Evaluation of Transport Processes over North China Plain and Yangtze River Delta using MAX-DOAS Observations

. Pollutant transport has a substantial impact on the atmospheric environment in megacity clusters. However, owing to the lack of knowledge of vertical pollutant structure, quantification of transport processes and understanding of their impacts on the environment remain inadequate. In this study, we retrieved the vertical profiles of aerosols, nitrogen dioxide (NO 2 ), and formaldehyde (HCHO) using multi-axis differential optical absorption spectroscopy (MAX-DOAS) and analyzed three typical transport phenomena over the North China Plain (NCP) and Yangtze River Delta (YRD). We found the following: (1) The 25 main transport layer (MTL) of aerosols, NO 2 and HCHO along the southwest–northeast transport pathway in the Jing-Jin-Ji


Introduction
With rapid economic development, urbanization in China has increased.Many cities of different scales have recently emerged, forming many megacity clusters such as Jing-Jin-Ji (JJJ) and the Yangtze River Delta (YRD).With the this rapid pace of urbanization, air pollution has become one of the most serious environmental threats that China must address.Heavy air pollution adversely impacts affects every aspect of human life, including climate, air visibility, and human health (Pokharel et al., 2019;Gao et al., 2017;Su et al., 2020a;Li et al., 2017a).
Currently, air pollution sources can be roughly broadly classified into as direct emissions, secondary production, and transport.
Transport remarkably contributes significantly to pollution in some megacities.Firstly, transportation carries large amounts of pollutants, directly deteriorates deteriorating the environment through the production and emission of a large number of pollutantsair quality.Regional transport plays an important predominant role in pollution formation in many major cities in China, such as Beijing, Shanghai, Guangzhou, Hong Kong, Hangzhou, and Chengdu, contributing more than 50% of the particulate matter, PM2.5 , during polluted periods (Sun et al., 2017).For In the JJJ region, regional transport from southwest to northeast, driven by the southwestsouthwesterly winds, is an the important factor influencingdominant influence on the daytime increase in PM2.5 and ozone (O3) concentrations (Ge et al., 2018).In addition to regional transport, cross-regional transport has a important significant impacts.For example, from 2014 to 2017, intra-and inter-regional transport accounted for 25% and 28% of the total PM2.5 in the JJJ region, respectively,, from 2014 to 2017, while the local contributions was 47% (Dong et al., 2020).During the 2019 National Day parade, cross-regional dust contributed more than 74% of Beijing's particulate matter (PM) concentrations below 4 km (Wang et al., 2021).Furthermore, under certain conditions, some transported pollutants can interact with the planetary boundary layer (PBL) and create an environment favorable for direct emission accumulation and secondary formation enhancement, thereby indirectly amplifying the impacts of pollution (Li et al., 2017b;Wilcox et al., 2016;Petaja et al., 2016).A typical example is the aerosols from the YRD being transported to the upper PBL over the North China Plain (NCP), which decreases the PBL heights, and increasesing pollutant accumulation (Huang et al., 2020).The movement of warm and humid air masses likely increases secondary aerosol formation by aggravating aqueous and heterogeneous reactions (Huang et al., 2014).Hence, we must understand the air pollutant transport that occurs in megacity clusters by using an appropriate measurement method.
The cCurrent technological means of monitoring and analyzing air pollution mainly include in situ measurements, satellite observations, model simulations, and ground-based remote sensing monitoring.By 2021, the number of China National Environmental Monitoring Centers (CNEMCs), which that provide in situ measurements, had extended to 2734, forming a complete comprehensive and mature air quality monitoring network.The CNEMCs monitor many kinds of pollutants, including sulfur dioxide (SO2), carbon monoxide (CO), nitrogen dioxide (NO2), PM10, PM2.5, and O3.However, the pollutant concentrations monitoring monitored by the CNEMCs are limited to the ground levelsurface.Characterize Characterizing 带格式的: 非上标/ 下标 pollutants in the upper-level air column using surface observations is difficult (Huang et al., 2018b) because various factors, including local emissions, regional transport, geographical factors, and meteorological conditions, need tomust be considered (Tao et al., 2020;Che et al., 2019).Therefore, the vertical distribution of pollutants cannot be diagnosed using only tthe CNEMC dataset alone.Satellite observations can be used to investigate the horizontal distribution of vertical column densities (VCDs) of NO2, formaldehyde (HCHO), O3, and aerosols on a global scale, to provideproviding technological support for horizontal pollutant transport analysis.However, because of their limited temporal and spatial resolutions, the data from satellite datas cannot be used for the continuous monitoring of a specific area (Bessho et al., 2016;Veefkind et al., 2012).
Furthermore,I satellite remote sensing data cannot be used tot is difficult to monitor characterize the vertical features distribution of and variations in the atmospheric composition using only satellite remote sensing or CNEMC data (Yumimoto et al., 2016;Su et al., 2020b;Bessho et al., 2016).The cChemical transport models can be used to simulate pollutant distribution, and they is anare also important tools for monitoring, forecasting, and analyzing atmospheric quality (Huang et al., 2018a).
However, considerableLarge uncertainties remain in estimating pollutant distribution estimation using model simulations, primarily owing to the effects of emission inventories, meteorological fields, and of hypothetical conditionsassumptions made (Grell et al., 2005;Huang et al., 2016;Xu et al., 2016;Zhang et al., 2017).Moreover, model simulations are incapable ofcannot completely characterizing characterize the air composition profiles because of the inadequate modeling of atmospheric pollutants in the vertical direction.To meet the need for to understanding the vertical distribution of air pollutants, some technical monitoring methods have been developed, such as light detection and ranging (LiDAR) (Collis, 1966;Barrett and Ben-Dov, 1967;Wang et al., 2020) and in situ monitoring instruments carried by aircraft, balloons, or unmanned aerial systems (UASs) (Corrigan et al., 2008;Tripathi et al., 2005;Ferrero et al., 2011).Nevertheless, the number of detectable pollutants is limited for a single LiDAR device, and a single set is expensive.Alternatively, monitoring based on moving platforms requires substantial labor and material resources, which prevents continuous observations.The differential optical absorption spectroscopy (DOAS) technique (Platt and Stutz, 2008) is a cutting-edge and promisingwellestablished and reliable method for the quantitative analysis of many crucial atmospheric gases.The DOAS method uses highfrequency molecular absorption structures in the ultraviolet (UV) and visible regions of the spectrum.Multiaxis differential optical absorption spectroscopy (MAX-DOAS), which employs the DOAS technique at multiple elevation angles, is a tool used for long-term atmospheric quality monitoring (Hönninger et al., 2004).Combined with radiative transfer modeling, MAX-DOAS can be used to retrieve the vertical profiles of aerosols and trace gases based on scattered sunlight signals from multiple elevation angles (Frieß et al., 2006).This method has been widely used for to retrieveing aerosols, HCHO, NO2, O3, and glyoxal (CHOCHO) concentrations (Hönninger and Platt, 2002;Meena, 2004;Wagner et al., 2004;Frieß et al., 2006;Hönninger et al., 2004;Irie et al., 2008;Xing et al., 2020;Hong et al., 2022b).Compared with the above techniques, MAX-DOAS is high resolution anddoes not require radiometric calibration and has many other advantages such as simple design, low power demand, possible automation, low cost, and its operation is automaticminimal maintenance.Moreover, the MAX-DOAS is capable of regularly operating regularly in harsh environments, such as those over the Tibetan Plateau (Xing et al., 2021a).On this basis, we have established nearly approximately 30 MAX-DOAS stations covering seven regions in China (north, east, south, northwest, southwest, northeast, and cCentral China) to form a mature ground-based hyperspectral stereoscopic remote sensing network (Xing et al., 2017;Liu et al., 2021;Hong et al., 2022a).The data provided by theThis monitoring network successfully meets the actual demands for vertical observations, and the network providesproviding powerful technical support for analyzing pollution sources and transport (Liu et al., 2022).
In this study, we aimed to understand the vertical distribution characteristics of air pollutants during the transport process and analyze their possible impacts on and between regions.The remainder of this paper is structured as follows.: Section 2 describes the stations, instruments, algorithms, and ancillary data we used in the study.In Section 3, we discuss three typical transport processes (regional, dust, and transboundary long-range transport).Finally, we present a the summary and conclusions in Section 4. We think that our results will help the public gain a comprehensive understanding of the pollutant transport process and provide reference values for optimization policies and regulations.

