Open Access
Add to BibSonomy
Abstract
Aerosol intensive optical properties, including lidar ratio and particle depolarization ratio, are of vital importance for aerosol typing. However, aerosol intensive optical properties at near-infrared wavelength are less exploited by atmospheric lidar measurements, because of the comparably small backscatter cross section of Raman-scattering and a low efficiency of signal detection compared to what is commonly available at 355 nm and 532 nm. To obtain accurate optical properties of aerosols at near-infrared wavelength, we considered three factors: Raman-spectra selection, detector selection, and interference-filter optimization. Rotational Raman scattering has been chosen for Raman signal detection, because of the higher cross-section compared to vibrational Raman scattering. The optimization of the properties of the interference filter are based on a comprehensive consideration of both signal-to-noise ratio and temperature dependence of the simulated lidar signals. The interference filter that has eventually been chosen uses the central wavelength at 1056 nm and a filter bandwidth (full-width-at-half-maximum) of 6 nm. We built a 3-channel 1064-nm rotational Raman lidar. In this paper two methods are proposed to test the temperature dependence of the signal-detection unit and to evaluate the quality of the Raman signals. We performed two measurements to test the quality of the detection channel: cirrus clouds in the free troposphere and aerosols in the planetary boundary layer. Our analysis of the measured Raman signals shows a negligible temperature dependence of the Raman signals in our system. For cirrus measurements, the Raman signal profile did not show crosstalk even for the case of strong elastic backscatter from clouds, which was about 100 times larger than Rayleigh scattering in the case considered here. The cirrus-mean extinction-to-backscatter ratio (lidar ratio) was 27.8 ± 10.0 sr (1064 nm) at a height of 10.5-11.5 km above ground. For the aerosols in the planetary boundary layer, we found the mean lidar ratio of 38.9 ± 7.0 sr at a height of 1.0-3.0 km above ground.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Aerosol particles play a vital role for Earth’s climate system and different aerosol types have distinctive climatic impacts [1]. Biomass-burning aerosols are highly light-absorbing and thus can heat the atmosphere. In contrast, inorganic aerosols originating from anthropogenic sources are less light-absorbing and can cool the surface by scattering solar radiation [2]. Aerosols have a short lifetime and large spatiotemporal variability in the atmosphere. Knowing details of optical, chemical, microphysical and morphological properties of aerosol types and their concentrations on a temporal and vertical scale can help with more accurate assessments of aerosol radiative forcing.
Aerosol intensive optical properties, such as the linear particle depolarization ratio and lidar (extinction-to-backscatter) ratio, can be used as “fingerprints” for classifying aerosol types [3]. The linear particle depolarization ratio is a good indicator of particle shape as an enhanced particle depolarization ratio always shows the presence of nonspherical particles, such as for dust [4,5], volcanic ash or dry sea salt [6]. The lidar ratio is an indicator of the capacity of particle light-absorption and particle size [7]. Aerosol extensive optical properties, such as aerosol backscatter/extinction coefficients are direct measures of particle (surface area) concentration and can serve as an indicator of particle number/mass concentration [8–10].
Atmospheric lidar is one of the most important tools for aerosol profiling with high spatiotemporal resolution [11]. The intensity of Rayleigh scattering is proportional to λ-4, where λ describes the wavelength of the emitted laser beam. For that reason light at 1064 nm becomes less attenuated by atmospheric molecules with increasing measurement height (as a result of decreasing molecule concentration) than laser light at 355 nm and 532 nm [12,13].
In the near-infrared band, high-quality parameters of particle linear depolarization ratio and lidar ratio measured by polarization Raman lidar are important for an improved understanding of the formation of (cirrus) clouds and a better characterization of aerosols (aerosol types). In addition, the lidar ratio obtained from Raman lidar observations is required for deriving extinction coefficients from the inversion of the Mie-scattering lidar equation [14]. For example, the lidar aboard CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations) was used for monitoring the global distribution of aerosols. The aerosol extinction and backscatter coefficients were derived by assuming fixed lidar ratios for different aerosol types [15]. Quantitative, accurate measurements of aerosol optical properties were limited due to the highly variable lidar ratio. Thus, the lidar ratio observed by the CALIPSO lidar was one of the largest sources of uncertainty in the retrievals of aerosol extinction [16].
The development of lidar technology allowed for establishing two lidar techniques for comparably accurate measurements of extinction and backscatter coefficients of aerosols: the Raman lidar technique [17] and the High Spectral Resolution Lidar (HSRL) technique [18]. The application and development of HSRL lidar has been limited, owing to its high structural complexity and the need for significant expert knowledge to build and operate HSRL. Raman lidar has advantages in regard to long-term ground-based measurements, due to its simple setup, for example with regard to robust emitting and receiving components [19,20].
However, literature on the topic of Raman lidar technology at near-infrared wavelength still is quite limited. Haarig et al. (2016) used for the first time multiple rotational Raman lines at 1058 nm for directly measuring vertical profiles of 1064-nm extinction coefficients [21]. For the first time, particle backscatter coefficients and extinction coefficients and linear particle depolarization ratios at 355, 532 and 1064 nm (denoted as 3α + 3β + 3$\delta $ configuration) were measured by the advanced and sophisticated lidar system BERTHA (Backscatter Extinction lidar-Ratio Temperature Humidity profiling Apparatus) [22].
It is well known that the accuracy of lidar-derived data products is determined by two parts: hardware and inversion algorithms. The hardware is the basis for signal detection accuracy. For near-infrared Raman lidar systems, hardware parameters such as energy and repetition frequency of the laser, photon detection efficiency (PDE) and dark count rate (DCR) of the detector unit, the aperture and field-of-view of the signal-receiving telescope, and the center wavelength (CWL) and bandwidth of the interference filters (IF) have a direct impact on the accuracy of aerosol and cloud measurements.
