We developed a multiple-field-of-view multiple-scattering polarization lidar (MFMSPL) to study the microphysics of optically thick clouds. Designed to measure enhanced backscattering and depolarization ratio comparable to space-borne lidar, the system consists of four sets of parallel and perpendicular channels mounted with different zenith angles. Depolarization ratios from water clouds were large as observed by MFMSPL compared to those observed by conventional lidar. Cloud top heights and depolarization ratios tended to be larger for outer MFMSPL channels than for vertically pointing channels. Co-located 95 GHz cloud radar and MFMSPL observations showed reasonable agreement at the observed cloud top height.
© 2016 Optical Society of America
It has been widely accepted that lidar cannot penetrate optically thick clouds. The typical detectable optical thickness is approximately 3 for ground-based lidar, such that it generally detects the bottom of water clouds when they occur . The optical thickness limitation appears to be different for space-borne lidar observations. Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) carries dual-wavelength polarization lidar and has a much larger footprint size (~90 m) compared to that of conventional ground-based lidar (about 1 m at 1 km height) . Because of this large footprint size, it can collect more multiple scattered signals and the resulting optical thickness measurable from space-borne lidar is larger than those from ground-based lidar observations .
When a single scattering process is dominant, spherical particles produce zero depolarization ratio. In contrast, randomly oriented ice particles produces large depolarization ratio(~40%). Therefore depolarization ratio has been used for cloud phase classification in case of ground-based polarization lidar observations . However, similar discrimination is not straightforward for space-borne lidar observations, since observed depolarization ratios often well exceed 30% when water clouds are observed as for the case of CALIPSO lidar . Therefore there is a fundamental difficulty in discriminating water from ice by depolarization ratio alone. It was shown that cloud particle phase can be determined by the combined use of depolarization ratio and the ratio of the attenuated backscattering coefficients for two vertically consecutive layers . Results from Monte Carlo simulations indicates depolarization values tend to increase as extinction increases, which can be explained by large contributions from multiple-scattering, and the values reflect cloud microphysics . Thus, depolarization ratio together with attenuated backscattering measurements is useful for obtaining cloud phase and cloud microphysical properties.
There have been several attempts to extend the ground-based conventional lidar. For example, “Off-beam lidar” or “multiple-scattering lidar” have been developed [7,8] in which signals from a direction non-parallel to that of the emitted laser beam have been observed using a larger field of view (FOV) to detect lidar signals affected by multiple-scattering . Observations made by such lidar systems have successfully captured signals from optically thick clouds. Airborne multiple-scattering lidar  and multiple FOV lidar [10,11] have also been developed for cloud measurements. Most of these lidar systems can only be operated at night and cannot observe depolarization ratios, with the exception of one system, in which the largest FOV is limited to 12 mrad to maximize polarimetric purity .
In this study, we introduce a lidar system that can detect backscattering coefficients and depolarization ratios similar to those observed by space-borne lidar. To overcome the limitations of ground-based lidar observations, we developed a multiple-field-of-view multiple-scattering polarization lidar (MFMSPL) system by combining four channels with different zenith angles, such that each receiver channel has a 10 mrad FOV and a polarization function with a total FOV of 70mrad. One very early result obtained by the MFMSPL has already been reported . In this study, we describe the MFMSPL settings and synergetic observations with 95 GHz cloud radar, conducted at the National Institute of Environmental Studies (NIES) in Tsukuba, Japan, to evaluate the system. In addition, we describe the MFMSPL specifications and calibration process for all of the channels and present results from observations of low-level water clouds, including the depolarization ratio. To evaluate MFMSPL performance, a co-located 95 GHz cloud radar FALCON-1  and MFMSPL were simultaneously operated.
