China Aerosol Raman Lidar Network (CARLNET)—Part I: Water Vapor Raman Channel Calibration and Quality Control

Abstract

Water vapor is an active trace component in the troposphere and has a significant impact on meteorology and the atmospheric environment. In order to meet demands for high-precision water vapor and aerosol observations for numerical weather prediction (NWP), the China Meteorological Administration (CMA) deployed 49 Raman aerosol lidar systems and established the first Raman–Mie scattering lidar network in China (CARLNET) for routine measurements. In this paper, we focus on the water vapor measurement capabilities of the CARLNET. The uncertainty of the water vapor Raman channel calibration coefficient (Cw) is determined using an error propagation formula. The theoretical relationship between the uncertainty of the calibration coefficient and the water vapor mixing ratio (WVMR) is constructed based on least squares fitting. Based on the distribution of lidar in regions with different humidity conditions, the method of real-time calibration and quality control based on radiosonde data is established for the first time. Based on the uncertainty requirements of the World Meteorological Organization for water vapor in data assimilation, the calibration and quality control thresholds of the WVMR in regions with different humidity conditions are determined by fitting real-time lidar and radiosonde data. Lastly, based on the radiosonde results, the calibration algorithm established in this study is used to calibrate CARLNET data from October to December 2023. Compared with traditional calibration results, the results show that the stability and detection accuracy of the CARLNET significantly improved after calibration in regions with different humidity conditions. The deviation of the Cw decreased from 12.84~18.47% to 5.41~11.54%. The inversion error of the WVMR compared to radiosonde decreased from 1.05~0.46 g/kg to 0.82~0.34 g/kg. The reliability of the improved calibration algorithm and the CARLNET’s performance have been verified, enabling them to provide high-precision water vapor products for NWP.

1. Introduction

Atmospheric water vapor is an important parameter for describing the physical state of the atmosphere. It exhibits significant temporal and spatial variation, making it one of the most important greenhouse gases [1]. The World Meteorological Organization (WMO) has published relevant requirements for the accuracy of water vapor detection in numerical forecasting. Therefore, obtaining continuous, high-precision, and high-temporal-resolution humidity profiles is of great significance for meteorology and environmental research [2]. At present, the main methods employed for monitoring atmospheric water vapor include GPS remote sensing monitoring, meteorological satellite monitoring, balloon-borne, and lidar monitoring. However, GPSs possess poor spatial resolution, and satellites are more expensive than ground-based lidar systems, making them unsuitable for large-scale launches for high-precision detection of water vapor [3,4]. In comparison, as an active remote sensing tool, water vapor Raman lidar possesses high vertical resolution and temporal resolution. Ground-based lidar is more suitable for continuous long-term observation of the same research area [5,6,7,8]. In addition, it can achieve unmanned continuous observation throughout the day. Raman lidar systems are suitable for use in ground-based and airborne environments [6,9]. Raman lidar systems have been widely used in environmental research over the past few decades and excellent observational results have been achieved. In 1969, Melfi et al., from NASA first studied the feasibility of using Raman scattering lidar to detect atmospheric water vapor [10]. This type of lidar enables the detection of the vibration Raman scattering echo signals of nitrogen and water vapor. Using this type of lidar, the water vapor profile at a height of about 2 km was successfully obtained [11]. In 2010, Whiteman et al. first used airborne Raman scattering lidar to simultaneously measure atmospheric water vapor and aerosols. They detected parameters such as the water vapor mixing ratio (WVMR) and aerosol optical parameters below 8 km [12]. Compared to efforts abroad, research on atmospheric water vapor detection in China started relatively late but developed rapidly [13,14,15,16,17,18,19,20]. Research involving Raman scattering lidar in China was first introduced by researchers at the Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences [21]. In 2015, Wang Yufeng and colleagues from Xi’an University of Technology designed a Raman lidar system for the simultaneous detection of temperature, humidity, and aerosols [22].

Effective quality control of Raman lidar results is crucial for improving the accuracy of the WVMR. Newsom et al., incorporated both the photomultiplier analog output voltage and photon counts into a single signal with improved dynamic range [23]. The effect of modifications to an algorithm was evaluated by comparing profiles of the water vapor mixing ratio from the lidar with radiosonde measurements over a six-month period. The modifications that were implemented resulted in a reduction in the mean bias in the daytime water vapor mixing ratio from a 3% dry bias to well within 1%. Turner et al. applied dead time and background subtraction, overlap correction, and aerosol transmittance correction to improve the data quality of Raman lidar systems [24]. They calibrated the Raman lidar system’s water vapor mixing ratio by adjusting the calibration value to achieve agreement in total precipitable water vapor with that retrieved from a dual-channel microwave radiometer. Sakai analyzed the influencing factors that cause changes in calibration coefficients and performed quality control of their Raman lidar observation results [25], which improved the accuracy of the WVMR. In terms of numerical forecasting and data assimilation, Raman lidar observation data have a significant and positive impact on numerical weather forecasting. The initial conditions of the water vapor field can be improved, thereby improving rainfall forecasting [26,27,28,29,30,31]. At present, observation of and research on water vapor primarily focus on using a single ground Raman lidar system. However, a single-station lidar system cannot monitor the transport of water vapor over a large scale. Therefore, its contribution to regional weather monitoring and numerical forecasting is limited.

From 2021 to the present, the China Meteorological Administration has built 49 ground-based Raman aerosol lidar systems. For the first time in China, a Raman–Mie scattering lidar observation network (CARLNET) for the three-dimensional monitoring of water vapor and aerosol has been established. This study represents an early attempt in the field of data quality control and calibration for the CARLNET. A method for the real-time calibration and quality control of the WVMR based on radiosonde observation for lidar network measurement has been established for the first time. In addition, real-time calibration of the observational data of CARLNET from October to December 2023 has been completed.