Geographical locations and selected stations
Our study focused mainly on the transport phenomena in the NCP and YRD, two of the main plains within ChinaThe analyzed transport phenomena mainly occurred in the NCP and YRD, which are two of the main plain areas in China.The NCP is partially enclosed by Mt.Taihang, Mt.Yan, and the Bohai Sea, whereas the YRD is close to the Yellow Sea and East China Sea.Many megacities are located in these two regions (i.e., Beijing and Tianjin in the NCP, and Shanghai and Nanjing in the YRD).Beijing, Tianjin, and the entire Hebei Province form large megacity clusters within the NCP, named the Beijing-Tianjin-Hebei (BTH) region or the JJJ region.Owing to the numerous industrial industries, combined withfactories and traffic emissions, the JJJ region is one of the most polluted areas in China.In addition, the JJJ has a typical continental monsoon climate, indicating that wind plays an important role in the local climate and environment.The semi-basin geographical features and continental monsoon climate indicate that regional transport is a significant factor affecting local air quality in the JJJ region.The regional transport of pollutants is prevalent within the JJJ region, which exerts serious effects on local air quality.Similarly, the YRD is affected by many several local pollution sources and by pollutant transport.Therefore, we selected eight MAX-DOAS stations in the NCP and YRD to explore the corresponding transport phenomena.

Instrument setup
We operated a eight commercial MAX-DOAS instruments (Airyx SkySpec-1D, Heidelberg, Germany) from January 1 to March 31, 2021, which consistinged of three major components: two spectrometers inside a thermoregulated box, a telescope unit, and a computer for instrument control and data storage.One spectrometer covered the UV wavelength range (296-408 nm), and the other worked in the visible region (420-565 nm), with a spectral resolution of 0.45 nm.The sScattered sunlight was was collected by a telescope and then directed to the spectrometer through a prism reflector and quartz fiber.The instrument automatically recordeded the spectra of scattered sunlight at sequences that we set to 11 elevation angles, that is,: 1°, 2°, 3°, 4°, 5°, 6°, 8°, 10°, 15°, 30°, and 90°.The duration of each measurement sequence was was approximately 5-15 min, depending on the received radiance.Additionally, the setup only collected collected scattered sunlight in theduring daytime, whereas the dark current and offset spectra were were automatically measured at night and removed from all the measured spectra before data analysis.To avoid the strong impact influence of stratospheric absorbers, we filtered the measured spectra measured with a solar zenith angle (SZA) of > 75° (Supplementary Sect.S1).(Aliwell, 2002);

Data processing 160
We analyzed the spectra using DOAS Intelligent System (DOASIS) spectral fitting software, which is based on the least squares algorithm (Kraus, 2006).Spectral analysis derives the slant column densities (SCDs), i.e., the integrated concentration along the light path.FirstSubsequently, we measured calculated the differential slant column densities (DSCDs), which is are defined as the difference between the off-zenith and zenith slant column densities (SCDs).Furthermore, we calculated the ring spectrum as the measured spectrum, considering the contribution of the stratosphere to the DSCDs.We analyzed the DSCDs 165 of the oxygen dimer (O4) and NO2 in the interval between 338 and 370 nm, and we used the 322.5-358 nm and 335-373 nm wavelength intervals for HCHO and nitrous acid (HONO) absorption analysis, respectively (Xing et al., 2020;Xing et al., 2021b).We have used similar retrieval settings in our previous studies (Xing et al., 2020;Xing et al., 2021b).Table 1 2 lists the detailed DOAS fitting settings for O4, NO2, HCHO, and HONO in detail.The Ring spectrum was added to the fitting settings to remove the influence of the stratosphere on the DSCDs.In the fitting process, wWe included a small shift or squeeze of the wavelengths in the fitting process to compensate for the possible instability caused by small thermal variations in the spectrograph.Figure .S1 shows a typical DOAS fit for the four species.To ensure the validity of the retrieved data, we removed the DOAS fit results with a root mean square (RMS) larger than 1.0 × 10 -3 .After applying the RMS threshold, the results for O4, NO2, HCHO, HONO remained at 69.8%, 71.6%, 64.8%, and 73.1% respectively.We assumed two times the fitting error RMS as the DSCD detection limits (Wang et al., 2017;Lampel et al., 2015), which were 7 × 10 41 (molec 2 • cm -5 ), 1.6 × 10 15 , 3.6 × 10 15 , and 5.8 × 10 14 molec• cm -2 for O4, NO2, HCHO, and HONO, respectively.Moreover, to remove the effects of clouds, we only used only data with slowly varying O4 DSCDs and intensities for the vertical profile retrieval (Supplementary Sect.S2).The detailed filtering methods is provided by Chan et al. (2019).The inversion algorithm we used for aerosols and trace gases is was based on the optimal estimation method (OEM), and; we selected the radiative transfer model (RTM) library for radiative transfer (libRadtran) as the forward model (Mayer and Kylling, 2005).We selectedBy minimizing the cost function  2 , to we determined the maximum a posteriori state vector :.
where (, ) is the forward model;  is denotes the meteorological parameters (i.e.g., atmospheric pressure and temperature profiles);  is the measured DSCDs;   is the a priori vector that serves as an additional constraint;   and   are the covariance matrices of  and   , respectively.We classified the retrieval of vertical profiles of aerosols and trace gases in two steps.As O4 absorption is closely linked to the optical properties of aerosols, our first step was to retrieveing the vertical aerosol profiles based on the measured retrieved O4 DSCDs at different elevation angles (Wittrock et al., 2003;Frieß et al., 2006;Wagner et al., 2004).In the previous study, we proposed a new method to quickly semi-quantify the aerosol scattering and absorption properties directly based on the observed UV and visible O4 absorption (Xing et al., 2019).In the second step, using the retrieved aerosol extinction profiles as the input parameter to the RTM, we obtained the NO2, HCHO, and HONO vertical profiles.For this retrievalIn this study, we separated the atmosphere into 20 layers from 0 to 3 km with a vertical resolution of 0.1 km under 1 km, and of 0.2 km from 1 to 3 km.For the aerosol profile retrieval, we selected an exponentially decreasing profile with a scale height of 0.5 km as a priori and set its aerosol optical density (AOD) to 0.4.For aerosol profile retrieval, we selected exponentially decreasing a priori with a scale height of 0.5 km.For the a priori trace gas profile, We we set the bottom layer concentration to 8 × 10 10 molec•cm -3 , and set the VCD to 15 × 10 15 , 15 × 10 15 , and 5 × 10 14 molec•cm -2 for NO2, HCHO, and HONO, respectively.a priori surface aerosol extinction to 0.2 km -1 .We set all the a priori uncertainties of the aerosols and trace gases, NO2, and HCHO to 50%, and HONO to 100%, set with the a correlation height to of 0.5 km.
During the retrieval, we employed a fixed set of aerosol optical properties with a single-scattering albedo of 0.95, an asymmetry parameter of 0.70, and a surface albedo of 0.04. Figure S2 shows the averaging kernels, which indicate that the retrieval profile was sensitive to the layers within 0-1.2 km.The sum of the diagonal elements in the averaging kernel matrix is the degrees of freedom (DOF), which denotes the number of independent pieces of information that can be measured.A more detailed description of the retrieval process can be found in previous studies (Chan et al., 2019;Chan et al., 2018).The profiles of aerosols and trace gases were filtered out when the degree of freedom (DOF) was less than 1.0 and the retrieved relative error was larger than 50% (Tan et al., 2018).About 0.5 %, 10.7 %, and 11.6 % of all measurements were discarded for aerosol, NO2 and HCHO profile retrievals, respectively.A more detailed description of the retrieval process can be found in previous studies (Chan et al., 2019;Chan et al., 2018).

