Ozone Continues to Increase in East Asia Despite Decreasing NO2: Causes and Abatements

Space-borne ozone (O3) measurements have indicated consistent positive trends across the entire Asia–Pacific region despite the considerable reduction of NOx since 2000s. The rate of increase in O3 derived from lower free tropospheric column measurements was observed to be 0.21 ± 0.05 DU/decade during 2005–2018. Our space-borne-based diagnosis of the nonlinear photochemical formation regimes, NOx-limited and NOx-saturated, show that O3 chemistry is undergoing a transitional process to the NOx-limited regime throughout most of the Asian region. Nevertheless, NOx-saturated conditions persist at present in and over eight major megacities. These NOx-saturated conditions in megacities contribute to the increased O3 due to NOx reduction, which could also affect the enhanced O3 concentrations throughout the Asia–Pacific region via long-range transport. This indicates that VOC limits along with NOx reductions are needed in megacities in the short term to reduce O3 levels. Moreover, NOx saturation in major megacities will continue until 2025, according to the forecast emission scenarios from the Intergovernmental Panel on Climate Change (IPCC). These scenarios gradually shift nearly all cities to the NOx-limited regime by 2050 with the exception of few cities under IPCC RCP8.5. Thus, continued reductions in NOx will be a key factor in reducing O3 in the long term.


Introduction
Surface ozone (O 3 ) is harmful to humans [1] and ecosystems [2] at elevated concentrations, whereas O 3 present in the upper troposphere acts as a greenhouse gas [1,3]. It is produced by the photochemical reaction of directly emitted nitrogen oxide precursors (NO x = NO + NO 2 ) and volatile organic compounds (VOCs). Long-range transport of tropospheric O 3 is possible such that it can reach the United States (U.S.) from East Asia [4][5][6][7][8][9][10][11][12][13][14][15]. In the 2000s, major East Asian countries began actively regulating air pollutants. South Korea and Japan have been enforcing stringent regulations on NO x emissions since the early 2000s [16][17][18]. In China, NO x emissions have also been on a downward trend since mid-2010s due to tougher restrictions on automobile emissions.
The O 3 formation mechanism is based on the O 3 -NO x -VOC photochemical reaction chain, identified as either NO x -limited or NO x -saturated (VOC-limited) depending on the VOC/NO x ratio, and is relevant to O 3 reduction policies focused on the NO x or VOC emission control priority. At high VOC/NO x ratios, O 3 production is NO x -limited, and at low VOC/NO x ratios, the production is VOC-limited [19][20][21]. On the global scale, O 3 production is largely dominated by the NO x -limited regime, though heavily polluted urban/metropolitan areas with high NO x emissions are frequently under the NO x -saturated condition [21,22]. NO x reductions could lead to decreasing ambient O 3 under the NO xlimited regime as NO x emission reductions reduce the NO 2 photolysis related to primary production of free oxygen atoms reacting with O 2 into O 3 . Under the VOC-limited regime, a reduction in NO x emissions has resulted in increased O 3 [20] due to weakened NO + O 3 titration. Thus, as NO x emissions decline, O 3 levels will continue to increase until the regime shifts to NO x -limited. Thus, surface O 3 does not respond to a linear reduction in NO x .
Numerous O 3 studies have been carried out to identify the characteristics of its changing trends and precursors in East Asia, in which the focus has been on high-emission areas, including urban regions [23][24][25][26][27][28]. However, few studies have examined the O 3 regime sensitivity over a large spatial scale in areas covering the Asia-Pacific region or in the many border areas of this region due to the lack of in situ measurements covering international areas or near-abroad joint data of the countries in this region [29][30][31]. Instead, studies have been limited to local measurement stations or city-scale precursor analyses [32][33][34][35]. Space-borne tropospheric O 3 regime identification has been carried out over large spatial areas; however, the ability to infer surface O 3 levels from these studies has been limited by the uncertainties in O 3 precursors from tropospheric column integration quantities that are based on precursor distribution probabilities [36,37]. However, with the help of sophisticated satellite retrieval algorithms, it has recently become possible to extract lower free tropospheric O 3 precursors from satellite signals [38][39][40][41][42], enabling large-scale spatiotemporal analysis of lower tropospheric O 3 concentrations.