2. Raman–Mie Scattering Lidar Observation Network

2.1. CARLNET Distribution

From 2021 to 2022, the China Meteorological Administration began to implement services for meteorological monitoring, early warning, and remedial measures, primarily considering the demand for disaster weather monitoring in the central and western regions. There are 49 ground-based Raman aerosol lidar systems being deployed at present, collocated with high-altitude profiling stations. The lidar system can operate continuously for several months without human intervention, measuring the vertical distribution of the WVMR in the lower troposphere of different climate regions and surface types. The average distance between each lidar system is 600 km, with the distance in the southwest region reaching 350 km. The construction of this station network has improved the country’s catastrophic weather vertical monitoring capabilities and gradually alleviated the need for the assimilation of high-altitude observation data in numerical forecasting models. The distribution map of the 49 Raman aerosol lidar systems is shown in Figure 1.

In order to evaluate the operational performance of the CARLNET in regions with different humidity conditions, in this study, we used the average annual precipitation data from 2014 to 2023 to partition the dry and wet regions of China. The regions were ultimately divided into humid areas (annual precipitation amount ≥ 800 mm), semi-humid areas (600 mm ≤ annual precipitation amount < 800 mm), semi-arid areas (400 mm ≤ annual precipitation amount < 600 mm), and arid areas (annual precipitation amount < 400 mm). The results of dry and humid zoning in the different Chinese regions are shown in Figure 2, and the distribution of the 49 Raman lidar systems in the regions with different humidity conditions is shown in Table 1.

2.2. Raman–Mie Scattering Lidar

The structure of the Raman aerosol lidar system is shown in Figure 3. The system’s emission laser wavelengths are 355 nm, 532 nm, and 1064 nm. The lidar system consists of eight receiving channels, including a 355 nm channel, a 532 nm channel, and a 1064 nm channel for measuring aerosol and cloud optical properties. In addition, the 355 nm and 532 nm channels possess vertical polarization channels, which can obtain polarization information to classify aerosols and clouds. The 607 nm, 407 nm, and 387 nm Raman channels are used to obtain Raman scattering signals of water vapor and nitrogen in the atmosphere. The lidar system has a compact structure and is easy to deploy. It can perform functions such as remote operation, automatic data transmission, and one-click start. The main parameters of the aerosol Raman lidar system are shown in Table 2. The 49 sites use the same type of lidar system. Zhang et al. used aerosol Raman lidar data to study the relationship between atmospheric haze and water vapor content. The results showed that atmospheric water vapor content was mainly concentrated below 5 km, accounting for 64% to 99% of the total water vapor below 10 km. Furthermore, water vapor content in air pollution exhibits distinct stratification characteristics with altitude, especially within the height range of 1–3 km, where significant water vapor variation layers exist, showing spatial consistency with inversion layers [32,33].

3. Methodology

3.1. Water Vapor Raman Channel Calibration

The laser emitted by lidar is attenuated and scattered by molecules (e.g., N₂ and H₂O) and aerosol particles in the atmosphere. The final signal returned to the telescope includes elastic scattering and inelastic scattering. The echo signal received by the inelastic scattering channel can be expressed using the following lidar equations:

(1)$P_{{\lambda }_N}\left(z\right)=K_{{\lambda }_N}{\displaystyle \frac{O_N\left(z\right)}{z^2}}{N}_N(z){\sigma }_{{\lambda }_N}(\pi )\mathrm{e}\mathrm{x}\mathrm{p}\left\{-{\int }_0^z\left[{\alpha }_{{\lambda }_0}^{aer}\left(r\right)+{\alpha }_{{\lambda }_0}^{mol}\left(r\right)+{\alpha }_{{\lambda }_N}^{aer}\left(r\right)+{\alpha }_{{\lambda }_N}^{mol}\left(r\right)\right]\mathrm{d}r\right\}$

(2)$P_{{\lambda }_H}\left(z\right)=K_{{\lambda }_H}{\displaystyle \frac{O_H\left(z\right)}{z^2}}{N}_H\left(z\right){\sigma }_{{\lambda }_H}\left(\pi \right)\mathrm{e}\mathrm{x}\mathrm{p}\left\{-{\int }_0^z\left[{\alpha }_{{\lambda }_0}^{aer}\left(r\right)+{\alpha }_{{\lambda }_0}^{mol}\left(r\right)+{\alpha }_{{\lambda }_H}^{aer}\left(r\right)+{\alpha }_{{\lambda }_H}^{mol}\left(r\right)\right]\mathrm{d}r\right\}$

where ${\lambda }_N$ is the nitrogen Raman wavelength, which is 386 nm and 607 nm for this device. ${\lambda }_H$ is the water vapor Raman wavelength, which is 407 nm for this device. $P_{{\lambda }_x}\left(z\right)$ is the energy of the atmospheric backscattered echo signal at height z. $K_{{\lambda }_x}$ is the system constant of the respective channel, which is related to the emitted laser energy, the optical efficiency of the system, the receiving range of the telescope, the spectral efficiency of the photomultiplier tube, etc. $O_x\left(z\right)$ is the overlap factor. ${\alpha }_{{\lambda }_x}^{aer}$ and ${\alpha }_{{\lambda }_x}^{mol}$ are the aerosol and molecular extinction coefficients, respectively. $N_N\left(z\right)$ and $N_H\left(z\right)$ are the molecular number density profiles of the nitrogen and water vapor, respectively. ${\sigma }_{{\lambda }_x}\left(\pi \right)$ is the Raman backward-scattering cross-section at the corresponding wavelength.

