High-Precision Rayleigh Doppler Lidar with Fiber Solid-State Cascade Amplified High-Power Single-Frequency Laser for Wind Measurement

Abstract

We introduce a novel Rayleigh Doppler lidar (RDLD) system that utilizes a high-power single-frequency laser with over 60 W average output power, achieved through fiber solid-state cascade amplification. This lidar represents a significant advancement by addressing common challenges such as mode hopping and multi-longitudinal mode issues. Designed for atmospheric wind and temperature profiling, the system operates effectively between altitudes of 30 km and 70 km. Key performance metrics include wind speed and temperature measurement errors below 7 m/s and 3 K, respectively, at 60 km, based on 30 min temporal and 1 km spatial resolutions. Observation data align closely with ECMWF reanalysis data, showing high correlation coefficients of 0.98, 0.91, and 0.94 for zonal wind, meridional wind, and temperature, respectively. Continuous observations also reveal detailed wind field variations caused by gravity waves, demonstrating the system’s high resolution and reliability. These results highlight the RDLD system’s potential for advancing meteorological monitoring, atmospheric dynamics studies, and environmental safety applications.

1. Introduction

Information on atmospheric wind and temperature fields in the upper stratosphere and lower mesosphere (USLM: ~30–70 km altitude) is crucial for both scientific research and practical applications, especially in climate and atmospheric dynamics research and meteorological modeling, where high-precision and high-resolution wind and temperature profiles are required [1,2,3,4]. Atmospheric physical processes in this region are complex, including gravity wave propagation and atmospheric temperature inversion, which have significant impacts on climate change, environmental protection, aerospace activities, and radio communication. Therefore, obtaining accurate wind and temperature profile data at these altitudes is vital for advancing aerospace technology, ensuring public safety, and promoting sustainable development [5,6,7,8].

However, studying atmospheric elements in the USLM region faces technical challenges. Most existing atmospheric detection technologies, such as radiosondes, weather radars, and meteorological satellites, are limited to altitudes below 30 km or above 70 km. Wind field detection, in particular, can only be achieved by lidar or by launching meteorological rockets [9,10,11,12]. Rocket-based detection, however, is costly and unable to provide continuous measurements, making lidar nearly the only viable method for routine wind field observations in the USLM region. When using Rayleigh Doppler lidar for atmospheric sounding at these altitudes, the SNR (signal-to-noise ratio) of the Rayleigh scattered signal is significantly reduced, primarily due to the low atmospheric density and long detection distances. Therefore, improving the stability and accuracy of the Rayleigh Doppler lidar system is crucial for reliable detection.

Zahn et al. used two lasers, each with an average power of 14 W and a repetition frequency of 30 Hz, stabilized by an iodine vapor cell. They measured residual frequency fluctuations of 0.84 MHz over 66 h in the RMR lidar system at the ALOMAR Observatory in northern Norway [13,14]. The system had a line-of-sight wind speed uncertainty of about 10 m/s at 80 km altitude, which translated to a horizontal wind uncertainty of about 20 m/s with a 2 h temporal and 2 km vertical resolution [4]; Hauchecorne et al., in the LIOvent Doppler lidar system at the OHP Provence Observatory, used a laser with an average power of 24 W and a repetition frequency of 30 Hz, stabilized through seed injection, achieving emission frequency stability with an RMS below 10 MHz. At 60 km altitude, with a 5 h temporal and 2 km vertical resolution, the system had a horizontal wind speed measurement error of about 8 m/s [11,15]; Similarly, Khaykin et al. used a lidar with the same specifications at the OPAR Meadow Observatory on Reunion Island, where a 3 h observation at 45 km altitude resulted in a detection error of 20 m/s [16]; Yan et al. developed a vehicle-mounted Rayleigh Doppler lidar system using a laser with an average power of 15 W and a repetition frequency of 30 Hz, achieving a horizontal wind speed error of 10 m/s at 70 km altitude, with a 2 h temporal and 3 km vertical resolution [17]; Dou et al. achieved 1.8 MHz RMS locking accuracy in 2 h using a 355 nm UV laser with an average power of 17.5 W and a repetition frequency of 30 Hz in a mobile Rayleigh Doppler lidar system. At 60 km altitude, with a temporal resolution of 30 min and a vertical resolution of 1 km, the wind speed error reached up to 9.2 m/s [18]; Gerding et al. used a 532 nm laser with an average power of 50 W and a repetition frequency of 100 Hz in the Doppler RMR system in Kühlungsborn, simultaneously serving two observation directions [19]; Sun et al. used a seed-injected laser with 15 W output power in their Three-Frequency Rayleigh Doppler Lidar (TFRDL). At 70 km, with 1 km spatial and 1 h temporal resolution, the system had temperature and wind speed uncertainties of 16.3 K and 8.1 m/s [20]; Zhao et al. developed a Doppler lidar system mounted on a rotating platform, featuring a single-frequency laser with an average power of 10.5 W and a frequency stability of 14.8 MHz, capable of reaching a detection altitude of 30 km [1,21]; Additionally, Lux et al. studied and analyzed ALADIN, the world’s first space-borne Doppler wind lidar, where the system used a seed-injection laser with an average power of about 4 W and a pulse energy of 80 mJ. After more than two years of space service, the inter-pulse frequency variation was measured at 10 MHz RMS [22,23,24].