In this study, space-borne data were used to retrieve lower tropospheric columns of O 3 , NO 2 , and formaldehyde (HCHO) between 900 and 700 hPa, covering the height of the atmospheric boundary layer. Interdecadal-scale past and current spatiotemporal variations in near-surface O 3 -NO 2 -HCHO relations were identified over the entire Asia-Pacific region. The resultant causes of increasing O 3 were diagnosed. Future abatement strategies are proposed from the perspective of NO x versus VOC emission reduction strategies in the Asia-Pacific region. In addition, the future prospect of changes in O 3 levels based on the Intergovernmental Panel on Climate Change Representative Carbon Pathway (IPCC-RCP) emission scenarios were analyzed to characterize the outcomes of the proposed O 3 abatement strategies. In view of the growing interest in long-range transport of O 3 between East Asia and North America and the role of background O 3 in the air quality of the U.S. [14], this work is relevant for both Asian and North American policymakers.

OMI Satellite Measurements
The Ozone Monitoring Instrument (OMI) onboard NASA's Aura satellite has been providing global observations of key atmospheric pollutant gases since October 2004, including ozone, nitrogen dioxide, sulfur dioxide, formaldehyde, and glyoxal, as well as aerosols [43]. The OMI is the first of a new generation of space-borne imaging spectrometers, with the capabilities of the unprecedented spatial resolution of 13 × 24-48 km 2 at nadir and near-daily global coverage at an equator crossing time of 1:45 pm local time. Compared to other similar space-borne UV instruments, the OMI has maintained long-term stability over its mission with low optical degradation (1-2% in radiance, 3-8% in irradiance) and high wavelength stability (0.005-0.020 nm).
However, there has been concern over row anomaly effects that appeared in 2007 and became serious in early 2009, damaging the global coverage that is currently obtained in approximately two days [44]. Therefore, in this study, for the purpose of improvement of the OMI product analysis, we excluded row anomaly-affected pixels, as well as cloudy pixels (cloud fraction >0.3) where the retrievals of the tropospheric gases were not convinc- ing, and derived monthly gridded data from the OMI level 2 product to fill data gaps and smooth out noise errors for individual measurements.

OMI Tropospheric O 3 Partial Columns
Monthly gridded tropospheric ozone data at the 1 • × 1 • resolution were derived from the OMI Ozone Profile (OMPROFOZ) research product publicly available from the Aura Validation Data Center at the NASA Goddard Space Flight Center (https://avdc.gsfc. nasa.gov/pub/data/satellite/Aura/OMI/V03/L2/OMPROFOZ/: last accessed on 1 June 2021). The OMPROFOZ is based on the retrieval of the partial ozone columns (DU) for a 24-layer atmosphere (~2.5 km-thick) from the surface to 65 km based on the Smithsonian Astrophysical Observatory's (SAO) algorithm using an optimal estimation as an inverse method [45] and the vector linearized discrete ordinate radiative transfer model (VLIDORT) as a forward model [46], with a spectral range of 270-310 nm in the UV1 channel and 310-330 nm in the UV2 channel.
This algorithm was initially reported by Liu et al. (2010) [47] and several modifications were made by Kim et al. (2013) [48] to produce the OMPROFOZ. OMI measurements were spatially coadded to match the different spatial resolutions between UV1 and UV2, as well as to meet computational costs. As a result, the OMPROFOZ dataset was produced at a spatial resolution of 52 × 48 km 2 at nadir. The retrieval errors due to precision (instrument random noise) and smoothing errors (insufficient vertical resolution) ranged from 6 to 35% in the troposphere and from 1 to 6% in the stratosphere [47]. In this inversion, zonal mean climatological a priori information was used to complement insufficient measurement information, especially in the lower troposphere, due to reduced retrieval sensitivity. Therefore, OMI retrievals are more influenced by a priori information in the lower troposphere. Despite greater a prior influence, several studies have demonstrated that the retrievals can observe lower tropospheric ozone enhancement over East Asia [37,38]. However, our comparisons of the O 3 retrievals and a priori climatological partial columns (shown in Figure S1) demonstrate that the transported pattern of the ozone amount is extracted from the OMI measurements based on the fact that the latitudinal variation can only be observed in the corresponding a priori map. The retrieval of ozone profiles near the surface can also be strongly constrained toward a priori information due to less photon penetration into the lower atmosphere. Therefore, the partial columns were integrated from 900 to 700 hPa for use in this study. In spite of the overall long-term stability of the OMI, ozone retrievals are susceptible to instrument degradation, but the magnitude of trend errors is relatively smaller in the lower troposphere (~0.5%/year) compared to the upper troposphere and lower stratosphere (~2%/year) [49]. Furthermore, the trend of OMI retrieval errors is opposite to increasing trends of lower tropospheric ozone, affected by increasing emission or ineffective emission controls which induce the sensitivity of ozone production to VOCs analyzed in this paper and in the referenced literature.