The WVMR is defined as the ratio of the mass of water vapor, $m_H$, to the mass of dry air, $m_{dry}$, in the same volume, in units of kg/kg. Based on the definition of the WVMR, there is also a mass equal to the product of the amount of matter $N/N_A$ ($N_A$ is Avogadro’s constant (mol-1)) and the molar mass $M$. Therefore, the WVMR can be expressed as follows:

(3)$w(z)={\displaystyle \frac{N_H\left(z\right)M_H}{N_{dry}\left(z\right)M_{dry}}}$

where $N_H\left(z\right)$ and $N_{dry}\left(z\right)$ are the water vapor and dry air molecule numbers. $M_H\left(z\right)$ and $M_{dry}\left(z\right)$ are the molar mass of water vapor and dry air (g/mol). As the content of nitrogen in the atmosphere is relatively stable, $N_{dry}\left(z\right)=N_N\left(z\right)/0.78$. Therefore, the WVMR can be calculated by using the nitrogen and water vapor Raman echo equation. The WVMR can be determined according to the ratio of Equations (1) and (2) as follows:

(4)$w(z)=C_w{∆}_q^w(z_0,z){\displaystyle \frac{P_{{\lambda }_H}\left(z\right)}{P_{{\lambda }_N}\left(z\right)}}$

where ${∆}_q^w(z_0,z)$ is the ratio of the transmittance of two wavelengths (also known as the transmittance correction function) and $C_w$ is the calibration coefficient.

The water vapor Raman channel can be calibrated by comparing the WVMR obtained from other devices, such as radiosondes. Radiosonde does not directly provide WVMR parameters; however, they can be calculated based on its temperature, pressure, and relative humidity data. To ensure high accuracy, it is necessary to calibrate the lidar system under stable atmospheric conditions. After meeting the atmospheric conditions, an appropriate altitude range must be selected to calibrate the WVMR measured by the lidar system and determine the calibration coefficient. The calculation method for the calibration coefficient $C_w$ is as follows:

(5)$C_W={\displaystyle \frac{{\sum }_{n\left(z\right)=n\left(z_1\right)}^{n\left(z_2\right)}\frac{w_{RS}\left(z\right)}{\left[\raisebox{1ex}{$P_H\left(z\right)$}\!\left/ \!\raisebox{-1ex}{$P_N\left(z\right)$}\right.\right] · T(z_0,z)}}{\left[n\right(z_2)-n(z_1)+1]}}$

where $w_{RS}$ is the WVMR obtained by the radiosonde. $n$ is the number of points corresponding to the calibration range. $P_H\left(z\right)$ and $P_N\left(z\right)$ are the returned optical signals for the water vapor and nitrogen channels. $T(z_0,z)$ is the atmospheric transmittance correction term from the height of $z_0$ to z.

3.2. Error Analysis

Defining the lidar WVMR before calibration as $w_{lidar}$, the uncertainty of the WVMR at height $z_i$ can be expressed as follows:

(6)${\delta }_{w_{lidar}}(z_i)=\sqrt{w_{lidar}\left(z_i\right)}\sqrt{{\displaystyle \frac{1}{{SNR}_{407}^2}}+{\displaystyle \frac{1}{{SNR}_{386}^2}}}$

where ${SNR}_{407}$ and ${SNR}_{386}$ denote the signal-to-noise ratio of the 407 nm channel and 386 nm channel, respectively. $w_{lidar}$ and $w_{RS}$ satisfy the relationship of the following equation:

(7)$w_{lidar}\left(z_i\right)=C_w^{\prime }{ · w}_{RS}\left(z_i\right)$

The fitted slope $C_w^{\prime }$ and the corresponding uncertainty ${\delta }_{C_w^{\prime }}$ can be derived through least squares fitting and satisfy the following relationship [34]:

(8)$C_w^{\prime }={\displaystyle \frac{S·S_{xy}-S_x·S_y}{S·S_{xx}-S_x^2}}$

(9)${\delta }_{C_w^{\prime }}={\displaystyle \frac{S}{S·S_{xx}-S_x^2}}$

Among them,

(10)$$\begin{gathered} S={\sum }_{i=1}^N{\displaystyle \frac{1}{{{\delta }_{w_{lidar}}\left(z_i\right)}^2}}, S_x={\sum }_{i=1}^N{\displaystyle \frac{w_{RS}\left(z_i\right)}{{{\delta }_{w_{lidar}}\left(z_i\right)}^2}}, S_y={\sum }_{i=1}^N{\displaystyle \frac{w_{lidar}\left(z_i\right)}{{{\delta }_{w_{lidar}}\left(z_i\right)}^2}} \\ {S}_{xx}=\sum _{i=1}^N{\displaystyle \frac{{w_{RS}\left(z_i\right)}^2}{{{\delta }_{w_{lidar}}\left(z_i\right)}^2}}, S_{xy}=\sum _{i=1}^N{\displaystyle \frac{w_{lidar}\left(z_i\right)w_{RS}\left(z_i\right)}{{{\delta }_{w_{lidar}}\left(z_i\right)}^2}} \end{gathered}$$

In practical operations, the calibration coefficient $C_w$ is the reciprocal of the fitted slope $C_w^{\prime }$, i.e.,

(11)$C_w={C_w^{\prime }}^{-1}$

Thus, the calibration coefficient uncertainty is

(12)$\delta C_w=\sqrt{{\displaystyle \frac{{\delta }_{C_w^{\prime }}}{{C_w^{\prime }}^2}}}=\sqrt{{\displaystyle \frac{S(S{S}_{xx}-{S_x}^2)}{{(S{S}_{xy}-S_x{S}_y)}^2}}}\propto {\displaystyle \frac{1}{\sqrt{\overline{w_{RS}}}}}$

In this study, we constructed a nonlinear function between the uncertainty of the calibration coefficients and the water vapor mixing ratio in regions with different humidity conditions. We determined the calibration threshold for the WVMR in different dry and wet regions.