The design schemes for single-frequency lasers in current Rayleigh Doppler lidars almost universally adopt seed injection, where the amplification process of the slave laser is initiated or guided by a single-frequency seed laser. In practice, regardless of whether the seed injection scheme involves Q-pulse minimization, ramp-fire, or ramp-hold-fire, frequency instability issues still arise. Typical problems include laser mode hopping and multi-longitudinal mode output, usually caused by system instability, device performance degradation, or environmental changes—many of which are difficult to avoid. Additionally, system detection accuracy is primarily characterized by detection errors, which are largely influenced by the SNR. To improve SNR and reduce errors, researchers often sacrifice spatial and temporal resolution to enhance data quality in high-altitude wind detection. For lidar systems requiring specific resolution, the SNR can typically only be improved by upgrading hardware parameters.

For the first time, we developed a Rayleigh Doppler lidar system using a fiber solid-state cascade amplification laser scheme with single-frequency seeded laser intensity modulation. This scheme provides the highest known average laser power output among Rayleigh Doppler lidars and avoids mode hopping and multi-longitudinal mode issues commonly found in traditional seed injection schemes, thereby improving detection accuracy and system reliability. The observation results demonstrate this system’s excellence in frequency stability, temporal and spatial resolution, and data quality, offering high-precision data for atmospheric research. Section 2 details the system hardware with a focus on the laser design; Section 3 analyzes the observation data and validates it with ECMWF (European Centre for Medium-Range Weather Forecasts) data; Section 4 demonstrates how high-resolution observations capture detailed wind field variations induced by gravity waves; and Section 5 summarizes the study.

2. RDLD System

The RDLD system is designed to detect atmospheric temperature and wind profiles at altitudes ranging from 30 to 70 km. Figure 1 illustrates the schematic diagram of the RDLD, which consists of several subsystems that can be broadly divided into a laser emission system and a signal acquisition system from a hardware perspective.

Table 1. Key parameter specifications of the RDLD.

Subsystem	Parameter	Value
Laser emission system	Wavelength/nm	532.259
	Pulse energy/mJ	>300
	Repetition rate/Hz	200
	Linewidth/MHz	<200
Signal acquisition system	Telescope aperture/mm	1000
	Field of view/mrad	0.4
	Optical filter FWHM/nm	0.15
	Quantum efficiency of PMT/%	40
	Max photon counting rate/MHz	800

lists the main parameters of the RDLD. This section focuses on the laser emission system, the key component of the RDLD, and provides a brief overview of the signal acquisition system.

2.1. Laser Emission System

In recent years, single-frequency continuous-wave seed lasers, modulated into pulsed light and amplified through fiber amplifiers, have become common as single-frequency pulsed laser sources in MOPA laser systems, which makes the optical frequencies of seed and pulsed lasers highly consistent. As shown in Figure 2, the laser emission system is divided into three main modules: frequency-stabilized seed laser, fiber solid-state cascade amplifier, and frequency conversion and emission. First, the 1064 nm single-frequency laser generated by a DFB source is amplified and split. One beam is directed to an iodine vapor cell for frequency locking, while the other is sent to the fiber solid-state cascade amplifier for power amplification. During fiber amplification, an acousto-optic modulator (AOM) is used for intensity modulation, enabling pulse chopping. After hybrid cascade amplification, frequency doubling with an LBO crystal produces a 532 nm narrow-linewidth laser with over 300 mJ pulse energy and a 200 Hz repetition frequency. Finally, the final laser emission is realized by power control, time-division multiplexing, and other optical path designs.

