Dual-wavelength Mie-scattering Scheimpflug lidar system developed for the studies of the aerosol extinction coefficient and the &#x00C5;ngstr&#x00F6;m exponent

Liang Mei; Zheng Kong; Teng Ma

doi:10.1364/OE.26.031942

1. Introduction

Aerosol particles have a potential climate importance, and the effects strongly depend on their optical or microphysical properties such as spectral-dependent extinction coefficient, optical depth, Ångström exponent, and particle size distribution [1]. A number of techniques have been developed to measure these parameters for climate and environmental usages, e.g., visibility meter, nephelometer and sun photometer [2]. Lidar, as a powerful active remote sensing technique, is capable of detecting backscattering profile of aerosol particles with high temporal and spatial resolution. The optical or microphysical properties of aerosol particles can thus be retrieved from range-resolved lidar measurements [3]. However, single-wavelength Mie-scattering lidar system can only measure aerosol extinction/backscattering coefficient at a specific wavelength, which is difficult to extrapolate the microphysical properties such as size distribution of aerosol particles [4,5]. On the contrary, dual or multiple wavelength lidar techniques can measure aerosol extinction/backscattering coefficients at several independent wavelengths, allowing spectroscopic analysis of the aerosol extinction coefficient and thus the Ångström exponent that is related with the size distribution of aerosol particles [6].

During the past decades, dual-wavelength lidar techniques have attracted considerable interest and have been extensively developed for ground-based lidar stations or satellite observations to study the hygroscopic growth, Ångström exponent, particle size distribution of aerosol particles, etc [7]. Nishizawa et al. have measured vertical and temporal distributions of water-soluble, sea salt, and dust aerosols by using a dual-wavelength polarization lidar system and confirmed that the relative humidity (RH) contributes to the hygroscopic growth of aerosol particles in 2006 [8]. Sugimoto et al. found out that coarse non-spherical particles (Asian dust) almost always existed in the background at the AD-Net (the Asian Dust and aerosol lidar observation Network) station in Seoul by employing a polarization-sensitive two-wavelength lidar in 2014 [9]. Di et al. studied the vertical distribution of optical and microphysical properties of smog aerosols such as volume size distribution and effective radius based on a multi-wavelength polarization lidar in Xi’an, China in 2016 [10]. Gao et al. have shown that coarse mode particles were predominant in a dust storm over the Loess Plateau through combined observations of a dual-wavelength lidar and an aethalometer in 2017 [11]. The dual-wavelength lidar system, installed in the CALIPSO (Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observation) satellite, can provide vertical profiles of aerosols and clouds as well as their optical properties in a global scale [12–14]. Extensive work has been devoted into the evaluations of the data measured by the CALIPSO lidar, e.g., Ångström exponent studies, investigation on particle size distribution of aerosol particles [15].

Recently, the Scheimpflug lidar (SLidar) technique has shown a great potential in atmospheric aerosol monitoring [16–21]. Atmospheric backscattering signal is retrieved from the pixel intensities of a tilted image sensor based on the Scheimpflug principle. The system cost can be significantly reduced by employing low cost compact high-power laser diodes and CCD/CMOS detectors. Moreover, the robust and stable laser diodes also greatly facilitate system maintenance during operation. Thus, it is of great interest to develop a dual-wavelength SLidar system to meet the requirements for continuous and network monitoring of aerosol particles. In this work, a dual-wavelength SLidar system based on the Scheimpflug principle is developed by utilizing a 407-nm and an 808-nm high power continuous-wave laser diodes as light sources and two CMOS cameras as detectors. The system performance has been validated through a two-week continuous atmospheric measurement campaign in May 2018 at Dalian, which is a coast city in northern China. In the meanwhile, the dependencies of the aerosol extinction coefficient and the Ångström exponent on the RH and particle concentrations are also studied.

2. Instrumentations and methods

2.1. Dual-wavelength Mie-scattering Scheimpflug lidar system

The system schematic and the detailed specifications of the dual-wavelength Mie-scattering Scheimpflug lidar system are given in Fig. 1 and Table 1, respectively. An 808-nm laser diode and a 407-nm laser diode were employed as the light sources. Each laser diode was housed by a customized aluminum mount with a thermoelectric cooler (TEC) to control the case temperature. Therefore, the center emission wavelength can be accurately controlled to match the transmission bands of interference filters used in the detection system by tuning the case temperature and the driving current. The laser beams from the two laser diodes were combined by a dichroic mirror. The combined laser beams were transmitted into the atmosphere through a F6 refractor telescope. The full width at half maximum (FWHM) divergence of the 808-nm laser beam was about 6°∥ (slow axis) × 8°⊥(fast axis), while the 407-nm laser diode had much larger divergence (1/e²), i.e., about 13°∥ (slow axis) × 45° ⊥(fast axis). Thus, a cylindrical lens pair was designed to improve the transmission efficiency through the F6 refractor telescope. As shown in Fig. 1, a convex cylindrical lens and a concave cylindrical lens that were confocal with the F6 refractor were placed in front of the 407-nm laser diode [20]. Thus, the divergence along the fast axis was reduced to about ± 4.6° that matched the acceptance angle of the F6 refractor telescope, while the divergence along the slow axis had no change. The fast-axis of the 808-nm laser diode and the slow-axis of the 407-nm laser diode were aligned parallel to the Scheimpflug plane to optimize the effective range resolution [17]. Apart from the focuser of the telescope, the 407-nm laser diode can also be focused independently through a customized linear translation mount. Besides, an x-y linear translation mount was employed to adjust the axial position of the 808-nm laser diode. Thus, the two laser diodes can be collimated independently, and the two laser beams can be overlapped in the far end by adjusting the linear translation mount.