3.3. Quality Control Strategy

During the operation of the CARLNET, the water vapor Raman channel calibration coefficient may drift due to various factors, such as filter aging, change in the detector gain ratio, etc., hence the necessity of real-time calibration. However, due to matters relating to weather, personnel, and other factors, the drift of the constant may not be detected in a timely manner. Therefore, an automatic calibration method based on daily radiosonde data at 08:00 and 20:00 (China Standard Time, CST) was developed. In the first stage, the uncalibrated WVMR is derived. Based on the observation results of the radiosonde, the WVMR calibration coefficients are obtained through least squares fitting. There are many factors that can cause changes in calibration coefficients, such as the aging of Raman lidar hardware, the horizontal drift of sounding balloons, cloud interference, and insufficient signal-to-noise ratios of lidar signals. Therefore, in order to improve the accuracy of water vapor mixing ratio observation, data quality control is essential. The quality control flow used in this study is shown in Figure 4, with the key processes marked in red.

The entire quality control process can be divided into traditional and advanced calibration processes.

3.3.1. Traditional Calibration Process

In this process, the raw echoes of Raman aerosol lidar are processed while controlling the SNR quality and the WVMR result $w_L$ is inverted. The WVMR $w_r$ based on the radiosonde is linearly fitted with the lidar inversion result $w_L$. The calibration coefficient $C_w$ from the traditional method is obtained. By using the error propagation formula, the uncertainty of the calibration coefficient ${∆C}_W$ is determined.

3.3.2. Advanced Calibration Process

In order to improve the data quality and observation accuracy of the CARLNET, quality control of the raw data is required. Before calibrating the Raman aerosol lidar measurements, it is necessary to determine the minimum WVMR calibration thresholds for different dry and wet regions. In this module, the average WVMR of the radiosonde in the altitude range of 0.5 km to 2 km is calculated. Thereafter, the average WVMR with the uncertainty of the calibration coefficient can be derived through nonlinear fitting. The nonlinear equations were constructed for the calibration coefficient uncertainty and WVMR in regions with different humidity conditions. The minimum calibrated WVMR threshold for regions with different humidity conditions was calculated. The threshold method was used to control the quality of the Raman aerosol lidar inversion results, with an advanced calibration coefficient $C_w^{\prime }$ ultimately obtained. As shown in Figure 5, the calibration coefficient uncertainty of Raman aerosol lidar sites in regions with different humidity conditions is nonlinearly fitted to the mean value of the radiosonde WVMR. Thereafter, the minimum WVMR thresholds for different wet and dry regions are determined.

From Equation (12), it can be seen that there is a nonlinear relationship between the calibration coefficient uncertainty and the WVMR. The $y=a{x}^{-0.5}$ theoretical model is used to fit the measured lidar and radiosonde data to construct the functional relationship between the uncertainty of calibration coefficients and the WVMR in different humid regions. From the fitting results, it can be seen that the calibration coefficient uncertainty is larger in the humid region due to its susceptibility to precipitation, clouds, and other weather conditions, and the value of the fitting coefficient is larger than that of the other three regions. The calibration threshold of the WVMR for regions with different humidity conditions was determined according to the accuracy requirements published by the World Meteorological Organization (WMO) for water vapor during numerical assimilation. The minimum WVMR threshold for the arid region is 0.91 g/kg, the minimum WVMR threshold for the semi-arid region is 1.50 g/kg, the minimum WVMR threshold for the semi-humid region is 1.13 g/kg, and the minimum WVMR threshold for the humid region is 4.00 g/kg. The fitting coefficients of the stations in the different humidity zones were summed, and the results are shown in Figure 6.

As can be seen from the results displayed in Figure 6, the arid region is characterized by low and stable water vapor content due to low precipitation throughout the year. The Raman aerosol scattering lidar systems in this region work stably and the uncertainty of the calibration coefficients varies to a lesser extent, which correspondingly leads to a smaller fitting coefficient between the stations in this region, with a mean value of 4.86 ± 1.02. On the contrary, the water vapor content in the humid region is large and unstable. Humid regions often have low clouds, which leads to a large jitter in the uncertainty of the calibration coefficients, with a larger value for the fitting coefficients and the standard deviation of the region. The fitting coefficients of the semi-arid and semi-humid regions are close to one another, with values of 6.23 ± 1.15 and 6.02 ± 1.77, respectively.

4. Results and Discussion

4.1. Traditional Calibration Results

Due to the influence of radiosonde horizontal drift and weather-related factors such as clouds and rainfall, Raman aerosol lidar systems generate significant observational errors during long-term observation. Using the error propagation formula, the uncertainty of calibration coefficients for each Raman aerosol lidar station can be calculated. The WVMR inverted by the Raman aerosol lidar system was compared with the radiosonde measurement results under different calibration coefficient uncertainties, as shown in Figure 7.

When the uncertainty of the calibration coefficient is 15%, there is no correlation between the WVMR of the Raman aerosol lidar inversions and soundings. The average difference between the two is 1 g/kg. When the uncertainty of the calibration coefficient is 10%, the correlation coefficient between the WVMR of the Raman aerosol lidar inversions and the results of the soundings is 0.75, and the average difference between the two is 0.99 g/kg. When the uncertainty of the calibration coefficient is 5%, the correlation coefficient between the WVMR inverted by the Raman aerosol lidar system and the measurement results of the radiosonde is 0.92, and the average difference between the two is 0.68 g/kg. When the uncertainty of the calibration coefficient is 2%, the correlation coefficient between the WVMR inverted by the Raman aerosol lidar system and the measurement results of the radiosonde is 0.97, and the average difference between the two is 0.41 g/kg. It can be seen that when the uncertainty of the calibration coefficient is greater than 5%, there are significant differences in the calibration results of traditional calibration methods, which cannot meet the water vapor accuracy requirements for data prediction. The above constraints thus emphasize the urgent need for a high-precision data calibration algorithm.