2.1.1. Frequency-Stabilized Seed Laser

We employed an iodine vapor cell and a servo feedback system to achieve frequency locking for the seed laser, and the principle is shown in Figure 3. First, a 1064 nm DFB with a linewidth of 0.5 MHz and an output power of 40 mW was used as the seed laser. The output of the seed laser is connected to a self-developed fiber amplifier via FC/APC and amplified to 400 mW. The amplified laser is split into beams via a fiber with a beam split ratio of 3:1, with one beam used for frequency locking and the other for cascade amplification.

The 1064 nm laser beam, split to enable frequency locking, is frequency doubled to generate a 532 nm laser. Frequency conversion is achieved using a single-pass PPLN crystal from a fiber-coupled frequency doubling module. Laser frequency locking is achieved using an iodine molecule single-edge locking scheme. The iodine vapor cell has a compact structure, with assembled components measuring approximately 120 mm × 45 mm × 34 mm. The operating temperature of the iodine vapor cell is maintained at 50 °C, with a temperature control accuracy of ±0.01 °C. The laser frequency is tuned to the center of the short-wavelength edge of the iodine spectral absorption line 1109, and the iodine vapor cell converts changes in light frequency into variations in signal intensity. The servo feedback system adjusts the seed laser frequency in response to these intensity changes to achieve frequency locking.

2.1.2. Fiber Solid-State Cascade Amplifier

Figure 4 illustrates the working principle of the fiber solid-state cascade amplifier module. After splitting, the seed laser passes through a three-stage tandem fiber amplifier and two AOM intensity modulators. The combination of the fiber amplifier and AOM modulators converts the ~100 mW continuous-wave laser into a pulsed laser with ~2 μJ energy and ~15 ns pulse width. During the fiber amplification process, the first AOM modulates the continuous-wave laser into pulses with a repetition frequency of 20 kHz, using CW laser diode pumping for enhanced stability and efficiency. The second AOM then reduces the repetition frequency to 200 Hz while compensating for the frequency shift introduced by the first AOM.

The laser was then amplified by a five-stage, series-connected LD end-pumped Nd:YVO₄/Nd:YAG rod crystal. Each of the first three stages was amplified by a separate 878.6 nm fiber-coupled LD end-pumped Nd:YVO₄ amplifier, with a peak power of 120 W and a pump duration of 150 μs. Then, each of the two following stages of Nd:YAG end-pumped amplification was performed using a separate 808 nm fiber-coupled LD, also with a peak power of 120 W and a pump duration of 250 μs. Finally, the 1064 nm laser pulse was amplified to 15 mJ with a repetition frequency of 200 Hz and shaped into a Φ2.5 mm beam by a beam expander. The use of Nd:YVO₄ and Nd:YAG crystals optimizes power handling in the initial stage and efficiency in the later amplification stages. Nd:YVO₄ provides higher laser efficiency and better thermal conductivity, while Nd:YAG offers superior thermal management performance and a higher damage threshold under high-power conditions.

In the next amplification stage, due to the higher energy requirements, we use side-pumped slab amplifiers for their excellent thermal management and high efficiency. The three-stage, series-connected, side-pump slab amplifiers employ a symmetric conduction cooling design and a slab geometry. Each amplifier stage is optimized for thermal management and amplification efficiency. The first Nd:YAG slab crystal in the three-stage system has dimensions of 5 mm × 5 mm × 112 mm, a doping concentration of 1.0%, and its end face is cut at a Brewster angle. It is pumped by a 12 G-Stack (48 Bars) with a total peak power of 7200 W and a pump pulse width of 150 μs. The second slab crystal has dimensions of 6 mm × 8 mm × 121 mm, a doping concentration of 0.8%, and is pumped by a 14 G-Stack (84 Bars) with a total peak power of 12,600 W and a pump pulse width of 150 μs. The third-stage slab crystal measures 8 mm × 10 mm × 118 mm, with a doping concentration of 1.0%, and is pumped by a 12 G-Stack (72 Bars) with a total peak power of 10,800 W and a pump pulse width of 150 μs. The end faces of the last two slab crystals are cut at a 45° angle, with a 3° angular deviation between the two ends to prevent self-oscillation. Following amplification, two cylindrical mirrors compensated for the thermal focal length of the slab crystals in both horizontal and vertical directions, with a polarizer between the amplifiers to enhance laser polarization. Additionally, the optical elements in the amplification chain are angled to avoid self-excited oscillations. Ultimately, the laser energy was efficiently amplified from 15 mJ to 600 mJ by the three stages of tandem slab amplifiers, achieving optical-optical efficiencies of 7.1%, 17.3%, and 27%, respectively.

