A convolutional neural networks method for tropospheric ozone vertical distribution retrieval from Multi-AXis Differential Optical Absorption Spectroscopy measurements.

The vertical distribution of tropospheric ozone (O3) is crucial for understanding atmospheric physicochemical processes. A Convolutional Neural Networks (CNN) method for the retrieval of tropospheric O3 vertical distribution from ground-based Multi-AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) measurements to tackle the issue of stratospheric O3 absorption interference faced by MAX-DOAS in obtaining tropospheric O3 profiles. Firstly, a hybrid model, named PCA-F_Regression-SVR, is developed to screen features sensitive to O3 inversion based on the MAX-DOAS spectra and EAC4 reanalysis O3 profiles, which incorporates Principal Component Analysis (PCA), F_Regression function, and Support Vector Regression (SVR) algorithm. Thus, these screened features for ancillary inversion include the profiles of temperature, specific humidity, fraction of cloud coverage, eastward and northward wind, the profiles of SO2, NO2, and HCHO, as well as season and time features to serve as sensitive factors. Secondly, the preprocessed MAX-DOAS spectra dataset and the sensitive factor dataset are utilized as input, while the O3 profiles of the EAC4 reanalysis dataset incorporating the surface O3 concentrations are employed as output for constructing the CNN model. The Mean Absolute Percentage Error (MAPE) decreases from 26 % to approximately 19 %. Finally, the CNN model is applied for inversion and comparison of tropospheric O3 profiles using independent input data. The CNN model effectively reproduces the O3 profiles of the EAC4 dataset, showing a Gaussian-like spatial distribution with peaks primarily around 950 hPa (550 m). Since the reanalysis data used for model training has been smoothed, the CNN model is insensitive to extreme values. This behavior can be attributed to the MAPE loss function, which evaluates Absolute Percentage Errors (APEs) of O3 concentration at all altitudes, resulting in varying retrieval accuracy across different altitudes while maintaining overall MAPE control. Temporally, the CNN model tends to overestimate surface O3 in summer by around 20 µg/m3, primarily due to the influence of the temperature feature in the sensitivity factor dataset. In conclusion, leveraging MAX-DOAS spectra enables the retrieval of tropospheric O3 vertical distribution through the established CNN model.

A CNN-SVR model for NO2 profile prediction based on MAX-DOAS observations: The influence of Chinese New Year overlapping the 2020 COVID-19 lockdown on vertical distributions of tropospheric NO2 in Nanjing, China.

In this study, a hybrid model, the convolutional neural network-support vector regression model, was adopted to achieve prediction of the NO2 profile in Nanjing from January 2019 to March 2021. Given the sudden decline in NO2 in February 2020, the contribution of the Coronavirus Disease-19 (COVID-19) lockdown, Chinese New Year (CNY), and meteorological conditions to the reduction of NO2 was evaluated. NO2 vertical column densities (VCDs) from January to March 2020 decreased by 59.05% and 32.81%, relative to the same period in 2019 and 2021, respectively. During the period of 2020 COVID-19, the average NO2 VCDs were 50.50% and 29.96% lower than those during the pre-lockdown and post-lockdown periods, respectively. The NO2 volume mixing ratios (VMRs) during the 2020 COVID-19 lockdown significantly decreased below 400 m. The NO2 VMRs under the different wind fields were significantly lower during the lockdown period than during the pre-lockdown period. This phenomenon could be attributed to the 2020 COVID-19 lockdown. The NO2 VMRs before and after the CNY were significantly lower in 2020 than in 2019 and 2021 in the same period, which further proves that the decrease in NO2 in February 2020 was attributed to the COVID-19 lockdown. Pollution source analysis of an NO2 pollution episode during the lockdown period showed that the polluted air mass in the Beijing-Tianjin-Hebei was transported southwards under the action of the north wind, and the subsequent unfavorable meteorological conditions (local wind speed of < 2.0 m/sec) resulted in the accumulation of pollutants.

McPrA - A new gas profile inversion algorithm for MAX-DOAS and apply to 50 m vertical resolution.

Air pollutants represent an environmental and health risk, and the methods for their effective assessment are of the greatest importance. The MAX-DOAS method is a reliable retrieval algorithm, enabling a vertical gas profile analysis. However, the current MAX-DOAS retrieval algorithm heavily relies on the a priori profile, limiting its accuracy. To address this issue, we introduced a novel MAX-DOAS trace gas profile inversion algorithm called McPrA, which is less dependent on the a priori profile. It employs the Monte Carlo method to resolve the problem of optimal estimation of trace gases. The gas vertical column density is obtained from the air mass factor calculated by SCIATRAN. Afterward, the trace gas vertical distribution is retrieved by combining the weight function with the a priori profile. A normalization process is also included to improve the matching of the weight function and the a priori profile. The McPrA algorithm enables greater flexibility in grid modification to achieve a higher vertical resolution of up to 50 m, while sensitivity experiments contribute to determining the optimal configuration of retrieval parameters, with a degree of freedom of over 3.0. Comparative verification experiments indicate that the McPrA algorithm accurately retrieves gas profiles, with a correlation coefficient of over 0.89 for NO2 in the first layer compared to in situ data. Furthermore, comparisons with WRF-Chem and the simulation of synthetic data demonstrate the effectiveness of the McPrA algorithm in accurately retrieving gas profiles.