Error analysis
For profile retrieved results, we conducted the an error analysis on the trace gas VCDs and aerosol optical density (AOD), and near-surface (0--100 m) trace gas concentrations and aerosol extinction coefficients (AECs).The error sources considered are listed below and the final results are summarized in Table 23.Detailed demonstrations and calculation methods can be found in Supplementary Sect.S3.
a. Smoothing and noise errors refers to the errors caused by the limited vertical resolution of profile retrieval, and the fitting error of DOAS fits, respectively.By calculating the averaged error of retrieved profiles, we obtaineded the sum of smoothing and noise errors on near-surface concentrations and column densities, which are were 24 and 5 % for aerosols, 11 and 19 % for NO2, and 42 and 25 % for HCHO.
b. Algorithmic error (points to an imperfect minimum of the cost function, i.e., the difference between the measured and modeled DdSCDs) arises from an imperfect minimum of the cost function.This error is a function of the viewing angle.
However, it is difficult to assign discrepancies between the measured and modeled DSCDs at each profile altitude.
Therefore, the algorithm error on the near-surface values and column densities cannot be realistically estimated.Given that measurements at 1° and 30° elevation angles are sensitive to the lower and upper air layers, respectively, the average relative differences between the measured and modeled DSCDs for a 1 and 30° elevation angles can be used to estimate the algorithm errors on the near-surface values and column densities, respectively (Wagner et al., 2004).Based on the fact that measurements at 5° and 30° elevation angles are sensitive to the lower and upper air layers, respectively, the average relative differences between measured and modeled dSCDs for a 5 and 30° elevation angle are usually utilized to estimate the algorithm error on the near-surface values and column densities, respectively.Notably, algorithm error cannot be realistically estimated because it is difficult to assign the differences between measured and modelled dSCDs to each height of the profile.Considering its trivial role in the total error budget, we estimated these errors on the near-surface values and the column densities at 4 and 8 % for aerosols, 3 and 11 % for NO2, and 4 and 11 % for HCHO, respectively, according to Wang et al. (2017).
d.The errors related to the temperature dependence of the cross sections can be estimated as followswere then estimated.
We multiplied the amplitude changes of the cross sections per Kelvin by the maximum temperature difference to quantify this systematic error.With two cross sections at two temperatures, we firstly calculated the amplitude changes of the cross sections per kelvin.Subsequently, we multiplied this with the variation magnitude of the surrounding temperature.Due toGiven that the measurement period (from January 1 to March 31, 2021) was is at the end of winter and the beginning of the springin the winter-spring season, we conduct a rough estimation ofroughly estimated the maximum temperature gap difference as to be 45 K.The corresponding errors of for O4 (aerosols), NO2, and HCHO are were around approximately 10, 2, and 6 %, respectively.
e.The trace gas retrieval errors, associated witharising from the uncertainty in effects of aerosol layersaerosol retrieval, wereare estimated as the total error budgets of the aerosols retrievals.These estimations are bBased on a a linear propagation of the aerosol errors, the errors of trace gases were roughly estimated at 27% for VCDs and 14% for nearsurface concentrations for the two trace gases on the trace gas retrievals, which is a rough assumption.The perturbations of trace gas concentrations at each altitude caused by aerosol profile retrieval uncertainty resulted in a slight change in the profile shape.According to Friedrich et al. (2019), trace gas concentrations at 1.5-3.5 km respond most sharply to perturbations in the AEC profile, especially oscillations in the AEC below 0.5 km.The trace gas profile below 1.5 km shows a low sensitivity to AEC variation.Therefore, in this study, we focus mainly on the concentration variation below 1.5 km.Given the uncertainties on aerosol properties and profiles, additional sensitivity tests should be performed for different observation geometries to derive more realistic error estimates.
We calculated tThe total error is calculated by combining all the error terms in the Gaussian error propagation, and listed which are listed in the bottom row of Table 23.We can find that sSmoothing and noise errors plays played a dominant role in the total error estimation.

Transport flux calculation and main transport layer definition
Owing to the semi-basin topography, southwesterly or southerly winds play a dominant role in Given the major role of pollutant transport in the JJJ region.,In this study, we mainly focused on pollutant transport in the southwest-northeast direction, and thus selected four different stations along this pathway, namely, Shijiazhuang (SJZ), Wangdu (WD), Nancheng (NC), and Chinese Academy of Meteorological Sciences (CAMS) (Fig. 1).We calculated the hourly transport fluxes of each layer (  ) and column transport fluxes (  ) at each station were calculated to illustrate the dynamic transport process of pollutants along the southwest-northeast pathway.The detailed calculation methods was are described belowbelow.: First, the wind speed projection (WS) in the southwest-northeast direction (WS) was was calculated as follows: where  and  represented the modeling meridional wind component and zonal wind components, respectively;. above zero meant means that the wind came from the southwest and blew northeast, while whereas  below zero had has the opposite meaning.
Then, with , the   could can be obtained: Here,   denoted denotes the aerosolAEC or trace gas concentration at the altitude of the corresponding wind speed, while whereas   represented represents the wind speed in layer i from southwest to northeast.A flux above zero indicated indicates that the air pollutant was is transported from southwest to northeast, while whereas a flux less than zero meant means that the transmission direction was is from northeast to southwest.
Finally, we calculatedd the   per unit width by summing the   multiplied by the height of each layer: ( ) where   was is the height of each layer i.
For the convenience of description, the layer with the highest transport flux waswas defined as the main transport layer (MTL) for the corresponding pollutants.Equation 3 demonstrated thatAccording to the definition and Eq. 3, we know that the MTL was is determined by the concentration and wind speed in the corresponding layer.Due Owing to the large discrepancy differences in the vertical distributions of various pollutantstheir vertical distribution, the their MTLs of various pollutants were were bound to have different varying characteristics.Some calculation details and error analysis methods are provided in Supplementary Sect.S4.

Ancillary data
We obtained the surface NO2, PM2.5, CO, and O3 concentrations from the CNEMCs with a sampling resolution of 1 h (https://quotsoft.net/air/).We validated the MAX-DOAS measurements by comparing the lowest layer results from the MAX-DOAS observations with the CNEMC data.Using the CO and O3 concentrations, we performed source apportionment of ambient HCHO to identify the contribution ratios of primary and secondary HCHO.Moreover, we depicted a spatiotemporal distribution of PM2.5, reflecting surface PM2.5, and concentration variations during the transboundary transport process.We obtained the spatial distributions of NO2 and HCHO from TROPOMI at a spatial resolution of 3.5 × 7.0 km (Veefkind et al., 2012), and the spatial distributions of AOD and dust from Himawari-8 with a 0.5 ×2.0 km spatial resolution and a 10 min temporal resolution (Bessho et al., 2016).Satellite observations helped identify the pollutant transport phenomena because transport tends to cause large-scale continuous distribution of pollutants that can be detected by satellite measurements.
We simulated the wind speed and direction using the Weather Research and Forecasting Model, Version version 4.0 (WRF 4.0).See Supplementary Sect.S1 S5 which detailss the model and parameter settings.In terms of wind speeds and pollutant mixing ratios in different layers, we calculated transport fluxes at different heights to reflect the dynamic transport processes of various pollutants.In addition, we used wind-field information to reveal the transport direction at different altitudes.
We obtained the spatial distributions of NO2 and HCHO from TROPOMI at a spatial resolution of 3.5 × 7.0 km (Veefkind et al., 2012), and we obtained the spatial distributions of aerosol optical density and dust from Himawari-8 with a 0.5 ×2.0 km spatial resolution and a 10 min temporal resolution (Bessho et al., 2016).We calculated the 24-h backward trajectories of the air masses using the hybrid Hybrid singleSingle-particle Particle Lagrangian integrated Integrated trajectory Trajectory (HYSPLIT) model (Supplementary Sect.S2S6).In our study, the 24-h backward trajectories were calculated to investigate the dust origins and pathways that reached the NCP on March 15, 2021.

