key: cord-269423-p73bszlq
authors: Zhao, Fei; Liu, Cheng; Cai, Zhaonan; Liu, Xiong; Bak, Juseon; Kim, Jae; Hu, Qihou; Xia, Congzi; Zhang, Chengxin; Sun, Youwen; Wang, Wei; Liu, Jianguo
title: Ozone profile retrievals from TROPOMI: Implication for the variation of tropospheric ozone during the outbreak of COVID-19 in China
date: 2020-10-13
journal: Sci Total Environ
DOI: 10.1016/j.scitotenv.2020.142886
sha: 
doc_id: 269423
cord_uid: p73bszlq

During the outbreak of the coronavirus disease 2019 (COVID-19) in China in January and February 2020, production and living activities were drastically reduced to impede the spread of the virus, which also caused a strong reduction of the emission of primary pollutants. However, as a major species of secondary air pollutant, tropospheric ozone did not reduce synchronously, but instead rose in some region. Furthermore, higher concentrations of ozone may potentially promote the rates of COVID-19 infections, causing extra risk to human health. Thus, the variation of ozone should be evaluated widely. This paper presents ozone profiles and tropospheric ozone columns from ultraviolet radiances detected by TROPOospheric Monitoring Instrument (TROPOMI) onboard Sentinel 5 Precursor (S5P) satellite based on the principle of optimal estimation method. We compare our TROPOMI retrievals with global ozonesonde observations, Fourier Transform Spectrometry (FTS) observation at Hefei (117.17°E, 31.7°N) and Global Positioning System (GPS) ozonesonde sensor (GPSO3) ozonesonde profiles at Beijing (116.46°E, 39.80°N). The integrated Tropospheric Ozone Column (TOC) and Stratospheric Ozone Column (SOC) show excellent agreement with validation data. We use the retrieved TOC combining with tropospheric vertical column density (TVCD) of NO2 and HCHO from TROPOMI to assess the changes of tropospheric ozone during the outbreak of COVID-19 in China. Although NO2 TVCD decreased by 63%, the retrieved TOC over east China increase by 10% from the 20-day averaged before the lockdown on January 23, 2020 to 20-day averaged after it. Because the production of ozone in winter is controlled by volatile organic compounds (VOCs) indicated by monitored HCHO, which did not present evident change during the lockdown, the production of ozone did not decrease significantly. Besides, the decrease of NOx emission weakened the titration of ozone, causing an increase of ozone.

In early 2020, a novel coronavirus SARSCoV-2 (COVID-19) emerged in the city of Wuhan, China, and has spread all over the world by March 2020 (World Health Organization, 2020) . To curb the transmission of COVID-19 in China, many policy interventions, such as the travel restrictions, the shutdown of factories and halting mass transportation, have been implemented by the Chinese government, which was marked by the lockdown of Wuhan on January 23, 2020. These measures have also caused a great impact on the environment (Wang and Su, 2020; Xu et al., 2020; Zambrano-Monserrate et al., 2020) . Many previous researches indicate that the concentrations of primary air pollutants, such as like NO 2 , SO 2 , have sharply reduced due to large decrease of emissions . However, the changes of secondary pollutants like secondary aerosol and ozone were various, and their concentrations in some cities even increased due to the change of chemical reactions (Chinazzi et al., 2020; Le et al., 2020; Shi and Brasseur, 2020) .

Ozone has different effects at different altitudes in the atmosphere. Most of the ozone distributed in the stratosphere can absorb strong ultraviolet radiation that can cause health problems in humans (Bell et al., 2004; Norval et al., 2007; Su et al., 2017; Nuvolone et al., 2018; Zhang et al., 2020) . A small amount of ozone in the troposphere, which is mainly produced from the photochemical reaction between volatile organic compounds (VOCs) and nitrogen dioxides (NO x ) (Littman and Magill, 1953; Sillman, 1995) , shows harmful effects to humans. Therefore, the vertical distribution of ozone is very important for us to understand the physicochemical process of ozone. Especially, Zoran J o u r n a l P r e -p r o o f Journal Pre-proof 3 et al. (2020) found that ambient ozone may increase the COVID-19 infections, in addition to direct harm to human health. However, previous studies were all confined in the scale of city. Spatial distribution of ozone on regional and national scales should be obtained due to complex ozone chemistry in different regions.