2. Description of MFMSPL
The MFMSPL has eight channels (four parallel, four perpendicular) with detectors. The FOV of each telescope was chosen to be 10 mrad to maximize polarimetric purity. Four parallel-and perpendicular channel telescopes were inclined with different angles from the vertical direction along a line. Odd-numbered channels detected parallel signals; the tilt angles of channels 1, 3, 5, and 7 from the vertical direction were 0, 10, 20, and 30 mrad, respectively. Even-numbered channels were used to detect perpendicular signals; channels 2, 4, 6, and 8 had tilt angles of 0, 10, 20, and 30 mrad, respectively. An angle meter (Wixey digital angle gauge and level, model WR365) was used to mount each telescope with the accuracy of ± 0.1 deg. ( = 1.7mrad). We investigated the impact of this error on the observed attenuated backscattering coefficient (βatt) and depolarization ratio based on Monte Carlo simulation . The errors in (βatt) and depolarization ratio were estimated to be smaller than 1% and about 3%, respectively,
The schematic of the MFMSPL system is shown in Fig. 1, and the configuration of the MFMSPL is shown in Fig. 2. Each receiver channel consisted of a polarizer, interference filter, 5 cm diameter lens, and photomultiplier tube (PMT) with 16-bit capability as a detector. In addition, neutral-density (ND) filters were also adopted for channels 1 and 2 to prevent saturation. Transmittance of the ND filters for channels 1 and 2 were determined to be 1% and 7%, respectively. A neodymium-doped yttrium aluminum garnet (Nd:YAG) laser was used. Beam divergence was equal to or smaller than 0.1 mrad. Pulse repetition frequency was 10 Hz. A second harmonic wavelength of 532 nm was detected and laser peak power was 165 mJ/pulse for 532nm. Laser spot size was 30mm at the exit of beam expander. The temporal and vertical resolutions of the received data were originally 10 s and 6 m, respectively. MFMSPL specifications are summarized in Table 1.
3. Calibration procedures
There are two steps used for calibration of the MFMSPL to obtain attenuated backscattering coefficient (ßatt) for each channel. The first is absolute calibration, which is performed by comparing signals observed by Mie-type NIES vertical-pointing lidar and those observed from channel 1 (vertical-pointing). Calibration of the NIES lidar was performed in advance, according to the procedure described in . The NIES lidar was necessary for calibration because the MFMSPL cannot detect signals from altitudes higher than 8 km during the day, due to its limited sensitivity, and self-calibration was not possible.
Clear sky data were selected for channel 1 calibration to minimize contributions from multiple-scattering when both lidar systems observed aerosols and molecules signals. Thus, attenuated backscattering coefficients should be the same between the two systems after calibrating MFMSPL channel 1. The NIES lidar was operated for 5 min with a 10 min intermissions. Both the MFMSPL and NIES lidar data were first averaged for 5 min at a horizontal and 30 m vertical resolution.
Calibration was performed with approximately 2 h data obtained between 300 m and 900 m in altitude. Then a regression line between the two data sets was determined by applying least-squares curve fitting to obtain slope and intercept. Standard deviations of the calibration constant were approximately 10%. Relative calibration for channels 2 to 8 with respect to channel 1 was performed. For this purpose, all of the eight telescopes were pointed vertically and the polarizers for the perpendicular channels were removed to detect the parallel components of the signals. A regression line between signals from channels 1 and 2 was derived using clear sky data between 200 and 800 m for 500 min. Data were averaged for 5 min and 30 m in the vertical direction. Similarly, six regression lines corresponding to remaining channels (channels 3–8) were estimated against channel 1. Standard deviations for the calibration constant for the seven channels were estimated to range from 0.03% to 0.3%. Thus, the overall accuracy of the MFMSPL was approximately 10%.
4. Water cloud observations
4-1. Attenuated backscattering coefficients for the MFMSPL parallel channels
Observations of attenuated backscattering coefficient βatt for channels 1, 3, 5, and 7 obtained on 5 March, 2015, are shown in Fig. 3. The data were plotted for 14000 s and 86400 s, corresponding to 3:53 Coordinated Universal Time (UTC) and 24:00 UTC, respectively.
Low-level clouds were observed between 2 and 3 km altitude. A two-layer structure was found for records between 60000 and 86400 s. The on-beam channel (channel 1) showed larger βatt near the cloud bottom than channels 3, 5, and 7, because only channel 1 detected single scattering photons, whereas off-beam channels did not.
