Measurements of aerosol phase function and vertical backscattering coefficient using a charge-coupled device side-scatter lidar

Transcription

1 Measurements of aerosol phase funtion and vertial baksattering oeffiient using a harge-oupled devie side-satter lidar Zongming Tao, 1,2 ong Liu, 2,* Zhenhu Wang, 2 Xiaomin Ma, 1 Qinge Zhang, 1 Chenbo Xie, 2 Guangyu Bo, 2 Shunxing Hu, 2 and Yingjian Wang 2 1 New Star Institute of Applied Tehnology, Hefei, Anhui, 2331, China, 2 Key Laboratory of Atmospheri Composition and Optial Radiation, Anhui Institute of Optis and Fine Mehanis, Chinese Aademy of Sienes, Hefei, Anhui, 2331 China * dliu@aiofm.a.n Abstrat: By using a harge-oupled devie (CC) as the detetor, sidesatter lidar has great potential appliations in the near range atmospheri detetion. A new inversion method is proposed for CC side-satter lidar (Clidar) to retrieve aerosol phase funtion and vertial baksattering oeffiient. Case studies show the retrieved results from Clidar are in good agreements with those obtained from other instruments. It indiates that the new proposed inversion method is reliable and feasible and that the Clidar is pratiable. 213 Optial Soiety of Ameria OCIS odes: (1.111) Aerosols; (28.364) Lidar; (29.582) Sattering measurements. Referenes and links 1. R. J. Charlson, S. E. Shwart, J. M. Hales, R.. Cess, J. A. Coakley, Jr., J. E. Hansen, and. J. Hofmann, Climate foring by anthropogeni aerosols, Siene 255(543), (1992). 2. A. Ansmann,. Althausen, U. Wandinger, K. Franke,. Miiller, F. Wagner, and J. Heintenberg, Vertial pro ling of the Indian aerosol plume with six-wavelength lidar during INOEX: a rst ase study, Geophys. Res. Lett. 27(7), (2). 3.. M. Winker, J. Pelon, and M. P. MCormik, The CALIPSO mission: Spaeborne lidar for observation of aerosols and louds, Pro. SPIE 4893, 1 11 (23). 4. F. Mao, W. Gong, and J. Li, Geometrial form fator alulation using Monte Carlo integration for lidar, Opt. Laser Tehnol. 44(4), (212). 5. J. E. Barnes, S. Bronner, R. Bek, and N. C. Parikh, Boundary layer sattering measurements with a hargeoupled devie amera lidar, Appl. Opt. 42(15), (23). 6. J. E. Barnes, N. C. Sharma, and T. B. Kaplan, Atmospheri aerosol profiling with a bistati imaging lidar system, Appl. Opt. 46(15), (27). 7. F. G. Fernald, Analysis of atmospheri lidar observations: some omments, Appl. Opt. 23(5), (1984). 8. Z. Tao,. Liu, C. Xie, X. Ma, X. Meng, S. Hu, and Y. Wang, A numerial inversion method for CC sidesatter lidar, in Proeedings of International Conferene on Remote Sensing, Environment and Transportation Engineering (Aademi, 213), pp T. Nakajima, G. Tonna, R. Rao, P. Boi, Y. Kaufman, and B. Holben, Use of sky brightness measurements from ground for remote sensing of partiulate polydispersions, Appl. Opt. 35(15), (1996). 1. Introdution Aerosol plays a strong role in global limate hange both diretly, by affeting Earth s radiation balane, and indiretly, by affeting loud properties [1]. Understanding their ontribution to limate hange requires the vertial distribution of the aerosol [2]. As aerosol types, hemistry, onentrations are highly variable and strongly altitude-dependent, measurements of aerosol properties as a funtion of altitude are espeially important. The majority of aerosol exists in a layer of few kilometers near the Earth s surfae whih has a very important impat to human beings. This aerosol-rih layer is generally alled the atmospheri (or planetary) boundary layer. Remote sensing tehniques using a variety of instruments have been applied to aerosol detetion. Sun photometers provide the total olumn aerosol ontent, indiative of the aerosol properties integrated through the entire atmosphere. # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 1127