Currently, detectors are less efficient in the near-infrared band than in the visible and UV bands. DCR is an important factor for evaluating the detector performance [23]. Signal-to-noise ratio (SNR) and temperature dependence are the main parameters that need to be evaluated in the context of Raman-channel signal-quality. The parameters need to be comprehensively assessed in the design of Raman channel parameters [24–26]. The cross-section of rotational Raman scattering is temperature dependent and could lead to a large error in the retrieval of extinction coefficients if it is not properly considered in data analysis [26]. Reducing the temperature dependence of the total cross section of a selected rotational Raman spectrum is of high importance for achieving high-quality measurements of aerosol optical properties.
The SNR and temperature dependence are also closely related to the parameters of the interference filter. Therefore, it is necessary to choose a suitable detector, as well as the optimal parameters of the interference filter to meet the requirements of high-precision measurements of optical properties of aerosols and clouds.
In this article, we report on the design of a 1064-nm Raman lidar system which has been optimized for measuring important optical properties, such as the lidar ratio of clouds and aerosols. This paper is structured as follows: in section 2 we explain the optimization of the CWL and bandwidth of the narrowband interference filters. Section 3 shows the system design of the 1064-nm rotational Raman lidar. Section 4 presents the results of test measurements. Section 5 presents the conclusion of the paper.
2. Optimization of Raman channel at near-infrared wavelength
In the paper, we described the design of a Raman measurement channel that allowed for measurements of the extinction coefficient at 1064 nm. We aimed at achieving high detection efficiency of the signals that are needed for measurements of the Raman signals in the near IR. The quality of the Raman signals was determined by two factors: selection of the Raman spectra and parameter optimization of the interference filter (IF) that is used for the Raman channel. The selection of the Raman spectra was primarily about making a choice between the rotational Raman (RR) band and the vibration-rotation Raman (VRR) bands. The parameter optimization of the IF which is the core of the Raman-channel design, mainly targeted the choice of its central wavelength (CWL) and bandwidth, i.e., full width at half maximum (FWHM).
2.1 Spectral selection
Figure 1 shows the molecular backscatter spectrum for a laser excitation wavelength of 1064.1 nm. The Raman backscatter cross-section depicted in this figure quantitatively describes the characteristics of the Raman spectra. Nitrogen and oxygen serve as reference gases. Their atmospheric concentrations are known in principle from measurements of temperature and pressure. The Rayleigh scattering spectrum, and the RR spectra and VRR spectra of nitrogen and oxygen molecules are included in Fig. 1 [27,28]. The Stokes and anti-Stokes lines of the VRR spectra of nitrogen and oxygen are also shown in Fig. 1. The black line represents the Rayleigh spectrum. The red and blue lines represent the Raman spectra of nitrogen and oxygen molecules, respectively.
Fig. 1. Raman backscatter spectra of N2 and O2, excited by an Nd: YAG laser that emits light pulses at 1064.1 nm for ambient pressure of 1 atm and temperature of 220 K. The respective relative volume abundances are 0.78 for N2 and 0.21 for O2.
The section of the spectrum that is most suitable for our instrument design depends on 3 individual parts: the selection of VRR and RR spectra, the selection of the most appropriate detector and a thorough understanding of the temperature dependence of the hardware and the Raman signals.
The intensity of the spectral lines is an important basis for the selection of the RR spectra and then VRR spectra. As shown in Fig. 1, the cross-sections of Stokes and anti-Stokes VRR scattering are 2 and 4 orders of magnitude lower compared to the cross-sections of the RR spectra. The intensity of the Raman signals of the anti-Stokes VRR spectra is comparably low and may not meet the requirements with regard to achieving the necessary detection distance and/or temporal-spatial resolution. The Stokes VRR of N2 is within the absorption band of water vapor and thus is not suitable. Since O2 is less abundant in the atmosphere than N2, the Stokes VRR of O2 molecules has usually not been considered as a source of Raman signals. For the same reason we also did not consider O2 in our work. In summary of the above analyses, we consider the RR spectra as a more suitable choice for a Raman channel at near infrared wavelength.
The efficiency of the detector that can be used of RR spectra was another issue that had to be considered. The RR spectra are in the near-infrared band and photon detection efficiency (PDE) of different detectors varies widely in this band range. Currently technologically mature single photon detectors (SPAD) that are commercially available are mainly Silicon-SPAD and InGaAs-SPAD. The parameters limiting the performance of SPAD include their photon detection efficiency (PDE), the dark count rate (DCR) and the after pulsing (AP) probability [29]. The dark count rate of InGaAs-SPAD is more than an order of magnitude higher than that of Si-SPAD. Although the average DCR can be subtracted from the signal count rates, the time-dependent fluctuations in the DCR constitute a noise contribution in the SPAD performance [30]. For single-photon detection, DCR is the main factor influencing signal detection. After-pulsing is also more severe in InGaAs-SPAD than in Silicon devices. Considering these effects, we used a Si-SPAD for the detection of signals from the RR spectra.
The PDE curve given by the manufacturer of the detector does not completely cover the entire RR spectrum that we need for our work. We obtained a complete PDE curve by extrapolating the data that were provided by the manufacturer in the data sheet. To obtain a more accurate curve, we introduced a zero-response point. This point describes from a theoretical point of view the cutoff wavelength (of the silicon material) at which the detection efficiency is 0.
