To enhance the observational capability of the vertical profiling network in Zhejiang Province, the Zhejiang Meteorological Bureau has deployed multiple Raman-Mie aerosol LiDARs. However, the quality of current aerosol retrieval products from these LiDARs still needs improvement, and there is limited application and analysis of existing LiDAR data in Zhejiang, making it difficult to fully support refined meteorological services and pollution prevention efforts.

This study aims to optimize the aerosol retrieval algorithm for Raman-Mie LiDARs, improve the quality of aerosol products, and enhance their operational application. Given the general lack of auxiliary observations for aerosol LiDARs deployed in Zhejiang, especially the difficulty in accurately obtaining the LiDAR ratio for Mie data retrievals, this study proposes a method to derive aerosol LiDAR ratios based on Raman channel data during nighttime and early morning periods. This provides an effective support for improving the retrieval accuracy of aerosol optical properties based on Mie channel data.

In addition, by integrating meteorological elements and air mass backward trajectory data, a comprehensive analysis is conducted on a LiDAR observation case from January 6 to January 10, 2024. This provides a technical support for studying the formation mechanism of regional aerosol pollution and for developing pollution control strategies.

Considering the higher quality of Raman channel data during nighttime and early morning and the fact that these data are more suitable for extracting true signals using wavelet transform, this study prioritizes the use of wavelet transform to process Raman data collected at night and in the early morning. High-precision aerosol backscatter coefficient profiles are retrieved from these processed Raman data.

Based on these profiles above, a range of LiDAR ratios from 1 to 120 (with a step size of 1) is set. Multiple LiDAR ratios are then individually applied to retrieve the aerosol backscatter coefficient profiles from the corresponding Mie channel data at the same time. Using the retrieved aerosol backscatter coefficient profile based on Raman data as a reference, the mean relative error (MRE) is calculated for each retrieval based on Mie data. The LiDAR ratio corresponding to the minimum MRE is regarded as the optimal value for that time.

Furthermore, the optimal LiDAR ratios determined for multiple time points during the night and early morning are averaged. The mean LiDAR ratio from 00:00 to 06:00 is applied to retrieve the aerosol optical parameters from Mie data during 06:00—12:00 of the same day, while the mean LiDAR ratio from 18:00 to 24:00 is used for retrievals during 12:00—18:00. In this way, the retrievals of aerosol extinction coefficient and backscatter coefficient profiles for both daytime and nighttime are achieved.

A comparative analysis of wavelet denoising effects on Raman signals shows that using the wavelet basis function of sym7 with a decomposition level of 3 results in smoothed Raman signals while preserving more detailed information (Fig. 3 and Fig. 4). This way achieves a better balance between smoothness and effective information retention in the denoising of Raman signals.

In the optimized method for aerosol optical parameters retrieved from Mie channel data, multiple LiDAR ratios are preset to retrieve the corresponding aerosol backscatter coefficient profiles (Fig. 5). These profiles are then compared with those retrieved from Raman data, and the MRE between them is calculated (Fig. 2). The retrieval result based on Mie data with the lowest MRE is considered optimal, and the corresponding LiDAR ratio is taken as the optimal value for that time (Fig. 6).

Time series plots of aerosol extinction coefficients, backscatter coefficients, and depolarization ratios reveal that the near-surface aerosol extinction coefficients are relatively high between 18:00 on January 6 and 18:00 on January 7, 2024, and between 18:00 on January 9 and 08:00 on January 11, 2024. In both periods, the extinction coefficients generally exceed 1 km?1 and the corresponding depolarization ratios (mostly below 0.2) are low, indicating heavy aerosol pollution near the surface and predominantly spherical aerosol particle shapes (Fig. 8).

A correlation analysis between the near-surface aerosol extinction coefficient retrieved by the optimal LiDAR ratio and the ground-based PM_2.5 mass concentration shows a correlation coefficient of 0.7, indicating a high level of consistency (Fig. 10). By combining the meteorological elements with the air mass backward trajectory data, it is found that during the periods from 18:00 on January 6 to 18:00 on January 7, 2024 and from 18:00 on January 9 to 08:00 on January 11, 2024, wind speeds are relatively high, mostly exceeding 2 m/s, and the prevailing wind direction is northwesterly (Fig. 11). All air mass backward trajectory clusters from January 6 to January 10, 2024, originate from the northwest. The mass concentration weighted trajectory values in the eastern Shandong, the northeastern Anhui, and parts of Jiangsu are relatively high, all exceeding 75 μg/m3, indicating that the northwesterly winds transport aerosol pollutants to the Huzhou area (Fig. 12).

This study takes full advantages of the high-accuracy aerosol optical parameters retrieved from Raman channel data during nighttime and early morning, providing a more reasonable LiDAR ratio selection strategy for retrieving aerosol properties from Mie channel data. The proposed method markedly improves the accuracy of aerosol retrievals based on Mie LiDAR data and effectively addresses the inherent limitations of single-instrument configurations and the absence of complementary observational data. It also presents a practical technical pathway for promoting the application of similar LiDAR systems in regional aerosol pollution monitoring.

Using the optimized method, the near-surface aerosol extinction coefficients retrieved from Mie LiDAR data exhibit a strong correlation with ground-based PM_2.5 mass concentrations, thereby confirming the reliability of the retrieval results.