The channel 1 βatt decreased with altitude from strong attenuation, due to large water cloud extinction with small amounts of compensation from multiple-scattering . Conversely, off-beam channel βatt showed smaller differences in the vertical direction compared to channel 1, because this channel detected more secondary, multiple scattered photons, and the clouds was not affected by extinction. Consequently, cloud top heights detected by the off-beam channels were higher than those detected by channel 1. Interestingly, only channel 1 detected aerosol signals below clouds; off-beam channels did not detect such signals below the cloud base when there was no drizzle. Channels 1 and 3 observed drizzle at about 30000 s and also between 71000 and 82000 s, whereas the channel 1 βatt value was much larger than that from channel 3. This occurred because multiple-scattering due to drizzle is generally weak due to the small optical thickness of the clouds, such that single scattering dominates. This was also observed by channels 5 and 7, neither of which detected the drizzle.
To illustrate the performance of the MFMSPL, NIES lidar data are shown in Fig. 3(e). Cloud top height by NIES lidar was lowest relative to the MFMSPL channels. We therefore conclude that the MFMSPL can observe larger cloudy areas compared to the NIES lidar due to its ability to detect greater multiple-scattering contribution, as expected. The MFMSPL may overestimate the cloud top boundary due to pulse stretch; that is, some photons may remain in a cloud top layer longer than single scattered photons, which leads to a higher and artificial cloud top. Thus, we conducted synergistic observations of the cloud radar and the MFMSPL to evaluate cloud boundary information obtained by the MFMSPL.
4-2. Depolarization ratio due to multiple-scattering
The depolarization ratio for the four tilt angles was estimated using eight channels (Fig. 4). Depolarization ratio for four different angles (mrad) was estimated by
Smaller depolarization ratios were obtained from the on-beam channel than the off-beam channels . This can be explained as follows. As described in subsection 4-1, only channel 1 observed single scattering photons, and the single scattering depolarization ratio of spherical water particles should be 0%. Thus, the resultant was minimized; was close to 0% near the cloud bottom and reached 60% near the cloud top for the on-beam channel (Fig. 4(a)).
In general, the observed depolarization ratio increased as cloud altitude increased, likely due to multiple-scattering (Fig. 4(a)). The depolarization ratio observed by the NIES lidar was also very small value near the cloud bottom (Fig. 4(e)). was already large (about 20–30%) near the cloud bottom and reached 80% near cloud top at 10 mrad (Fig. 4(b)).
Cloud depolarization ratios do not always monotonically increase as altitude because the value depends on vertical structure of cloud type, number concentration, effective size and phase. For example, of the lower cloud layer between 60000 and 75000 s was about 20% near the cloud bottom at 2.2 km and about 60% near the cloud top at 2.4 km. The value for the second upper cloud layer was 30% at 2.6 km, near the cloud bottom, which was smaller than the cloud top value of the first cloud layer; the depolarization ratio finally reached 80% at 2.8 km, near the cloud top. Thus, the depolarization ratio reflected the inhomogeneity of the cloud microphysics and was affected by the vertical profile of any resultant multiple-scattering .
5. Comparison of MFMSPL and 95 GHz cloud radar
Co-located 95 GHz cloud radar FALCON-1 and MFMSPL observations were conducted at Tsukuba to evaluate MFMSPL performance in detecting optically thick clouds. When cloud radar and conventional lidar both detected the same clouds, cloud top height as detected by the cloud radar was often much higher than that detected by the lidar. Therefore, we used cloud radar to assess the cloud vertical extent detected by the MFMSPL.
Cloud radar tended to miss water clouds when they did not contain drizzle, but were able to detect clouds when drizzle appeared because the size of drizzle was usually larger than water cloud particles . On the other hand, lidar detects liquid water clouds regardless of the presence of drizzle . For this evaluation, we selected the same time periods as those previously shown, when both MFMSPL and FALCON-1 detected the same cloud at the same observation time. Minimum sensitivity of FALCON-1 is −30dBZ at 5km.
FALCON-1 detected clouds between 1.8 km and 2.9 km (Fig. 5(a)). Channel 1 of the MFMSPL showed lower cloud top heights compared to FALCON-1, due to strong attenuation of the lidar signals. The cloud top height detected by channel 1 was the lowest among all of the channels and cloud radar. The cloud top heights for channels 1, 5, and FALCON-1 were approximately 2.5 km, 2.9 km, and 2.9 km, respectively.
