Cite this Article
Mou Fusheng, Luo Jing, Li Suwen, Shan Wei, Hu Lisha. Vertical profile of aerosol extinction based on the measurement of O4 of multi-elevation angles with MAX-DOAS. Chinese Physics B, 2019, 28(8): 084212  
Permissions
Vertical profile of aerosol extinction based on the measurement of O4 of multi-elevation angles with MAX-DOAS
Mou Fusheng, Luo Jing, Li Suwen ‡, Shan Wei, Hu Lisha
School of Physics and Electronic Information, Huaibei Normal University, Huaibei 235000, China

 

† Corresponding author. E-mail: swli@chnu.edu.cn

Abstract

A method for aerosol extinction profile retrieval using ground-based multi-axis differential optical absorption spectroscopy (MAX-DOAS) is studied, which is based on a look-up table algorithm. The algorithm uses parametric method to represent aerosol extinction profiles and simulate different atmospheric aerosol states through atmospheric radiation transfer model. Based on the method, aerosol extinction profile was obtained during six cloud-free days. The O4 differential air mass factor (dAMF) measured by MAX-DOAS is compared with the corresponding model results under different atmospheric conditions ( ). The aerosol optical thickness, aerosol weight factor in boundary layer, and the height of the boundary layer are obtained after the process of minimization and look-up table method. The retrieved aerosol extinction in boundary layer is compared with PM2.5 data measured by ground point instrument. The diurnal variation trends of the two methods are in good agreement. The correlation coefficients of the two methods are 0.71 when the aerosol optical thickness is smaller than 0.5. The results show that the look-up table method can obtain the aerosol state of the troposphere and provide validation for other instrument data.

PACS: 42.68.wt;78.40.-q;87.64.k-;95.75.Rs
Keyword:vertical profile of aerosol extinction;look-up table;O4;differential air mass factor;multi-axis differential optical absorption spectroscopy
1. Introduction

With the development of industry and the acceleration of urbanization, the concentration of particulate matter and secondary particulate matter in the atmosphere has increased substantially, which has an important impact on atmospheric environmental quality and terrestrial gas system radiation budget.[1,2] Due to the uncertainty of aerosol in morphology, composition, physical and chemical properties, and optical properties, it is difficult to accurately assess the impact of aerosol on the environment and global climate. Therefore, the spatial and temporal distribution, optical properties, and spectral distribution characteristics of aerosol have become a research hotspot in atmospheric environmental monitoring and related climate effects. It is very important to obtain aerosol optical properties for understanding the evolution process of atmospheric pollution, calibration of model, and satellite data.[3]

In recent years, differential optical absorption spectroscopy (DOAS) is used in the field of atmospheric environmental monitoring.[4] The multi-axis DOAS (MAX-DOAS) composed of zenith direction and low observation elevation greatly improves the observation sensitivity of the tropospheric atmosphere. Trace gas profiles in the troposphere can be obtained by combining with the atmospheric radiation transfer model. In 2004, Wagner et al. proposed to retrieve the atmospheric scattering light path by using the characteristics of oxygen dimer (O4) vertical profile which is basically unchanged in the atmosphere and evenly distributed horizontally.[5] Since the scattering energy of aerosol strongly affects the photonic optical path of the atmosphere, the aerosol optical thickness and profile information can be obtained by observing the slant column density (SCD) of O4 through ground-based MAX-DOAS. In order to achieve the rapid, accurate, and real-time acquisition of aerosol profile information, a series of studies have been carried out by different research groups.[6–8] At present, two main algorithms of the optimal estimation method and look-up table method are used to obtain aerosol optical properties based on the SCD of O4 through multi-elevation angles with MAX-DOAS. The optimal estimation method can obtain a continuous profile with a high vertical resolution. However, the algorithm relies on prior information and singular values could appear in the retrieval.[9] The look-up table method carries out parameter quantization assumption for aerosol profile, and the retrieval of aerosol profile is relatively fast in look-up table algorithm after the establishment of database. The results are relatively stable because they come from the given situation of database.[10]

In 2010, look-up table method was first used to retrieve the extinction profile of aerosol by Li et al.[11] The results of different time periods were compared with the ground-based in situ data. However, the complete weight factor was not included in the profile hypothesis (0.5–1 proportion of aerosols in the boundary layer) and the effect of extraneous transport on aerosol was not considered. In 2013, Wu et al. used look-up table method to retrieve the aerosol extinction profile and compared it with aerosol optical thickness obtained by the Lidar.[12] However, only four hours of measurement results were obtained in the noon period of a day (9:00–13:00).