Results and discussion
For validation, we comparedd the surface NO2 concentrations and aerosol extinction coefficientAECs from MAX-DOAS measurements from January to March 2021 with in situ NO2 and PM2.5 collected by the CNEMCs.We calculated the O4 effective optical path as the distance threshold (~ 5 km) to exclude some MAX-DOAS stations from the correlation analysis 14 (Supplementary Sect.S7).Table S1 lists the selection conditions.Furthermore, we filtered the "abnormal values" of MAX-DOAS and in situ measurements before comparison, which was favorable to lessen the effects of occasional extreme conditions and improve the correlation (Supplementary Sect.8).As displayed in Fig. 12, we found found good agreement between MAX-DOAS and in situ data, with Pearson correlation coefficients (R) of 0.615 752 and 0.674 for aerosol and NO2, respectively.
The fine correlation between AOD from MAX-DOAS and AERONET also confirmed the reliability of this measurement, with R reaching 0.941 and 0.816 for CAMS and XH, respectively (Fig. S3).We excluded some stations from the comparison analysis because their distance from the CNEMCs exceeded 10 km.Table S1 lists the detailed selection conditions.
Furthermore, tThe aerosol information that we directly obtained directly from the MAX-DOAS was is the AECs.Under most conditions, aerosol mass concentration is approximately proportional to the extinction coefficient (Charlson, 1969;Robert et al., 1968).However, they were are not completely equivalent; relative humidity (RH) influenced influences their correlation (Lv et al., 2017), which could explain the weaker aerosol correlation.

Dynamic transport processes of NO2, HCHO, and aerosol
Impacted by the semi-basin topography and continental monsoon climate, intra-regional transport in the JJJ region is frequent and is an important dominant factor influencing the environmental air quality of many cities.Based on in situ measurements in the JJJ region, southwest-to-northeast regional transport was found to play an important significant role in increasing PM2.5 and O3 levels (Ge et al., 2018).In addition, a south-to-north transport belt exists in this region (Ge et al., 2012).Using WRF-Chem simulations, Wu et al. (2017b) successfully evaluated the contributions of regional transport to elevated PM2.5 and O3 concentrations in Beijing during summer.Based on vertical LiDAR observations, Xiang et al. ( 2021) revealed that PM2.5 was transported to Beijing via the southwest pathway.However, it is difficult to fully understanding regional transport using model simulations or in situ measurements is difficult.Therefore, we combined MAX-DOAS measurements with WRF simulations to more accurately describe the regional transport processes of aerosols, NO2, and HCHO more accurately.According to tThe TROPOMI results, we found that indicated that NO2 was was continuously homogeneously distributed between SJZ and WD, whereas and that, on February 5, 2021, thea HCHO distribution belt connected connected NC with CAMS on February 5, 2021 (Fig. S2S4).Figure .S3 S5 displays shows the regional wind information in different layers (0-20, 200-400, 400-600, and 600-800 m), with the wind direction being mainly southwest-northeast among the four stations (i.e., SJZ, WD, NC, and CAMS).These findings revealed a typical regional transport process along the southwest-northeast pathway.Figure 3 presents We also identified this southwest-northeast regional transport process in the temporal variations in the vertical distributions of aerosols, NO2, and HCHO during this regional transport(Fig.3).At the CAMS and NC stations, the aerosol, NO2, and HCHO concentrations were consistently high near the surface, primarily because of the heavy traffic flow and dense factory emissions in Beijing (Zhang et al., 2016;Li et al., 2017).Previous studies have suggested that urban air pollution in Beijing is dominated by a combination of coal burning and vehicle emissions, which results in severe particulate pollution (Wang and Hao, 2012;Wu et al., 2011).At SJZ, the NO2 concentration was high (~ 12 ppb) in the morning and late afternoon, whereas the concentration was lowest (~ 6 ppb) near noon, which is explained by the morning and evening rush hour.Comparatively, the overall AEC and NO2 levels were relatively low at the WD station, whereas a continuous high-value HCHO distribution (> 2 ppb) occurred at 0-1500 m between 11:00 and 16:00.This occurred because the WD station is located in a farm field with less traffic flow and high vegetation coverage; therefore, large amounts of HCHO are directly emitted by biogenic sources and secondarily produced by natural and anthropogenic volatile organic compound (VOC) photolysis (Wang et al., 2016;Wu et al., 2017a).
Nevertheless, Fig. 3  displayexhibiteded different spatiotemporal characteristics.Although surface and high-altitude (400-800 m) AECs both remained at a relatively high level (> 0.3 km -1 ) at CAMS during 12:00-17:00 (Fig. 3), there was a large discrepancy between their corresponding   values (Fig. 4).The aerosol near-surface   was ~ 1 km -1 • m• s -1 after 12:00, while   in layers of 400-800 m all exceeded 1.2 km -1 • m• s -1 , and even reached ~ 2 km -1 • m• s -1 around 12:00.At the SJZ station, the AECs at surface and 300-1000 m layer mostly ranged from 0.3 to 0.4 km -1 , especially after 10:00 (Fig. 3).However, the MTLs of aerosols were mostly at 400-800 m throughout the day, with many transport fluxes in those layers even reaching ~ 2 km -1 • m• s -1 (Fig. 4).At the WD station, the highest   also tended to occur at high layers (400-800 m), with maximum   exceeding 1.7 km -1 • m• s -1 at 400-500 m at 15:00.This suggested that aerosol transport occurred mainly in the upper layers.In the late afternoon, aerosols gradually accumulated towards the surface, and triggered a variation in the distribution of   .After 16:00, the shift in the high-AEC air mass caused the transport fluxes in the lower layers (100-200 m) to increase to > 1.1 and ~ 2 km -1 • m• s -1 for the CAMS and SJZ stations, respectively.Surface aerosol accumulation is closely linked to the collapse of the mixing layer and formation of a stable nocturnal boundary layer (Ding et al., 2008;Ran et al., 2016).Remarkably, high-altitude aerosol air masses began to mix with near-surface aerosols after 14:00 at the NC station (Fig. 3), triggering a variation in the MTL (Fig. 4).This might be explained by enhanced vertical mixing due to the heating of the surface during the course of the day (Castellanos et al., 2011;Wang et al., 2019).With aerosol as an indicator, we found a high-AEC (~0.3 km -1 ) air mass started at 400-800 m around 10:00 at SJZ (Fig. 3).Driven by the southwest wind, the aerosols were transported in the northeast direction, and we detected a relatively high AEC (~0.2 km -1 ) over 300 m at WD around 11:00.Over time, we measured a subtle increase in AEC from ~0.2 to ~0.3 km -1 at 400-1000 m at NC around 12:00.Finally, we discovered an extremely high AEC (>0.5 km -1 ) at 200-800 m at CAMS at 12:15.Generally, tThe MTL of aerosols was mainly situatedd at 400-800 m during the daytime, where and variations in the boundary layer and increased vertical mixing can gradually dropped below 400 m in the late afternooninfluence the MTL (Fig. 4).In contrast to aerosols, After 16:00, the high-extinction air mass shifted MTL from to 300-1000 m toward the surface at SJZ, with the AEC gradually exceeding 0.5 km -1 (Fig. 3).We observed a similar phenomenon at NC around 16:00 and at CAMS after 17:00.Surface aerosol accumulation is closely linked with the collapse of the mixing layer and the formation of a stable nocturnal boundary layer (Ding et al., 2008;Ran et al., 2016), triggering a descending tendency in the MTL.wWe found found that a high-value NO2   frequently occurredred in the 0-400 m layer because NO2 was mainly sourced from near-surface traffic emissions (Fig. 4).Except that   reached the highest level of ~ 50 ppb• m• s - 1 in the 400-600 m layer at 16:00 at the SJZ station, the other highest   all occurred below 400 m at any station and at any 带格式的: 上标 带格式的: 上标 time.This indicated that the MTL of NO2 was 0-400 m.Near-surface NO2 emission sources (e.g., vehicle and factory emissions) might be the main reason for this phenomenon.At the CAMS and NC stations, the NO2 concentration was consistently high near the surface (Fig. 3), primarily because of the heavy traffic flow in Beijing.At SJZ, NO2 concentration was high in the morning and late afternoon, whereas the concentration was lowest near noon, which we explained by the morning and evening rush hour.The variation in high-concentration NO2 (Fig. 3) agreed well with the shift in the corresponding MTLs (Fig. 4).In addition, we determined a short-distance transport of HCHO between the NC and CAMS stations.At approximately 09:30, ~2 ppb HCHO air masses first emerged in the 300-1000 m layer in the NC.Approximately half an hour later, we measured an enhanced HCHO signal (>4 ppb) at the same height at CAMS.This high-altitude transport process corroborated the HCHO regional transport belt observed by the satellite between NC and CAMS (Fig. S2b)Compared with aerosols and NO2, we found that The MTL of HCHO ranged from 400 to 1400 m, and was much more likely tohighvalue HCHO   extend toextended to higher altitudes than that of aerosols or NO2.Taking CAMS as an example, we found found the a strongest HCHO   constantly emerging at 6001000-12400 m, which lasted 4 h ( from 108:00 to -143:00, and averaging 9.18 ppb• m• s -1 ) (Fig. 4).During the same period, surface HCHO   only averaged 6.44 ppb• m• s -1 .However, at the CAMS station, the surface HCHO concentration was much higher than that of the 1000-1200 m layer between 8:00 and 13:00 (Fig. 3), proving that high-altitude transport contributed more to overall HCHO transport.After 10:00, we found that the highest HCHO   gradually increased from ~ 8 to ~ 20 ppb• m• s -1 at WD, with the MTL of HCHO ranging from 400 to 1000 m.LikewiseAt station SJZ, the the MTL of HCHO increased to over 800 m between 11:00 and 16:00 at the WD stationstrongest HCHO   increased from ~ 10 to ~ 16 ppb• m• s -1 during 11:00-17:00, with the highest transport fluxes occurring mostly at 400-800 m.These findings indicated that the MTL of HCHO was mainly 400-1200 m.The sharp variation in the MTL at the NC station might be caused by atmospheric vertical mixing (Castellanos et al., 2011;Wang et al., 2019).As shown in Fig. 3, hHigh HCHO concentrations tended to appear at higher altitudes than those of aerosols and NO2 (Fig. 3).A possible explanation is that the precursor compounds of HCHO are transported to higher layers and converted into HCHO through photochemical reactions, resulting in elevated HCHO concentrations at higher altitudes (Kumar et al., 2020).
Furthermore, strong high-altitude winds were were more conducive to HCHO transport (Fig. S53), which further increasing increased the corresponding transport flux (Fig. 4).Notably, HCHO   was was enhanced around noon because the increased solar radiation promotes d the secondary generation of HCHO.Long-term observations have revealeded that secondary HCHO formation through VOCs photolysis plays a large significant role in Beijing (Liu et al., 2020;Zhu et al., 2018).Furthermore, high concentrations of NO2 and HCHO likely enhance the AEC at the corresponding heights (Fig. 3).
Taking the CAMS results as an example, NO2 exceeded 6 ppb at 400-800 m between 12:00 and 13:00, followed by an extreme AEC (>0.5 km -1 ) at the same height after 12:00.At WD, high-concentration HCHO (~2.5 ppb) triggered a belt of high-AEC (>0.2 km -1 ) distribution in the 300-1200 m layer at 14:00-16:00.The increased AEC was closely linked to the generation of 带格式的: 下标 带格式的: 下标 带格式的: 下标 secondary aerosols, with HCHO and other VOCs being the main precursor compounds of organic aerosols, and NO2 being the main precursor of inorganic aerosols.Based on the above results, we discovered that secondary aerosol generation always accompanied the regional transport process.In addition, We we discovereded a wide discrepancy in the   among the stations for various pollutants (Fig. S45).The average aerosol   decreasedd in the following order: SJZ (3.21 × 10 3 km -1 •m 2 •s -1 ) > NC (2.69 × 10 3 km -1 •m 2 •s -1 ) > CAMS (2.43 × 10 3 km -1 •m 2 •s -1 ) > WD (1.42 × 10 3 km -1 •m 2 •s -1 ).The largest aerosol   was possibly linked to the higher wind speed at SJZ and the many surrounding pollution sources, resulting in aerosol transport.For NO2 transport, the average   values at SJZ (5.69 × 10 4 ppb• m 2 •s -1 ), NC (4.42 × 10 4 ppb• m 2 •s -1 ), and CAMS (6.16 × 10 4 ppb• m 2 •s -1 ) were were substantially higher than those at WD (2.04 × 10 4 ppb• m 2 •s -1 ), which we attributed to higher traffic flow in Shijiazhuang and Beijing.Conversely, the average   of HCHO was was the highest at in WD (3.21 × 10 4 ppb• m 2 •s -1 ), whereas the   values at in SJZ, NC, and CAMS were were 1.76 × 10 4 , 2.01 × 10 4 , and 1.94 × 10 4 ppb• m 2 •s -1 , respectively.This occurred because the WD station was located in a farm field with high vegetation coverage, so large amounts of HCHO were directly emitted by biogenic sources and secondarily produced by natural and anthropogenic VOC photolysis (Wang et al., 2016;Wu et al., 2017a).To some extent, the   of various pollutants can reveal the dominant pollution species in a regionIn terms of the relative locations of stations (Fig. 1) and the   results, we considered that SJZ was an important source of transported aerosol and NO2, and WD was one of the main HCHO sources during this regional transport, which largely affected the air quality of cities along the southwest-northeast transport pathway.The corresponding error distributions of   and   were provided in Figs.S6 and S7.The superiority of MAX-DOAS in monitoring the vertical distribution of various pollutants was highlighted by this analysis.Combined with wind field information, we identified diverse pollutant transport processes and their corresponding MTLs, which can enable the public to gain a comprehensive understanding of pollutant transport.