The laser is cooled by two separate water-cooling units, which control the temperature of the laser baseplate and slab amplifiers, with cooling capacities of 0.65 kW and 4 kW, respectively. The cooling fluid used is ethylene glycol, selected to prevent damage to the components in the event of an unexpected power failure.

2.1.3. Frequency Conversion and Emission

As shown in Figure 5, the cascade-amplified 1064 nm single-frequency laser is frequency-doubled by an LBO crystal to produce a single-frequency, high-energy 532 nm laser. The fundamental and frequency-doubled lasers are then separated by a dichroic mirror, with the fundamental laser being recovered by a water-cooled absorber. The output power of the 532 nm laser required for the lidar system is tuned by a combination of a half-wave plate and a polarizer. Specifically, the half-wave plate is mounted in a piezoelectric-controlled rotating mirror frame, allowing control of the laser’s linear polarization direction. The polarizer selects the horizontally polarized component to adjust the output power continuously, while the vertically polarized component is recovered by the water-cooled absorber. Compared to controlling the pump power directly, this method of output power control maintains other optical parameters and improves safety. A second polarizer is positioned to compensate for optical path offsets, further enhancing the horizontal polarization purity of the outgoing laser.

In particular, the frequency-doubling crystal is a class I phase-matched LBO with dimensions of 4 mm × 4 mm × 15 mm, with a cutting angle of θ = 90° and an incidence angle of φ = 0° relative to the optical axis. The phase-matching temperature of the LBO crystal is 46.5 °C. The pulse width was compressed to 4.8 ns, producing a 532 nm single-frequency laser output with more than 300 mJ pulse energy and a beam quality factor of about 4.7. The laser output reached an average power of 63.87 W, with a power standard deviation of 0.21 W and a power instability of 0.3% within 8 h of operation. This source currently achieves the highest average laser output power among known Rayleigh Doppler lidars, providing the lidar system with more robust detection capabilities and enabling more flexible observations.

The abundant laser power allows the RDLD to use a time-division multiplexing scheme for detection in three directions, with only one laser for the whole system. The time-division multiplexing unit consists of two half-wave plates and two polarization beam splitters (PBS). The half-wave plates are mounted on a piezoelectric rotating mirror frame, which adjusts their angles to switch the laser’s polarization direction between vertical and horizontal. The laser is either completely reflected or transmitted after passing through the polarization beam splitters. In practice, the piezoelectric rotating mirror frame is controlled to adjust the rotation angle of the two half-wave plates according to the timing control, so the laser is emitted sequentially in the vertical, westward, and northward directions. Due to the limitations of integration time, which are primarily determined by the system’s SNR and the required measurement accuracy, and since the vertical channel has relatively lower detection requirements, the integration times for the three detection directions have been set to 5 min, 12.5 min, and 12.5 min after considering all factors. The output laser is compressed through a 10× beam expander to reduce the full divergence angle to within 0.15 mrad and is finally transmitted into the atmosphere through a mirror mounted on the telescope barrel wall.

In addition to the abundant laser power, the laser achieves excellent frequency stability through iodine vapor cell frequency stabilization. Traditional seed injection schemes in other Rayleigh Doppler lidars often experience mode hopping and multi-longitudinal mode issues during extended operation. A 1 GHz mode-hopping frequency spacing can result in significant wind speed inversion errors, with mode hopping in 1% of the laser pulses causing approximately 2.67 m/s errors. Multi-longitudinal mode outputs not only increase random errors but also reduce system sensitivity due to spectral broadening. Calculations show that when the excitation linewidth broadens from 200 MHz to 50 GHz due to seed injection failure, system sensitivity decreases by two orders of magnitude, leading to significant wind speed inversion errors.

The RDLD laser uses intensity-modulated fiber solid-state cascade amplification, which amplifies the seed laser power without altering its frequency and avoids the sensitive resonant cavity issues associated with traditional spatial light paths. A frequency test lasting eight hours was conducted using the WS7-60 wavelength meter from HighFinesse, Tübingen, Germany, collecting five laser frequency data points per second. The test results, shown in Figure 6, indicate a standard deviation of 0.70 MHz and a peak-to-peak variation of 5.29 MHz, demonstrating that the system’s laser output is highly stable.