2 Baksatter lidar is apable of providing range-resolved atmospheri profiles ontinuously, making them valuable tool in the remote sensing of atmospheri aerosol [3]. But the baksatter lidar system has a shortoming in the lower hundreds of meters beause of the overlap fator aused by the onfiguration of the transmitter divergene and reeiver s fieldof-view(fov) at ranges lose to the instrument [4]. This limits the baksatter lidar to be applied to the near-range measurement, espeially for the fixed vertial-pointing apparatus. Appling a CC amera detetor, a side-satter lidar, in whih the amera is spatially separated to look the laser beam at a ertain distane, is a different onfiguration ompared with the baksatter lidar. The Clidar solves the overlap problem aused by baksatter lidar, and is espeially suitable to measure aerosol in the near range [5,6]. This paper is foused on how to obtain the aerosol phase funtion and how to solve the side-satter lidar equation. In the following paragraph, a numerial inversion and experimental method for Clidar is proposed. The skyradiometer retrieved aerosol phase funtion is used to make omparison. Combining the baksattering lidar measurement simultaneously, the vertial aerosol baksattering oeffiient obtained by Clidar is also retrieved and validated. 2. Apparatus The diagram of Clidar is shown in Fig. 1. For the side-satter Clidar onfiguration, the lidar equation an be desribed as [5] PKA P( θ ) = β ( θ) TTd r θ (1) where P( θ ) is the reeived power at height and sattering angle θ by a pixel, K is a system onstant inluding the optial and eletroni effiieny and A is the aperture of the reeiving optis, T and T r are the total (aerosol and moleular) atmospheri transmittane from the laser to distane and from distane along the slant path R to the CC reeiver, respetively. β ( θ ) is the total atmospheri side-sattering oeffiient. Beause of θ = π /2+ γ, dθ (equal to dγ ) is FOV of one pixel. The speifiations of the Clidar system are summaried in Table 1. Fig. 1. The measurement onfiguration of the Clidar. # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 1128

3 Table 1. The key speifiations of the Clidar system Laser (Quantel Brilliant) Nd:YAG Wavelength 532 nm Pulse energy 2 mj Repetition rate 1 H ivergene.5 mrad etetor (SBIG) ST-83M Pixel array Pixel sie (μm) Angle per pixel (deg/pix) ~.2 ark (eletrons/s/pixel) <.5e at C Full well apaity (eletrons) 25,5 A/ onveter (bits) 16 Cooling Single TE, Ative Fan, ~35 C elt Temperature regulation ( C) Closed loop, ±.1 Wide-angle lens Walmexpro f/2.8 Lens foal length (mm) 14 CC sensor Kodak KAF-83 Quantum effiieny (532nm) ~55% Interferene Filter (Semrok orporation) Bandwidth (nm) 25.6 Peak transmittane ~95% Computer (Lenovo 2) For the CC reeiver, side-satter lidar signals reord by different pixels. From the geometry, the spatial resolution d is expressed as 2 R d = d θ (2) For the Clidar system, the FOV of eah pixel dθ is measured by alibration experiments; the measured dθ is an approximate onstant, so the spatial resolution d inreases with the slant distane R. By omparing the Eqs. (1) and (2) with orresponding baksatter lidar equations, Clidar has two main differene aspets with the baksatter lidar. One is that the spatial resolution is 2 not onstant, the other is reeived power is not 1/R dependene. The anellation of the 2 1/R dependene has a benefiial onsequene for the dynami range of the Clidar system. For the short range detetion, suh as the atmospheri boundary layer, Clidar an provide a fine vertial resolution due to the elaborate optis design. 3. Methodology As to the baksatter lidar equation, Fernald provided an interative inversion solution for two-omponent atmosphere [7]. But to the Clidar equation, Fernald s inversion method is not suitable, due to the differene of the atmospheri transmittane and the adoption of the aerosol phase funtion. The side-sattering oeffiient β ( θ ) and the total sattering oeffiient β s have the relationship as β ( θ) = pf ( θ) β s (3) where pf ( θ ) is the aerosol phase funtion. If the baksattering oeffiient is known, the side-sattering oeffiient β ( θ ) an be expressed as β ( π ) β ( θ) = ( ) ( ) ( ) pf ( ) pf θ = f θ β π (4) π f( θ) = pf( θ)/ pf( π) is the relative aerosol phase funtion. The Clidar equation in forwardsattering ( θ 9 ), an be written as # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 1129