An important factor that needed to be considered was the temperature dependence of the RR spectra of N2 and O2 molecules. Figure 2 shows the rate of change of the backscatter cross-sections per unit temperature (expressed as temperature variation factor) if the temperature changes from 230 K to 300 K. The RR spectra show a positive and negative region of change. This behavior indicates that the intensities of the RR-spectra lines are temperature dependent.
Fig. 2. (a) Backscatter cross-section for nitrogen and oxygen at different temperatures (230 K and 300 K). (b) Temperature variation factor of the RR spectra when temperature (TVF) is varied from 230 K to 300 K. Rotation-Raman spectra of N2 and O2, excited by a laser at 1064.1 nm for ambient pressure of 1 atm. The respective relative volume abundances are 0.78 for N2 and 0.21 for O2.
The workable method considered in this paper was to select suitable portions of temperature-dependent RR spectra lines that are opposite to one another [21,26]. The selection of the RR lines was achieved by the use of a specially parameter-designed IF which transmits the desired portion of the RR lines [26]. By applying this approach, the temperature dependence of the RR scattering cross-section can theoretically be reduced to an adequate level for tropospheric measurements. This specific approach is described in detail in another section of this paper.
2.2 Interference filter parameter optimization
The specially designed IF needs to suppress the elastic-scattering component from entering the RR channel as much as possible. The main aspect that affects the out-of-band rejection performance of the IF is the line shape of the transmittance curve.
Current manufacturing skills do not allow for creating the ideal rectangular shape of the transmittance curve. Different manufacturers will use different manufacturing processes and therefore the transmittance curve will also be different. In order to improve the accuracy of the subsequent optimization results, the line function of the modified Gaussian [31] was determined by applying a linear least-squares regression analysis to the data that have been provided by the manufacturer.
The modified Gauss function is defined as:
Figure 3 shows the modified Gaussian function that was used for fitting the data acquired by the manufacturer. We find a correlation coefficient of determination, expressed in terms of R2 of 0.88.
Fig. 3. Modified Gaussian function fitted to data provided by the manufacturer of the IF. The correlation coefficient R2 = 0.88. Vertical axis shows numbers (unitless) on logarithmic scale.
The leakage of the elastic signals into the Raman channel needs to be suppressed. The ability of the Raman channel to suppress the elastic signals depend on the quality, i.e., optimization of the interference filter. The intensity of the Raman signal and the ability to suppress elastic signal needed has significant influence on the quality of tropospheric measurements. In the simulations described here we selected targets at a distance of 10 km and a backscatter ratio of 10. That number is typical for cirrus clouds.
An important quantity in lidar measurements of aerosols and clouds is the backscatter ratio (denoted in the following text as R) [32]. The parameter R is defined as the ratio of the total backscatter coefficient (aerosol particles βa plus molecules βm) to Rayleigh (molecular) scattering βm [33]. We can use the backscatter cross-sections of Mie and Rayleigh scattering instead of their respective backscatter coefficients to calculate the backscatter ratio [32].
For the case of R equal to 10, the interference filter parameters need to be optimized with respect to signal-to-noise ratio (SNR) and temperature dependence. The temperature dependence can be quantified by the temperature variation factor (TVF). Figure 4 shows the flowchart of the optimization process. We briefly describe the processing steps.
- (1) Calculation of the simulated lidar signals for different assumptions of CWL and FWHM. The following numbers were used: the FWHM was set from 1 to 12 nm. The step size of the FWHM to the next higher level was 1 nm. The range of the CWL covers the complete rotational Raman spectrum. The step size of the CWL is 0.5 nm. The simulated lidar signal intensity was calculated for different CWLs and FWHMs.
- (2) Calculation of the SNR and TVF.
- (3) Selection of the upper limiting region of the SNR. This region is defined as the region where the SNR is greater than 95% of the maximum SNR.
- (4) Selection of the lower limiting region of the TVF. The lower limiting region is defined as the region where the TVF is lower than 5% of the minimum TVF.
- (5) Identification of the region where the upper limiting region and the lower limiting region overlap each other. We denote this region as the candidate region.
- (6) In this candidate region, we choose suitable values of the CWL and FWHM as the final result of the optimization.
2.2.1 Optimization condition 1: signal-to-noise ratio
In general, the total Raman channel signal (${S_{Total}}$) consists of two parts: the Raman signal (${S_{Raman}}$) and the leakage of the Rayleigh and Mie signals (${S_{Leakage}}$). In our simulations, we expressed the signal intensity in terms of the effective backscatter cross section (${\delta ^{eff}}$). The parameter ${\mathrm{\delta }^{\textrm{eff}}}$ describes the convolution of the backscattering cross-section ($\delta $), the photon detection efficiency (PDE) and the transmittance function (${f_{IF}}$) of the interference filter. The RR backscattering cross-section (${\delta _{RR}}$) can be calculated as the sum of the contributions from the individual rotational lines of nitrogen and oxygen. The total cross-section (${\delta _{Total}}$) is the combination of the RR backscattering cross-section (${\delta _{RR}}$) and the cross-section of the elastic backscatter.
The parameters ${S_{Total}}$ and ${S_{Raman}}$ are calculated by using Eqs. (2) and (3). The Raman signal intensity is proportional to the FWHM of the interference filter. The intensity of the Raman signal shows a peak in the anti-Stokes and Stokes spectral bands.
We calculate the SNR by Eq. (4). We select the upper-limit region of the SNR on basis of the condition that the maximum SNR is greater than 95%. Figure 5(a) shows the results. The SNR is not linearly related to the variation of both bandwidth and wavelength, but shows a bimodal pattern. The SNR is maximized at the location of the CWL (1055 nm) and FWHM (7 nm). Figure 5(b) shows the SNR normalized to its maximum value. The upper-limit region is marked by a white dashed line.