Fig. 1 Architecture of the dual-wavelength Scheimpflug lidar system. The lidar system can be carried by an equatorial mount.

Download Full Size | PDF

Table 1. System specifications of the dual-wavelength Scheimpflug lidar system.

View Table | View all tables in this article

A Newtonian telescope was employed to collect atmospheric backscattering signals. Backscattering signals at the two wavelengths were separated by a dichroic mirror and then detected by two CMOS cameras, respectively. The two CMOS cameras, connected with the dichroic mirror mount, were installed on the Newtonian telescope with 45° titled to its optical axis. The refractor telescope and the Newtonian telescope were installed on an aluminum-alloy bar with approximately 806 mm separation to each other to meet the requirement of the Scheimpflug principle. An 808-nm interference filter (3 nm) and a long-pass color filter (RG715) were used to suppress the sunlight background for the 808-nm channel, while a 407-nm interference filter (1.7 nm) and a 400-nm interference filter (25 nm) were used for the 407-nm channel. The two laser beams of 808 nm and 407 nm were imaged in the region of interests (ROI) of the two CMOS sensors, respectively.

In order to eliminate the sunlight background, the two laser diodes were intensity-modulated. The 808-nm channel CMOS camera generated an exposure synchronization signal, which was used to trigger a 2-bit Johnson counter. A modulation signal was then generated and fed to the current driver of the two laser diodes. Thus, the two laser diodes can be turned on and off synchronously. In the meanwhile, the two CMOS cameras can capture real-time images simultaneously based on a multithread LabVIEW-based program. Each CMOS camera recorded the laser beam image (“on” image) and the background image (“off” image) successively in the ROI (2048 × 240 pixels) as the corresponding laser diode was turned on and off, alternatively. Thus, the modulation frequency of the laser diode is inversely proportional to the twice of the exposure time, e.g., the modulation frequency of the laser diode is 25 Hz when the exposure time is 20 ms. The recorded “on” and “off” images were vertically binned and background subtracted by signal interpolation to obtain a single lidar recording [17]. The laser beam image can be fully captured, the geometric overlap factor was one even in the near range and the minimum measurement distance can reach to about 80 m. A small telescope (f = 400 mm, ∅70 mm) with a black-white camera, which was not presented in Fig. 1, was also mounted on the aluminum bar to monitor the measurement site and the facula from a distant object/building during system alignment.

2.2. SLidar equation and extinction coefficient retrieval

The atmospheric backscattering signal of the SLidar technique is independent of the distance square factor, and the lidar equation is described by [17]

P (λ, z) = K [β_{a} (λ, z) + β_{m} (λ, z)] \exp [- 2 \int_{0}^{z} [α_{a} (λ, z') + α_{m} (λ, z')] d z']

here

z

is the measurement distance to the lidar system,

P (λ, z)

is the backscattering intensity,

K

is the system constant,

β_{m} (λ, z)

and

α_{m} (λ, z)

are the molecular backscattering and extinction coefficients, respectively,

β_{a} (λ, z)

and

α {}_{a}{(λ, z)}

are the aerosol backscattering and extinction coefficients, respectively. The Rayleigh scattering from molecules can often be neglected in the infrared region, but should be considered in the short wavelength regions, e.g., 355 nm or 407 nm, particularly in a clean atmospheric condition. Thus, the Fernald approach is utilized in this work to retrieve the aerosol extinction/backscattering coefficient [22]. The aerosol extinction coefficient of the SLidar technique is given by [23]

α_{a} (λ, z) = - \frac{S_{a}}{S_{m}} α_{m} (λ, z) + \frac{P (λ, z) \exp [2 (\frac{S_{a}}{S_{m}} - 1) \int_{z}^{z_{c}} α_{m} (λ, ζ) d ζ]}{\frac{P (λ, z_{c})}{α_{a} (λ, z_{c}) + \frac{S_{a}}{S_{m}} α_{m} (λ, z_{c})} + 2 \int_{z}^{z_{c}} P (λ, ζ) \exp [2 (\frac{S_{a}}{S_{m}} - 1) \int_{ζ}^{z_{c}} α_{m} (λ, z') d z'] d ζ}

here

z_{c}

is the calibration (boundary) distance,

S_{m}

and

S_{a}

are the molecular and aerosol lidar ratio, respectively, and

α_{a} (λ, z_{c})

is the boundary value of the aerosol extinction coefficient at distance

z_{c}

. The molecular lidar ratio and the backscattering coefficient can be calculated from the atmospheric model with known atmospheric pressures and temperatures. The molecular lidar ratio is 8π/3. For typical urban aerosols, the aerosol lidar ratio at 808 nm and 407 nm are set to 40 sr and 50 sr, respectively [24,25]. The boundary value

α_{a} (λ, z_{c})

can be retrieved by linearly fitting the log-scale lidar signal in a subinterval range, where the atmosphere is assumed homogeneous according to the slope method. The range-resolved aerosol extinction coefficient can then be obtained.

3. Atmospheric remote measurements

3.1 Measurements

Atmospheric remote measurement campaign has been carried out from May 18^th to 31^st, 2018 (in total two weeks) in the urban area of Dalian, which is a coastal city near the Yellow Sea. The SLidar system was located on the ground floor of the School of Physics and the laser beam was about 3-m height above the out-door ground at the experimental site. The lidar system was pointed to the sky with an elevation angle of about 6.4° on the southwest-west direction, as shown in Fig. 2. An air quality sensor (Microair A108, Fairsense) was employed to measure weather parameters such as the RH and the temperature once an hour. The Microair sensor was mounted on a lamppost with 3 m above the ground, which was 170 meters away from the lidar system. In the meanwhile, PM2.5/PM10 concentrations were reported by local national pollution monitoring station (Qixianling Station). The monitoring station was located in the southwest-south direction of the lidar system with a distance of about 2.5 km. The pixel-distance relationship was calibrated by measuring the lidar echo signal of a tall building located at approximately 971 m away from the lidar system.

