The systematic error is primarily due to the ozone-measurement error caused by aerosols. Based on the dual-wavelength differential absorption algorithm, the ozone-measurement error estimated using aerosol optical parameters retrieved from Mie lidar data requires the assumption of aerosol parameters (aerosol AE, aerosol backscatter ratio, and aerosol extinction-backscatter ratio). Compared with the ozone-measurement error contributed by an aerosol extinction-backscatter ratio of 50, an aerosol AE of 1.0, and an aerosol backscatter ratio of 1.01 in this experiment, the maximum difference in ozone concentration caused by an aerosol backscatter ratio of 1.002 or 1.05 for the signals of the 287/299 nm pair is (4.82±0.47) μg/m³ (Fig. 6). The maximum difference in ozone concentration due to a deviation of 50% in the aerosol extinction-backscatter ratio is (21.15±1.77) μg/m³ (Fig. 5). The maximum difference in ozone concentration due to a deviation of 50% in the aerosol AE is (30.11±2.12) μg/m³ (Fig. 5). The results show that the deviations of the aerosol AE and aerosol extinction-backscatter ratio significantly affect ozone-concentration retrieval, whereas the deviation of the aerosol backscatter ratio affect ozone-concentration retrieval less.

To reduce the deviations of the aerosol AE and aerosol extinction-backscatter ratio, we proposed a new idea involving two steps. In step 1, we assumed that the aerosol extinction-backscatter ratio in the experiment was stable. Thus, the AERONET data near the lidar measurement point was interpolated to obtain the aerosol extinction-backscatter ratio at 396 nm (average was 72.75±11.53), and the average value was considered as the accurate value to retrieve the aerosol extinction coefficient and aerosol backscattering coefficient. In step 2, we acquired the aerosol AE profile, as shown in Fig.7. In this experiment, the tropospheric ozone concentration retrieved from the aerosol AE profile was considered the actual value. The root mean square error of the tropospheric ozone concentration retrieved based on the assumed aerosol AE of 1.0 is 15.67 μg/m³ (Fig. 10).

To address the severe ozone pollution in China in recent years, monitoring tropospheric ozone concentration has become extremely important for effectively controlling tropospheric ozone pollution. Lidar is suitable for the in-depth study of regional tropospheric ozone because of its high spatial resolution, fast real-time operation, and large dynamic range. The dual-wavelength differential absorption algorithm is widely used to retrieve ozone concentrations measured using lidar. Most researchers retrieve aerosol optical parameters using the assumed value of the aerosol extinction-backscatter ratio or the aerosol angstrom exponent (AE), after which these aerosol optical parameters are used to estimate the ozone-measurement error caused by aerosols. This results in a significant error in the retrieval of the tropospheric ozone concentration. To obtain a relatively accurate ozone concentration using the dual-wavelength differential absorption algorithm, we combine ultraviolet multiwavelength lidar (276, 287, 299, and 396 nm) data with aerosol extinction-backscatter ratios provided by AERONET to retrieve the tropospheric ozone concentration. This approach can reduce the deviation of the aerosol optical parameters.

Based on the dual-wavelength differential absorption algorithm, we retrieved the ozone concentrations in the aerosol zone using four-wavelength ultraviolet (UV) lidar (276, 287, 299, and 396 nm) data. A more accurate ozone-measurement error estimated using aerosol optical parameters retrieved from lidar data at 396 nm requires a more accurate aerosol AE and aerosol extinction-backscatter ratio. To obtain a more accurate aerosol extinction-backscatter ratio, we interpolated multichannel data from AERONET to obtain aerosol extinction-backscatter ratios at 396 nm using a cubic spline function. These ratios were combined with the 396 nm channel data measured using the UV multiwavelength lidar to retrieve the aerosol optical parameters (aerosol extinction coefficient and aerosol backscattering coefficient). Subsequently, the aerosol optical parameters at 276, 287, and 299 nm were obtained for different aerosol AEs. The ozone concentrations were retrieved by the signals of 276/287, 276/299, and 287/299 nm wavelength pairs. The difference between the ozone concentrations retrieved by the signals of two wavelength pairs was calculated, and the minimum difference was used to determine the aerosol AE. A relatively accurate ozone concentration obtained was the mean of the two ozone concentrations corresponding to the aerosol AE.

When ultraviolet multiwavelength lidar data are used to retrieve the ozone concentration in the aerosol zone, the errors in the ozone concentration are primarily statistical and systematic errors. The statistical errors arise primarily from the signal quantum noise and background noise. To reduce the statistical error, We perform wavelet transform denoising on the original signal with a resolution of 7.5 m, which resulted in ?4.81% for the mean of relative errors of ozone concentrations under 2 km. This indicates that the statistical error of the ozone-concentration profile below 2 km is significantly reduced after wavelet denoising (Fig.2).

By applying the dual-wavelength differential absorption algorithm, when ultraviolet multiwavelength lidar (276, 287, 299, and 396 nm) data are used to measure the ozone concentration in the aerosol zone and then combined with the aerosol extinction-backscatter ratio from AERONET, we successfully estimate the ozone-concentration measurement error contributed by the aerosol extinction coefficient and atmospheric backscattering coefficient more accurately. Additionally, the deviation of the aerosol optical parameters in the ozone measurement is reduced and more accurate aerosol AE and ozone concentrations in the aerosol zone are retrieved.