OMI observations are performed in the early afternoon (~1:45 p.m.) when photochemical reactions peak and O 3 concentrations are typically near their peak near the surface. Moreover, the altitude of the atmospheric boundary layer is also characterized by strong vertical mixing processes during the passage of the Aura satellite, as well as the observation of transported O 3 over downwind areas.

Tropospheric NO 2 and HCHO Columns
The tropospheric NO 2 data are taken from the NASA OMI standard product (version 4.0) which is available from the NASA Goddard Earth Sciences Data and Information Services Center (https://aura.gesdisc.eosdis.nasa.gov/data/Aura_OMI_Level2/OMNO2 .003/: last accessed on 1 June 2021) [50]. The NASA standard product retrieval algorithm derives NO 2 slant columns from the OMI reflectance spectra in the 402-465 nm window using the differential optical absorption spectroscopy (DOAS) method [51,52].
The separation between the stratospheric and tropospheric slant columns is calculated by assimilating OMI observations into a global chemical transport model, i.e., the GMI Remote Sens. 2021, 13, 2177 4 of 17 with the spatial resolution of 1 • × 1.25 • . Then, the vertical NO 2 column is taken from the tropospheric slant column applied with an air mass factor (AMF). The tropospheric AMF is derived to be a function of the GMI model-based monthly a priori NO 2 profile shapes and single scattering weight (SW) calculated using a forward radiative transfer model (TOMRAD). We used the monthly mean tropospheric NO 2 column, which is calculated from the daily values at a 0.25 • × 0.25 • grid spacing, to investigate spatial and temporal variability of ozone precursors.
The spatial and temporal variations in the tropospheric HCHO columns for our analysis are based on the OMI HCHO level 2 product (Collection 3) retrievals [53] from the SAO (OMI-SAO HCHO column data acquired from https://aura.gesdisc.eosdis.nasa.gov/ data/Aura_OMI_Level2/OMHCHO.003/: last accessed on 1 June 2021). The HCHO slant column densities were obtained using backscattered solar radiation within the spectral window of 328.5-356.5 nm. In the SAO algorithm, the AMF is computed using the LIDORT radiative transfer model and vertical profile information from the HCHO and aerosols simulated by the GEOS-Chem model. Then, tropospheric HCHO columns are calculated with the air mass factors and slant columns. We re-gridded the monthly tropospheric HCHO column data with a 0.1 • × 0.1 • horizontal resolution in this study.

HCHO-to-NO 2 Ratio (FNR) Analysis
The photochemical O 3 formation regime is investigated by the VOC/NO x ratio [19][20][21]. In order to examine the long-term variations of the O 3 formation regimes for wide areas over East Asia, we employed the satellite measurements of tropospheric NO 2 and HCHO columns to calculate the HCHO-to-NO 2 ratio. HCHO, a short-lived oxidation product of many VOCs, is approximately proportional to the summed rate of reactions of VOCs with peroxy radicals [21,54,55] and is detected by satellites. NO x can be approximated from satellite-observed NO 2 columns due to short chemical lifetime of NO x [56]. The FNR is only calculated for the summer peak O 3 season (June-July-August) when both VOCs and NO x have relatively short lifetimes (approximately a few hours) [57][58][59].