Fig. 2 Experimental site map. Microair sensor was located nearby the lidar system (170 m) and the Qixianling national monitoring station was about 2.5 km away from the lidar system.

Download Full Size | PDF

During this period, the weather was mostly sunny or cloudy, and the RH changed between 30%-90%. Moreover, the airborne PM10 (atmospheric dynamics equivalent diameter ≤10 µm) and the PM2.5 (atmospheric dynamics equivalent diameter ≤2.5 µm) particle concentrations significantly varied between 12 µg/m³ to 166 µg/m³ and 5 µg/m³ to 57 µg/m³, respectively. The varying atmospheric conditions provided the possibilities to study the dependencies of the aerosol extinction coefficient and the Ångström exponent on the RH and the aerosol particle concentrations.

3.2 Lidar signal processing

The exposure time of the CMOS camera was automatically changed between 20 ms-500 ms according to the variation of sunlight intensity to achieve the best performance. During daytime measurements with 20 ms exposure time (image sensor could be saturated for longer exposure time under sunshine), the noise level is dominated by the sunlight background noise when no digital averaging is performed. In the meanwhile, the lidar signals were averaged for about 45 s to improve the signal-to-noise ratio (SNR) for all exposure times. Besides, the lidar signals were also smoothened by a Savitzky-Golay (SG) filter. A special feature of the SLidar technique is that the range resolution is proportional to the square of the measurement distance, leading to centimeter-level range resolution in the near range and significant larger range resolution in the far range. Since centimeter-level range resolution is unnecessary for typical atmospheric measurements in the near range, the lidar signals in the near region (<700 m) were resampled by the weighted average of each subset signal with about 3-m range, leading to a 3-m range resolution for the near range lidar signal. As a result, the SNR can be improved and the data amount can be reduced as well [26].

The backscattering intensities measured in several typical atmospheric conditions are shown in Fig. 3. The backscattering intensity at 808 nm was much stronger than that of the 407 nm as the power of the 407-nm laser diode was much smaller than that of the 808-nm laser diode. Moreover, the atmospheric extinction was generally larger in short wavelength. The SNRs of the near-range (≈100 m) lidar signals were in the region between 300 and 500 during nighttime, while the SNR at the 808-nm and 407nm channels varied from 250 to 400 and from 100 to 250 in the daytime, respectively. However, it should be emphasized that the SNR does not decrease with the square of the measurement distance, which is the case in conventional pulsed lidar techniques. The boundary value of the aerosol extinction coefficient is estimated by automatically searching a linear region between 1 and 10 km based on the least square fitting. Although signal resampling is employed, the range resolution in the near range is still higher than that of the far range. Thus, the number of fitting data points increases as the fitting region is closer, which is given by an inverse function $[4 \times (z_{\max} / z)]$ . Here $z_{\max}$ is the maximum retrieval distance, i.e., 7 km, and $[]$ is an rounding operator that converts the value to the nearest integer. The number of the fitting points between 7 and 10 km was set to 4 due to the limited range resolution in this region. Finally, the aerosol extinction coefficient can be obtained according to Eq. (2). The boundary solution of the lidar curves are indicated by solid black curves shown in Fig. 3. The error bars shown in Figs. 3(b) and 3(d) indicate the error due to the estimation uncertainties of the boundary value.

Fig. 3 Typical atmospheric lidar signals and the aerosol extinction coefficients. (a) Atmospheric backscattering signals and (b) the aerosol extinction coefficients at 808 nm channel; (c) atmospheric backscattering signals and (d) the aerosol extinction coefficients at 407 nm channel.

Download Full Size | PDF

Figure 4 shows the time-range map of the aerosol extinction coefficients retrieved by the Fernald method. During 12:00 on May 19^th to 03:00 on May 23^rd and 00:00 to 12:00 on May 29^th, the weather was rainy, cloudy or foggy every now and then. Thus, the far end boundary value of the aerosol extinction coefficient could not be retrieved in some periods due to strong return signals resulted from either low-altitude cloud or fog from the Yellow Sea. The measurement data were then removed as shown by the white-stripe regions in Fig. 4. Localized peak signals within 1 km region were mainly due to the emissions of road traffic, barbecue, etc. However, the atmosphere tended to be homogeneous in the far end. This was because local emissions were mainly concentrated in the ground level and the range-resolution of the SLidar technique decreased with the increasing of the measurement distance. Besides, the area within 1.4 km belongs to the urban region, while the area beyond 1.4 km is mountainous region in Fig. 2. Figure 5 shows the weather parameters, particle concentrations, as well as the retrieved aerosol extinction coefficients that were spatially averaged by taking the median value along the laser beam path. The aerosol extinction coefficients of the two wavelengths were also time-averaged for 10 times in Fig. 5(c). As can be seen from Fig. 5(a), the rising and the falling of the RH were opposite to the variations of the local temperature due to the solar evaporation effect.

Fig. 4 Time-range map of aerosol extinction coefficients retrieved by the Fernald method at the (a) 808 nm and (b) 407 nm channels.

Download Full Size | PDF

Fig. 5 (a) Temperature and RH, (b) PM2.5/PM10 concentrations measured by a local national pollution monitoring station, and (c) aerosol extinction coefficients at 808 nm and 407 nm retrieved by the Fernald method, which were averaged along the measurement path and in time scale (10 times).