Effects of dust transport on regional air pollution
The Himawari-8 satellite results observations revealeded that a dust storm had occurred in northern China on March 15, 2021 (Fig. S5S8), with the NCP being one of the most severely affected.Combined with the aerosol 24 -h backward trajectories (Fig. S6S9), we found found that the dust storm originated originated in Mongolia, and its major transport pathway was was Mongolia-Inner Mongolia-NCP.According to the selection standards described in Supplementary Sect.S3S9, we confirmed that March 15 was was a dusty day; we chose March 6 and 22 as comparison benchmarks because they were were the nearest clean days before and after the dust storm, respectively.The dDust could can suppress dissipation and aggravate the local pollution accumulation.On the dust storm day, the AEC and HCHO concentrations substantially increasedd, especially near the surface (Fig. S7 S10 and and S9S12), while.Meanwhile, NO2 concentrations also increasedd a lotsignificantly at in SJZ and Dongying (DY) (Fig. S8S11).As shown in Fig. 56, we roughly classified the vertical profiles at all four stationsshapes displayed peaks in the high layers into multiple-peak and Gaussian shapes on clean days.According to Liu et al. (2021), vertical profile shapes can be used to perform an overall evaluation of pollutant sources and meteorological conditions in a certain area.In the JJJ region, we selected three stations (i.e., NC, SJZ, and XH) and found that high-altitude peaks occurred at 300-500 m for aerosols, NO2, and HCHO (Table S3-S5).For example, the AEC vertical distribution at SJZ had displayed a high peak of 0.75 km -1 a two-peak shape at 500 m on March 6, with a low surface peak of 0.33 km -1 and a high peak of 0.75 km -1 at 500 m (Table . S3).At the NC station For the Gaussian-shaped vertical profiles, the surface concentrations remained at relatively high levels.Taking the NC results as an example, the AEC vertical distribution exhibiteded a Gaussian shape, with the only peak (0.70 km -1 ) emerging at 300 m, and the surface AEC was relatively high (0.38 km -1 ).According to Liu et al. (2021), vertical profile shapes can be used to perform an overall evaluation of pollutant sources and meteorological conditions in a certain area.As discussed in Section 3.1,This may be explained by the prevalent regional transport, which strongly influenced influences the air quality in the JJJ region (Ge et al., 2018;Wu et al., 2017;Xiang et al., 2021), corresponding to the occurrence of high-altitude peaks.As discussed in Section 3.1, high-altitude transport phenomena trigger high-values of pollutant distribution in the high layers.The surface peaks on clean days were possibly caused by dense traffic and factory emissions in the JJJ region (Qi et al., 2017;Zhu et al., 2018;Yang et al., 2018;Han et al., 2020).In contrast to the JJJ region, the DY station was is situated in a rural area surrounded by many open oil fields and was is adjacent to the Bohai Sea (Guo et al., 2010).The transport of sea salt is an important significant source of local aerosols (Kong et al., 2014), which might be the main reason for the high-altitude AEC peaks occurring at 300-900 m (Table S3).High-concentration surface pollutants are mainly attributed to dense near-surface emission sources (e.g., traffic emissions and coal combustion).In addition, the occurrence of high-altitude NO2 and HCHO peaks may be attributed to the surrounding high-elevation point emission sources (e.g., chemical plants) (Kong et al., 2012).During dusty periods, aerosol, NO2, and HCHO concentrations notably increased, particularly near the surface, at most stations (Fig. 6).The high-layer peaks dropped to lower altitudes and even disappeared (Tables S3-S5).For example, on the dusty day, we found that high-altitude peaks disappeared and the only peak emerged at the surface for aerosol, NO2, and HCHO vertical profiles at the NC, XH, and SJZ stations (Tables S3-S5).Meanwhile, the NO2 and HCHO concentration peaks both dropped to the 100 m layer at the DY station (Tables S4, S5).These changes might trigger variations in the vertical profile shapes and convert 带格式的: 下标 带格式的: 下标 many vertical profile shapes (e.g., AEC vertical profiles at all stations) into an exponential shape (Fig. 6).On dusty day,This is because elevated dust concentrations weaken turbulence and decrease PBL heights on the dusty day, mostly through surface cooling and upper PBL heating (Mccormick and Ludwig, 1967;Li et al., 2017b;Mitchell and Jr., 1971).Unfavorable meteorological conditions not only impede pollutant dissipation and transport, but also favor the accumulation of locally produced pollutants (including direct emissions and secondary production).Moreover, some accumulating components (e.g., NO2, SO2, and VOC) are important precursors of aerosols, providing favorable conditions for secondary aerosol formation (Behera and Sharma, 2011;Volkamer et al., 2006;Huang et al., 2014).Therefore, during dusty periods, aerosol, NO2, and HCHO concentrations notably increased and peaked at ground level at most stations (Table .S3, S4, S5), forming an exponential profile (Fig. 5).Moreover, some accumulating components (e.g., NO2, SO2, and VOC) further promoted secondary aerosol formation (Huang et al., 2014).In contrast, we observed no changes in the vertical profile shapes of NO2 and HCHO at DY.The DY station was set in a wetland with low traffic flow, and NO2 was predominantly emitted from the surrounding higher-elevation point sources (e.g., chemical plants).Local HCHO was mainly sourced from VOC photochemical reactions that occurred at high altitudes (Chen et al., 2020).Based on the above analysis, we determined why NO2 and HCHO concentrations both peaked in the high layers, and their vertical profiles maintained a Gaussian shape.Under the influence of suppressed dissipation, the NO2 and HCHO concentration peaks both dropped to the 100 m layer on the dusty day (Table .S4, S5).However, we found a substantial decrease in near-surface NO2 concentrations at NC and XH on the dusty day (Fig. S6).Using the March 6 results as the benchmark, we found the near-surface NO2 maximum concentration was 76.7% (2.43 ppb) and 66.7% lower (2.57ppb) on the dusty day at NC and XH, respectively (Fig. 5).We think that this finding is related to the critical role of heterogeneous reactions in NO2 removal combined with optical variation; we provide detailed explanations and verifications in the paragraphs below.In addition to aggravating pollutant accumulation, transported dust can indirectly affect the environment and pollutant concentrations in other ways.regional pollution by weakening light intensity.To determine the impact of dust on the local environment based on optical intensity variation, we classified the four stations (XH, NC, SJZ, and DY) into two groups: bright group (BG) and dark group (DG), using 20 arbitrary units (A.U.) as a threshold.In terms of optical intensity levels, we assigned NC and XH to BG, and SJZ and DY to DG (Fig. 6 and S10).ForTo quantificationquantitatively demonstrate the impacts of dust on various pollutants, we introduced growth rate in the comparative analysis (Supplementary Section Sect.S3S9).For convenience, we defined the comparison of the results of March 6 and 15, 2021, as precomparison (PRE); ), and we defined the other comparison between March 15 and 22, 2021, as postcomparison (POST) using March 22, 2021 as the clean day.
As shown in Fig. S117A, the AEC noticeably increased increases at all stations on the dusty day, especially below 0.5 km.To quantitatively evaluate the impacts of increased dust and aerosols on light intensityThis occurred not only because local accumulation was enhanced due to the decrease in PBL height, but also because the secondary production of aerosols increased through aqueous-phase and heterogeneous chemical reactions (Ravishankara, 1997;Mcmurry and Wilson, 1983)., we averaged the optical signal intensities received at each channel of the spectrometer as the light intensity of each station.By subtracting the light intensity on clean days from that on the dusty day, we found that light intensity was substantially reduced at each station on the dusty day (Fig. S13).This is because enhanced aerosol concentrations, along with large amounts of dust, aggravate light attenuation and weaken light intensity.At certain wavelengths (e.g., 360 nm), aerosol extinction contributes more to light attenuation than dust (Wang et al. 2020).As described above, the advent of dust results in unfavorable meteorological conditions (e.g., decreased PBL height and more stable PBL) and enhances local pollutant accumulation, which boosts aerosol increase in the lower layer.Moreover, such meteorological conditions are always accompanied by higher levels of RH in the lower PBL (Huang et al., 2020), creating good conditions for enhanced secondary production of aerosols through aqueous-phase and heterogeneous chemical reactions (Ravishankara, 1997;Mcmurry and Wilson, 1983).These two factors could be the main reasons for the enhanced aerosol concentrations on the dusty day.The AECs at DG were higher than those at BG.We mainly attributed the reduction in optical intensity to aerosol extinction rather than dust because dust had less impact on light attenuation at 360 nm than aerosols (Wang et al., 2020).Therefore, our obtained coefficients reflected the level of aerosol extinction.In contrast with In contrast to aerosolss, we observedd large differences in the NO2 growth rates (Fig. 7B) between BG and DG.At DGSJZ and DY, NO2 concentrations exhibiteded a substantially increasing trend.The surface growth rates at the SJZ 555 and DY stations were were 6.97 and 17.50 in PRE, and 2.06 and 6.50 in POST, respectively (Fig. 7A 7B (c-1, c-2, d-1, d-2)).
In contrast, we observed decreased NO2 concentrations at almost every height at stations NC and XH.The near-surface growth rates at NC and XH were -0.81 and -0.76 in the PRE, and -0.30 and -0.59 in the POST, respectively (Fig. 7B (a-1, a-2, b-1, bof HCHO at the DY station (Chen et al., 2022;Chen et al., 2020).Thus, we assumedd that the explosive considerable increase in HCHO levels at station DY was was closely related to VOC accumulation.