In Situ Ground Observation Data
NO 2 and O 3 in situ measurement data in Seoul and Ganghwa Island were obtained from the publicly available Air Korea website (https://www.airkorea.or.kr/web: last accessed on 1 June 2021) based on the National Ambient Air Monitoring Information System (NAMIS) operated by the Korea Environment Corporation. Surface NO x and O 3 levels at Oki and Ogasawara were obtained from a ground-based surface monitoring system operated by the Acid Deposition Monitoring Network in East Asia (EANET) (https://www.eanet.asia/: last accessed on 1 June 2021). Since the EANET provides NO x concentrations, we analyzed the long-term trends for surface NO x instead of NO 2 in Japan. Air concentrations of gaseous species by both automatic monitors were calculated from the calibration factor obtained by the span/zero gas measurement of each constituent. Both EANET and NAMIS provided NO 2 and NO x concentrations detected by commercial NO x chemiluminescence instruments with a molybdenum converter. Its use in urban sites near emission sources may be acceptable for NO 2 measurements since the major components of NO x are NO 2 and NO in urban areas. For rural and remote sites, these instruments measure total NO y since its NO x mode also responds to HNO 3 and other organic nitrates non-specifically.

Climate and Air Quality Modeling Based on RCP Scenarios
In this study, the Integrated Climate and Air Quality Modeling System (ICAMS/NIER-SNU) [60] was employed for investigating behavior of future ozone and its precursors. ICAMS/NIER-SNU is developed by the National Institute of Environmental Research (NIER) and Seoul National University (SNU) for high-resolution climate information needed for an assessment of climate change impacts and adaptation measures and consists of four sub-model systems including the Hadley Centre Global Environment Model (HadGEM2-AO, v2.0) as the general circulation model (GCM), the Goddard Earth Observing System with Chemistry Model (GEOS-Chem, v8.3.1) as the global chemistry transport model (GCTM), the Fifth-Generation Penn State/NCAR Mesoscale Model (MM5, v3.7.4) as the regional climate model (RCM), and the EPA Models-3 Community Multiscale Air Quality Model (CMAQ, v4.6) as the regional chemistry transport model (RCTM).
Simulations of future air quality over East Asia with high spatial resolution were conducted using RCP emission scenarios as input information for the ICAMS/NIER-SNU model system. The RCPs are named according to the radiative forcing target level in 2100. In this study, we employed three RCPs-a mitigation scenario leading to a very low forcing level with the lowest emissions based on the most stringent climate policy (RCP2.6), medium stabilization scenarios (RCP4.5), and very high baseline emission scenarios without any climate policy (RCP8.5). Figure S2 shows NO x and total VOC emissions derived from the three RCP scenarios during the periods of 2025-2100. The ICAMS/NIER-SNU model results with 54 km × 54 km spatial resolution using RCP emissions over East Asia have been validated in many previous studies [61,62], and simulated surface O 3 , NO 2 , and HCHO concentrations were used to analyze the future HCHO/NO 2 ratio in 2025 and 2050. Furthermore, future FNRs were also obtained by extracting the simulation results. In order to follow the consistent data selection method with that applied to the OMI satellite observation data, we applied the identical time window of the OMI overpass time to the hourly atmospheric concentration dataset that was simulated using the ICAMS/NIER-SNU model.