In practice, the system is not equipped with a wavelength meter for laser frequency monitoring. Therefore, as shown in Figure 1, we introduced the single-frequency seed continuous laser into the receiver and used a chopper to control the timing. Long-term monitoring of the seed laser frequency is achieved by calibrating the stable transmission spectrum of the iodine vapor cell. As shown in Figure 7, we analyzed laser frequency monitoring data over a time span of more than one month (17 days of actual observations), with a time interval of about ten minutes for each data point. The standard deviation of the monitored frequency is about 2.94 MHz, and the peak-to-peak variation is about 18.24 MHz. Considering that the monitored frequency data are affected by structural deformation of key components in the receiver as well as environmental influences, the data shown in Figure 7 indicate a satisfactory level. Additionally, the system benefits from real-time monitoring of the zero-frequency reference, which corrected the wind speed inversion error caused by frequency drift, up to a maximum of 4.86 m/s.

2.2. Signal Acquisition System

As shown in Figure 1, the signal acquisition system consists of three telescopes, an iodine vapor cell, detectors, acquisition cards, and other components. Each telescope has an aperture of 1 m and a focal length of 1.5 m, aligned to the vertical, west, and north directions, respectively (with the westward and northward telescopes tilted at an angle of 30° from the zenith). To effectively integrate and process the echo signals from the three directions, the system uses a three-in-one fiber combiner, which combines the three signals, introducing them into the receiver via a common optical path, with an energy loss of approximately 7%. This design saves both cost and space while also avoiding inversion errors caused by performance differences in frequency discriminators across multiple receivers.

After beam combining, the optical signal first passes through an aspherical collimating lens for beam shaping, and background noise is filtered out using a narrow-band filter with a 0.15 nm bandwidth and more than 80% central wavelength transmittance. To fully utilize the linear dynamic response range of the detector, we truncate the atmospheric echoes below 30 km using a chopper. In the frequency discrimination optical path, the optical signal is divided into two paths by a beam splitter, with 30% of the signal used as the reference signal directly entering the photomultiplier tube (PMT), while the remaining signal is transmitted through the iodine vapor cell and then enters the other PMT, where the absorption spectrum of iodine molecules is used for optical frequency discrimination. The PMTs convert the optical signals into electrical signals, which are recorded by a high-speed data acquisition card with a maximum photon count rate of 800 MHz.

Additionally, as shown in Figure 1, the seed laser is introduced into the receiver via an optical fiber and is coupled into two PMTs using a beam splitter placed before the iodine cell. The timing of the seed laser signal hitting the PMTs is controlled by the chopper, ensuring it is separated from the atmospheric echo signal. This allows the same two PMTs to time-division multiplex the detection of both the seed laser signal and the atmospheric echo signal. This module is designed for both long-term laser frequency monitoring and zero-Doppler calibration, as well as for monitoring the operational health of the system.

3. Observations and Validation

3.1. Analysis of Measured Data

At altitudes above 30 km, the echo signal primarily originates from molecular Rayleigh scattering, so the atmospheric echo signal is proportional to the atmospheric molecular volume density. We select the altitude with a relatively high SNR as the reference altitude and obtain the absolute atmospheric density at this altitude from atmospheric models. The relative atmospheric density at other altitudes is then calculated using the ratio method. Based on this, we perform a backward integration of the Rayleigh scattering signal using the hydrostatic equilibrium assumption and the ideal gas law to derive the atmospheric temperature.

The principle for wind speed inversion is based on Rayleigh scattered light passing through an iodine cell. Due to selective absorption of different frequency components by the iodine cell’s absorption lines, the intensity of the transmitted Rayleigh scattered signal decreases. In the steep edge region of the absorption line, the relative transmittance of the Rayleigh scattering signal is monotonically related to the signal’s central frequency. Thus, we can obtain the Doppler shift and invert the wind speed.

From a system perspective, in addition to the high stability of the laser frequency that enhances system reliability and reduces lidar detection errors, its high-power characteristics also provide significant advantages in spatial and temporal resolution for RDLD, making data products richer and able to meet more diverse demand scenarios. The maximum spatial and temporal resolution of RDLD can be set to 15 min and 150 m, but to obtain high-precision data products, it is usually necessary to integrate time and space to improve the SNR of the raw data. Thanks to the high average power output of the laser, the default resolution of the RDLD can be set to 30 min and 1 km, ensuring high detection accuracy under normal weather conditions.