Spatiotemporal characteristics of aerosol during transboundary transport
Back-and-forth transboundary long-range transport between the NCP and YRD is common, especially during winter, which deteriorates the air quality of these two regions (Huang et al., 2020;Petaja et al., 2016).During the transport process, the aerosol-PBL interaction can amplify the overall haze pollution and deteriorate the air quality of these two regions (Petaja et al., 2016;Ding et al., 2016;Huang et al., 2014;Huang et al., 2018b).Based on model simulations, Huang et al. (2020) elaborated on this transport process and the haze-amplifying mechanism in three stages.First, air pollutants from the YRD are transported to the upper PBL over the NCP and substantially affect PBL dynamics.Subsequently, under the influence of aerosol-PBL interaction, local pollutant accumulation and secondary production of aerosols are enhanced, causing severe pollution in the NCP.Finally, strong weather patterns (e.g., cold fronts), can dissipate low-PBL pollutants in the NCP and transport them over long distances back to the YRD.Specifically, a weak southerly wind drives the air mass northward, causing severe pollution in the NCP (Cai et al., 2017;Callahan et al., 2019), whereas a strong cold front increases the dissipation of the pollutants in the NCP and pushes them to the YRD (Li et al., 2018;Ding et al., 2013).According to results ofMany model simulations have suggested that, the mechanism of aerosol-PBL interaction amplifies the overall haze pollution is amplified during the transport process, mostly through the mechanism of aerosol-PBL interaction (Petaja et al., 2016;Ding et al., 2016;Huang et al., 2014;Huang et al., 2018b).Using MAX-DOAS measurements, we investigated the spatiotemporal variation in aerosols along the transport pathway, and validated the haze-amplifying mechanism of this transboundary transport.
The Himawari-8 satellite observations revealed a substantial increase in aerosol concentrations within the NCP From from January 18 to 20, 2021, the NCP experienced a substantial increase in aerosol concentrations, with an overall AOD overpassing 0.9 (Fig. S14S16). .Subsequently, high-concentration aerosol air masses began toassumed a move southward movement tendency, gradually leaving the NCP and covering the YRD from on January 21-to 22, 2021.We attributed this phenomenon to the back-and-forth transport of aerosols between these two regions, which we validated using wind-field simulations.
According to tThe wind field and Himawari observations results indicated that, the wind blew towards the East China Sea at every altitude on January 18, 2021 (Fig. 8).Aa south-to-north transport belt gradually firstly formeded in the upper layers (500-1500 m) on January 19, 2021, which and lasted fored nearly two days (Fig. S14 and Fig. 8).Around 12:00 on January 21, the wind direction begaan to change, and the north-to-south transport trend strengthened strengthened in the 0-1000 m layer on January 22 (Fig. 8 and Fig. S19).The diurnal variation of wind fields in different layers on January 18-22, 2021, were provided in Figs.S17-S21.In terms of overall transport direction, we classified the MAX-DOAS monitoring results into four periods: west-to-eastWest-to-East, YRD to NCPYRD-to-NCP, transformationTransformation, and NCP to YRDNCP-to-YRD, 620 to further explore their vertical characteristics during the transport process (Fig. 99).On January 18, 2021, wind gathered toward the East China Sea at every altitude, and no direct long-range transport occurredred between the NCP and YRD (Fig. 8 and Fig. S15S17).These two regions both had acceptable air quality, with the maximum AEC at the four stations being less than 0.88 km -1 (Fig. 98) and the overall spatiotemporal average PM2.5 concentration less than 80approximately 43.71 g/m 3 (Fig. S20aS22a).The wind simulation indicated that south-to-north transport initially took 630 shape in layers of 500-1500 m on January 19 (Fig. 8 and Fig. S18), which we defined as the start of the YRD-to-NCP transport period.On January 19During this period, the overall AECs at all stations startbeganed to increase in varying degrees.We 带格式的: 字体: (中文) 宋体, 10 磅 notedd that a continuous high-AEC aerosol distribution transport beltsoccurred at XH in the 100-1400 m layer, at DY in the 200-1000 m layer, and at NB in the 100-700 m layer (Fig. 9), .with the peakThe maximum AECs reaching reached 6.10, 1.41, and 0.96 km -1 at for XH, DY, and NB, respectively (Fig. 9).This was closely related to the strong YRD-to-NCP transport.
The results ofAccording to the wind simulation for January 19-21, indicated that the YRD-to-NCP transport mainly occurred at 500-1500 mlasted until 12:00 on January 21 (Fig. 8 and Fig. S18-S20).During this period, large amounts of aerosol from the YRD werewere transported to the upper layers (500-1500 m) of the NCP.In addition, secondary particle formation intensified because the transport of warm and humid air masses favors aqueous and heterogeneous reactions (Huang et al., 2014).These factors jointly led to a sharp increase in the AECs in the high layers at stations XH, DY, and NB (Fig. 99).On January 19, the south-to-north wind direction formed in the 500-1500 m layer (Fig. S16).In contrast, the increase in the nearsurface AEC was slower than that in higher layers.On January 19, for instance, the surface AECs were mostly less than 0.6 km -1 from 10:00 to 16:00 in XH, while surface peak AECs in the morning and late afternoon could be explained by the diurnal variation in PBL height (Ding et al., 2008;Ran et al., 2016).At the DY station, the average surface AEC only increased from 0.61 km -1 on January 18 to 0.62 km -1 on January 19.Meanwhile,The reason is that surface transport was was mainly driven mainly by the east wind on January 19 (Fig. 8), resulting in the PM2.5 concentrations at many western CNEMCs exceeding 80 g/m 3 (Fig. S20bS22b).From January 20 to 21, 2021, the surface wind converted into a south wind, but becaame so weak that near-surface transport contributed contributed little to the local environmentNCP (Fig. 8 8 and Fig. S1917, S-S2018).This also explains why the increase in near-surface AEC was less than that in the higher layers.During this period, large amounts of aerosol from the YRD were transported to the upper layers of the NCP.In addition, secondary particle formation intensified because the transport of warm and humid air masses favors aqueous and heterogeneous reactions (Huang et al., 2014).These factors jointly led to a sharp increase in the AECs in the high layers at XH, DY, and NB (Fig. 9).Despite the minimal contribution from surface transportHowever, we continued to observed a substantial increase in AEC at the ground level on January 20-21, 2021 (Fig. 99).At station DY, for example, the average surface AEC increased increased from 0.61 km -1 on January 18 to 1.03 km -1 on January 20, which was was a 68.9% growth rate.A possible reason for this was was the strong dome effect caused by the high-layer aerosols.As a result of the aerosol-PBL interaction, PBL height decreased decreases while temperature and humidity increased in the lower PBL, which favorsed pollution accumulation and secondary aerosol production (Bharali et al., 2019;Huang et al., 2020;Petaja et al., 2016).Generally, this YRD-to-NCP transport intensifies local pollution in the NCP region, causing a substantial increase in aerosol concentrations on January 18-20 (Fig. S16).Around 12:00 on January 21, the wind direction startwent throughed to experience a half-day transformation period, changing from southerly to westerly (Fig. S18S20).The continuous high-AEC distribution belt at in NB brwas interruptedoke at 12:30-13:30 and 14:15-16:00 (Fig. 99), possibly owing to the western clean-air injection.The northerly wind finally formeded on January 22, and creating the NCP-to-YRD transport belt was created (Fig. S198).In contrast to YRD-to-NCP transport, NCP-to-YRD transport mainly occurred occurred at low altitudes (0-1000 m), with surface wind speeds rising to a relatively high level> 4 m• s -1 (Fig. 8 and Fig. S21).Influenced by strong cold fronts, large amounts of aerosols in the NCP startbeganed to disperse and gradually covereded the YRD (Fig. S14S16), increasing the average surface AEC at in NB by 183.33% (from 1.56 to 4.42 km -1 ) (Fig. 99).Furthermore, with 11:00 as a dividing line, high-AEC air masses (average AEC of 0-1.6 km layer: 2.60 km -1 ) suddenly abruptly vanisheded at the DY station, highlighting the effect of quick dispersion driven by cold fronts.The weak dispersion of aerosols at in XH may have been affected by the anticyclones in the JJJ region (Fig. S19S21).For cities in the area where the NCP and YRD overlap, their location determinesd that they suffered a longer dome effect period because the two transport processes (YRD-to-NCP and NCP-to-YRD) both passed by through these areas, producing a constant increase in pollution levels in the shallow PBL.For example, from January 18 to 22, the average AECs in the 0-1 km layer at in HNU were 0.35, 0.48, 0.75, 0.76, and 0.85 km -1 , respectively, showing a continuously increasing tendency (Fig. 99).Furthermore, on January 22, extreme PM2.5 values (> 200 g/m 3 ) were mostly concentratedd in the overlapping zone (Fig. S20eS22e).
Generally, we found found that transboundary long-range transport amplifiedd haze pollution within the NCP and YRD (Fig. 9 9 and Fig. S20aS22a-e), which agrees well with previous WRF-Chem simulation results (Huang et al., 2020).This MAX-DOAS measurements accurately demonstratedd the spatiotemporal characteristics of aerosols during transboundary transport.
Furthermore, from the a practical perspective of practical observations, we verified the haze-amplifying mechanism (Huang et al., 2020).