Study of SO2 measurement based on a dual optical path Fabry-Perot correlation spectroscopy.

This paper investigates a method for measuring SO2 concentration using Fabry-Perot interferometer correlation spectroscopy. In this method, the experimental system is designed as a separated beam, with the beam entering the F-P cavity at two incidence angles simultaneously to match the peak and valley positions of the SO2 absorption cross-section. The system achieves a 2σ detection limit of 28.2 ppm·m(15 cm) at a sampling frequency of 10 Hz. An outfield comparison experiment with the differential optical absorption spectroscopy method shows good agreement for the simultaneous measurement of SO2 concentration from sulfur combustion, with a correlation coefficient of R2 = 0.93. This study introduces a non-dispersive, highly accurate, and fast gas detection technique.

The vertical distribution and potential sources of aerosols in the Yangtze River Delta region of China during open straw burning.

To explore the impact of open straw burning on air quality in the Yangtze River Delta (YRD) and surrounding areas, three key cities in the YRD, namely Hefei, Nanjing, and Shanghai, were selected to observe changes in aerosol characteristics. Based on Multi-AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) observations from May to June 2021, the spatial-temporal distribution and potential sources of aerosol were studied. During the observation period, aerosol optical depth (AOD) in Shanghai was 55.15 % and 29.50 % higher than that in Hefei and Nanjing, respectively. For Shanghai, aerosols accumulated at night, and the aerosol extinction could reach 1.3 km-1 in the morning. The aerosol variations in Hefei and Nanjing were consistent due to the relative conformity of the surrounding environmental conditions (R = 0.84). The vertical distribution of aerosol in all three cities had the same Gaussian shape. The aerosol lifted layers in Nanjing and Shanghai were higher than that in Hefei, with heights of 0.2-0.8 km and 0.2-0.6 km, respectively. The averaged aerosol extinctions for these two cities were 0.34 km-1 and 0.49 km-1, respectively. Pollution source analysis was conducted based on wind field trajectory, satellite observation, and model simulation, taking Hefei as the recipient. The results showed that western Shandong Province, northern Anhui Province, northern Jiangxi Province, central Jiangsu Province, and the central YRD were the most important aerosols sources for Hefei. The contributions of central and southern Jiangsu Province were significantly higher than those of other potential sources, with a WCWTAOD (Meteoinfo concentration weight trajectory) between 1.2 and 3.0. The influence of fine particles produced by open biomass burning inside the YRD was significantly higher than that outside the region (outside contribution: 36.6 %). Regarding the influence between YRD cities, more aerosols were transported from Shanghai to Hefei and Nanjing, with similar transport contributions between Nanjing and Hefei.

Vertical characteristics of NO2 and HCHO, and the ozone formation regimes in Hefei, China.

The research on the mechanism of combined air pollution in the Yangtze-Huaihe region, which is characterized by unique meteorological and geographical conditions and pollution emission characteristics, is still insufficient. We performed an experiment on key pollutants and an ozone formation study in Hefei, which is a pivotal city in the Yangtze-Huaihe region, from September 1 to 20, 2020. The aerosols retrieved via two-dimensional Multi-axis Differential Optical Absorption Spectroscopy (2D-MAX-DOAS) with a Boltzmann-shaped a priori profile had the best agreement with the results of Light Detection and Ranging (LIDAR) and sun-photometer measurements among the three typical a priori profiles (Gaussian, Boltzmann, and exponential shapes). The correlation coefficients of the near-surface gas concentrations retrieved using both 2D-MAX-DOAS and in situ measurements were 0.86 (NO2) and 0.61 (HCHO). The high NO2 and HCHO concentrations were observed at azimuths of 180° and 315° at heights of 0.8-1.5 km, and they may have been emitted by aircrafts. Importantly, the ratio of HCHO to NO2 during a typical pollution episode revealed that the factors controlling the O3 formation changed with altitude: VOCs (surface) to NOx (0.4 km) to transition (1.0 km) to VOCs (1.6 km). Moreover, the effect of VOCs on the O3 generation was stronger than that of NOx, especially in the downtown area of Hefei. When the ratio of HCHO to NO2 was 3.55-7.46, the ozone concentration in Hefei could be controlled well, especially at the optimal value of 5.50.