4.2. Comparison and Verification

4.2.1. Calibration Coefficient

Sites that had already been built and that are geographically located in the middle of each region were selected as examples to compare the calibration coefficients before and after the improvement of the calibration methods. The Huhehaote site was selected for the arid region, the Yanan site was selected for the semi-arid region, the Ganzi site was selected for the semi-humid region, and the Quzhou site was selected for the humid region. The traditional calibration method was improved based on the results of the threshold method outlined in Section 3.3.2. When the water vapor content in the atmosphere is less than the WVMR threshold and the uncertainty of the calibration coefficient is greater than 5%, the data set is excluded. The results of the calibration coefficients before and after the improvement of the calibration method for stations in different wet and dry areas are shown in Figure 8, Figure 9, Figure 10 and Figure 11.

From the comparison results in Figure 8, Figure 9, Figure 10 and Figure 11, it can be seen that the stability of the calibration coefficients of the Raman aerosol lidar system was significantly improved by the improvement of the calibration method. The relative deviation of the calibration coefficients of the HOHHOT station in the arid region decreased from 17.79% to 9.56%, and the stability of the calibration coefficients improved by 8.23%. The relative deviation of the calibration coefficient of the YAN AN station in the semi-arid region decreased from 17.56% to 8.4%, and the stability of the calibration coefficient increased by 9.16%. The relative deviation of the calibration coefficient of the GANZE station in the semi-humid region decreased from 18.47% to 11.54%, and the stability of the calibration coefficient increased by 6.93%. The relative deviation of the calibration coefficient of the QU XIAN station in the humid region decreased from 12.81% to 5.41%, and the stability of the calibration coefficient increased by 7.4%. It is worth noting that the stability of the calibration coefficients may be related to factors such as the atmospheric aerosol content in addition to the lidar system.

4.2.2. WVMR Results

In order to verify the accuracy of Raman aerosol lidar inversion results, it is necessary to verify the measurements with the aid of other experimental instruments. The radiosonde is a direct measurement method, which is considered to be the closest measurement to the actual condition of the atmosphere and possesses high reliability. The WVMR inversion results of all lidar sites in different wet and dry regions shown in Table 1 for the three months from October to December 2023 are correlated with the observations from the radiosonde at the same site. The water vapor mixing ratio observation results within the altitude range of 500 m to 2000 m were used, and the comparison results are shown in Figure 12, Figure 13, Figure 14 and Figure 15.

From the correlation between the analyzed WVMR results of the two devices, it can be seen that the observation performance of the CARLNET in different dry and wet regions significantly improved after the improvement of the traditional calibration method. The correlation and the average effective detection distance in different dry and wet regions vary, which is due to the difference in water vapor content and the influence of precipitation and other weather-related factors. The detection performance of the CARLNET before and after improvement of the calibration method for different wet and dry regions is shown in Figure 16.

5. Conclusions

In this study, the CMA deployed 49 Raman aerosol lidar systems and established the first Raman–Mie scattering lidar network (CARLNET) in China to vertically monitor large-scale water vapor transport. In order to improve the data quality of the CARLNET, traditional calibration algorithms were improved and a high-precision calibration algorithm for the CARLNET was established. According to the distribution of annual precipitation, the CARLNET is divided into four different humidity zones, namely an arid zone, a semi-arid zone, a semi-humid zone, and a humid zone. Based on the observation results of the WVMR using radiosonde, the calibration coefficient uncertainty and WVMR nonlinear function were constructed for different dry and wet zones. The minimum WVMR calibration threshold for regions with different humidity conditions was determined, with thresholds of 0.91 g/kg for arid regions, 1.50 g/kg for semi-arid regions, 1.13 g/kg for semi-humid regions, and 4.00 g/kg for humid regions. Lastly, based on the observational radiosonde results, the calibration algorithm established in this study was used to perform real-time calibration of the CARLNET and compared with traditional calibration results. The results show that the improved calibration method demonstrates excellent consistency between the WVMR inverted by Raman aerosol lidar in regions with different humidity conditions and the radiosonde observation results. The correlation coefficient for arid regions is 0.89, for semi-arid regions, it is 0.93, for semi-humid regions, it is 0.91, and for humid regions, it is 0.93. In addition, compared with traditional calibration results, the calibration algorithm established in this study significantly improves the stability and detection accuracy of the CARLNET following calibration. The calibration coefficient deviation in arid areas decreased from 17.79% to 9.56%, the observation error of the WVMR decreased from 0.46 g/kg to 0.34 g/kg, and the average detection distance increased from 2.63 km to 2.74 km. The calibration coefficient deviation in semi-arid regions decreased from 17.56% to 8.4%, the observation error of the WVMR decreased from 0.72 g/kg to 0.47 g/kg, and the average detection distance increased from 2.35 km to 3.13 km. The calibration coefficient deviation in semi-humid areas decreased from 18.47% to 11.54%, the observation error of the WVMR decreased from 0.93 g/kg to 0.73 g/kg, and the average detection distance increased from 2.48 km to 3.0 km. The calibration coefficient deviation in humid areas decreased from 12.81% to 5.41%, the observation error of the WVMR decreased from 1.05 g/kg to 0.82 g/kg, and the average detection distance increased from 2.13 km to 3.01 km.

The results of this study validate the reliability of the improved calibration algorithm and the observational performance of the CARLNET, providing high-precision scientific data for the application of water vapor and aerosol data models. In the future, a second article will introduce the quality assurance system of the CARLNET, and a third article will focus on the operation of the CARLNET and evaluate and verify the quality of aerosol observation data.