Conclusions
On February 5, 2021, southwest-northeast regional transport of pollutants occurred within the JJJ region.By calculating the   and   at each selected station, we demonstrated the dynamic evolution of the MTL for aerosols, NO2, and HCHO.
The MTL of aerosols was was situated at 400-800 m during the daytime.,The variation in the PBL and enhanced vertical mixing could trigger a and we mainly attributed the subsequent decreasing trend attributed to aerosol surface accumulation caused by PBL variationin the MTL.Nevertheless, the MTL of NO2 was was below 400 m, possibly owing to near-surface traffic or factory emissions.The HCHO MTL was was 400-1400 1200 m, and extendeded to higher altitudes than aerosols and NO2.Higher HCHO concentrations and higher wind speeds at elevated altitudes jointly resulteded in much stronger transport fluxes at high altitudesa higher MTL.With respect to the results of   comparison result, we found found that the outflows of aerosols and NO2 at CAMS, NC, and SJZ were much larger than those at WDassumed high levels at the SJZ station.In contrast, the HCHO outflow at in WD far exceeded exceeded the thosethat at the other stations.Based on relative station locations and the   comparison results, we believed that SJZ and WD were two predominant pollutant sources during this regional transport, exerting a significant impact on cities along the southwest-northeast transport pathway.This is because the WD station was located in a farm field with high vegetation coverage, so NO2 and aerosol emissions were lower and more HCHO was produced from biogenic sources and VOC photolysis.
A severe dust storm occurred in the NCP along the Mongolia-Inner Mongolia-NCP pathway on March 15, 2021.By comparing the results of the dusty day (i.e., March 15, 2021) and clean days (i.e., March 6 and 22, 2021), On March 6 and 22, 2021, as clean days, we found that high-altitude concentration peaks dropped to a lower layer and even disappeared on the dusty day.We attributed this result to dust-induced being able to suppress dissipation, inhibition weakened pollutant transport, and intensified intensify local pollution accumulation, eventually converting the vertical profiles into an exponential shape.In addition to aggravating pollutant accumulation, transported dust could also affect the environment and pollutant concentrations in other ways.The vertical profiles of NO2 and HCHO at DY maintained a Gaussian shape, which we explained by the elevated NO2 emission sources and high-altitude secondary HCHO formation.In terms of optical intensity, we classified four stations into BG and DG to examine the impact of reduced solar radiation.High AEC growth rates occurwere observedred at all stations.Large amounts of dust and aerosols intensified light attenuation and weakened light intensityowing to the pollutant accumulation effect and enhanced secondary formation.Notably, dust and aerosols had different effects on NO2 concentration.At stations SJZ and DY, the NO2 concentrations assumed a high growth rate because dust and aerosolsMoreover, the reduced limited the received optical solar radiationintensity may haveand inhibiteded NO2 photolysis, which favorsing NO2 its accumulation for the DG stations.In addition, dust reduces turbulence and inhibits dissipation, eventually aggravating the surface NO2 accumulation.ContrarilyIn contrast, we observed a remarkable decreases in NO2 levels at the stations BG XH and NCstations.The increase in HONO levels confirmeded that heterogeneous reactions on dust and aerosol surfaces played 带格式的: 下标 带格式的: 下标 带格式的: 下标 a critical role in the decreases in NO2 levels, with NOx-to-HONO conversion being one of the main removal mechanisms.The results of source apportionment analysis of ambient HCHO levels revealeded that the total contributions of primary and background HCHO exceeded 75% at in SJZ, with primary sources playing a dominant role.The reduced solar radiation weakened weakens the photolysis of the primary and background HCHO levels, favoring HCHO level increases at in SJZ.We mainly attributed the The substantial increase in HCHO levels at the DY station to may be associated with VOC accumulation.
According to the results of the WRF simulation, we found found that the YRD-to-NCP transport mainly occurred occurred in the upper layers (500-1500 m).High-altitude transport triggered triggered a substantially enhanced AEC above the NCP , which might be attributed to direct aerosol injection and secondary particle formationby directly injecting aerosols and secondary particle formation.We mostly attributed tThe increased in near-surface AEC was probably due to the dome effect produced caused by the upper aerosol-PBL interactions.Subsequently, NCP-to-YRD transport, situated at 0-1000 m, dispersed dispersed the haze over the NCP and transferred transferred low-level aerosols to the YRD, causing a considerable increase in the surface AEC at station NB at ground level.During the entire transport process iIn the overlapping zone where between the NCP and YRD overlap, the AEC continuously increasedassumed a continuously increasing tendency in the shallow PBL during the entire transport process, possibly owing to the longer exposure to the dome effect.Generally, this transboundary long-range transport amplifiedd the air pollution in these two regions.,Based on practical observations, we investigated the spatiotemporal variation in aerosols and validated the haze-amplifying mechanism of transboundary transport.verifyingthe haze-amplifying mechanism determined based on practical observations.In summary, accurate quantification of the vertical distribution of pollutants in the air is a key requirement for understanding the pollutant transport process.Using the MAX-DOAS network, we successfully analyzed three typical transport types (regional, dust, and transboundary long-range transport), emphasizing the unique advantages provided byof the network in monitoring pollutant transport.We think believe that our findings can provide the public with a thorough understanding of pollutant transport phenomena and can serve as a reference for designing collaborative air pollution control strategies.