The look-up table method including complete weight factor (0.1–1) and different time periods (8:00–16:00) is used to retrieve the aerosol profile in this paper. Field experiments are carried out to study the ability of the algorithm to obtain the profile of aerosol extinction under different atmospheric aerosol conditions. The boundary layer aerosol extinction by MAX-DOAS is compared with PM2.5 data obtained by ground-based point instruments.

2. Methods and principles
2.1. MAX-DOAS system and settings

The one-dimensional MAX-DOAS system was used in the field, which was developed by Anhui Institute of Optics and Fine Mechanics.[13] The experimental site was located in the northern suburb of Huaibei, and continuous observation was carried out in the cloud-free days from September 29, 2018 to October 4, 2018. The telescope pointed downtown (azimuth angle 177°, north is 0°). Figure 1 shows the position of the MAX-DOAS and schematic diagram of the measurement principle. The spectral retrieval band was 338 nm–365.5 nm and the spectrometer had a resolution of 0.5 nm. To reduce the impact of the temperature drift on spectral measurements, the spectrometer was put in the temperature control box and the temperature was set to 27 °C. The measurement spectra in different directions can be obtained by changing the direction of the telescope. The zenith spectrum was used as the reference spectrum, and the retrieval result was the difference of O4 slant column density between the measurement spectrum and reference spectrum, which was called the differential slant column density (dSCD). To facilitate comparison with the model, the concentration divided by O4 vertical column density is converted into the differential air mass factor (dAMF). The point instrument can measure the concentration information of near-surface aerosol particles, which can provide comparative verification for the MAX-DOAS retrieval results in the field campaign.

Fig. 1. (a) Position of the MAX-DOAS and (b) schematic diagram of the measurement principle.
2.2. Algorithm for aerosol profile retrieval based on O4 absorption

For the aerosol profile, a two-layer setup for vertical aerosol distribution is used in the troposphere (0 km–15 km), which can be described by a limited set of parameters. The aerosol is assumed to be homogeneous in a layer near the ground surface (called boundary layer), and the aerosol extinction coefficient (unit: km−1) in the layer aloft (i.e., the free troposphere) decreases exponentially with height. This type of aerosol distribution is frequently observed in polluted regions and verified by Lidar observation. The extinction profile can be described as two layers in the range from 0 km to 15 km where z is the height above ground, t is the aerosol optical thickness from 0 km to 15 km, and w is the weight factor of total extinction t in the boundary layer. h is aerosol boundary layer height, and A is the normalizing constant for exponentiation function and here, a constant of 5 km is adopted. The norm A is calculated from the integral of E(z) over the entire tropospheric , leading to The parameters are input into the model to obtain the O4 dAMF simulation values of nine elevation angles at different aerosol states (see Table 1). After obtaining the simulation results and measurement results in the measurement cycle, to optimize the parameters ( ), the cost function ( ) is minimized The simulated and measured results are Simα and Measα, respectively. And σ (Measα) is the standard deviation of the measured values. If the minimum result of nine angles in a measurement cycle is the smallest, the corresponding profile parameters can be considered to be more consistent with the real situation. To reduce the uncertainty of a single measurement, the hourly mean of the measured results is carried out at each elevation angle. The output error is bigger when the solar zenith angle (SZA) is greater than 75° and the processing data is selected from 8:00 to 16:00 each day.

Table 1.

Parameters for look-up table. SZA: solar zenith angle. SRAA: solar relative azimuth angle.

.

The parameters of aerosol optical thickness t, aerosol boundary layer height h, aerosol extinction weight within the boundary layer w, and other atmospheric parameters (such as single scattering albedo) constitute the look-up table.[14] The simulated O4 dAMF of each elevation angle under different atmospheric conditions is calculated through the atmospheric radiation transmission model. The model results and measurement results are compared to obtain the aerosol profile which best conforms to the real atmospheric conditions through the minimization process. The algorithm diagram is shown in Fig. 2.

Fig. 2. Schematic diagram of aerosol profile retrieved with the absorption of O4 observed by MAX-DOAS.