Fig. 5. (a) SNR for different central wavelengths (CWLs) and bandwidths (FWHMs). (b) Normalized SNR, and upper limit region (marked by a white dashed line)
2.2.2 Optimization condition 2: temperature dependence
The scattering cross-sections in the RR-spectra lines are temperature dependent. The RR measurements can be used for calculating the extinction and backscatter coefficients. The temperature dependence of the RR spectra of the N2 and O2 molecules needs to be considered.
The approach considered in this paper is to select a portion of the RR spectrum by the use of a specifically designed interference filter. This filter only transmits the desired portion of the RR-spectra lines. Here, we consider temperature variations in the range of 230 K to 300 K, which represents the typical temperature changes below 15 km height.
The temperature dependence is characterized by the rate of change of the signal strength per unit temperature change. This dependence can be simplified in terms of a temperature variation factor (TVF).
The TVF is calculated by Eq. (5):
Figure 6 shows the results.
Fig. 6. (a) Temperature variation factor for different CWLs and FWHMs, (b) normalized TVF, upper-limit region of SNR (marked by the white dashed line) and region of the lower limit of the TVF (marked by the dark-gray dashed lines). The best optimization is marked by a red spot.
As shown in Fig. 6(a), for different FWHMs, there are always wavelength values where the corresponding TVF value is zero. On both sides of the wavelengths where TVF is zero, the value of TVF becomes progressively larger. The TVFs normalized to their maximum values are shown in Fig. 6(b). The lower limit region is marked by a white dashed line.
We considered the technical capability of the manufacturer of interference filter. When we selected the optimum parameters of the IF, the CWL and the FWHM we used steps of 0.5 nm and 1 nm, respectively. All in all, the optimum choice of the CWL is 1056 nm and the optimum choice of the FWHM is 6 nm (marked with a red spot, in Fig. 6(b)).
3. 1064 nm rotational Raman lidar: hardware set-up
Figure 7 shows the schematic of the 1064-nm rotational-Raman lidar. The lidar has been designed with a co-axial transceiver so that there is a relatively small blind area, i.e., range of incomplete overlap between emitted laser beam and receiver field of view. This lidar has three receiving channels: parallel polarization (P), cross polarization (S), and Raman (R) channel. Table 1 shows some technical parameters of the system.
Table 1. Main parameters of the 1064 nm polarization Raman lidar
The lidar system is composed of three main units: the laser emitting unit, the signal receiving unit, and the signal acquisition & controller unit. The laser is a diode-pumped solid-state (DPSS) laser with air cooling (Bright Solutions). This laser has a linewidth of 0.7 cm-1. Laser pulses are transmitted into the atmosphere after passing through a beam expander (BE). The magnification factor of the beam expander is 5. A 30-cm-diameter Cassegrain telescope collects the backscattered light. The field-of-view (FOV) of the signal-receiving unit is 1 mrad.
The signal receiving unit consists of two subunits: the receiving collimator unit and the spectroscopic unit. The main function of the receiving collimator unit is to collect and collimate the backscattered light. The receiving collimator unit includes a telescope, a field stop and a collimating lens (CL). The Cassegrain telescope is composed of special materials that results in a comparably small temperature drift. The telescope design ensures the consistency of the optical performance in case large temperature changes occur. After the scattered light passes through the collimator, it enters the spectroscopic unit.
The spectroscopic unit, which is the core part of the lidar system, is used for the spectral separation of the received signals. The collimated light first passes through the beam splitter (BS). The reflected light enters the Raman channel. The transmitted light enters the elastic channel.
The reflected light passes through the interference filter (IF1) of the Raman channel and then enters the fiber collimator (FC). The light emitted from the FC is coupled to the SPAD through an optical fiber. Due to the low intensity of the Raman signal, no neutral density filter is used in the 1056 nm Raman channel.
The transmitted light passes through the interference filter (IF2) of the elastic channel and then passes through the polarizing beam splitter (PBS). The PBS separates the light into two components, the parallel-polarized and the perpendicular-polarized signals. The light then enters the FC and the SPAD.
The signal acquisition & controller unit mainly consists of two parts, the data acquisition board and the control board. The control board is designed for controlling the laser, the power supply of the APDs, the data acquisition board, and the servo-motors which are used for steering the positions of the depolarizer and attenuators, respectively. The data acquisition board is operated in photon counting mode with an optical trigger, and records backscatter signals with a resolution of 15 m and 1 min, respectively. The three measurement channels are operated in photon counting mode, detected by an SPCM-NIR (Excelitas).
4. Results
In this section, we focus on the results we obtained from testing the temperature dependence and assessing the quality of the Raman signals. We carried out two measurements, one of which targeted the detection of cirrus cloud and the other one aerosol in the planetary boundary layer (PBL). We propose a new method for testing the temperature dependence. This new method will be described in detail in the following sections of this paper. The quality of the Raman signals was assessed by the relative deviation of cloud optical depth (COD) of the cirrus cloud calculated by two different methods. We also calculated particle backscatter coefficients, extinction coefficients and corresponding extinction-to-backscatter ratios (lidar ratio) at 1064 nm of the cirrus and the aerosols on the PBL.
The lidar we used for this study is on the top floor of the Laser Remote Sensing Laboratory of Wuhan University (30°31.7614’ N, 114°21.1304’ E, 80 meters above sea level (asl)). The examples correspond to measurements on 9 Sep 2023 and 9 July 2023. The laser beam pointed to the zenith during the time of observation. We did not apply a correction of the incomplete overlap between emitted laser beam and receiver-field-of-view of the receiver telescope (overlap correction) below 200 m asl.