Author Contributions

Conceptualization, N.S., Q.W., Z.B. and X.W.; data curation, Y.D.; formal analysis, Q.W.; funding acquisition, Z.B.; investigation, Q.W.; methodology, Z.Y.; project administration, Y.C.; supervision, X.W.; writing—review and editing, Q.W. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.

Acknowledgments

The authors acknowledge Wang Xuan for providing writing assistance, Li Zhigang for providing language assistance, and Mo Zusi and Dong Pingyi for proofreading the article.

Conflicts of Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to have influenced the work reported in this paper.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. Distribution of the network of 49 Raman aerosol lidar stations. The green stations represent those completed in 2021 and the red stations represent those completed in 2022. The color of the base map represents the height of the terrain.

Figure 2. Wet and dry zones in China.

Figure 3. Structure of the Raman aerosol lidar system.

Figure 4. WVMR calibration and validation route.

Figure 5. Raman aerosol lidar site calibration coefficient uncertainty and WVMR correlation analysis, where (a) shows the arid region, (b) shows the semi−arid region, (c) shows the semi−humid region, and (d) shows the humid region. The black dashed line in the figure represents the constructed nonlinear theoretical function. The pink dashed line represents the threshold line for the WVMR and calibration coefficient uncertainty. The pink numbers represent the calibration coefficient uncertainty and WVMR threshold.

Figure 6. Results of fitting coefficients for regions with different humidity conditions.

Figure 7. Comparison of WVMR measurements from the lidar system with radiosonde at different calibration coefficient uncertainties using the conventional calibration method, where figure (a) shows the results at 15% calibration coefficient uncertainty, figure (b) shows the results at 10% calibration coefficient uncertainty, figure (c) shows the results at 5% calibration coefficient uncertainty, and figure (d) shows the results at 2% calibration coefficient uncertainty.

Figure 8. Calibration coefficients of Huhehaote station in the arid region before and after the improvement of the calibration method. The red color represents the calibration coefficients before the improvement of the calibration method, and the green color represents the calibration coefficients after the improvement of the calibration method.

Figure 9. Results of calibration coefficients before and after the improvement of the calibration method for Yanan station in the semi-arid region. The red color represents the calibration coefficients before the improvement of the calibration method, and the green color represents the calibration coefficients after the improvement of the calibration method.

Figure 10. Results of calibration coefficients before and after the improvement of the calibration method at Ganzi station in the semi-humid region. The red color represents the calibration coefficient before the improvement of the calibration method, and the green color represents the calibration coefficient after the improvement of the calibration method.

Figure 11. Results of calibration coefficients before and after the improvement of the calibration method for Quzhou station in the humid region. The red color represents the calibration coefficient before the improvement of the calibration method, and the green color represents the calibration coefficient after the improvement of the calibration method.

Figure 12. Verification results of the lidar WVMR before and after improvement in the arid region. Figure (a) shows the validation results before calibration method improvement and figure (b) shows the validation results after improvement.

Figure 13. Verification results of the lidar WVMR before and after improvement in the semi-arid region. Figure (a) shows the validation results before the calibration method is improved and figure (b) shows the validation results after improvement.

Figure 14. Verification results of the lidar WVMR before and after improvement in the semi-humid region. Figure (a) shows the validation results before calibration method improvement and figure (b) shows the validation results after improvement.

Figure 15. Verification results of the lidar WVMR before and after improvement in the humid region. Figure (a) shows the validation results before calibration method improvement and figure (b) shows the validation results after improvement.

Figure 16. Detection performance of the Raman aerosol lidar network before and after improvement of different wet and dry region calibration methods. Figure (a) shows the correlation coefficient result, figure (b) shows the absolute deviation, figure (c) shows the relative deviation, and figure (d) shows the average effective detection distance.

Table 1

Distribution of Raman aerosol lidar sites in different wet and dry regions.

Dry and Wet Partition	Station	Longitude E/°	Latitude N/°
Arid	Jiuquan	100	40
	Mangnai	91	38
	Doulan	98	36
	Karamay	85	45
	Minfeng	82	37
	Wen-Su	80	41
	Hami	94	43
	Ejin Qi	101	42
	Hohhot	111	41
Semi-Arid	Anqin	112	37
	Tingri	87	28
	Yan An	105	33
	Yushu	109	36
	Tianjin	97	33
	Zhengzhou	117	39
	Changchun	114	35
	Harbin	125	49
	Taiyuan	127	46
Semi-Humid	Yichang	111	30
	Garze	117	31
	Darlag	100	31
	Hongyuan	100	34
	Nyingchi	102	32
	Qamdo	94	29
	Sheyang	97	31
	Wudu	120	34
	Hezuo	103	35
	Shenyang	124	42
	Batang	99	30
	Yibin	105	29
	Wenjiang	104	31
Humid	Guiyang	106	26
	Lijing	100	27
	Simao	101	22
	Weining	104	27
	Puan	105	26
	Qu Xian	119	29
	Chongqing	106	30
	Qingyuan	113	24
	Heyuan	115	24
	Shantou	117	23
	Shaowu	117	27
	Fuzhou	119	26
	Ankang	109	33
	Kunming	103	25
	Mengzi	103	23
	Xichang	102	28
	Da Xian	108	31
	Baise	107	24

Table 2

Main parameters of the Raman aerosol lidar system.