Download Full Size | PDF

3.3 Dependencies of the extinction coefficient on the mass concentration and the relative humidity

As can be seen from Fig. 5, the aerosol extinction coefficient was not only determined by the particle concentrations, but also significantly influenced by the RH. The relationship between the aerosol extinction coefficient and the RH is shown in Fig. 6. As hygroscopic particles took up water with the increasing RH, the aerosol extinction coefficient increased, particularly in the case of high RH, e.g., the RH was beyond 80% on May 22^nd and 28^th. It can be noted that the enhancement of the aerosol extinction coefficient was quite different for different particle concentrations. To better understand the effect of aerosol hygroscopic growth on the aerosol extinction coefficient, the measurement data can be divided into two categories according to the PM10 concentrations, i.e., the low-concentration category (<50 μg/m³) and the high-concentration category (>50 μg/m³). As shown in Fig. 6(a), the aerosol extinction coefficients at 808 nm were almost unchanged when the RH increased from 30% to 80% for the low-concentration category, indicating that the hygroscopic growth was not significant in this region. As the RH exceeded 80%, the aerosol extinction coefficient increased rapidly. However, the aerosol extinction coefficient started to increase when the RH was larger than 60% for the high-concentration situation. This implied that the aerosol particles in polluted atmospheric condition had a larger humidification factor and hydrophilic particles were dominant. Similar phenomenon can also be observed for the 407-nm aerosol extinction coefficient shown in Fig. 6(b). However, the hygroscopic growth of the 407-nm aerosol extinction coefficient was faster than that of the 808-nm.

Fig. 6 The relationship between the aerosol extinction coefficient and the RH (a) 808 nm and (b) 407 nm, enhancement factor at (c) 808 nm and (b) 407 nm. The measurement data presented in Figs. 6(c) and 6(d) was extracted from 12:00 May 26^th to 05:00 May 28^th, when the particle concentrations remained almost unchanged.

Download Full Size | PDF

During the period from 12:00 on May 26^th to 05:00 on May 28^th, the variation of particle concentrations were not significant, while the RH increased from 30% to 90%. Thus, the experimental data measured during this period was extracted to study the effect of hygroscopic growth quantitatively, which was often quantified by the enhancement factor of the backscattering coefficient

f_{β} (RH, λ) = β_{a} (RH, λ) / β_{a} (dry, λ)

where

β_{a} (dry, λ)

and

β_{a} (RH, λ)

represent backscattering coefficients at wavelength

λ

in dry conditions and a defined RH, respectively. The backscattering coefficient at the dry condition can be retrieved from the mean value of the backscattering coefficients at 30-50% RH where the hygroscopic growth was not significant. The relationship between

f_{β} (R H, λ)

and the RH can be modeled by [27]

f (RH) = 1 + a \times {RH}^{b}

Here a determines the maximum value of the enhancement factor at 100% RH and b indicates the curvature of the enhancement curve [28–30]. As can be seen from Figs. 6(c) and 6(d), the enhancement factors at 808 nm and 407 nm with 80% RH were 3.3 and 6.3, respectively, implying a wavelength dependency of the enhancement factor. Clearly, the enhancement factor of the short wavelength was larger than that of the long wavelength for the urban aerosols off the Yellow Sea coast. Moreover, strong wavelength dependencies were also reported by Kotchenruther et al. when studying the enhancement factor for urban/industrial aerosols off the east coast of the United States [28]. However, it is worth to mention that the wavelength dependency of

f_{β} (R H, λ)

also relies on the aerosol types [27].

Apart from the dependency on the RH due to the hygroscopic growth, the aerosol extinction coefficient also relied on particle concentrations. In order to investigate the relationship between the aerosol extinction coefficient and the particle concentrations, the RH was divided into three regions: <60% (low RH), 60%~75% (moderate RH) and >75% (high RH). As shown in Fig. 7, the aerosol extinction coefficient generally had a good linear relationship with the PM10 concentrations. The correlation coefficient in different RHs were within the region from 0.70 to 0.89, as shown in Table 2. It was also found that the linear fitting slope increased with the increasing of the RH, which was mainly attributed to the hygroscopic growth as has been discussed above. In general, the slope increased more than 6 times when the RH changed from below 60% to beyond 75%.

Fig. 7 Scatter plot of PM10 mass concentration and the aerosol extinction coefficient at the two measurement wavelengths under different RHs, (a) 808 nm, RH>75%, (b) 407 nm, RH>75%, (c) 808 nm, 60%<RH≤75%, (d) 407 nm, 60%<RH≤75%, (e) 808 nm, RH≤60%, (f) 407 nm, RH≤60%.

Download Full Size | PDF

Table 2. Correlation coefficients between the aerosol extinction coefficient and the PM2.5/PM10 particle concentrations under different RHs.

View Table | View all tables in this article

3.4 Studies on the relationship between the Ångström exponent, the mass concentration and the relative humidity

The Ångström exponent is often used as a qualitative indicator of aerosol particle size [31]. The value is connected with the mean radius of aerosol particles and is generally within the range of 0 to 2. The Ångström exponent ( $σ$ ) can be calculated by:

σ (z) = \frac{ln [α_{a} (λ_{1}, z) / α_{a} (λ_{2}, z)]}{ln(λ_{2} / λ_{1})} .

Here

λ_{1}

and

λ_{2}

are referred to as 407 nm and 808 nm, respectively. An Ångström exponent less than 1 indicates that the size distribution is dominated by coarse mode aerosol particles, while an Ångström exponent larger than 2 means that the size distribution is dominated by fine mode aerosol particles [32–34]. On the other hand, hygroscopic growth significantly influences the mean radius of aerosol particles. As a result, the Ångström exponent may also strongly influenced by the RH depending on the size distribution of aerosol particles.