Figure 8 shows the measurement from 9 Sep 2023, 19:14 local time (LT) to 10 Sep 2023, 05:31 LT. The temporal resolution and vertical resolution of the raw signal are 1 minute and 15 m, respectively. Figure 8 show the space-time distribution of the range-corrected signals (RCS). The temporal variation of aerosol concentration within the boundary layer below 1.0 km asl can be clearly seen. The comparably faint layers (in terms of optical depth) are visible between 1 km and approximately 5 km height asl. A cirrus cloud was present between 10-12 km asl, indicated by strong backscatter signals (dark red). This cirrus remained visible in the lidar signals for about one hour. We analyzed the lidar signals for two scenarios, i.e., one cloud-free and the one with the cirrus cloud, respectively. These two scenarios correspond to the white and dark-gray rectangular boxes in Fig. 8.
Fig. 8. Height-time distribution of the elastic RCS signals (1064 nm) using the RR lidar at Wuhan University. Shown are the signals collected between 9 and 10 Sep 2023, 19:14 - 05:31 LT. The temporal resolution and spatial resolution are 1 minute and 15 m, respectively. The two scenarios of cloud-free and cirrus cloud are marked with white and dark-gray rectangular boxes, respectively. The laser beam pointed into vertical direction during the time of observation.
4.1 Case 1: test of temperature dependence
The temperature dependence can be evaluated by comparing the signals to a temperature independent signal. This temperature independent signal is the Rayleigh backscatter signal measured at the same wavelength but with the rotational Raman channel at 1056 nm [25,34]. The measured Raman signal would deviate from the Rayleigh backscatter signal if there was temperature-dependence effect. Moreover, the extinction coefficient obtained by the Raman method would also be affected which would be indicated by a general systematic error at all range bins [25].
In this paper, we present a method for testing the temperature dependence. The temperature dependence was tested in terms of the uncertainty of the Raman signal, the relative deviation between the Raman and molecular signals, and the extinction coefficient. This method is described in detail as follows: (1) The Raman and atmospheric molecular signal were matched by Rayleigh-fit algorithm and the reference height range was recorded. (2) The relative deviations between the Raman and atmospheric molecular signals and the uncertainty of the Raman signals were calculated. Within the reference height range, if the relative deviation is smaller than the uncertainty, then it can be stated that there is no obvious systematic error in the Raman signal itself. (3) We also calculated the extinction coefficients of the aerosols using the Raman method. The extinction coefficients were compared to the value 0 within the reference height range. If the extinction coefficients are scattered around this value 0 it means that there is only statistical error and no systematic error. If both conditions (2) and (3) are satisfied simultaneously, we can state that the Raman channel signal has no significant temperature dependence.
In order to obtain accurate atmospheric molecular signals, Rayleigh scattering needs to be accurately determined from a-priori knowledge of atmospheric temperature and pressure [35]. These data can be obtained from a radiosonde which preferably should be launched near the lidar. For our test, we used radiosonde data from the observation station #57494 (Wuhan, 30°36’ N, 114°3’ E) managed by the Wuhan Meteorological Bureau which is approximately 30 km to the northwest of the observation site.
Figure 9 shows a measurement example of a cloud-free atmosphere. The profiles of the RCS have been derived from the elastic and Raman signals on 9 Sep 2023 between 19:14-20:14 LT. We averaged the lidar signals for one hour. The vertical resolution is 60 m. An aerosol layer was present below 7 km, see Fig. 9. Utilization of the Rayleigh fit algorithm with signal segmentation [35–37], the reference height range within which we carried out this fit was set to of 7-14 km asl.
Fig. 9. RCS of the elastic (red line) and Raman (black line) signals measured on 9 Sep 2023, 19:14 - 20:14 LT. The signal profiles are averaged over 1 h measurement time. The vertical resolution is 60 m. The blue dashed lines correspond to the Rayleigh profiles to which the measured signals were fitted. The green portion of the signal profile indicates the height range in which the Rayleigh fit was applied.
The relative deviation between the Raman signals and the molecular signals, as well as the uncertainty of the Raman signals are shown in Fig. 10(a). We smoothed the Raman signal profiles (averaged across one hour measurement time) with a vertical window length of 200 m. We calculated the uncertainty of the Raman signals and the relative deviation. Figure 10(a) shows that the relative deviation is less than the uncertainty of the Raman signals within the reference height range.
Fig. 10. (a) Relative deviation of the Raman signals and the molecular signals. The relative deviation is marked with a solid light-blue line. The uncertainty is indicated by the shaded areas. (b) Extinction coefficient for an aerosol-free atmosphere. The dark-blue shaded area represents the standard error of the extinction coefficient. The black dashed box indicates the range of the reference height that was used for the Rayleigh fits.
The profile of the extinction coefficient (indicated by a solid black line) is shown in Fig. 10(b). The blue-shaded area represents the error of the extinction coefficient. The values of the aerosol extinction coefficients are scattered around the value 0. This behavior indicates the absence of systematic errors in the extinction coefficients. The SNRs of the Raman signals decrease with increasing detection height and the statistical noise increases with height. In summary of the above descriptions, we conclude that the temperature dependence can be neglected.
4.2 Case 2: cirrus cloud
Cirrus provides an almost optimum scenario for checking the performance of the 1056 nm RR lidar [21]. In principle cloud optical depth (COD) can be derived from the transmissivity of clouds. The transmissivity of clouds can be obtained by calculating the extinction coefficient (denoted as extinction method) or by comparing the backscatter signals at the cloud base (Sbase) and cloud top (Stop) (denoted as signal method) [38]. In this paper, we used the relative deviation of COD obtained from using these two methods to evaluate the quality of the Raman signals.