Satellite-based remote sensing can monitor several species of trace constituents in the atmosphere on the global scale, which is an ideal means to analyse the spatio-temporal distribution of air pollutants. Total ozone column has been measured using backscattered UV radiation since 1970 with instruments like Backscatter UltraViolet (BUV) and Solar Backscatter Ultra Violet (SBUV) instruments (Bhartia et al., 1996) , as well as since 1995 by the Global Ozone Monitoring Experiment (GOME) (Hoogen et al., 1999) ). The SBUV make ~ 10 monochromatic measurements at the spectral resolution of ~ 1.0 nm. Therefore, the profile information is limited to the stratosphere with SBUV and the tropospheric column ozone is estimated using residual-based methods with a lot of approximations. The GOME first measures the full UV/VIS spectrum with a moderate spectral resolution of 0.2-0.4 nm and ozone profile retrievals down to the troposphere have been implemented (van der A et al., 2002; Liu et al., 2005) . Ozone Monitoring Instrument (OMI) on aboard the EOS Aura spacecraft, GOME-2 aboard the Metop and Ozone Mapping and Profiler Suite (OMPS) aboard the Suomi National Polar-Orbiting Partnership (SNPP) continued observation of atmospheric ozone as successors of GOME (Dirksen et al., 2006; Seftor et al., 2014) .

Ozone profiles retrieved from OMI has been extensively validated using ground-based data (Huang et al., 2017a; Xing et al., 2017; Zhang et al., 2019) and satellite data (Liu et al., 2010a; Huang et al., 2017b; Su et al., 2017) . The OMI instrument provides a favorable opportunity to contribute to the understanding of chemical and physical functions especially in the troposphere due to its high spatial resolution (13 km along the track × 24 km across the track at nadir). However, since the occurrence of OMI row anomaly in 2007 and worsened in 2009, a portion of cross-track positions have been affected and the data from these positions can no longer be used for scientific research (Huang et al., 2017a) . The OMPS is designed to measure total column and stratospheric ozone profile with spatial resolution of 50 x 50 km 2 , but Bak et al. (2017) retrieves the tropospheric column ozone with careful measurement calibration in spite of insufficient spectral information.

TROPOMI, which is the only payload of Sentinel-5 precursor mission, was launched on 13 October 2017.

The TROPOMI works on a 817 km sun-synchronous polar orbit with a mean Local Solar Time of 13:30 at Ascending Node (Veefkind et al., 2012) . TROPOMI has a similar instrument concept to OMI but has higher spatial resolutions and extend spectral range than those of OMI. Details on TROPOMI instrument parameters are discussed in later sections. The purpose of this paper is to describe the inversion algorithm of TROPOMI ozone profile and validate the results using various in-situ measurements as well as satellite data, which is further used to evaluate the variation of tropospheric ozone in different regions in China. To the best of our knowledge, TROPOMI ozone profiles have not been published in the literature. For this purpose, we adopt an optimal estimation based the algorithm which has been implemented for GOME, GOME-2, OMI and OMPS instruments (Liu et al., 2010b; Cai et al., 2012; Bak et al., 2017) . Accurate forward model simulation, absolute wavelength/radiometric calibration and good knowledge of a priori are essential for ozone profile retrievals. The adopted algorithm needs to be modified for TROPOMI, mainly with respect to the fitting windows, instrument slit function, wavelength/radiometric correction. Therefore, to improve the fitting accuracy, we thoroughly characterize the radiometric calibration spectrum from the comparison between simulated and measured radiances. We present a validation of the retrievals including ozone profile, SOC and TOC. This paper proceeds as follows: section 2 introduces the retrieval algorithm and TROPOMI L1b data, description of the forward model, retrieval scheme and examination of different slit function impacts on the reduction of fitting residuals and radiometric calibration and evaluate the effect of soft calibration by comparison of spectral fitting residuals with and without correction at all latitudes. In section 3, we talk about retrieval characterization. The validation results of retrievals are discussed in section 4, the impact of COVID-19 on tropospheric ozone are discussed in section 5 and conclusions are described in section 6.