The spatial-temporal variation of the Ångström exponent can be calculated according to Eq. (5), as shown in Fig. 8. As can be seen, the Ångström exponent varied significantly during the measurements. The variation of the Ångström exponent along the measurement path also indicated that the particle size distribution was inhomogeneous in atmosphere, in spite that the measurements were performed on a near horizontal path. The Ångström exponent was first spatially averaged by taking the median value along the laser beam path and then time-averaged in a time scale of one hour. The temporal distribution of the Ångström exponent was shown in Fig. 9(b). In order to investigate the dependency of the Ångström exponent on the particle size distribution, the mass concentration ratio between the PM2.5 (fine mode particle) particle concentration and the PM10 particle concentration was calculated, which was referred to as the mass concentration fraction of the fine-mode particles ( $c_{f i n e}$ ). A large value of $c_{f i n e}$ implied that the atmosphere is dominated by the fine mode particles. Figure 9 showed the relationship between the Ångström exponent, the RH and $c_{f i n e}$ . Generally speaking, the temporal variation of the Ångström exponent followed with the variation of $c_{f i n e}$ , implying a good correlation between the Ångström exponent and $c_{f i n e}$ . However, it can also be found that the Ångström exponent deviated from the expected values when the RH significantly changed, as shown in Fig. 9(b). The dependencies of the Ångström exponent on $c_{f i n e}$ and the RH will be discussed following.

Fig. 8 Time-range map of Ångström exponent.

Download Full Size | PDF

Fig. 9 The temporal variations of the (a) RH, (b) the Ångström exponent and $c_{f i n e}$ .

Download Full Size | PDF

The measurement was started at 09:00 on May 18^th when the PM2.5 concentration was only about 9 µg/m³ and the PM10 concentration was about 56 µg/m³. Clearly, the atmosphere was dominated by the coarse mode particles, and the Ångström exponent was only about 0.15. From 09:00 May 18^th to 10:00 on May 19^th, the Ångström exponent increased to 1.0 with the rapid increasing of $c_{f i n e}$ . In the midnight on May 23^rd, the atmosphere was very clean and dominated by the fine mode particles. The value of $c_{f i n e}$ was 0.6 and the Ångström exponent was about 2.0. Since the early morning, atmospheric pollutants began to accumulate and the coarse mode particles became dominated, leading to a rapid decreasing of $c_{f i n e}$ (0.6→0.19) as well as the Ångström exponent (2.0→0.1). As the atmospheric pollutants were dissipating, the value of $c_{f i n e}$ increased to about 0.48 and consequently the Ångström exponent slowly increased to 1.0 in the evening on May 25^th. It can be concluded that the primary pollutants in Dalian during the measurement periods were the coarse mode particles. From 06:00 on May 27^th to 12:00 on May 29^th, the RH varied frequently from 60% to 90% and a severe haze occurred. As a result, the aerosol extinction coefficients varied significantly during this period owing to the high particle concentrations as well as the hygroscopic growth effect as show in Fig. 5(c). Nevertheless, the value of $c_{f i n e}$ was almost constant. Thus, there was no significant change for the Ångström exponent during this period, which confirmed that the Ångström exponent was mainly related with mass concentration faction of the fine mode particles but not the absolute particle concentrations. According to the above discussions, the data measured during the periods from 09:00 on May 18^th to 10:00 on May 19^th and 00:00 on May 22^nd to 16:00 on May 29^th were extracted to investigate the relationship between the Ångström exponent and $c_{f i n e}$ , as shown in Fig. 10(a). A very high correlation coefficient was found, i.e., 0.74, which was very promising compared with previous studies [34].

Fig. 10 (a) Scatter plot of the mass concentration fraction of the fine-mode particles ( $c_{f i n e}$ ) and the Ångström exponent, the data were retrieved during the periods from 09:00 on May 18^th to 10:00 on May 19^th and 00:00 on May 22^nd to 16:00 on May 29^th. (b) Scatter plot of the RH and the Ångström exponent utilizing all measurement data.

Download Full Size | PDF

Strong correlation between the Ångström exponent and the RH were observed on May 21^st, 22^nd, 23^rd, 26^th and 30^th. The measurement data measured on May 26^th was discussed in detail as an example. As can be seen in Fig. 9(b), the value of $c_{f i n e}$ was nearly constant, i.e., 0.3 ± 0.02, while the Ångström exponent showed a strong positive correlation with the RH as it was decreasing and increasing. Considering the humidification factor of the 407 nm was much larger than that at 808 nm, a positive correlation was not a surprising result. The correlation coefficient was about 0.68 between the Ångström exponent and the RH by analyzing all the measurement data in Fig. 10(b). The positive correlation implied a bimodal or multimodal size distribution of the aerosol particles [35]. It has been noted that negative value of the Ångström exponent appeared at around 12:00 on May 26^th, when the air was clean and very dry. The aerosol extinction coefficient were as low as 0.1 km⁻¹ for 407-nm channel and less than 0.1 km⁻¹ for 808-nm channel. The possible reasons for the negative Ångström exponent could be attributed to the systematic uncertainties resulted from the assumption of lidar ratio and the retrieval of the boundary value, which became significant for the low aerosol extinction coefficient. Moreover, negative Ångström exponents were also founded in the presence of relatively larger particles during low aerosol concentration and low scattering by Pandolfi et al. [36], which is very similar to the present case.

4. Conclusions

In this work, a dual-wavelength Scheimpflug lidar system has been developed for the studies of aerosol extinction coefficients and the Ångström exponent. An 808-nm laser diode and a 407 nm laser diode were employed as the light sources. Backscattering signals at two different wavelengths were collected utilizing a Newtonian telescope and then detected by two CMOS cameras, respectively. The lidar signal of the 407 nm channel was much weaker than that of the 808 nm channel due to the low output power of the 407 nm laser diode. The sensitivity and the SNR of the 407 nm channel should be further improved by employing laser diodes with even higher output power or larger apertures of the receiving telescope. Nevertheless, atmospheric remote measurements performed for two weeks in May 2018 have successfully demonstrated the capability of the dual-wavelength Scheimpflug lidar system for continuous atmospheric aerosol measurements.