Figure 11(a) shows a measurement example of the RCS at 1064 nm (elastic signal) and 1056 nm (Raman signal) in the presence of cirrus clouds that were observed for several hours. The base of the cirrus layer remained comparably stable at 10 km asl during the time of this event. The top height of the cirrus was detected at 12 km asl. Figure 11(b) shows an enlargement of the signal profiles shown in Figure.11(a). The backscatter intensity increased by 2 orders of magnitude at the cloud base, according to the 1064 nm elastic backscatter signal. The Raman signal profiles do not show any interference by elastic backscatter by cirrus.
Fig. 11. Profiles of RCS. Signals are averaged over a 1-h measurement period, i.e., from 2:30 - 3:30 LT. Data were taken on 10 Sep 2023 at Wuhan University. The vertical resolution of the profiles is 60 m. The observation zenith angle of this experiment is 90 degrees. (a) Signal profiles of cirrus observation, (b) Enlargement of the signal profiles in the 7-15-km height interval of (a). For display purposes the signal profile at 1056 nm has been magnified five times. Sbase and Stop are the matched signals at the cloud base and cloud top, respectively.
Figure 12 shows the height profiles of the particle backscatter, extinction coefficients and the lidar ratio in the cirrus. The results were obtained from averaging a one-hour segment of the lidar signals. The values of the extinction coefficient are around 95 Mm−1 in 10.5 - 11.5 km height agl (inside the cirrus). The lidar ratio varies from 17.5 - 37.0 sr in the same height range. The cirrus-mean lidar ratio was 27.8 ± 10.0 sr for this measurement case. The shaded error bars were calculated by error propagation [39].
Fig. 12. Particle extinction coefficients (a), backscatter coefficients (b), and lidar ratio (c). The vertical resolution of the raw signals was 15 m. We used 60 minutes of measurement time for calculating the profiles. We smoothed these temporally averaged signal profiles with a vertical window length of 300 m.
The COD of the cirrus cloud was calculated to be 0.19 ± 0.02 by the extinction method. The COD of the cloud was calculated to be 0.16 by the signal method at 1064 nm. The relative deviation of COD obtained from these two methods is less than 16%.
4.3 Case 3: aerosols within planetary boundary layer
Finally, a measurement of aerosols in the planetary boundary layer (PBL) was used as another test of the instrument. Observations were carried out from 9 July 2023, 22:45 local time (LT) to 11 July 2023, 05:41 LT. The temporal resolution and spatial resolution of the raw signals are 1 minute and 15 m, respectively.
Figure 13 shows the temporal evolution of the elastic and Raman signals. Aerosol concentration varied significantly within the boundary layer below1.5 km asl during the observation time. There was a geometrically thin aerosol layer at about 3.5 km height above ground. We selected the observations that correspond to gray rectangular boxes in Fig. 13(a). Figure 14 shows a 2-hour segment (22:45 - 00:45 LT, 9 -10 July 2023) that was used for a more detailed data analysis. We averaged the lidar signals for these two hours to obtain the lidar profiles. The vertical resolution is 60 m.
Fig. 13. Height-time distribution of the RCS. Shown are the elastic (a) and Raman (b) signals taken between on 9 and 10 July 2023, 22:45 to 03:57 LT. The laser beam pointed into vertical direction. The temporal resolution and vertical resolution are 1 minute and 60 m, respectively. The time segment marked by gray rectangular dashed box was used for further data analysis.
Fig. 14. RCS averaged over a 2-h period from 22:45 to 00:50 LT on 9 July 2023. The temporal resolution and spatial resolution are 1 minute and 60 m, respectively.
Figure 15 shows the height profiles of the particle backscatter and extinction coefficients, and the lidar ratio. We used a gliding 500-m window for vertical smoothing of the Raman signals in the calculation of the particle extinction coefficient. The extinction values vary between 30 and 100 Mm−1 (1056 nm) in the height range between 1.0 - 3.0 km. The lidar ratio varies between 35 - 50 sr in that height range. The mean value of the lidar ratio in the PBL in that height range is 38.9 ± 7 sr (1064 nm).
Fig. 15. Particle extinction coefficients (a), backscatter coefficients (b), and lidar ratio (c). The resolution of the backscatter profile is 60 m. The signals were smoothed with a 500-m gliding average for the calculation of the extinction coefficients. The vertical resolution of the lidar ratio is 60 m, too. Shaded error bars show the uncertainty caused by signal noise during the lidar measurements.
5. Conclusions
In this article, we present results obtained with optimized interference filters that are used for the Raman channel of a lidar that is used for measurements of particle extinction at 1064 nm.
We first investigated the photon detection efficiency and the interference filter function to determine what optimization steps can be done. The photon detection efficiency of the detector used in our study should cover the entire wavelength range of the rotational Raman spectrum. The real photon detection efficiency values that we obtained from the datasheet of the manufacturer does not cover the entire rotational Raman spectrum. Therefore, we extrapolated the PDEs for the required individual points within the entire rotational Raman spectrum.
The modified Gaussian function was used to model the transmittance of the interference filter.
The optimization of the parameters of the modified Gaussian function is based on a real transmittance that we obtained from data provided by the manufacturer We applied a linear least-squares regression analysis to obtain the best parameter of the modified Gaussian function.