The S-5P is a polar orbiting satellite, which was launched on 13 October 2017. The S-5P will fill the gap between the end of OMI and the Sentinel-5 mission. The TROPOMI is the only payload of the S-5P (Veefkind et al., 2012) . There are four different spectrometers, each with its own optics and detector: medium-wave UV, longwave UV combined with visible (UVIS), NIR, and SWIR (Kleipool et al., 2018) . TROPOMI is a double channel push-broom imaging spectrometer measuring radiance and irradiance in the ultraviolet (band-1: 267-300 nm, band-2: 300-332 nm), visible (band-3: 305-400 nm, band-4: 400-499 nm), near-infrared (band-5: 661-725 nm, band-6:725-786 nm), and the shortwave infrared (band-7: 2300-2343 nm, band-8: 2343-2389 nm) that provide daily global trace gases concentration information. The spectral resolution varies from 0.5 nm in UV, UVIS, and NIR bands, to 0.23 nm in SWIR band. Ozone profile information is contained in band 1-3 measurements. But we found that band-1 radiances have opposite systematic biases between tropics and mid/high J o u r n a l P r e -p r o o f Journal Pre-proof 5 latitudes and band-2 noise is larger than band-3 (Fig. 2) . Therefore, we retrieve ozone profile using optimal estimation technique from the band-3 (314-340 nm) where typical resolution is 0.54 nm.

TROPOMI has a wide field-of-view (108º) across the track and a small field-of-view along the track and the instantaneous swath width is 2600 km at the earth's surface. There are 450 cross-track positions used for spatial coverage and hence a smaller pixel size of 7 km (along the track, now reduced to 5.5 km) × 3.5 km (across the track) is acquired compared to its precursors OMI and OMPS.

TROPOMI is commanded to perform a solar irradiance measurement every 15 orbits. If no solar measurements are available in the data granule being processed, no irradiance product will be generated. The solar irradiance measurement follows the same binning scheme as the Earth radiance measurements. The impact of the reflectance degradation on retrieved ozone profiles is shown to be large . The comprehensive long-term analysis of the reflectance degradation for TROPOMI shows that the irradiance degrades 5 % around 314 nm and by 2 % around 340 nm from April 2018 through April 2019. However, we found that the reflectance changes less over time by comparing with soft calibration spectrum (more details in section 3) of different months in 2018. Therefore, the radiance shows similar attenuation and long-term changes as the irradiance and we use the daily solar irradiances to normalize the earthshine radiances.

The OMPS-LP ozone product measures the vertical distribution of ozone in the stratosphere and lower mesosphere. The algorithm derives ozone profile values along with errors in the UV from 29.5 km and 52.5 km, and in the visible from 12.5 km to 37.5 km (Deland, 2017) .

The FTS observation station, located in Hefei city in central-eastern China, has continuously measured ozone profile since April 2014. Thus it is significant to validate the retrieved profile in this region. The precision of the tropospheric ozone column (0-12 km) is approximately ~3 % and the accuracy is estimated to be ~8 %. A detailed description of the FTS station can be see in Sun et al. (2018) and Tian et al. (2018) . A total of 129 days of data from 1 March 2018 to 1 May 2019 are available. 6 Ozonesonde data during 2018-2019 are obtained from the World Ozone and Ultraviolet Data Center (WOUDC, http://www.woudc.org) and the Southern Hemisphere ADditional OZonesondes (SHADOZ, https://tropo.gsfc.nasa.gov/shadoz/) Witte et al., 2017; Sterling et al., 2018; Witte et al., 2018) . Ozonesonde data with vertical resolution ~100-150 m and typically 5-10 % accuracy and 3-5% precision have been widely used and validated (Huang et al., 2017a; Xing et al., 2017; Sterling et al., 2018) .