Transmitter
Laser	Nd: YAG
Wavelength (nm)	354.7 nm/532.1 nm/1064.1 nm
Pulse energy (mJ)	354.7: ≥0.6 mJ/532.1: ≥1.5 mJ/1064.1: ≥1 mJ
Repetition frequency (Hz)	1000
Beam divergence (mrad)	354.7: ≤0.3 mrad/532.1: ≤0.3 mrad/1064.1: ≤0.5 mrad
Receiver
Telescope type	Cassegrain system
Diameter of the primary mirror (m)	0.3
Field of view (mrad)	1
Detectors
PMT	Hamamatsu H7421-40
PMT	Hamamatsu H10721P-210
APD	Excelitas SPCM-AQRH-CD3773-FC
Data acquisition
ALA-LARA
Detection specifications	Raman water vapor	Raman nitrogen		Elastic
Center wavelength (nm)	407.6	386.7	607.6	354.7
Bandwidth (nm)	1	0.5	1	0.5
Peak transmission (%)	90%	80%	90%	70%
Rejection at 355 nm	OD6	OD7	OD7	OD7

References

1. Wang, Y.; Cao, X.; Zhang, J.; Tang, L.; Song, Y.; Di, H.; Hua, D. Detection and Analysis of All-Day Atmospheric Water Vapor Raman Lidar Based on Wavelet Denoising Algorithm. Acta Opt. Sin.; 2018; 38, 0201001. [DOI: https://dx.doi.org/10.3788/AOS201838.0201001]

2. Goldsmith, J.E.M.; Blair, F.H.; Bisson, S.E.; Turner, D.D. Turn-key Raman lidar for profiling atmospheric water vapor, clouds, and aerosols. Appl. Opt.; 1998; 37, pp. 4979-4990. [DOI: https://dx.doi.org/10.1364/AO.37.004979]

3. Shi, D.-C.; Hua, D.-X.; Ning, L.; Fei, G.; Li, W.; Qing, Y.; Yi, Z. Research of Solar-Blind Ultraviolet Raman Lidar for Water Vapor Measurement Technology. Spectrosc. Spectr. Anal.; 2018; 38, pp. 108-114.

4. Fei, G.; Ning, L.; Bo, H.; Zhu, Q.-S.; Shi, D.-C.; Li, W.; Yu-feng, W.; Qing, Y.; Liu, J.-J.; Hua, D.-X. Design of Polychromator and Technical Implementation of Daytime Raman Lidar for Atmospheric Water Vapor Measurement. Acta Photonica; 2018; 47, pp. 204-212.

5. Han, Y.; Wang, T.J.; Rao, R.Z.; Wang, Y.J. Progress in the study of physic-optics characteristics of atmospheric aerosols. Acta Phys. Sin.; 2008; 57, pp. 7396-7407.

6. Wang, Y.; Gao, F.; Zhu, C.; He, T.; Hua, D. Raman Lidar for atmospheric temperature, humidity and aerosols up to troposphere height. Acta Opt. Sin.; 2015; 35, 0328004. [DOI: https://dx.doi.org/10.3788/AOS201535.0328004]

7. Whiteman, D.N. Examination of the traditional Raman lidar technique. II. Evaluating the ratios for water vapor and aerosols. Appl. Opt.; 2003; 42, pp. 2593-2608. [DOI: https://dx.doi.org/10.1364/AO.42.002593]

8. Zhen, S.; Xie, C.; Zhong, Z.; Wang, B.; Wang, Z.; Zhao, M.; Tan, M.; Liu, D.; Wang, Y. Raman lidar for measurement of tropospheric water vapor. Infrared Laser Eng.; 2016; 45, 1211003. [DOI: https://dx.doi.org/10.3788/IRLA20164512.1211003]

9. XiaoMing, C. Detection and Analysis of Rrecipitable Water Vapor Based on All-Time Raman Lidar. Ph.D. Thesis; Xi’an University of Technology: Xi’an, China, 2017.

10. Melfi, S.H. Remote measurements of the atmosphere using Raman scattering. Appl. Opt.; 1972; 11, 1605. [DOI: https://dx.doi.org/10.1364/AO.11.001605]

11. Melfi, S.H.; Lawrence, J.D.; McCormick, M.P. Observation of Raman scattering by water vapor in the atmosphere. Appl. Phys. Lett.; 1969; 15, pp. 295-297. [DOI: https://dx.doi.org/10.1063/1.1653005]

12. Whiteman, D.N.; Rush, K.; Rabenhorst, S.; Welch, W.; Cadirola, M.; McIntire, G.; Russo, F.; Adam, M.; Venable, D.; Connell, R. et al. Airborne and Ground-Based Measurements Using a High-Performance Raman Lidar. J. Atmos. Ocean. Technol.; 2010; 27, pp. 1781-1801. [DOI: https://dx.doi.org/10.1175/2010JTECHA1391.1]

13. XiuXiang, W. Construction of a Differential Absorption Lidar System. Ph.D. Thesis; South China Normal University: Guangzhou, China, 2014.

14. Shao, J. Lidar Observation of Aerosols in Boundary Layer and Free Troposphere at Wuhan, China. Ph.D. Thesis; Wuhan University: Wuhan, China, 2020.

15. Liu, H. Retrieval of Ocean Optical Parameters Based on Lidar Return. Ph.D. Thesis; Zhejiang University: Hangzhou, China, 2020.

16. Zhang, Q. Technical Study of Boundary Layer Ozone Detection Lidar Applications. Ph.D. Thesis; University of Science and Technology of China: Hefei, China, 2020.

17. Yang, J. Profiling Atmospheric SO₂, NO₂ and CO₂ by Lidar. Ph.D. Thesis; University of Science and Technology of China: Hefei, China, 2019.

18. Zhao, Q. Atomspheric Turbulence Measurement with Scintillation Lidar. Ph.D. Thesis; University of Science and Technology of China: Hefei, China, 2018.

19. Liu, Y. Research on Laser Lidar Detection Technology of Boundary Layer Temperature and Humidity Profile. Ph.D. Thesis; University of Science and Technology of China: Hefei, China, 2018.