We simulated Raman signals based on different center wavelengths and FWHMs. Parameter optimization of the center wavelength and FWHM of the interference filter of the Raman channel was performed for the scenario of a backscatter ratio of 10. Signal-to-noise ratio and temperature dependence are the two main criteria in the optimization process. The temperature variation factor allows for quantifying the temperature dependence. We select the final optimization result based on the optimization schemes described in this article. After comprehensive analysis, we selected an interference filter with a center wavelength of 1056 nm and a FWHM of 6 nm as the final optimization result.
In the final stage of this work, we built a 3-channels 1064-nm rotational-Raman lidar. We evaluated the performance of this lidar based on three measurement situations. In the first case we tested the temperature dependence of the detection unit. We used atmospheric temperature and pressure data from a nearby radiosonde launch to calculate the atmospheric molecular signal at the time of lidar observation. We analyzed the Raman signal and extinction coefficient based on the testing method proposed in this article. There was no obvious temperature dependence of the Raman signal in the actual measurement process of the lidar. We assessed the quality of the Raman signals by using the relative deviation of the cloud optical depth of a cirrus cloud calculated by the extinction and signal methods, respectively. The relative deviation is less than 16%.
We measured particle extinction coefficient, backscatter coefficient and lidar ratio profiles for cirrus clouds and aerosols in the PBL in a second and third test measurement. Within a cirrus cloud, the values of the extinction coefficient are around 95 Mm−1 and the mean lidar ratio is 27.8 ± 10.0 sr at 1064 nm. We find sufficient suppression of the elastic signals by the interference filters. In the case of urban pollution in the PBL, we find a mean lidar ratio of 38.9 ± 7.0 sr.
Funding
National Key Research and Development Program of China (2023YFC3007800); National Natural Science Foundation of China (42205130, 62105248, 62275202).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data will be made available on request.
References
1. Z. Yin, F. Yi, F. Liu, et al., “Long-term variations of aerosol optical properties over Wuhan with polarization lidar,” Atmos. Environ. 259, 118508 (2021). [CrossRef]
2. I. S. Stachlewska, M. Samson, O. Zawadzka, et al., “Modification of local urban aerosol properties by long-range transport of biomass burning aerosol,” Remote Sens. 10(3), 412 (2018). [CrossRef]
3. V. Amiridis, D. Balis, E. Giannakaki, et al., “Optical characteristics of biomass burning aerosols over Southeastern Europe determined from UV-Raman lidar measurements,” Atmos. Chem. Phys. 9(7), 2431–2440 (2009). [CrossRef]
4. S. Zhang, Z. Huang, M. Li, et al., “Vertical structure of dust aerosols observed by a ground-based Raman lidar with polarization capabilities in the center of the Taklimakan Desert,” Remote Sens. 14(10), 2461 (2022). [CrossRef]
5. Q. Dong, Z. Huang, W. Li, et al., “Polarization lidar measurements of dust optical properties at the junction of the Taklimakan Desert–Tibetan Plateau,” Remote Sens. 14(3), 558 (2022). [CrossRef]
6. M. Saito and P. Yang, “Advanced bulk optical models linking the backscattering and microphysical properties of mineral dust aerosol,” Geophys. Res. Lett. 48(17), e2021GL095121 (2021). [CrossRef]
7. A. Kolgotin, D. Müller, E. Chemyakin, et al., “Improved identification of the solution space of aerosol microphysical properties derived from the inversion of profiles of lidar optical data, part 1: theory,” Appl. Opt. 55(34), 9839–9849 (2016). [CrossRef]
8. M. Tesche, A. Ansmann, D. Müller, et al., “Vertical profiling of Saharan dust with Raman lidars and airborne HSRL in southern Morocco during SAMUM,” Tellus Ser. B-Chem. Phys. Meteorol. 61(1), 144–164 (2022). [CrossRef]
9. Y. He, Y. Zhang, F. Liu, et al., “Retrievals of dust-related particle mass and ice-nucleating particle concentration profiles with ground-based polarization lidar and sun photometer over a megacity in central China,” Atmos. Meas. Tech. 14(9), 5939–5954 (2021). [CrossRef]
10. RE. Mamouri and A. Ansmann, “Fine and coarse dust separation with polarization lidar,” Atmos. Meas. Tech. 7(11), 3717–3735 (2014). [CrossRef]
11. J. Guerrero-Rascado, B. Ruiz, and L. Alados-Arboledas, “Multi-spectral Lidar characterization of the vertical structure of Saharan dust aerosol over southern Spain,” Atmos. Environ. 42(11), 2668–2681 (2008). [CrossRef]
12. Q. Chen, S. Mao, Z. Yin, et al., “Compact and efficient 1064 nm up-conversion atmospheric lidar,” Opt. Express 31(15), 23931–23943 (2023). [CrossRef]
13. J. H. Xian, D. Sun, S. Amoruso, et al., “Parameter optimization of a visibility LiDAR for sea-fog early warnings,” Opt. Express 28(16), 23829–23845 (2020). [CrossRef]
14. F. G. Fernald, B. M. Herman, and J. A. Reagan, “Determination of aerosol height distributions by lidar,” J. Appl. Meteor. 11(3), 482–489 (1972). [CrossRef]