4 PKA P( θ) = [ β1( π) f1( θ) + β2( π) f2( θ)]exp{ [ ( α 1( ) + α2( )) d + ( α ( ) + α ( )) d /os θ]} dθ 1 2 for bakward-sattering ( 9 θ 18 ), the equation an be written as PKA P( θ) = [ β1( π) f1( θ) + β2( π) f2( θ)]exp{ [ ( α 1( ) + α2( )) d + ( α ( ) + α ( )) d / os( π θ)]} dθ 1 2 where α ( ) is the extintion oeffiient, subsripts 1 and 2 stand for aerosol and moleule, respetively. The first transmission terms without sattering angle θ in both Eqs. (5a) and (5b) indiate the path from the laser to the satter region. The other parts of the transmission terms indiate the path from the satter region to the CC amera. In general, for Eqs. (5a) and (5b), there are three unknown variables, i.e. phase funtion, baksatter and extintion oeffiient of aerosol. This is the similar senario one met as the retrieval of the baksatter lidar [7]. A prior assumption has to be given, i.e. lidar ratio (extintion-to-baksatter ratio) of the aerosol, in order to redue the unknowns. The value of 5 Sr is used as lidar ratio at 532 nm wavelength in our algorithm. Thus, if the phase funtion is determined, the baksatter and extintion oeffiient of the aerosol an be retrieved by Clidar. 3.1 Phase funtion retrieval Using horiontal pointing onfiguration of Clidar and baksatter lidar, a retrieval method of the aerosol phase funtion is proposed as the following. Let Clidar and baksatter lidar point horiontally at the same time. In this ase, the onstant aerosol extintion oeffiient an be assumed due to the homogeneity. Let θ as a referene point, the moleular baksatter oeffiient, extintion oeffiient and phase funtion an be determined from a standard atmosphere. Aerosol baksatter oeffiient and extintion oeffiient an obtained from baksatter lidar in the simultaneous experiment. At the referene point θ, Eqs. (5a) and (5b) beome PKA P f f d (6a) ( θ) = [ β1( π) 1( θ) + β2( π) 2( θ)]exp{ [( α1+ α2) + ( α1+ α2)( )/os θ]} θ PKA P f f d (6b) ( θ ) = [ β ( π) ( θ ) + β ( π) ( θ )]exp{ [( α + α ) + ( α + α )( ) / os( π θ )]} θ is the distane orresponding to θ. C an be determined by (5a) (5b) P( θ ) C = [ β ( π) f ( θ ) + β ( π) f ( θ )] dθ (7) The attenuated side-satter, β ( θ ), an then be omputed P( θ ) β ( θ) = (8) C # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 113

5 From Eqs. (6a) and (6b), Eq. (8) an be written as β ( θ) = [ β ( π) f ( θ) + β ( π) f ( θ)]exp{ [( α + α )( ) ( α + α )(( )/os θ ( )/os θ )]} dθ 1 2 β ( θ) = [ β ( π) f ( θ) + β ( π) f ( θ)]exp{ [( α + α )( ) ( α + α )(( )/os( π θ) ( )/os( π θ ))]} dθ 1 2 (9a) (9b) Firstly, θ 18 is seleted as the referene point, i.e. f 1 ( θ ) = 1. Using the numerial inversion method [8], the relative phase funtion of aerosol within the satter angles from 9 to 18 an retrieved from Eq. (9b). Then, setting θ = 9 as the referene point, using above method, the relative phase funtion of aerosol within the satter angles from to 9 an retrieved from Eq. (9a). There is a proportional onstant between the relative phase funtion and the phase funtion of aerosol, this proportional onstant ould be omputed by normaliation or by saled to math the known phase funtion of aerosol from other instruments. In our algorithm, this onstant is omputed by saled to math the known phase funtion of aerosol from POM2 measurement at 5 nm wavelength. 3.2 Vertial baksatter oeffiient retrieval Aerosol phase funtion is determined by sie distribution, refrative index and omponent, is not dependent on the onentration. Assuming the aerosol phase funtion as uniform in the planetary boundary layer, the baksatter oeffiients loud be retrieved from Clidar as shown in the following way. Let Clidar point vertially, and keep the polariation angle the same for both the vertial and horiontal measurements. In this ase, the aerosol phase funtion measured by Clidar in horiontal pointing is used to retrieve the vertial aerosol baksatter oeffiient. Seleting θ as a referene point, the aerosol baksatter oeffiient value at this point an be obtained from the data of baksatter lidar in the simultaneous experiment. The moleular baksatter, extintion oeffiients and phase funtion an be determined from a standard atmosphere. Using above numerial inversion method, the baksatter and extintion oeffiient of aerosol an be retrieved from Eqs. (5a) and (5b) if the lidar ratio is given. 4. Case study and validation In the evening on Sep.9, 213, the aerosol phase funtion was measured with Clidar pointed horiontally. The distane between the CC and the laser beam was 7.91 m. The sky ondition of this evening was overast, the aerosol extintion oeffiient was.88 km 1 near the surfae, the orresponding visibility was around 4.5 km. Beause of the total view angle of the CC amera is about 64 (64 in horiontal, 52 in vertial), the enter axis of the CC amera was put on three diretions in turn in order to get to 18 sattering signals. First, CC enter axis was pointed to 15 sattering angle, with the exposure time of 1s. Seond, CC enter axis was pointed to 9 sattering angle, with the same exposure time. Last, CC enter axis was pointed to 3 sattering angle, with the exposure time of about 1s. Figure 2(a) shows the signals measured at three diretions. The blak solid line in Fig. 2(b) is the retrieved aerosol phase funtion using the new proposed method disussed above. In order to validate the results, the aerosol phase funtion measured before the sunset by skyradimeter (POM2) at the same loation is plotted in red solid line in Fig. 2(b). The skyradimeter is an important instrument of SKYNET projet, and the SKYRA.pak version 4.2 software [9] was used to retrieve aerosol optial and mirophysial properties inluding aerosol phase funtion at the wavelengths of 34nm, 38nm, 4nm, 5nm, 675nm, 87nm, and 12nm by using a radiative transfer ode and an inversion sheme. # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 1131