The corresponding morphology of cloud top boundaries between the cloud radar and channel 5 showed that channel 5 detected real cloud signals. Agreement between channel 5 and FALCON-1 on cloud top height indicated that the off-beam channel of the MFMSPL was indeed able to detect cloud top height.
In some records, the cloud top height detected by channel 5 was even higher than that detected by cloud radar (e.g., at 32000 s). This might be explained by the presence of small cloud particles near the cloud top regions; such small particles were only detected by off-beam channels such as channel 5, whereas the sensitivity of cloud radar was not great enough to detect them.
The bottom boundaries of hydrometeors (i.e., clouds and drizzle) were similar between FALCON-1 and channel 1, and signals due to drizzle were less apparent for channel 3 and were diminished for channel 5. Cloud radar signals were largely contaminated by drizzle; thus, it was difficult to determine cloud bottom altitude. Channel 5 did not detect signals due to drizzle (small β) below the cloud bottom (large β) and only channel 1 detected drizzle and the cloud bottom. Thus, unlike the cloud radar, combinations of on-beam and off-beam channels of the MFMSPL could be used to distinguish the cloud base from drizzling regions.
Similar comparisons between FALCON-1 and the MFMSPL were conducted when two-layer cloud systems were observed by the MFMSPL. Cloud radar showed the difficulty of distinguishing cloud and drizzle (Fig. 6(a)). Channel 1 detected the lower cloud bottom at 2.3 km and higher cloud bottom at 2.7 km before 76000 s and also detected signals due to drizzle below these cloud bottoms (Fig. 6(b)). Drizzle signals became weaker for channel 3 compared to channel 1 (Fig. 6(c)). The boundaries of the two-layer clouds were most clearly seen in channel 5 (Fig. 6(d)) and the clear detection of cloud boundaries may also make it possible to use the depolarization ratio from off-beam channels (Fig. 7(b), 7(c)).
The capability to detect cloud top boundary with the MFMSPL system was examined based on Monte Carlo simulations. We considered single-layer cloud with 600-m geometrical thickness assuming similar situation with Fig. 5 where cloud bottom and top heights are 2400m and 3000m. Liquid water content and effective radius are assumed to be 0.01g/m3 and 10µm, respectively. There are layers of molecules above cloud top. The vertical profile of the simulated βatt and depolarization ratio were shown in Figs. 8(a) and 8(b). There are distinct differences between βatt (more than 3 orders) and depolarization ratio within the cloud layer (below 3000m) and those at cloud-free region above cloud top. The simulation suggests that cloud top detected by the MFMSPL system is likely to be the actual cloud top or that at least restricted by the minimum detectable sensitivity of the system, and not artificial.
For the MFMSPL system, minimum sensitivity of βatt for Ch1 and CH2 were estimated to be about 10−7 [1/m/sr] and about 10−9 [1/m/sr] for off-beam channels, respectively, and the maximum altitude detected by on-beam channels are considered to be lower than other channels due to their lower sensitivity.
In addition, there are consistent features found in the observation (Fig. 5) and simulation;
- ・Depolarization ratio for off-beam channels are larger than those for on-beam channels. Maximum depolarization ratio was found near cloud top and the values were about 0.8.
- ・Outer channels show smaller βatt compared with channels with smaller tilt angles when the penetration depth from cloud base is small.
Since overlapping cloud regions detected by MFMSPL and FALCON-1 were large and depolarization information were obtainable in these regions, a radar and lidar algorithm used for ice microphysics analysis could be used to derive cloud microphysics with proper modification . In this regard, synergetic use of MFMSPL and cloud radar is a promising tool to study cloud and drizzle microphysics and this will be the focus of a future study.
We have developed a multiple-scattering lidar that has polarization function for four tilt angles. It consists of eight channels (i.e., four parallel channels and four perpendicular channels). The MFMSPL has been making continuous observations of clouds, day and night, since June 2013. The system was designed to detect lidar signals from optically thicker parts of clouds that a conventional lidar could not penetrate. It is also designed to study similar lidar signals that are heavily influenced by multiple-scattering detected by space-borne lidars such as CALIPSO .