Figure 1 .
Figure 1.Study area location, topography, and distribution of MAX-DOAS stations:.The black points represent the stations;.The solid red contour line red contour:indicates the JJJ region;.The solid blue contour line contour:shows the YRD region.The dashed purple contour line contour:indicates the NCP region.

Figure 2 .
Figure 2. (a) Correlation analysis of in situ measured PM2.5 and surface AECs (0--100 m) retrieved from CAMS, HNU, NC, and SJZ MAX-DOAS stations from January to March, 2021 and (b) their corresponding NO2 comparing comparative results.The black line denotes the linear least-squares fit to the data;.R denotes: Pearson correlation coefficient;.N: denotes the number of valid data.
cannot reveal the exact layers in which the main transport phenomena occur.For instance, at the CAMS station, the AEC at the surface and upper layers both reached ~ 0.5 km -1 around noon, making it difficult to determine the layer 带格式的: 下标 in which transport was more obvious.To further demonstrate the dynamic transport process of different pollutants, we calculated the hourly   and   , and defined the MTL.As shown in Fig.. 4, the a positive   indicatesd that NO2, HCHO, and aerosols were all transported from southwest to northeast at the four stationsall transport flux projections in the southwest-northeast pathway are from southwest to northeast at the four stations.The MTLs of the aerosols, HCHO, and NO2

Figure 6 .
Figure 6.The daily optical intensity level at NC, XH, SJZ and DY stations on March 15, 2021.Note that the bottom and top of the box represented the 25th and 75th percentiles, respectively; the red line within the box represented the median; the yellow dot represented the mean; the whiskers below and above the box stood for the 10th and 90th percentiles.

Figure 77 .
Figure 77.The growth rates of (A) AEC, (B) NO2 and (CB) HCHO at different altitudes at (a) NC, (b) XH, (c) SJZ, and (d) DY stations.The dashed line was a dividing line.The stations on the left of the dashed line represented the bright group (BG) while two 550

Figure 88 .
Figure 88.The wWind fields in layers of surface, 500, 800, 1000 and 1500 m at 12:00 from January 18 to 22, 2021.The red arrow in the first subgraph represents the wind speed of 8 m/s.625

Figure 99 .
Figure 99.Temporal and vertical variations in aerosol distribution at XH, DY, HNU, and NB stations from January 18 to 22, 2021.

Table 1 .
The names (codes), latitudes and longitudes of stations and their corresponding regions.

Table 23 .
Averaged error estimation (in %) of the retrieved near-surface (0-100 m) trace gas concentrations and AECs, and trace gas VCDs and AOD.