The aerosol extinction coefficients measured by the lidar systems were promising and showed good correlation with particle concentrations as well as the RH. Hygroscopic studies revealed that the enhancement factors of the backscattering coefficient at the 808 nm and 407 nm were 3.3 and 6.3, respectively, implying a clear wavelength dependency. Moreover, the shorter wavelength had a larger enhancement factor for the aerosol particles off the Yellow Sea coast. Experimental result revealed that the Ångström exponent, which varied between 0 and 2 during the measurement season, was influenced by the mass concentration fraction of fine model particles (PM2.5) and the RH. A correlation coefficient of 0.74 was obtained between the Ångström exponent and the mass concentration fraction of fine model particles. The primary pollutants of Dalian city during the measurement season were found to be the coarse mode particles. Besides, the Ångström exponent was positive correlated with the RH, implying a bimodal or multimodal size distribution of aerosol particles. The promising result demonstrated in this work shows a great potential of using the dual-wavelength Scheimpflug lidar system for the studies on Ångström exponent as well as the qualitative analysis of particle size distributions.

Funding

National key research and development program of China (2016YFC0200600); National Natural Science Foundation of China (61705030); Natural Science Foundation of Liaoning Province, China (201602163); Fundamental Research Funds for the Central Universities (DUT18JC22).

Acknowledgment

The authors greatly acknowledge the help of Zhi Liu and Lishan Zhang for the experimental work.

References

1. C. Toledano, V. E. Cachorro, A. Berjon, A. M. de Frutos, M. Sorribas, B. A. de la Morena, and P. Goloub, “Aerosol optical depth and Angstrom exponent climatology at El Arenosillo AERONET site (Huelva, Spain),” Q. J. R. Meteorol. Soc. 133(624), 795–807 (2007). [CrossRef]

2. S. Li, T. J. Wang, M. Xie, Y. Han, and B. L. Zhuang, “Observed aerosol optical depth and angstrom exponent in urban area of Nanjing, China,” Atmos. Environ. 123, 350–356 (2015). [CrossRef]

3. R. E. Mamouri, A. Papayannis, V. Amiridis, D. Muller, P. Kokkalis, S. Rapsomanikis, E. T. Karageorgos, G. Tsaknakis, A. Nenes, S. Kazadzis, and E. Remoundaki, “Multi-wavelength Raman lidar, sun photometric and aircraft measurements in combination with inversion models for the estimation of the aerosol optical and physico-chemical properties over Athens, Greece,” Atmos. Meas. Tech. 5(7), 1793–1808 (2012). [CrossRef]

4. A. Comerón, C. Muñoz-Porcar, F. Rocadenbosch, A. Rodríguez-Gómez, and M. Sicard, “Current research in lidar technology used for the remote sensing of atmospheric aerosols,” Sensors (Basel) 17(6), 1450 (2017). [CrossRef] [PubMed]

5. B. Rosati, E. Herrmann, S. Bucci, F. Fierli, F. Cairo, M. Gysel, R. Tillmann, J. Gross, G. P. Gobbi, L. Di Liberto, G. Di Donfrancesco, A. Wiedensohler, E. Weingartner, A. Virtanen, T. F. Mentel, and U. Baltensperger, “Studying the vertical aerosol extinction coefficient by comparing in situ airborne data and elastic backscatter lidar,” Atmos. Chem. Phys. 16(7), 4539–4554 (2016). [CrossRef]

6. Y. Hara, T. Nishizawa, N. Sugimoto, K. Osada, K. Yumimoto, I. Uno, R. Kudo, and H. Ishimoto, “Retrieval of aerosol components using multi-wavelength Mie-Raman lidar and comparison with ground aerosol sampling,” Remote Sens. (Basel) 10(6), 937 (2018). [CrossRef]

7. I. Veselovskii, D. N. Whiteman, A. Kolgotin, E. Andrews, and M. Korenskii, “Demonstration of aerosol property profiling by multiwavelength lidar under varying relative humidity conditions,” J. Atmos. Ocean. Technol. 26(8), 1543–1557 (2009). [CrossRef]

8. T. Nishizawa, N. Sugimoto, I. Matsui, A. Shimizu, X. Liu, Y. Zhang, R. Li, and J. Liu, “Vertical distribution of water-soluble, sea salt, and dust aerosols in the planetary boundary layer estimated from two-wavelength backscatter and one-wavelength polarization lidar measurements in Guangzhou and Beijing, China,” Atmos. Res. 96(4), 602–611 (2010). [CrossRef]

9. N. Sugimoto, T. Nishizawa, A. Shimizu, I. Matsui, and H. Kobayashi, “Detection of internally mixed Asian dust with air pollution aerosols using a polarization optical particle counter and a polarization-sensitive two-wavelength lidar,” J. Quant. Spectrosc. Radiat. Transf. 150, 107–113 (2015). [CrossRef]

10. H. Di, H. Hua, Y. Cui, D. Hua, T. He, Y. Wang, and Q. Yan, “Vertical distribution of optical and microphysical properties of smog aerosols measured by multi-wavelength polarization lidar in Xi’an, China,” J. Quant. Spectrosc. Radiat. Transf. 188, 28–38 (2017). [CrossRef]

11. X. Gao, X. Cao, P. Tian, L. Zhang, Z. Huang, and T. Zhou, “Combined observation of a dust storm over the Loess Plateau using a dual-wavelength lidar and an aethalometer,” Atmos. Pollut. Res. 8(6), 1103–1112 (2017). [CrossRef]