The basic findings are as follows.
Cloud top height of low-level clouds detected by the MFMSPL was higher than that detected by NIES lidar. This is the first time that the off axis lidar depolarization ratio of clouds has been observed by a multiple-scattering lidar. Depolarization ratio from the MFMSPL was larger for off-beam channels than that by the on-beam ones. Based on Monte Carlo simulations for a single layer water cloud, the impact of miss-alignment of the channels on the observed attenuated backscattering coefficient and depolarization ratio turned out to be small.
The depolarization ratio values of low-level water clouds often exceed 60% when impacted by the multiple-scattering, which appears to be as large as those detected by Cloud-Aerosol Lidar with Orthogonal Polarization . Thus, this instrument can be used to simulate and test algorithms for space-borne lidar.
Co-located observations of the 95 GHz cloud radar and the MFMSPL were conducted to evaluate the performance of MFMSPL cloud detectability. Reasonable agreement between cloud radar and MFMSPL was achieved for the detection of cloud top height. Off-beam channels showed higher cloud top height compared to on-beam channels, and the off-beam channels showed even higher cloud top height than cloud radar. Monte Carlo simulation supported that the MFMSPL system can detect the actual cloud top boundary.
The MFMSPL might be a useful tool to study aerosol-cloud radiation interaction, because it detects signals from optically thicker parts of the clouds and can clearly distinguish among aerosols, clouds, and precipitation using multiple FOV information. These would be essential information to estimate radiative effects of clouds.
The MFMSPL will be used to evaluate an algorithm for cloud mask  and for cloud particle type  for CALIPSO lidar data. Furthermore, synergy use of the MFMSPL and 95 GHz cloud radar will be used to test algorithms for cloud microphysics using CloudSat and CALIPSO data . Details of the cloud mask and particle type classifications schemes that can be applied to the MFMSPL will be reported in future reports.
As a further extension of the system, implementation of circular polarization is worth to consider .
JSPS Kakenhi (JP25247078 and JP15K17762); Collaborated Research Program of Research Institute for Applied Mechanics, Kyushu University (Fukuoka, Japan).
Backward Monte Carlo code used in this study was originally developed by H. Ishimoto at Meteorological Research Institute.
References and links
1. K. Sassen and H. Zhao, “Lidar multiple-scattering in water droplet clouds: toward an improved treatment,” Opt. Rev. 2(5), 394–400 (1995). [CrossRef]
2. D. M. Winker, W. H. Hunt, and M. J. McGill, “Initial performance assessment of CALIOP,” Geophys. Res. Lett. 34(19L19803), L19803 (2007). [CrossRef]
3. J. E. Yorks, D. L. Hlavka, M. A. Vaughan, M. J. McGill, W. D. Hart, S. Rodier, and R. Kuehn, “Airborne validation of cirrus cloud properties derived from CALIPSO lidar measurements: Spatial properties,” J. Geophys. Res. 116(D19), D19207 (2011). [CrossRef]
4. K. Sassen, “The polarization lidar technique for cloud research: A review and current assessment,” Bull. Am. Meteorol. Soc. 72(12), 1848–1866 (1991). [CrossRef]
5. Y. Hu, “Depolarization ratio-effective lidar ratio relation: Theoretical basis for space lidar cloud phase discrimination,” Geophys. Res. Lett. 34(11), L11812 (2007). [CrossRef]
6. R. Yoshida, H. Okamoto, Y. Hagihara, and H. Ishimoto, “Global analysis of cloud phase and ice crystal orientation from Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) data using attenuated backscattering and depolarization ratio,” J. Geophys. Res. 115, D00H32 (2010). [CrossRef]