The "soft-calibration" spectra can be derived by comparing measured radiance and simulated radiance with "true" atmosphere as an input. Following the method proposed by Liu et al. (2010b) , the Vector LInearized Discrete Ordinate Radiative Transfer (VLIDORT) is used to simulate radiances and weighting functions (Liu et al., 2005; Liu et al., 2010b; Cai et al., 2012) . The Ring spectrum is directly characterised by a single scattering RRS model has been used in previous ozone profile retrievals (Sioris and Evans, 2000; Liu et al., 2010b; Cai et al., 2012) .The radiance is simulated for a Rayleigh atmosphere which excludes the effects of water vapor and aerosols. The surface and clouds are assumed to be Lambertian equivalent reflections. We treated clouds as a Lambertian surfaces at cloud-top with a fixed reflectivity of 80%. We use the Independent Pixel Approximation (IPA) to assume that partial clouds are a mixture of clear and cloudy scene. In the IPA, the cloud-top pressure we used are derived from the TROPOMI FRESCO algorithm (Coldewey-Egbers et al., 2018) . We acquired an initial cloud fraction at 347 nm based on surface albedo derived from the OMI surface climatology. High-resolution ozone absorption cross section is determined from Brion et al. (1993) according to the recommendation made in Liu et al. (2007) and Liu et al. (2013) . Daily temperature profiles and surface pressures derived from National Centers for Environmental Protection (NCEP) final (FNL) operational global reanalysis data (1°×1°) (https://rda.ucar.edu/) are used in our retrievals. Table 1 summarizes the VLIDORT inputs used in simulation. (Kleipool et al., 2008) ozone cross sections (Brion et al., 1993) A priori ozone profile (Bak et al., 2013) J o u r n a l P r e -p r o o f 

The Instrument Spectral Response Function (ISRF) has a crucial influence on the high fitting precision. We 

The retrieval algorithm, that has been applied to GOME (Liu et al., 2005) , GOME-2 , OMI (Liu et al., 2010b) , and OMPS (Bak et al., 2017) , is based on principle of optimal estimation method (Rodgers, 2000) . The algorithm simultaneously minimize the difference between calculated and measured radiances, and between a priori (X a ) and retrieved ( ) state vectors during iteration. The state vectors are constrained by a priori error covariance matrix and measurement error covariance matrix. The cost function and a posteriori solution can be given as:

(1)

Where and +1 are the previous and current state vectors contains the ozone profile and other parameters, respectively. is the measured radiance normalized to the daily solar irradiance; is the radiation transfer model; ( ) is the computed radiances using the radiation transfer model and normalized with ;

is the diagonal covariance matrix calculated from the measurement error. is defined as the weighting function matrix, = ⁄ ; Partial ozone column densities in 24 layers are divided by 25 pressure levels. The J o u r n a l P r e -p r o o f Journal Pre-proof 8 state vector consists of 33 parameters, including 24 partial ozone columns, one cloud parameter and three reflectivity parameters. It also includes two wavelength shift parameters, one Ring effect scaling parameter, BrO and HCHO vertical column densities (VCDs). Table 2 summarize the fitting variables, a priori and corresponding a priori errors.

The measurement errors are derived from TROPOMI level 1b random-noise errors. The a priori ozone profile and a priori error covariance matrix is derived from the ozone climatology calculated based on tropopause that has been use in retrievals by (Bak et al., 2013) . It can improve the precision of retrieved ozone profile, especially in the upper troposphere and lower stratosphere (UTLS). In the OMI retrievals, invariable floor noise errors are used as preliminary measurement constraint (Liu et al., 2010b) . For TROPOMI, we directly use the TROPOMI radiance random-noise errors constrain our retrievals. 

The high quality of reflectance is required to extract the independent information on the ozone profile, especially for the tropospheric ozone. The reflectance however is strongly affected by instrument performance and pre-launch calibration. Similar to OMI, 2-dimensional CCD array is used to measure TROPOMI radiance and irradiance and the CCD detector array is susceptible to systematic wavelength registration errors and cross-J o u r n a l P r e -p r o o f Journal Pre-proof 9 track dependent biases, which are likely to cause striping in ozone and other trace gas retrievals. To evaluate and eliminate such systematic biases, the "soft calibration," implemented in OMI (Liu et al., 2010b) , OMPS (Bak et al., 2017) and GOME-2 ozone profile retrievals, is also applied to TROPOMI radiances in this paper. This soft calibration is in general based on the comparison of measured radiances with expected values.

