acquisition and control partcombined a powerful analog-to-digital (A/D) converter(12 bit at 20 MHz) with a 250-MHz fast photon countingsystem. In the Raman channel, the signal is weak, so wechose the photon counting method. In the Mie channel,

an A/D converter worked when the signal was from thelow atmosphere while photon counting worked when thesignal was from the upper atmosphere.

Our lidar station was at Wuhan University (30◦32’N,114◦21’E) located at the center of Wuchang Town. It isan industrial city and is densely populated. From the an-nual average air pollution indices[9], industrial pollutionemissions are still important contents of the atmosphere,and dust emissions from people’s daily life also make largecontributions to the air aerosols. Inhalable particulatesexceed the allowed standard every year and the concen-

Table 1. Specification Parameters of Raman/MieLidar

Transmitter:Nd:YAG Laser

Wavelength 532 nm

Pulse Energy 200 mJ

Pulse Repetition Rate 10 Hz

Pulse Width ∼8 ns

Receiver:Schmidt Cassegrain Telescope

Optical Diameter 356 mm

Focal Length 3556 mm

Detector:PMT Hamamatsu 7400

Mie Channel

CWL 532 nm

FWHM 5±1 nm

Out-of-Band Rejection <10−4

Minimum Transmission 50%

Raman Channel

CWL 607 nm

FWHM 3±0.5 nm

Out-of-Band Rejection <10−4

Minimum Transmission 50%

Data Acquisition

Analog Acquisition 12 bit, 20 MHz

Photon Counting 250 MHz

June 10, 2010 / Vol. 8, No. 6 / CHINESE OPTICS LETTERS 535

Fig. 1. (a) R-corrected Mie backscattering at 532 nm, and(b) R-corrected Raman backscattering at 607 nm observed at21:40 local time on September 12, 2009.

Fig. 2. Comparison of the Raman method and the Mie Fer-nald solution. (a) Aerosol extinction coefficient αaer, (b)backscattering coefficient βaer, and (c) aerosol extinction tobackscatter coefficient ratio S1 profiles at 532 nm observed at21:40 local time on September 12, 2009.

Fig. 3. Comparison of the Raman method and the Mie Fer-nald solution. (a) Aerosol extinction coefficient αaer, (b)backscattering coefficient βaer, and (c) aerosol extinction tobackscatter coefficient ratio S1 profiles at 532 nm observed at21:50 local time on September 12, 2009.

trations of both SO2 and acid rain have increasing trends.On the whole, the concentration of aerosols produced byindustry and human life is high and the particles’ size isrelatively large.

The vertical profiles of the aerosol extinctioncoefficient, backscatter coefficient and lidar ratio weremeasured by our multi-channel Raman/Mie lidar sys-tem. These optical properties of aerosol can reflect thequantity, size, and other information. of the particles atdifferent heights in the atmosphere.

Figure 1 shows the R-corrected Mie and Ramanbackscatter signals at 21:40 local time on September 12,2009. In Fig. 2, the solid and dashed lines represent theretrieved profiles by the Raman method and Mie Fernaldsolution, respectively. The vertical range resolution is 7.5m, and the signals are smoothed with a window length of

150 m. The aerosol extinction coefficient profile (solidline) shown in Fig. 2(a) is retrieved by Eq. (1) using thenitrogen Raman scattering signals at 607 nm. In orderto calculate backscatter coefficient in Eq. (2), the heightof 6 km is chosen as the reference r0. The lidar ratioprofile retrieved above by the Raman method is shownin Fig. 2(c). We can see that it varies with height andthe values are mostly between 20 and 60 sr from 0.5 to3 km. During the start of September, rainfall is rarein Wuhan. When temperature decreases at night, theair is heavy with moisture in the near-surface. Coupledwith the recent straight air currents in high altitudes,a thick inversion layer is very unfavorable to pollutantdispersion. The content of air particulates increases andthe size is bigger, so the lidar ratio is relatively smaller.This result is very close to the observation taken in thePearl River Delta in southern China in 2004[10]. Theprofile of the lidar ratio is used as the prerequisite inthe Mie Fernald method. The dashed lines in Figs. 2(a)and (b) present the aerosol extinction coefficient profileand backscatter coefficient profile by Mie data analyses.Figure 3 shows the retrieved results at 21:50 local time.From this experiment, we can see that there is a goodagreement between them. The correlation shows that op-tical filtering in the Raman and Mie channels is perfect,and the two channels receive the same field of view. Thisalso illustrates that our multi-channel Raman/Mie lidarsystem has good performance and is in well adjustment.

Figure 4 shows all the observations at 21:00 − 22:30local time on September 12, 2009. The change of aerosolsduring this period is clear. From Figs. 4(a) and (b), adownfall movement of aerosols in the atmosphere from2.3 km is obvious, and we can also see the variation of thelidar ratio in Fig. 4(c). At a reference height, the valueof the lidar ratio becomes bigger as time passes. Before22:00, most lidar ratios are below 50 sr, and after 22:00the values are becoming bigger to about or above 50 sr.

Fig. 4. (a) Aerosol extinction coefficient, (b) backscatter-ing coefficient, and (c) lidar ratio at 532 nm observed at21:00−22:30 local time on September 12, 2009.

536 CHINESE OPTICS LETTERS / Vol. 8, No. 6 / June 10, 2010

This means that larger sized aerosols are dropping andthe atmosphere becomes clearer as the night is falling.

In conclusion, we use the multi-channel Raman/Mie li-dar system to observe the profiles of aerosol optical prop-erties by Raman data and Mie data analyses in lower tro-posphere at Wuhan. The experimental results illustratethat this system has good performance. It will be used toperform transient and long-term observations for aerosolparticle activities. This can help us to understand theeffect of aerosols on the atmospheric visibility, formationof cloud and rain, solar radiation, and even the wholeclimate system.

This work was supported by the National “973”Program of China (No. 2009CB723905), the National“863” Program of China (No. 2009AA12Z107), andthe National Natural Science Foundation of China(Nos. 10978003 and 40871171).

References

1. Y. Sasano, E. V. Browell, and S. Ismail, Appl. Opt. 24,3929 (1985).

2. J. D. Spinhirne, J. A. Reagan, and B. M. Herman, J.Appl. Meteorol. 19, 426 (1980).

3. C. Y. She, R. J. Alvarez II, L. M. Caldwell, and D. A.Krueger, Opt. Lett. 17, 541 (1992).

4. A. Ansmann, M. Riebesell, and C. Weitkamp, Opt. Lett.15, 746 (1990).

5. D. N. Whiteman, Appl. Opt. 38, 3360 (1999).

6. A. Albert, W. Ulla, R. Maren, W. Claus, and M. Wal-fried, Appl. Opt. 31, 7113 (1992).

7. F. G. Fernald, Appl. Opt. 23, 652 (1984).

8. H. Liu, Z. Ge, Z. Wang, W. Huang, and J. Zhou, ActaOpt. Sin. (in Chinese) 28, 10 (2008).

9. H. Li, Q. Yu, and J. Shen, Industrial Safety and Envi-ronmental Protection (in Chinese) 33, 2 (2007).

10. M. Tesche, A. Ansmann, D. Miller, D. Althausen, R.Engelmann, M. Hu, and Y. Zhang, Appl. Opt. 46, 6302(2007).