7. N. I. Polonsky, S. P. Love, and A. B. Davis, “Wide-Angle Imaging Lidar development at the ARM Southern Great Plains site: Intercomparison of cloud property retrievals,” J. Atmos. Ocean. Technol. 22(6), 628–648 (2005). [CrossRef]
8. A. B. Davis, “Multiple-scattering lidar from both sides of the clouds: Addressing internal structure,” J. Geophys. Res. 113(D14), D14S10 (2008). [CrossRef]
9. R. F. Caharan, M. McGill, and J. Kolansinski, “THOR-cloud thickness from offbeam lidar returns,” J. Atmos. Ocean. Technol. 22(6), 605–627 (2005). [CrossRef]
11. L. R. Bissonnette, G. Roy, L. Poutier, S. G. Cober, and G. A. Isaac, “Multiple-scattering lidar retrieval method: tests on Monte Carlo simulations and comparisons with in situ measurements,” Appl. Opt. 41(30), 6307–6324 (2002). [CrossRef] [PubMed]
12. G. Roy, L. Bissonnette, C. Bastille, and G. Vallée, “Retrieval of droplet-size density distribution from multiple-field-of-view cross-polarized lidar signals: theory and experimental validation,” Appl. Opt. 38(24), 5202–5211 (1999). [CrossRef] [PubMed]
13. H. Okamoto, K. Sato, T. Makino, T. Nishizawa, N. Sugimoto, Y. Jin, and A. Shimizu, “Depolarization ratio of clouds measured by multiple-field of view multiple-scattering polarization lidar,” EPJ Web of Conferences 119, 11007 (2016). [CrossRef]
14. J. Yamaguchi, T. Takano, T. Nakajima, T. Takamura, H. Kumagai, and Y. Ohno, Y. nakanishi, K. Akita, Y. Kawamura, H. Abe, K. Futaba, and S. Yokote, “Sensitivity of FMCW 95GHz cloud radar for high clouds,” 2006 Asia-Pacific Microwave Conference, Yokohama (2006). [CrossRef]
15. H. Ishimoto and K. Masuda, “A Monte Carlo approach for the calculation of polarized light: application to an incident narrow beam,” J. Quant. Spectrosc. Radiat. Transf. 72(4), 467–483 (2002). [CrossRef]
16. A. Shimizu, N. Sugimoto, I. Matsui, K. Arao, I. Uno, T. Murayama, N. Kagawa, K. Aoki, A. Uchiyama, and A. Yamazaki, “Continuous observations of Asian dust and other aerosols by polartization lidars in China and Japan during ACE-Asia,” J. Geophys. Res. 109(D19), D19S17 (2004). [CrossRef]
17. H. Okamoto, T. Nishizawa, T. Takemura, H. Kumagai, H. Kuroiwa, N. Sugimoto, I. Matsui, A. Shimizu, S. Emori, A. Kamei, and T. Nakajima, “Vertical cloud structure observed from shipborne radar and lidar: Midlatitude case study during the MR01/K02 cruise of the research vessel Mirai,” J. Geophys. Res. 112(D8), D08216 (2007). [CrossRef]
18. H. Okamoto, T. Nishizawa, T. Takemura, K. Sato, H. Kumagai, Y. Ohno, N. Sugimoto, A. Shimizu, I. Matsui, and T. Nakajima, “Vertical cloud properties in the tropical western Padific Ocean: Validation of the CCSR/NIES/FRCGC GCM by shipborne radar and lidar,” J. Geophys. Res. 113(D24), D24213 (2008). [CrossRef]
19. H. Okamoto, K. Sato, and Y. Hagihara, “Global analysis of ice microphysics from CloudSat and CALIPSO: Incorporation of specular reflection in lidar signals,” J. Geophys. Res. 115(D22), D22209 (2010). [CrossRef]
20. Y. Hagihara, H. Okamoto, and R. Yoshida, “Development of a combined CloudSat-CALIPSO cloud mask to show global cloud distribution,” J. Geophys. Res. 115, D00H33 (2010). [CrossRef]
21. K. Sato and H. Okamoto, "Refinement of global ice microphysics using spaceborne active sensors," J. Geophys. Res. 116, D20202 (2011).
22. X. Cao, G. Roy, R. Bernier, and N. Roy, “Comparison of the relationships between lidar integrated backscattered light and accumulated depolarization ratios for linear and circular polarization for water droplets, fog-oil, and dust,” Appl. Opt. 48(21), 4130–4141 (2009). [CrossRef] [PubMed]