The keys to soft calibration are the choice of regions, the a priori profile and the use of daily means. The a priori ozone profile from TOA(top of atmosphere)to 215 hPa are derived from the latest version 4 Aura Microwave

Limb Sounder (MLS) instrument (Waters et al., 2006) . This version and earlier versions have been extensively validated (Liu et al., 2010a; Huang et al., 2017b) and been used as a priori profile by Liu et al. (2010b) and . We obtain a priori profiles by merging ozone climatology from McPeters et al. (2007) for pressure greater than 215 hPa with zonal mean MLS ozone profiles below 215 hPa. The tropical measurements were typically characterized where ozone change is relatively small (Bak et al., 2017) , and cloud-contaminated pixels and extreme viewing-geometries are screened out to reduce the errors in radiative transfer calculation, with the criteria: latitude < 10°N/S, cloud fraction < 0.2, and solar zenith angle < 40°.

The correction spectrum is the ratio of measured radiances and simulated radiances. The correction can be written as follows:

=

where is the measured radiance at wavelength and is the corresponding simulated radiance. In this paper, one day of TROPOMI measurements on 3 April 2018 is used to characterize ratios of measured to calculated radiances in the spectral range 305-360 nm (band-3) for all cross-track positions (Fig.1) . At wavelengths longer than 314 nm, the mean differences steadily decrease from ~10% with fluctuations at 314 nm to a few percent at ~350 nm, and increase sharply at wavelengths below 314 nm to 40-50 %. The standard deviations vary from 0.01-0.5% at wavelengths longer than 346.7 nm to 0.5-9.5% between 305-346 nm.

These spectral variations do not change much with cross-track position. Therefore, we do not use wavelengths below 314 nm for ozone profile fitting.

The cross-track dependent correction spectra shown in Fig. 1(a) are applied to TROPOMI radiances before the fitting starts. Fig. 2 (c) and (d) show comparisons of fitting residuals both with and without correction at all latitudes for band-3. Before applying the correction spectrum to TROPOMI radiances, the mean fitting residuals are within 1%, 0.8% and 0.45% for high-latitude, mid-latitude and tropics regions, respectively. The soft calibration remarkably reduces the fitting residuals for all latitudes, especially for tropics and mid-latitudes.

Fitting residuals are improved to 0.1-0.2 % in the tropics and 0.2-0.4% in the mid-latitudes. However, there are large systematic measurement biases in high latitudes with the fact that the derived soft calibration does not account for the dependence of systematic fitting residuals on the solar zenith angle. Therefore, our retrievals are still susceptible to these errors that vary latitudinally and seasonally. To determine the temporal variation of the soft calibration spectra, we derived the correction spectrum every month between April 2018 and December 2018. The mean differences can vary by up to ~2 % at center cross-track position and ~1 % at the edge crosstrack position from month to month, respectively, but the overall structure is similar and there is no noticeable time-dependent degradation. Therefore, we do not account for the time dependence in the soft calibration. We also calculate correction spectrum from 300-330 nm of band-2 and 270-330 nm of combination of band-1 and band-2 ( Fig. 2 (a), (b) ). However, we found that there are large systematic biases in radiance at wavelength shorter 290 nm and the retrieval residuals in the 314-330 range of band-2 are larger than those of band-3 ( Fig.2 (c), (d))). We also compare the retrievals, using spectral in the wavelength range 302.5-330 nm of band-2, with ozonesonde profiles and ground total ozone taken from WOUDC. The difference of retrieved and ground Total column Ozone (TO) using band-2 spectral are larger than that using band-3. The retrieved profiles in the spectral range 302.5-330 nm show larger positive biases at higher altitudes than those using band-3 (not shown here).