6 Skyradiometer uses the Sun radiation, while Clidar uses laser. Skyradiometer measures the olumnar atmospheri aerosol properties, while the Clidar measures phase funtion of aerosol whih lies in the laser beam path. Figure 2(b) shows that the retrieved aerosol phase funtion from Clidar has a good agreement with the POM2 measurement. Within to 15 sattering angle, aerosol phase funtion hanges greatly with sattering angle. It is diffiult to detet aerosol phase funtion preisely. So for above and following two ases, the retrievals from to 15 in the sattering angle are omitted. Fig. 2. Measurement by Clidar on Sep. 9, (a) raw signal (b) retrieved aerosol phase funtion In the evening on Sep. 16, 213, the aerosol phase funtion was measured again for further omparison. The distane between the CC and the laser beam was 6.71 m. The sky ondition of this evening was lear, the aerosol extintion oeffiient was.33 km 1 near the surfae, the orresponding visibility was around 11.5 km. The CC enter axis was put on three diretions in turn. The detailed exposure time and CC pointers are the same as the last experiment on Sep.9, 213. Figure 3(a) shows the signals measured at three diretions. Comparison with POM2 in Fig. 3(b) again indiates a good agreement exept the data point at about 15 and 15 sattering angles. Fig. 3. Measurement by Clidar on Sep. 16, (a) raw signal (b) retrieved aerosol phase funtion In the evening on Ot.15, 213, the Clidar was pointed vertially to measure the aerosol baksatter oeffiient profile. The distane between CC and laser beam was 2.89 m, and the sky ondition of this evening was lear. The CC enter axis was put on 65 tilting angle, with the exposure time of about 1s. Meanwhile, two baksatter lidars worked horiontally and vertially respetively at the same plae and POM2 was also deployed at the same loation to provide the aerosol phase funtion. Figure 4 is the signal measured by the Clidar at 19:4 at loal standard time (LST), Fig. 4(a) is the image of the vertial transmitting laser beam, Fig. 4(b) is the signal profile with pixel number and Fig. 4() is the signal profile with altitude. Figure 5(a) and 5(b) show the retrieved aerosol baksattering oeffiients by different instruments at 19:4 and 2:4 LST, respetively. The blak solid lines in Figs. 5(a) and 5(b) are the retrieved aerosol baksatter oeffiient profiles using vertial-pointing baksatter lidar, the red solid lines in Figs. 5(a) and 5(b) are the aerosol baksattering oeffiient profiles from the Clidar using the above proposed method, and the blak stars are # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 1132