Changes in O 3 Precursors: NO 2 and HCHO
Via photochemical reactions, LFT O 3 forms from its two precursors, NO 2 and VOCs, which react in the presence of hydroxyl (OH) and hydroperoxyl (HO 2 ) radicals. Figure 2 presents time series of tropospheric NO 2 and HCHO column concentrations retrieved from space-borne measurements. In regions 1 and 2, NO 2 shows an inverted U-shape variation, with a 1.5-fold increase over the period from 2005 to 2013 and the opposite trend during the period from 2011 to 2013. By contrast, region 3 shows a decreasing NO 2 trend. These results are consistent with those of previous in situ ground observations [67,68] and are particularly relevant to NO x emission reduction. For example, in region 1 (East China), strict NO x regulations have been imposed on urban areas since 2012-2013 [69,70]. In regions 2 (South Korea) and 3 (Japan), stringent NO 2 regulations have been enforced since the early 2000s [18,71,72]. However, HCHO levels have exhibited a slight (but detectable) rise of about 5% in all the regions (Figure 2), due mainly to the increased VOC emission strength [70,73]. The eight megacities indicated in Figure 2 (Beijing, Tianjin, Hebei (Tangshan), and Shanghai in region 1, East China; Seoul and Incheon in region 2, South Korea; and Tokyo and Osaka in region 3, Japan) showed an increasing rate of O 3 production from 0.01 to 0.03 DU/year, as indicated by the angled arrows in Figure 2. This led to a rise in O 3 levels over most of the region. the early 2000s [18,71,72]. However, HCHO levels have exhibited a slight (but detectable) rise of about 5% in all the regions (Figure 2), due mainly to the increased VOC emission strength [70,73]. The eight megacities indicated in Figure 2 (Beijing, Tianjin, Hebei (Tangshan), and Shanghai in region 1, East China; Seoul and Incheon in region 2, South Korea; and Tokyo and Osaka in region 3, Japan) showed an increasing rate of O3 production from 0.01 to 0.03 DU/year, as indicated by the angled arrows in Figure 2. This led to a rise in O3 levels over most of the region.  In situ ground measurements were consistent with the abovementioned trends. Figure 3 shows site-by-site measurements of O 3 and NO 2 at two locations in region 2, one in Seoul, the capital of South Korea (a typical urban area), and one in Ganghwa (a typical rural area) located in the Yellow Sea between China and South Korea, as well as at two sites in region 3, Oki and Ogasawara. Oki is a typical background reference site in the Asia-Pacific Rim region, and Ogasawara represents a true Asia-Pacific oceanic site. Prior to the analysis of long-term trends of ground measurements of O 3 and NO 2 , we carried out comparative analysis of OMI results against ground measurement data obtained at four sites located in South Korea and Japan (see Figure S3). The results showed good agreement for both O 3 and NO 2 with the correlation coefficient of up to 0.89 for O 3 , especially over the megacity ( Figure S3). Long-term trends of in situ ground measurements indicated that, as indicated in Figure 3, NO 2 concentrations decreased at all sites (Seoul: −4.7 ± 0.7 ppbv/decade; Ganghwa: −2.2 ± 0.8 ppbv/decade; Oki: −0.4 ± 0.05 ppbv/decade; Ogasawara: −0.1 ± 0.03 ppbv/decade). NO 2 levels in Ganghwa decreased until 2003. Notably, NO 2 levels in Seoul decreased consistently due to regulations imposed by the government of South Korea in the early 2000s. Several studies examined the reduction of NO 2 levels based on in situ ground measurements taken in Seoul since the implementation of the Special Act on the Improvement of Air Quality in the Seoul Metropolitan Area [18,71]. O 3 and NO 2 trends in Oki and Ogasawara were similar to those in Seoul (Figure 3). Taken together, these results signify a steady increase in O 3 in Seoul, Ganghwa, Oki, and Ogasawara that has persisted despite the reduction in NO 2 precursor levels, especially since 2013 in regions 1 and 2. Thus, O 3 transport can have a large or small impact on the entire Asia-Pacific region.