The RDLD system was commissioned in late 2022 at the detection base of Nanjing University of Information Science and Technology (NUIST) and has been conducting automated continuous observations under weather conditions that met the observation requirements. We selected representative detection data to analyze the detection error. Figure 8 presents a comparison of horizontal wind speed and temperature detection errors under different integration time conditions based on a 1 km spatial resolution. These plots are derived from typical data collected in Nanjing and are used for comparing measurement errors under different conditions. The measurement error evaluation method is consistent with traditional lidar analysis approaches [17,25,26]. In Figure 8, different colors represent different integration times.

Figure 8 illustrates a steady decrease in wind speed and temperature errors as the signal integration time increases. At 60 km altitude, with a 30 min integration time and 1 km spatial resolution, the wind speed detection error is less than 7 m/s and the temperature detection error is less than 3 K.

The horizontal wind speed and temperature detection errors at 60 km altitude with different spatial and temporal resolution configurations are further analyzed, as shown in Figure 9. Different colors in the figure represent the horizontal wind speed and temperature detection errors at the corresponding spatial and temporal resolutions of the horizontal and vertical coordinates. These plots are based on the same set of raw data used in Figure 8 and are derived through measurement error analysis to illustrate the system’s performance under different conditions. The results show that RDLD is not only capable of making observations at high altitudes with high spatial and temporal resolution but also of producing highly accurate data in low spatial and temporal resolution scenarios.

3.2. Comparative Validation

In this subsection, we selected the valid data observed in Nanjing from 6 January to 15 February 2023, and used the ECMWF-provided wind speed and temperature reanalysis data as the reference source [27,28,29,30,31].

We first selected the first three days of valid data from the data source, i.e., typical data from the evening to early morning on 6, 7, and 10 January 2023, Beijing time (observations were not made on the 8th and 9th due to weather). For effective data comparison, we selected the altitude range of 30 km to 60 km, which is relatively accurate for both data sources. Figure 10 shows the comparison of zonal wind speed, meridional wind speed, and temperature profiles between RDLD and ECMWF data. The solid curves represent the RDLD data, while the dashed curves of the same color represent the ECMWF data. The three subfigures share a common legend, with the numbers in the legend representing the day of observation. The graphs show that the daily mean profile trends of RDLD and ECMWF data in the 30 km to 60 km range exhibit good consistency.

We analyzed the deviation maps between RDLD and ECMWF for the three aforementioned days. Figure 11 shows the deviations in zonal wind speed, meridional wind speed, and temperature of RDLD data relative to ECMWF data. The numbers on the horizontal axis represent the month, day, and time. The figure shows that the deviations in zonal wind speed, meridional wind speed, and temperature vary across different altitude layers and time periods. Zonal wind speed deviations primarily range from −20 m/s to +10 m/s, with marked fluctuations, especially in the 40–50 km altitude range. Meridional wind speed deviations range from −10 m/s to +15 m/s, with smaller variations over time and a more uniform altitude distribution. Temperature deviations are mainly concentrated between −10 K and +10 K, with considerable fluctuations in the 40–50 km altitude layer. Although the relative deviations are generally small, the atmospheric dynamical processes at different altitudes are complex. These results help identify and understand the differences between RDLD and ECMWF data in different atmospheric layers and further optimize and calibrate the data to improve the accuracy of atmospheric sounding and weather forecasting.

Finally, we compared and analyzed all valid data from RDLD and ECMWF for the period from 6 January to 15 February 2023. We first selected ECMWF data within ±1° latitude and longitude of the RDLD location, matching time and height, followed by a scatterplot comparison. Specifically, for time matching, since the update rate of RDLD data is 30 min and that of ECMWF data is 1 h, the RDLD data closest to the ECMWF whole-hour data were selected, with a time deviation of no more than 15 min. For height matching, the vertical resolution of the ECMWF data was interpolated linearly to match the 1 km vertical resolution of the RDLD. To ensure data quality, the SNR threshold of the RDLD data was set to 10, and all data with SNR below this value were excluded.