7 the aerosol baksattering oeffiient near the surfae by the horiontal-pointing baksatter lidar. Fig. 4. Signal measured by Clidar on Ot. 15th, (a) the image of laser beam, (b) the signal profile versus pixel number and () the signal profile versus altitude Fig. 5. Retrieved profile of aerosol baksattering oeffiient by vertial-pointing baksatter lidar, Clidar and baksatter lidar pointing horiontally. (a) measured at 19:4 LST, (b) measured at 2:4 LST For both Clidar and baksatter lidar retrieval methods, the aerosol extintion-tobaksatter ratio is assumed 5 Sr, the moleular extintion oeffiient and baksattering oeffiient are from the U. S. standard atmosphere model(1976). From Fig. 5(a), one an see that between.9 km and 1.5 km altitude, the aerosol baksattering oeffiients measured by the Clidar and vertial-pointing baksatter lidar are in a good agreement. From the surfae to about.9 km, the aerosol baksattering oeffiient retrieved from vertial-pointing baksatter lidar is not orret due to the overlap fator, so there is a big differene between vertial-pointing baksatter lidar and Clidar results. The Clidar results have a good agreement with the horiontal-pointing baksatter lidar retrieval at 5m above ground level (AGL). In pratie, for avoiding the obstrutions, the laser beam has an elevation angle of 2 degrees when making the horiontal measurement. The slant range orresponding to 5 m in vertial is about 1.5 km horiontally. Figure 5(b) shows the similar results as Fig. 5(a). The aerosol baksattering oeffiient in the near surfae hanged from.84 km 1 Sr 1 to.12 km 1 Sr 1 during this one hour duration. By ombining vertial-pointing baksatter lidar and Clidar measurement, retrieved baksattering oeffiient profile from surfae to 7 km AGL at 19:4 LST and 2:4 LST on Ot.15, 213 are plotted in Fig. 6. From Fig. 6, one an see that there were two aerosol layers separated at about 2 km in this evening. Within one hour, aerosol baksattering oeffiient in the above aerosol layer hanged little, while the baksattering oeffiient in the lower layer hanged greatly in the vertial shape. In the very lose surfae, there is still variation of the aerosol in the vertial struture whih an t be deteted by the ordinary baksattering lidar due to the dead one of the overlap fator, while the Clidar an reveal it. # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 1133

8 Fig. 6. Aerosol baksattering oeffiient profiles retrieved by ombined lidars 5. isussion and onlusions A new method of Clidar is proposed to detet the aerosol phase funtion. The aerosol phase funtion retrieved by a skyradiometer is used to make the omparison. In the omparison experiments, Clidar worked at nighttime in order to redue bakground noise, while the aerosol phase funtion retrieved from the skyradiometer is before the sunset. This means that the omparison experiments is not simultaneous, but at the same plae. From above two ases, aerosol phase funtion from Clidar and skyradiometer are onsistene within 15 to 18 in sattering angle whih indiates the omposition and sie distribution of aerosol not hanging greatly within that time. The errors in the aerosol phase funtion retrieval from Clidar are due to the bakground noise, interfering light, CC dark urrent level, the model moleular parameters and so on. Based on the Clidar Eq. (9), the impat of the moleular phase funtion is getting smaller while aerosol baksattering oeffiient is getting bigger. Aerosol baksattering oeffiient in Fig. 3(b) is smaller than that in Fig. 2(b), so the deviation in Fig. 3(b) may be from moleular phase funtion variation. The omparisons with POM2 in Figs. 2(b) and 3(b) show that the method in whih aerosol phase funtion is retrieved by Clidar is feasible, and the retrieved aerosol phase funtion is reliable. Figure 3(b) shows that aerosol baksattering oeffiient profile retrieved by Clidar is feasible and reliable too. The Clidar is a good supplement to the baksatter lidars. Combined these two tehniques, the vertial profile of aerosol baksattering oeffiient an be measured from the very beginning of the surfae and the overlap fator of the baksatter lidar an also be obtained for the fixed vertial-pointing lidars. Aknowledgments This work was supported by the National Natural Siene Foundation of China under Grant No , No and No The skyradiometer POM2 retrieval is provided by the SKYNET Hefei observatory. # $15. US Reeived 6 Nov 213; revised 1 e 213; aepted 13 e 213; published 1 Jan 214 (C) 214 OSA 13 January 214 Vol. 22, No. 1 OI:1.1364/OE OPTICS EXPRESS 1134