12. W. H. Hunt, D. M. Winker, M. A. Vaughan, K. A. Powell, P. L. Lucker, and C. Weimer, “CALIPSO lidar description and performance assessment,” J. Atmos. Ocean. Technol. 26(7), 1214–1228 (2009). [CrossRef]

13. S. A. Young, M. A. Vaughan, R. E. Kuehn, and D. M. Winker, “The retrieval of profiles of particulate extinction from Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) data: uncertainty and error sensitivity analyses,” J. Atmos. Ocean. Technol. 30(3), 395–428 (2013). [CrossRef]

14. G. Stephens, D. Winker, J. Pelon, C. Trepte, D. Vane, C. Yuhas, T. L’Ecuyer, and M. Lebsock, “CloudSat and CALIPSO within the A-Train: ten years of actively observing the earth system,” Bull. Am. Meteorol. Soc. 99(3), 569–581 (2018). [CrossRef]

15. M. Vaughan, Z. Y. Liu, Y. X. Hu, K. Powell, A. Omar, S. Rodier, W. Hunt, J. Kar, J. Tackett, B. Getzewich, and K. P. Lee, “Cloud-aerosol interactions: retrieving aerosol Angstrom exponents from CALIPSO measurements of opaque water clouds,” Epj Web Conf119(2016). [CrossRef]

16. L. Mei and M. Brydegaard, “Continuous-wave differential absorption lidar,” Laser Photonics Rev. 9(6), 629–636 (2015). [CrossRef]

17. L. Mei and M. Brydegaard, “Atmospheric aerosol monitoring by an elastic Scheimpflug lidar system,” Opt. Express 23(24), A1613–A1628 (2015). [CrossRef] [PubMed]

18. L. Mei and P. Guan, “Development of an atmospheric polarization Scheimpflug lidar system based on a time-division multiplexing scheme,” Opt. Lett. 42(18), 3562–3565 (2017). [CrossRef] [PubMed]

19. L. Mei, P. Guan, Y. Yang, and Z. Kong, “Atmospheric extinction coefficient retrieval and validation for the single-band Mie-scattering Scheimpflug lidar technique,” Opt. Express 25(16), A628–A638 (2017). [CrossRef] [PubMed]

20. L. Mei, Z. Kong, and P. Guan, “Implementation of a violet Scheimpflug lidar system for atmospheric aerosol studies,” Opt. Express 26(6), A260–A274 (2018). [CrossRef] [PubMed]

21. G. Zhao, E. Malmqvist, S. Török, P.-E. Bengtsson, S. Svanberg, and J. Bood, “Dual-band continuous-wave lidar system employed for particle classification,” Photon. J. (to be published).

22. F. G. Fernald, “Analysis of atmospheric lidar observations: some comments,” Appl. Opt. 23(5), 652–653 (1984). [CrossRef] [PubMed]

23. C. Böckmann, U. Wandinger, A. Ansmann, J. Bösenberg, V. Amiridis, A. Boselli, A. Delaval, F. De Tomasi, M. Frioud, I. V. Grigorov, A. Hågård, M. Horvat, M. Iarlori, L. Komguem, S. Kreipl, G. Larchevêque, V. Matthias, A. Papayannis, G. Pappalardo, F. Rocadenbosch, J. A. Rodrigues, J. Schneider, V. Shcherbakov, and M. Wiegner, “Aerosol lidar intercomparison in the framework of the EARLINET project. 2. Aerosol backscatter algorithms,” Appl. Opt. 43(4), 977–989 (2004). [CrossRef] [PubMed]

24. C. Zhang, L. Sun, and L. Chen, “Retrieval and analysis of aerosol lidar ratio at several typical regions in China,” Chin. J. Lasers 40(5), 0513002 (2013). [CrossRef]

25. H. Di, X. Hou, H. Zhao, L. Yan, X. Wei, H. Zhao, and D. Hua, “Detections and analyses of aerosol optical properties under different weather conditions using multi-wavelength Mie lidar,” Zhongguo Wuli Xuekan 63, 244206 (2014).

26. L. Mei, L. Zhang, Z. Kong, and H. Li, “Noise modeling, evaluation and reduction for the atmospheric lidar technique employing an image sensor,” Opt. Commun. 426, 463–470 (2018). [CrossRef]

27. R. A. Kotchenruther and P. V. Hobbs, “Humidification factors of aerosols from biomass burning in Brazil,” J. Geophys. Res. Atmos. 103(D24), 32081–32089 (1998). [CrossRef]

28. R. A. Kotchenruther, P. V. Hobbs, and D. A. Hegg, “Humidification factors for atmospheric aerosols off the mid-Atlantic coast of the United States,” J. Geophys. Res. Atmos. 104(D2), 2239–2251 (1999). [CrossRef]

29. C. M. Carrico, P. Kus, M. J. Rood, P. K. Quinn, and T. S. Bates, “Mixtures of pollution, dust, sea salt, and volcanic aerosol during ACE-Asia: radiative properties as a function of relative humidity,” J. Geophys. Res. Atmos. 108(D23), 8650 (2003). [CrossRef]

30. L. Zhang, J. Y. Sun, X. J. Shen, Y. M. Zhang, H. Che, Q. L. Ma, Y. W. Zhang, X. Y. Zhang, and J. A. Ogren, “Observations of relative humidity effects on aerosol light scattering in the Yangtze River Delta of China,” Atmos. Chem. Phys. 15(14), 8439–8454 (2015). [CrossRef]

31. A. Ångström, “On the atmospheric transmission of sun radiation and on dust in the air,” Geogr. Ann. 11, 156–166 (1929). [CrossRef]