Journal Pre-proof To screen out TROPOMI data for comparisons, we only use TROPOMI data under the following conditions:

cloud fraction < 30 %, solar zenith angle < 75°. The collocation criteria is similar to that used by Huang et al. (2017a) . we first selected TROPOMI data with cloud fraction < 0.3, a time difference within 6 hours and within ±1° latitude, ±1°longitude from the ground site and then obtain the nearest TROPOMI pixel for comparison. The ozonesonde profiles are interpolated to the TROPOMI grid and then TROPOMI averaging kernels are convolved with the interpolated profile to remove smoothing errors. After filtering out TROPOMI data that does not meet the above conditions, approximately 120 coincidence pairs are used in the comparison. Fig. 5 shows the comparisons of the retrieved, a priori and FTS TOC with and without TROPOMI AK. The integrated TOC shows good agreement with FTS measurements. The a priori TOC is much lower than the FTS TOC with the mean bias of -16.9 DU and standard deviation of 12.01 DU. The correlation coefficient between FTS and a priori TOC is 

We validate TROPOMI ozone profile, TOC and SOC retrievals with global ozonesonde data except for in the northern high latitudes from 2018 to 2019. There are no sufficient ozonesonde data for statistical analysis in the northern high latitudes. We only used ozonesonde profiles with correction factors between 0.8 and 1.15. The tropopause pressure derived from TROPOMI and ozonesonde burst pressure are used to calculate TOC and SOC.

The column concentration integrated from the surface to the tropopause is defined as TOC and the SOC is defined as the integral of ozone profile from the tropopause pressure to the burst pressure.

We selected 5 stations to compare ozone profiles between ozonesondes and the retrieved (Fig. 7) . The a priori profiles show a large negative bias, with a maximum negative bias of -80%, below 30 hPa at most stations.

TROPOMI agrees with ozonesonde to within ±5 % from 0.8-30 hPa and within ± 15 % below 30 hPa.

TROPOMI retrievals show significant reduction in mean biases over the climatological profiles below 30 hPa.

The retrieved ozone profiles above 20 hPa show worse agreement with ozonesonde than the a priori profile mainly due to not using information below 314 nm. (Sterling et al., 2018) . 20-day average of TROPOMI NO 2 (Copernicus Sentinel-5P (processed by ESA), 2018a), HCHO (Copernicus Sentinel-5P (processed by ESA), 2018b) TVCD and the retrieved TOC before and during the lockdown of Wuhan on January 23, 2020 (hereafter referred to as the "pre-lockdown" and "post-lockdown" periods) are shown in Fig. 8 . NO 2 TVCD over east of China fell by as much as 63 % and HCHO TVCD slightly reduced by 6 % from pre-lockdown period to lockdown period, respectively. However, the TROPOMI TOC over east of China increase by 10 % due to the temperature rise and the uncoordinated reduction of NO 2 and HCHO TVCD.

Tropospheric ozone is mainly formed through photochemical reactions of nitrogen oxidation (NOx=NO+NO 2 ) indicated by NO 2 and volatile organic compounds (VOCs) indicted by HCHO (Sillman, 1995) . However, the interrelations among ozone, NOx and VOCs are nonlinear. The chemical formation of ozone is controlled by NOx or VOCs, namely NOx-limited or VOC-limited regimes, depending upon which substance is inadequate in the reactions (Chameides et al., 1992) . Because ozone production in the winter in China is usually controlled by VOCs, slight decrease of VOCs in eastern China may not cause an obvious declining of ozone formation from photochemical reaction (Lu et al., 2019) . Besides, the decrease of NO x emission would reduce the concentration of NO, which can titrate ozone in the ambient (Li et al., 2016; Zou et al., 2019) , causing more accumulation of ozone. Especially, in the BTH region, the NO 2 TVCD sharply reduce by 77% much higher than the average decreasing amplitude in the eastern China; however, HCHO TVCD even increased by 18 %. This variation tendency of NO x and VOCs caused TOC in BTH region greatly increased by 17 %. The Yangtze River Delta (YRD) and Pearl River Delta (PRD) regions show similar ozone pollution characteristics to eastern China.