O 3 Continues to Increase Despite Declining NO 2 Levels
O 3 formation is an example of the nonlinear nature of the O 3 -NO x -VOC chemical reaction. Several indicator species [54,74] of O 3 can increase (or decrease), providing information on the ratio of VOCs/NO x . This ratio is widely used to diagnose O 3 sensitivity regarding indications of NO x -limited or NO x -saturated (or VOC-limited) conditions. In this study, for the space-borne measurement analysis, the ratio of HCHO (a marker of Remote Sens. 2021, 13, 2177 9 of 17 VOCs) to NO 2 (a marker of NO x ) (FNR) was applied. Conventionally, we considered FNR <1.0, >2.0, and 1.0-2.0 to signify the NO x -saturated regime, the NO x -limited regime, and the transitional regime, respectively [21,22,55,75]. Figure 4 shows the spatiotemporal distributions of the retrieved FNR, NO 2 , HCHO, and O 3 . NO 2 features similar to those shown in Figure 3 were evident in the data. An increase in NO 2 was observed up to 2011, followed by a sharp decline after mid-2010 (Figure 4a). HCHO showed a rather sharp increase until 2013, especially over urban areas, followed by slower growth after 2013 (Figure 4c). The domain-averaged FNR (Figure 4a  In urban megacities, including the eight megacities of Beijing, Tianjin, Hebei, and Shanghai (East China), Seoul and Incheon (South Korea), and Tokyo and Osaka (Japan), the lower NO x -saturated FNR retained a value of FNR ≤ 1 ( Figure 5). NO x -saturated regimes (FNR ≤ 1 or slightly greater) were found in the eight megacities before 2013; this was maintained even over the period of extensive NO x emission reduction from 2016 to 2018. The mean FNRs of the eight megacities ranged from 0.87 to 1.32 (Figure 5a): Beijing, Tenjin, Tangshan, and Shanghai in region 1 had an FNR of 0.96→1.01→1.32; Seoul and Incheon (region 2) had an FNR of 0.87→1.15→1.12; and Tokyo and Osaka (region 3) had an FNR of 0.89→1.14→1.11. The average for the three regions as a whole was 12.44 ± 2.40 (Figure 5b), due presumably to short lifetime for NO 2 and also the role of biogenic VOCs over rural or suburban areas. Thus, megacity-centered NO 2 emission reductions reinforced a relatively weak titration effect, deriving the increasing O 3 tendencies under NO x saturation. This also indicates megacity-centered VOC emission reductions for the abatement of current O 3 levels, according to the U.S. Environmental Protection Agency's empirical kinetic modeling approach [76].  Figure 1) are also shown.

Diagnoses of Future O 3 Abatements
For future O 3 abatement, the emission reductions of the three IPCC-RCP scenarios, RCP2.6, RCP4.5, and RCP8.5, were projected for the years 2025 and 2050 based on future climate and emission changes (https://tntcat.iiasa.ac.at/RcpDb: last accessed on 1 June 2021) provided by the International Institute for Applied Systems Analysis. As a projected modeling platform, the Integrated Climate and Air Quality Modeling System [60] was applied over the Asia-Pacific region.
The projected future NO x emissions showed declining trends in all the regions, and the VOC emissions in region 1 are expected to slightly increase until 2030 only in RCP8.5, whereas all other regions show a decline from 2025 ( Figure S2). In all three RCP scenarios, the FNRs in the eight megacities are highly likely to retain a VOC-limited regime up to 2025 ( Figure S4). RCP8.5 indicates a specifically lower FNR over regions 1 and 2 in the aftermath of a relatively small NO x reduction ( Figures S2 and S5). From 2025 to 2050, only region 2 in RCP8.5 retains a VOC-limited status due to the reduced HCHO levels in this region up to 2050 ( Figures S2 and S6), whereas the regime transitions under RCP2. 5 and RCP4.5 show NO x -limited conditions by 2050 over the entire domain ( Figure S4). Figure 6 illustrates past, present, and future FNRs for regions 1-3 in the Asia-Pacific region. During 2005-2018, the FNRs of megacities with NO x saturation were well-contrasted to the NO x -limited regime of the domain-averaged FNRs. All three future RCP scenarios show the NO x -saturated regime (i.e., FNR ≤1 or slightly greater) in all the eight megacities up to 2025; VOC reduction may be a key factor. RCP4.5 shows a tendency to change to the NO x -limited regime up to 2025 more easily by reflecting a relatively higher emission ratio of VOCs to NO x than others ( Figure S2). Three scenarios then achieve prospective turnarounds to the NO x -limited (i.e., FNR ≥2) regime up to 2050 with the only exception of Seoul and Incheon in region 2 under RCP8.5.