15. A. H. Omar, D. M. Winker, M. A. Vaughan, et al., “The CALIPSO automated aerosol classification and lidar ratio selection algorithm,” J. Atmos. Ocean. Technol. 26(10), 1994–2014 (2009). [CrossRef]
16. D. M. Winker, M. A. Vaughan, A. Omar, et al., “Overview of the CALIPSO mission and CALIOP data processing algorithms,” J. Atmos. Ocean. Technol. 26(11), 2310–2323 (2009). [CrossRef]
17. A. Ansmann, M. Riebesell, and C. Weitkamp, “Measurement of atmospheric aerosol extinction profiles with a Raman lidar,” Opt. Lett. 15(13), 746–748 (1990). [CrossRef]
18. J. Sroga, E. W. Eloranta, S. T. Shipley, et al., “High spectral resolution lidar to measure optical scattering properties of atmospheric aerosols. 2: Calibration and data analysis,” Appl. Opt. 22(23), 3725–3732 (1983). [CrossRef]
19. R. Engelmann, T. Kanitz, H. Baars, et al., “The automated multiwavelength Raman polarization and water-vapor lidar PollyXT: the neXT generation,” Atmos. Meas. Tech. 9(4), 1767–1784 (2016). [CrossRef]
20. S. Mao, A. Wang, Y. Yi, et al., “Polarization Raman lidar for atmospheric correction during remote sensing satellite calibration: instrument and test measurements,” Opt. Express 30(7), 11986–12007 (2022). [CrossRef]
21. M. Haarig, R. Engelmann, A. Ansmann, et al., “1064 nm rotational Raman lidar for particle extinction and lidar-ratio profiling: cirrus case study,” Atmos. Meas. Tech. 9(9), 4269–4278 (2016). [CrossRef]
22. M. Haarig, A. Ansmann, D. Althausen, et al., “Triple-wavelength depolarization-ratio profiling of Saharan dust over Barbados during SALTRACE in 2013 and 2014,” Atmos. Chem. Phys. 17(17), 10767–10794 (2017). [CrossRef]
23. J. P. Donnelly, E. K. Duerr, K. A. McIntosh, et al., “Design considerations for 1.06-(mu)m InGaAsP-InP Geiger-Mode Avalanche Photodiodes IEEE Journal of Quantum Electronics: A Publication of the IEEE Quantum Electronics and Applications Society,” IEEE J. Quantum Electron. 42(8), 797–809 (2006). [CrossRef]
24. J. A. Zenteno-Hernández, A. Comerón, A. Rodríguez-Gómez, et al., “A comparative analysis of aerosol optical coefficients and their associated errors retrieved from pure-rotational and vibro-rotational raman lidar signals,” Sensors 21(4), 1277 (2021). [CrossRef]
25. P. Ortiz-Amezcua, A. E. Bedoya-Velásquez, J. A. Benavent-Oltra, et al., “Implementation of UV rotational Raman channel to improve aerosol retrievals from multiwavelength lidar,” Opt. Express 28(6), 8156–8168 (2020). [CrossRef]
26. I. Veselovskii, D. N. Whiteman, M. Korenskiy, et al., “Use of rotational Raman measurements in multiwavelength aerosol lidar for evaluation of particle backscattering and extinction,” Atmos. Meas. Tech. 8(10), 6759–6795 (2015). [CrossRef]
27. F. Liu and F. Yi, “Lidar-measured atmospheric N2 vibrational-rotational Raman spectra and consequent temperature retrieval,” Opt. Express 22(23), 27833–27844 (2014). [CrossRef]
28. C. Weitkamp, Lidar: range-resolved optical remote sensing of the atmosphere, Springer Science & Business Vol. 102 (2006).
29. X. Jiang, M. A. Itzler, R. Ben-Michael, et al., “InGaAsP-InP avalanche photodiodes for single photon detection,” IEEE J. Sel. Top. Quantum Electron. 13(4), 895–905 (2007). [CrossRef]
30. N. Bertone, R. Biasi, and B. Dion, “Overview of photon counting detectors based on CMOS processed single photon avalanche diodes (spad), ingaas apds, and novel hybrid (tube+ apd) detectors,” Semiconductor Photodetectors II 5726, 153–163 (2005). [CrossRef]
31. M. Radlach, A. Behrendt, and V. Wulfmeyer, “Scanning rotational Raman lidar at 355 nm for the measurement of tropospheric temperature fields,” Atmos. Chem. Phys. 8(2), 159–169 (2008). [CrossRef]
32. C.-Y. She, “Spectral structure of laser light scattering revisited: bandwidths of nonresonant scattering lidars,” Appl. Opt. 40(27), 4875–4884 (2001). [CrossRef]
33. V. Freudenthaler, M. Esselborn, M. Wiegner, et al., “Depolarization ratio profiling at several wavelengths in pure Saharan dust during SAMUM 2006,” Tellus Ser. B-Chem. Phys. Meteorol. 61(1), 165–179 (2022). [CrossRef]
34. National Oceanic and Atmospheric Administration, “US standard atmosphere,” Atmosphere, U. S. (1976).
35. L. Wang, M. B. Mačak, S. Stanič, et al., “Investigation of aerosol types and vertical distributions using polarization Raman lidar over Vipava Valley,” Remote Sens. 14(14), 3482 (2022). [CrossRef]
36. H. Baars, T. Kanitz, R. Engelmann, et al., “An overview of the first decade of PollyNET: an emerging network of automated Raman-polarization lidars for continuous aerosol profiling,” Atmos. Chem. Phys. 16(8), 5111–5137 (2016). [CrossRef]
37. W. Gong, F. Mao, and S. Song, “Signal simplification and cloud detection with an improved Douglas-Peucker algorithm for single-channel lidar,” Meteorol. Atmos. Phys. 113(1-2), 89–97 (2011). [CrossRef]
38. W. N. Chen, C. W. Chiang, and J. B. Nee, “Lidar ratio and depolarization ratio for cirrus clouds,” Appl. Opt. 41(30), 6470–6476 (2002). [CrossRef]
39. J. Zhuang and F. Yi, “Nabro aerosol evolution observed jointly by lidars at a mid-latitude site and CALIPSO,” Atoms. Environ. 140, 106–116 (2016). [CrossRef]