32. D. L. Westphal and O. B. Toon, “Simulations of microphysical, radiative, and dynamical processes in a continental-scale forest fire smoke plume,” J. Geophys. Res. Atmos. 96(D12), 22379–22400 (1991). [CrossRef]

33. T. F. Eck, B. N. Holben, J. S. Reid, O. Dubovik, A. Smirnov, N. T. O’Neill, I. Slutsker, and S. Kinne, “Wavelength dependence of the optical depth of biomass burning, urban, and desert dust aerosols,” J. Geophys. Res. Atmos. 104(D24), 31333–31349 (1999). [CrossRef]

34. G. L. Schuster, O. Dubovik, and B. N. Holben, “Angstrom exponent and bimodal aerosol size distributions,” J. Geophys. Res. Atmos. 111(D7), D07207 (2006). [CrossRef]

35. N. G. Loeb and G. L. Schuster, “An observational study of the relationship between cloud, aerosol and meteorology in broken low-level cloud conditions,” J. Geophys. Res. Atmos. 113(D14), D14214 (2008). [CrossRef]

36. M. Pandolfi, M. Cusack, A. Alastuey, and X. Querol, “Variability of aerosol optical properties in the Western Mediterranean Basin,” Atmos. Chem. Phys. 11(15), 8189–8203 (2011). [CrossRef]

	Model	Specifications
Laser source	Unknown	λ₁ = 808 nm; Power: 4 W, CW; Mode: multimode; FWHM: ≈2 nm; Divergence: 6°║ × 8°⊥
Laser source	Ushio, HL40033G	λ₂ = 407 nm; Power: 1 W, CW; Mode: multimode; FWHM: ≈1.5 nm; Divergence: 13°║ × 45°⊥
Collimator	Tianlang, F6 refractor; Thorlabs, Cylindrical lens pair: LJ1918L1-A&LK1426L1-A	f = 600 mm, Ф100 mm Convex lens: f = 5.8 mm, height: 4 mm Concave lens: f = −25 mm, height: 10 mm
Receiver	Skywatcher, CFP200	f = 800 mm; Diameter: 200 mm
Detector	Lumenera CMOS camera Lt225NIR	Tilt angle: 45°; 2048 × 1024 Pixels; 170 fps; Bit: 12/8 bit; pixel: 5.5 µm × 5.5 µm; Quantum efficiency: 33% @407 nm, 42% @808 nm; ROI: 2048 × 240 pixels
Filters	Semrock LL01-407 & Edmund Optics 400 nm	407 nm: 1.7 nm FWHM; T>90% 400 nm: 25 nm FWHM; T>90%
Filters	808 nm interference filter Edmund Optics RG-715	808 nm: 3nm FWHM; T: >90% RG715:Longpass 715nm; T: >96%
Dichroic Mirror	DMLP490R	>97% T@808nm; >99% R @ 407nm

	Correlation coefficient at 808 nm		Correlation coefficient at 407 nm
	with PM2.5	with PM10	with PM2.5	with PM10
RH≤60%	0.86	0.89	0.86	0.81
60%<RH ≤75%	0.84	0.81	0.81	0.78
RH>75%	0.74	0.74	0.73	0.70

	Model	Specifications
Laser source	Unknown	λ₁ = 808 nm; Power: 4 W, CW; Mode: multimode; FWHM: ≈2 nm; Divergence: 6°║ × 8°⊥
Laser source	Ushio, HL40033G	λ₂ = 407 nm; Power: 1 W, CW; Mode: multimode; FWHM: ≈1.5 nm; Divergence: 13°║ × 45°⊥
Collimator	Tianlang, F6 refractor; Thorlabs, Cylindrical lens pair: LJ1918L1-A&LK1426L1-A	f = 600 mm, Ф100 mm Convex lens: f = 5.8 mm, height: 4 mm Concave lens: f = −25 mm, height: 10 mm
Receiver	Skywatcher, CFP200	f = 800 mm; Diameter: 200 mm
Detector	Lumenera CMOS camera Lt225NIR	Tilt angle: 45°; 2048 × 1024 Pixels; 170 fps; Bit: 12/8 bit; pixel: 5.5 µm × 5.5 µm; Quantum efficiency: 33% @407 nm, 42% @808 nm; ROI: 2048 × 240 pixels
Filters	Semrock LL01-407 & Edmund Optics 400 nm	407 nm: 1.7 nm FWHM; T>90% 400 nm: 25 nm FWHM; T>90%
Filters	808 nm interference filter Edmund Optics RG-715	808 nm: 3nm FWHM; T: >90% RG715:Longpass 715nm; T: >96%
Dichroic Mirror	DMLP490R	>97% T@808nm; >99% R @ 407nm

	Correlation coefficient at 808 nm		Correlation coefficient at 407 nm
	with PM2.5	with PM10	with PM2.5	with PM10
RH≤60%	0.86	0.89	0.86	0.81
60%<RH ≤75%	0.84	0.81	0.81	0.78
RH>75%	0.74	0.74	0.73	0.70

Dual-wavelength Mie-scattering Scheimpflug lidar system developed for the studies of the aerosol extinction coefficient and the Ångström exponent

Abstract

1. Introduction

2. Instrumentations and methods

2.1. Dual-wavelength Mie-scattering Scheimpflug lidar system

2.2. SLidar equation and extinction coefficient retrieval

3. Atmospheric remote measurements

3.1 Measurements

3.2 Lidar signal processing

3.3 Dependencies of the extinction coefficient on the mass concentration and the relative humidity

3.4 Studies on the relationship between the Ångström exponent, the mass concentration and the relative humidity

4. Conclusions

Funding

Acknowledgment

References

Cited By

Figures (10)

Tables (2)

Equations (5)

Optics Express