The full spherical Monte-Carlo atmospheric radiative transfer model, Monte Carlo atmospheric radiative transfer and inversion model (McArtim), is used in this study.[15] To improve the operational efficiency, the simulation wavelength is selected as 350 nm, which is the central wavelength of the retrieval band (337 nm–365.5 nm). Sensitivity studies have shown single scattering albedo ssa, asymmetric factor g, and surface albedo albedo have a small influence on air mass factor (AMF). Therefore, fixed values of ssa=0.92, g=0.72, and albedo=0.05 are adopted in the simulation. The specific settings of the model are shown in Table 1.

3. Results and analysis
3.1. Spectral analysis

The measurement spectrum is retrieved with DOAS method. The fitting structures are NO2 (298 K, vanDaele), O4 (296 K, Hermans), O3 (223 K, Bogumil), and Ring structure. The Ring structure is calculated by DOASIS software.[16] A fitting example of the 2° spectrum is shown in Fig. 3 on October 1, 2018 at 11:56. The zenith spectrum of each measurement cycle is selected as the reference spectrum. The O4 dSCD is , and the residual is 1.94 × 10−3, which is mainly derived from noise and unknown structure. Differential air mass factor of O4 is obtained according to this method from September 30, 2018 to October 4, 2018 (see Fig. 4). Due to the increase of aerosol content in the atmosphere, the measured results for all elevation angles decrease on October 3 and October 4.

Fig. 3. Fitting example of the measured spectrum. The black and red curves indicate the absorption structures and the fitted absorption structures, respectively.
Fig. 4. Differential air mass factor of O4 measured on cloud-free days.
3.2. Retrieval aerosol profile

Figure 5 shows the O4 dAMF at different elevation angles and the retrieval aerosol profile results in the case of low aerosol (10:00–11:00, October 1, 2018, t=0.3) and high aerosol (12:00–13:00, October 4, 2018, t=0.8). The measurement results and model results are in good agreement under low aerosol and high aerosol. For both the low aerosol (Fig. 5(a)) and high aerosol (Fig. 5(b)) situation, the largest O4 dAMF difference between simulation value and measured value is at 15° and the value is 5.3 × 10−2 and 1.2 × 10−1, respectively. The aerosol extinction profiles are obtained by look-up table method (see Figs. 5(c) and 5(d)). The dAMF mean values for all elevation angles are 1.07 and 0.65 in low and high aerosol, respectively. The effective absorption path of photons along the direction of the telescope is reduced because the high aerosol concentrations might be close to the ground. The differences of dAMF between different elevations are also decreased.[17] The linear correlation of O4 dAMF between simulation and measurement is shown in Fig. 6. Correlation analysis (R 2=0.78) shows that the retrieval results and measurement results are very close. The retrieval algorithm minimizes the cost function and the residual error of the inversion is very small.

Fig. 5. Examples for the retrieval of the aerosol extinction with high and low aerosol loads. Upper panels: comparison of measured and modeled O4 dAMF. Lower panels: retrieved profiles. Left: October 1, 2018, 10:00–11:00, SZA=43.03°. Right: October 1, 2018, 12:00–13:00, SZA=37.68°.
Fig. 6. O4 dAMF from MAX-DOAS observation and those derived from the model.

From the minimization process described in Subsection 2.1, the aerosol optical parameters (t, w, h) can be obtained from the iterative calculation of the O4 dAMF. These three parameters determine the aerosol extinction profile. Figure 7 shows the retrieval results of aerosol extinction profile for six days. It is found that the aerosol extinction coefficient near the ground on October 4 is larger. The aerosol optical thickness gradually increases from October 3, and the boundary layer height does not change much between 10:00 and 16:00, especially on September 30. From October 1 to October 4, the boundary layer has changed significantly, and the boundary layer is higher between 12:00 and 14:00. Due to the low height of the boundary layer, the extinction coefficient of aerosols near the ground is larger before 10:00 and after 15:00 every day. The MAX-DOAS system observed towards the urban area. And the point instrument (CEMS system of American thermal power company) was installed in the urban area to monitor the near-ground air quality. After obtaining the extinction coefficient of aerosol by the look-up table method, the extinction coefficient ( ) in the boundary layer can be obtained in the field. Figure 8 shows surface aerosol extinction and boundary layer height retrieved by MAX-DOAS for 6 days. PM2.5 data from surface point instrument are also shown in Fig. 8. Surface aerosol extinction E 0 by MAX-DOAS is in good agreement with PM2.5 for point instrument. Based on all data (N = 47) and (N = 34), the correlation coefficient is 0.64 and 0.71, respectively (as shown in Fig. 9).