20. Liu, F.; Yi, F. Spectrally resolved Raman lidar measurements of gaseous and liquid water in the atmosphere. Appl. Opt.; 2013; 52, pp. 6884-6895. [DOI: https://dx.doi.org/10.1364/AO.52.006884] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24085202]

21. Tao, L.; Qi, F.; Jin, C.; Yue, G.; Hu, H.; Zhou, J. Raman Lidar System for the Measurements of Water Vapor Mixing Ratio in the Atmosphere. Chin. J. Atmos. Sci.; 2000; 24, pp. 843-854.

22. Wang, Y.; Jia, L.; Li, X.; Fan, F.; Di, H.; Song, Y.; Hua, D. Compact fiber-optic spectroscopic design and its validation in atmospheric water vapor Raman lidar. J. Opt. Soc. Am. B; 2020; 37, pp. 941-948. [DOI: https://dx.doi.org/10.1364/JOSAB.384944]

23. Newsom, R.K.; Turner, D.D.; Mielke, B.; Clayton, M.; Ferrare, R.; Sivaraman, C. Simultaneous analog and photon counting detection for Raman lidar. Appl. Opt.; 2009; 48, pp. 3903-3914. [DOI: https://dx.doi.org/10.1364/AO.48.003903]

24. Turner, D.D.; Ferrare, R.; Brasseur, L.H.; Feltz, W.; Tooman, T. Automated retrievals of water vapor and aerosol profiles from an operational Raman lidar. J. Atmos. Ocean. Technol.; 2002; 19, pp. 37-50. [DOI: https://dx.doi.org/10.1175/1520-0426(2002)019<0037:AROWVA>2.0.CO;2]

25. Sakai, T.; Nagai, T.; Izumi, T.; Yoshida, S.; Shoji, Y. Automated compact mobile Raman lidar for water vapor measurement: Instrument description and validation by comparison with radiosonde, GNSS, and high-resolution objective analysis. Atmos. Meas. Tech.; 2019; 12, pp. 313-326. [DOI: https://dx.doi.org/10.5194/amt-12-313-2019]

26. Wulfmeyer, V.; Bauer, H.; Grzeschik, M.; Behrendt, A.; Vandenberghe, F.; Browell, E.V.; Ismail, S.; Ferrare, R.A. Four-Dimensional Variational Assimilation of Water Vapor Differential Absorption Lidar Data: The First Case Study within IHOP_2002. Mon. Weather Rev.; 2006; 134, pp. 209-230. [DOI: https://dx.doi.org/10.1175/MWR3070.1]

27. Grzeschik, M.; Bauer, H.; Wulfmeyer, V.; Engelbart, D.; Wandinger, U.; Mattis, I.; Althausen, D.; Engelmann, R.; Tesche, M.; Riede, A. Four-Dimensional Variational Data Analysis of Water Vapor Raman Lidar Data and Their Impact on Mesoscale Forecasts. J. Atmos. Ocean. Tech.; 2008; 25, pp. 1437-1453. [DOI: https://dx.doi.org/10.1175/2007JTECHA974.1]

28. Bielli, S.; Grzeschik, M.; Richard, E.; Flamant, C.; Champollion, C.; Kiemle, C.; Dorninger, M.; Brousseau, P. Assimilation of water-vapor airborne lidar observations: Impact study on the COPS precipitation forecasts. Q. J. R. Meteorol. Soc.; 2012; 138, pp. 1652-1667. [DOI: https://dx.doi.org/10.1002/qj.1864]

29. Lin, Y.; Ye, J.; Wang, Y.; Zhong, S. Water vapor retrieval method based on MODIS thermal infrared band and projection pursuit model. Remote Sens. Land Resour.; 2018; 30, pp. 120-127.

30. Flamant, C.; Chazette, P.; Caumont, O.; Di Girolamo, P.; Behrendt, A.; Sicard, M.; Totems, J.; Lange, D.; Fourrié, N.; Brousseau, P. et al. A network of water vapor Raman lidars for improving heavy precipitation forecasting in southern France: Introducing the WaLiNeAs initiative. Bull. Atmos. Sci. Technol.; 2021; 2, 10. [DOI: https://dx.doi.org/10.1007/s42865-021-00037-6]

31. Yoshida, S.; Sakai, T.; Nagai, T.; Ikuta, Y.; Shoji, Y.; Seko, H.; Shiraishi, K. Lidar Observations and Data Assimilation of Low-Level Moist Inflows Causing Severe Local Rainfall Associated with a Mesoscale Convective System. Mon. Weather Rev.; 2022; 150, pp. 1781-1798. [DOI: https://dx.doi.org/10.1175/MWR-D-21-0213.1]

32. Zhang, T.; Yin, Z.; Wei, Y.; Dai, Y.; Wang, L.; Dong, X.; Gao, Y.; Wei, L.; Zhang, Q.; Hu, D. et al. Vertical Distribution of Water Vapor During Haze Processes in Northeast China Based on Raman Lidar Measurements. Remote Sens.; 2024; 16, 3713. [DOI: https://dx.doi.org/10.3390/rs16193713]

33. Zou, W.; Yin, Z.; Dai, Y.; Chen, Y.; Bu, Z.; Li, S.; He, Y.; Hu, X.; Müller, D.; Lu, T. et al. Robust Lidar-Radar Composite Cloud Boundary Detection Method with Rainfall Pixels Removal. IEEE Trans. Geosci. Remote Sens.; 2024; 62, pp. 1-16. [DOI: https://dx.doi.org/10.1109/TGRS.2024.3476127]

34. Bevington, P.R. Data Reduction and Error Analysis for the Physical Sciences; McGraw-Hill: New York, NY, USA, 1969; Chapter 6