Based on the above comparison method, the comparison results between RDLD and ECMWF data are shown in Figure 12. It is found that RDLD agrees well with ECMWF data in detecting zonal wind speed, meridional wind speed, and temperature. The linear fitting equations, correlation coefficients (R), mean bias (MB), and root mean square error (RMSE) are displayed sequentially in the upper left corner of the figure, where the correlation coefficients for zonal wind speed, meridional wind speed, and temperature between the two data sources are 0.98, 0.91, and 0.94, respectively. The matched data samples in the figure consist of 9340 wind speed data points (both zonal and meridional) and 11,679 temperature data points. Although there are slight differences in correlations among different data products, the overall results demonstrate the high detection accuracy and reliability of RDLD.

4. Observational Applications

In this section, the advantages of high-resolution observation by the RDLD in atmospheric detection in the USLM region are further analyzed by comparing the wind speed observation data of the RDLD system at different temporal resolutions. We also selected the first three days of valid data for illustration. Figure 13 and Figure 14 show the meridional wind speed time sequences of RDLD at different temporal resolutions of 30 min and 2 h. The red rectangular box in the figure highlights the key comparison area, where finer perturbations can be clearly observed at higher temporal resolution.

Figure 13 shows that RDLD successfully captured the fine wind field structural changes induced by gravity waves at a high temporal resolution of 30 min. These fine changes include oscillations and phase shifts in wind speed, reflecting the complex dynamical processes in the atmosphere of the USLM region. Since the propagation of gravity waves in the atmosphere can cause small perturbations in atmospheric temperature and wind speed, RDLD can precisely record these perturbations through high temporal resolution observations, providing crucial data for studying the propagation characteristics of gravity waves in the atmosphere. In contrast, Figure 14 shows a sequence of meridional wind speed observations at a 2 h time resolution. At this resolution, although the overall trend of the wind field remains visible, many details of wind speed changes caused by gravity waves are averaged out. In this case, the observations are more suited for studying the overall circulation characteristics of the atmosphere and the dynamical processes on larger scales. While a 2 h temporal resolution cannot capture the fine structure on short time scales, the observations provide greater stability and reliability for medium- and long-term atmospheric analysis due to the higher SNR of the data. Therefore, RDLD can adapt flexibly to various research and application scenarios by providing high temporal resolution data and switching to low temporal resolution when needed.

The adjustable spatial and temporal resolution of RDLD gives it great flexibility in practical applications. In high-resolution modes, the system is well suited for studying small-scale dynamical processes in the atmosphere, such as gravity waves, turbulence, and local storms. In low-resolution modes, the system is more suitable for monitoring overall circulation changes in the atmosphere and large-scale wind field characteristics, providing reliable data for weather forecasting and climate research. This flexibility not only demonstrates the advanced capabilities of the RDLD system but also highlights its potential for wide application in atmospheric science research. By selecting appropriate observation modes, RDLD can maximize its effectiveness in different research fields and provide high-precision observation data for atmospheric dynamics, meteorology, and environmental science.

5. Conclusions

This paper presents a new Rayleigh Doppler Lidar (RDLD) system, which is, to the best of our knowledge, among the first wind lidars to achieve more than 60 W of average laser power output through fiber solid-state cascade amplification while avoiding the mode hopping and multi-longitudinal mode issues typical of traditional seed injection schemes. The system enables high-precision wind and temperature measurements from 30 km to 70 km altitude, with errors below 7 m/s and 3 K at 60 km, based on a 30 min temporal and 1 km spatial resolution. The laser’s frequency drift standard deviation is approximately 0.70 MHz, ensuring high accuracy in wind speed inversion, while the integration of a system health monitoring module enhances real-time frequency monitoring and system reliability. Between 6 January and 15 February 2023, the RDLD system conducted automated continuous observations in the Nanjing region of China, capturing detailed wind field variations induced by gravity waves. A comparison with ECMWF reanalysis data shows high correlation coefficients for zonal wind speed (0.98), meridional wind speed (0.91), and temperature (0.94) in the 30 km to 60 km altitude range, demonstrating the RDLD system’s high accuracy and reliability in atmospheric detection.

The RDLD system excels in frequency stability, spatial and temporal resolution, and data quality, providing robust support for meteorological monitoring, atmospheric dynamics research, and safe operations in the USLM region. Future work will focus on optimizing system performance for daytime observations, particularly by employing a large-aperture Fabry–Perot etalon to further suppress background light noise, ensuring a sufficient signal-to-noise ratio for reliable daytime detection. On this basis, the potential of the system will be explored for broader research and operational scenarios.