Fig. 7. The aerosol profile retrieved by MAX-DOAS.
Fig. 8. Surface aerosol extinction E 0 and boundary layer height BH retrieved by MAX-DOAS for 6 days. Also shown are PM2.5 data from surface point instrument.
Fig. 9. Comparison between surface aerosol extinction measured by MAX-DOAS and PM2.5 data from point instrument. Different colors represent different aerosol optical thicknesses.

The difference between the two instruments is largest in the morning. Considering the fact that MAX-DOAS measures the aerosol extinction averaged over a distance while the point instrument detects the aerosol information near the ground, the results of the linear regression demonstrate good agreement between the measurement results of these two instruments. The latter is more easily influenced by local emission source and could produce high values.[11] In addition, fog and high aerosols in the morning will reduce atmospheric visibility and the effective optical path at all elevation angles of the MAX-DOAS telescope, which restricts the measurement process by MAX-DOAS.[9]

4. Conclusion

The retrieval algorithm of aerosol extinction vertical profile is studied based on the O4 absorption of multi-elevation angles by MAX-DOAS. The look-up table method is used to obtain the optimal aerosol extinction profile by minimizing the cost function. By comparing the measured results of O4 dAMF with model results, it is found that this method can accurately obtain the aerosol information for low and high aerosol situation. The retrieved aerosol extinction in boundary layer is compared with PM2.5 data measured by ground-based point instrument. The diurnal variation trend of the two methods is in good agreement. The results show that the look-up table method can obtain the aerosol state of the troposphere and provide validation for other instrument data. However, due to the limitation of the profile shape, it is difficult to retrieve more complex atmospheric pollution state such as the uplift layer. Therefore, different profile shapes will be added to the model and the results will be compared with other algorithms in the future.

Reference
[1] Rosenfeld D 2006 Science 312 1323 https://doi.org/10.1126/science.1128972
[2] Che H Yang Z Zhang X Zhu C Ma Q Zhou H 2009 Atmos. Environ. 43 1093 https://doi.org/10.1016/j.atmosenv.2008.11.010
[3] Ramanathan V Crutzen P J Kiehl J T Rosenfeld D 2001 Science 294 2119 https://doi.org/10.1126/science.1064034
[4] Wei M H Tong M M Li S W 2015 Chin. J. Chem. Phys. 28 688 https://doi.org/10.1063/1674-0068/28/cjcp1504063
[5] Wagner T Dix B Friedeburg C V 2004 J. Geophys. Res. 109 D22205 https://doi.org/10.1029/2004JD004904
[6] Frieß U Baltink H K Beirle S 2016 Atmos. Meas. Tech. 9 3205 https://doi.org/10.5194/amt-9-3205-2016
[7] Frieß U Monks P S Remedios J J 2006 J. Geophys. Res. 111 D14203 https://doi.org/10.1029/2005JD006618
[8] Wagner T Beirle S Brauers T 2011 Atmos. Meas. Tech. 4 2685 https://doi.org/10.5194/amt-4-2685-2011
[9] Wang Y Lampel J Xie P H 2017 Atmos. Chem. Phys. 17 2189 https://doi.org/10.5194/acp-17-2189-2017
[10] Li X Brauers T Hofzumahaus A 2013 Atmos. Chem. Phys. 13 2133 https://doi.org/10.5194/acp-13-2133-2013
[11] Li X Brauers T Shao M 2010 Atmos. Chem. Phys. 10 2079 https://doi.org/10.5194/acp-10-2079-2010
[12] Wu F C Xie P H Li A 2013 Acta Opt. Sin. 33 0601002 https://doi.org/10.3788/AOS201333.0601002
[13] Wang Y Beirle S Lampel J 2017 Atmos. Chem. Phys. 17 5007 https://doi.org/10.5194/acp-17-5007-2017
[14] Mou F S Luo J Li S W Wang J W Shi R R Wei M H 2018 Acta Photon. Sin. 47 0701002 https://doi.org/10.3788/gzxb20184707.0701002
[15] Deutschmann T Beirle S Frieß U 2011 J. Quantum Spectrosc Radiat. Transfer 112 1119 https://doi.org/10.1016/j.jqsrt.2010.12.009
[16] Kraus S 2006 Ph. D. Dissertation Germany University of Mannheim https://hci.iwr.uni-heidelberg.de/publications/dip/2006/Kraus_PhD2006.pdf
[17] Wang S S Cuevas C A Frieß U 2016 Atmos. Meas. Tech. 9 5089 https://doi.org/10.5194/amt-9-5089-2016