Discussion
In the development of emission control policies in the Asia-Pacific region, strategies can be implemented to prioritize the control of target species (NO x or VOC) for emission reduction. In the NO x -limited regime, a decrease in NO x emissions reduces NO 2 photolysis and, in turn, O 3 formation. In the NO x -saturated (or VOC-limited) regime, reduced VOC emissions lead to a decrease in the production of OH, HO 2 , and other organic (RO 2 ) radicals, resulting in decreased cycling with NO x and less O 3 formation. However, in the NO x -saturated regime, a decrease in NO x emissions can promote O 3 production via a weaker NO titration effect. In the transitional range between the two regimes, O 3 formation is sensitive to both NO x and VOCs. In this study, we selected space-borne lower free tropospheric column-integrated O 3 using a sophisticated retrieval algorithm and analyzed decade-scale O 3 trends over the entire Asia-Pacific Rim area. FNR analysis provided information on the sensitivity of O 3 to photochemical regimes regarding the efficiency of emission reduction over the Asia-Pacific region.
Our results from satellite data of the lower free tropospheric (900-700 hPa) O 3 levels showed a steady increase in all the Asia-Pacific areas considered at a rate of 0.21 ± 0.05 DU/decade (0.57 ± 0.12 × 10 16 molecules/cm 2 /decade) from 2005 to 2018 despite the reduction in NO 2 , especially since 2013 when China initiated its massive emission reduction strategy. The estimated FNR distributions in major cities in East China, South Korea, and Japan showed NO x saturation in megacities and transitional (or NO x -limited) levels across the Asia-Pacific region as a whole.
Our results indicate that the megacity-centered NO 2 emission reductions enforced a relatively weak titration effect and derived increasing O 3 tendencies under the NO x -saturated regime. Therefore, VOC reduction along with NO x reductions are needed in megacities in the immediate short term to reduce O 3 levels. To date, there have been a few reports on this phenomenon for the warm season (e.g., from Beijing-Hebei and Chengdo, as well as from rural areas near the Yangtze River delta for a specific warm period) [77][78][79][80][81][82][83]. Here, warm and cold season features have also been completely characterized to explain the annual mean O 3 enhancement over the Asia-Pacific region. The O 3 growth rates for both summer and winter over the Asia-Pacific region are shown in Figure S5 (Supplementary Materials). The summer O 3 growth rate was 3.8 ± 0.4%/decade, with a much greater increase during 2005-2011 than during 2012-2018 ( Figure S7). In regions 1-2, the pronounced enhancement of summer O 3 was driven mainly by increases in both HCHO and NO 2 during 2005-2011, whereas a small enhancement of summer O 3 was found despite lower NO 2 levels during 2013-2018 (see Figure S7). However, winter O 3 did not show a significant response to changes in either HCHO or NO 2 , mainly due to the low photochemical transformation caused by decreased photolysis rate in winter, thereby longer chemical lifetimes of both NO 2 and HCHO in winter with lower photochemical O 3 production. In addition, reduced retrieval sensitivity in winter due to larger solar zenith angle is also one of the reasons for this result. Only a significantly reduced NO x titration effect was expected in winter [20,83] in the aftermath of the rapid decrease in NO 2 , yielding a more pronounced increase in winter O 3 during 2013-2018, as shown in Figure S7. These phenomena, the summer HCHO increase, especially in 2005-2012, and the winter NO 2 decrease in 2013-2018, both led to an enhancement of the annual average O 3 in regions 1-2, reflecting the NO x -saturated O 3 sensitivity. In regions 3-4, a slight rise in summer O 3 with a small decrease in NO 2 was evident, whereas there was only a slight increase in the winter O 3 response, due in part to a slight rise in HCHO levels, from 2013 to 2018 ( Figure S7). This is attributable to the impact of O 3 transport over the Asia-Pacific region, which can have a small or large effect.
As for future O 3 , three projected IPCC-RCP emission reduction scenarios (RCP2.6, RCP4.5, and RCP8.5) were explored and indicated that the future NO x -saturated regime in major megacities would be prolonged until 2025, but would gradually shift to NO x -limited conditions by 2050 in most of the region's megacities where NO x emission regulation will be the key factor in controlling and reducing O 3 levels in the area, except for RCP8.5 in Seoul and Incheon, South Korea. Accordingly, the short-term policy of NO x emission regulation in megacities in 2025-2050 will be the focus for controlling and reducing O 3 levels over the Asia-Pacific region. The current study contains hypotheses of future emission scenarios. It will be necessary to continuously monitor the behavior of O 3 and its two precursors VOC and NO x to develop a "measured" approach for the most effective O 3 control over the Asia-Pacific region.