The variation of tropospheric ozone in eastern China during the outbreak of COVID-19 was evaluated using satellite-based observation. We have adapted and modified the OMI ozone profile algorithm to retrieve TROPOMI ozone profiles. Compared to the derived on-orbit ISRFs assuming standard Gaussian and super Gaussian, the pre-flight ISRF measurements (TROPOMI ISRFs v3.0.0) is demonstrated as a best representative in term of the spectral fitting accuracy.

To eliminate the effect of systematic measurement errors in the TROPOMI radiances that depend on wavelength and cross-track positions, we apply "soft calibration" to TROPOMI radiances before the fitting starts.

J o u r n a l P r e -p r o o f Journal Pre-proof 21 After applying the correction spectrum to TROPOMI, the fitting residuals have been significantly improved for five latitude bands, especially in the tropics and mid latitudes. The fitting residuals reduce by ~71 % and 46 % in the low and middle latitudes, respectively.

We have validated our retrievals against FTS data and GPSO 3 ozonesonde data observations in China between March 2018 and December 2019. The integrated TOC shows excellent agreement with FTS and GPSO 3 TOC after the application of TROPOMI AKs, respectively. The comparisons of the retrieved and ozonesondes profiles show that the retrievals have better consistency with the ozonesonde profile than a priori at most of selected stations, especially in the troposphere. However, the mean biases and standard deviations between TROPOMI and ozonsondes profiles above 20 hPa are larger than those between a priori and ozonesondes profiles partly due to missing information below 314 nm. The retrieved SOC show good agreement with ozonesonde data.

The validation of integrated TOC with applying TROPOMI AKs against ozonesonde shows that TROPOMI is in very good agreement with ozonesonde within 2.4 DU at all latitudes. The correlations are greater than 0.922 except for 0.703 in the tropics partly due to a "drop off" in SHADOZ data after ~2014.

Based-on the retrieved TOC as well as TCVD of NO 2 and HCHO from TROPOMI, we evaluated the change of ozone in the lockdown period during which special measures were conducted to prevent the spread of COVID-19 and also caused large decrease of primary pollutant emission. The retrieved TOC over east of China increased by 10 % compared with that during the pre-lockdown period, due to the uncoordinated reduction of NO 2 and HCHO TVCD. In the BTH region, the NO 2 TVCD sharply reduce by 77 % but HCHO TVCD increased by 18 %.

This variation tendency of NO x and VOCs caused TOC in BTH region greatly increased by 17 %. It indicated that in the VOC-limited region, sharp reduction of NO x emission took no role on the mitigation of ozone pollution, but even caused the increase of ozone concentration.

Improvement of OMI ozone profile retrievals in the upper troposphere and lower stratosphere by the use of a tropopause-based ozone profile climatology

Characterization and Correction of OMPS Nadir Mapper Measurements for Ozone Profile Retrievals

Ozone and short-term mortality in 95 US urban communities

Algorithm for the estimation of vertical ozone profiles from the backscattered ultraviolet technique

High-resolution laboratory absorption cross section of O 3 . Temperature effect

Characterization and correction of Global Ozone Monitoring Experiment 2 ultraviolet measurements and application to ozone profile retrievals

Ozone precursor relationships in the ambient atmosphere

The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak

The Global Ozone Monitoring Experiment: Review of in-flight performance and new reprocessed

TROPOMI Level 2 Nitrogen Dioxide total column products. Version 01

TROPOMI Level 2 Formaldehyde Total Column products. Version 01

OMPS-NPP L2 LP Ozone (O 3 ) Vertical Profile swath daily 3slit V2

Prelaunch characterization of the Ozone Monitoring Instrument transfer function in the spectral domain

Ozone profiles from GOME satellite data: Algorithm description and first validation

Validation of 10-year SAO OMI Ozone Profile (PROFOZ) product using ozonesonde observations

Validation of 10-year SAO OMI Ozone Profile (PROFOZ) Product Using Aura MLS Measurements

Pre-launch calibration results of the TROPOMI payload on-board the Sentinel-5 Precursor satellite

Unexpected air pollution with marked emission reductions during the COVID-19 outbreak in China

Impacts of thermal circulations induced by urbanization on ozone formation in the Pearl River Delta region, China

Some Unique Aspects of Air Pollution in Los Angeles

The impact of using different ozone cross sections on ozone profile retrievals from OMI UV measurements

Impact of using different ozone cross sections on ozone profile retrievals from Global Ozone Monitoring Experiment (GOME) ultraviolet measurements

Validation of Ozone Monitoring Instrument (OMI) ozone profiles and stratospheric ozone columns with Microwave Limb Sounder (MLS) measurements

Ozone profile retrievals from the Ozone Monitoring Instrument

Overview on the spatial-temporal characteristics of the ozone formation regime in China

Ozone climatological profiles for satellite retrieval algorithms

The effects on human health from stratospheric ozone depletion and its interactions with climate change

The effects of ozone on human health

Inverse Methods for Atmospheric Sounding -Theory and Practice

Postlaunch performance of the Suomi National Polar-orbiting Partnership Ozone Mapping and Profiler Suite (OMPS) nadir sensors

The Response in Air Quality to the Reduction of Chinese Economic Activities During the COVID-19 Outbreak

The use of NOy, H2O2 and HNO 3 as indicators for ozone-NOx-hydrocarbon sensitivity in urban locations

Impact of rotational Raman scattering in the O2A band

Homogenizing and estimating the uncertainty in NOAA's long-term vertical ozone profile records measured with the electrochemical concentration cell ozonesonde

Ozone seasonal evolution and photochemical production regime in the polluted troposphere in eastern China derived from high-resolution Fourier transform spectrometry (FTS) observations, Atmospheric Chemistry and Physics

Comparisons With Satellites and Ground-Based Instruments

Characterisation of methane variability and trends from near-infrared solar spectra over Hefei, China

Ozone profile retrieval from recalibrated Global Ozone Monitoring Experiment data

TROPOMI on the ESA Sentinel-5 Precursor: A GMES mission for global observations of the atmospheric composition for climate, air quality and ozone layer applications

A preliminary assessment of the impact of COVID-19 on environment -A case study of China

Tropospheric ozone trend over Beijing from 2002-2010: ozonesonde measurements and modeling analysis

The Earth observing system sounder (EOS MLS) on the aura Satellite, IEEE Transactions on Geoscience and Remote Sensing

First reprocessing of Southern Hemisphere ADditional OZonesondes (SHADOZ) profile records

First Reprocessing of Southern Hemisphere ADditional OZonesondes Profile Records: 3. Uncertainty in Ozone Profile and Total Column

Observations of the vertical distributions of summertime atmospheric pollutants and the corresponding ozone production in

Possible environmental effects on the spread of COVID-19 in China

Indirect effects of COVID-19 on the environment

Satellite UV-Vis spectroscopy: implications for air quality trends and their driving forces in China during

First observation of tropospheric nitrogen dioxide from the Environmental Trace Gases Monitoring Instrument onboard the GaoFen-5 satellite

Development and preliminary evaluation of a double-cell ozonesonde

Assessing the relationship between ground levels of ozone (O 3 ) and nitrogen dioxide (NO 2 ) with coronavirus (COVID-19

The weekday/weekend ozone differences induced by the emissions change during summer and autumn in Guangzhou, China

Fei Zhao: Software, Writing-Original draft preparation

Methodology. Xiong Liu: Writing-Reviewing and Editing. Juseon Bak: Writing-Reviewing and Editing. Jae Kim: Writing-Reviewing and Editing. Qihou Hu: Writing-Reviewing and Editing. Congzi Xia: Data Curation. Chengxin Zhang: Data Curation. Youwen Sun: Data Curation. Wei Wang: Data Curation

for providing the FTS data, OMPS science team for providing OMPS-LP data. We thank WOUDC for proving the ozonesonde data. We are grateful to TROPOMI science team for providing level 1 data. J o u r n a l P r e -p r o o f Journal Pre-proof