127I in aerosols from Xi’an, China: influence of East Asian monsoon and heavy haze events

Aerosol iodine isotopes are pivotal links in atmospheric circulation of iodine in both atmospheric and nuclear sciences, while their sources, temporal change and transport mechanism are still not well understood. This work presents the day-resolution temporal variation in iodine-129 (129I) and iodine-127 (127I) concentrations in aerosols from Xi’an, north-west China, during 2017/18. Both iodine isotopes have significant fluctuations with time, showing the highest levels in winter, approximately 2–3 times higher than in other seasons, but the correlation between 129I and 127I concentrations reflects that they have different sources. Aerosol 127I concentrations are found to be noticeably positively correlated with air quality index and five air pollutants. Enhanced fossil fuel combustion and inverse weather conditions can explain the increased concentrations and peaks of 127I in winter. The change in 129I concentrations confirms that the source and level of 129I in the monsoonal region were alternatively dominated by the 129I-enriched East Asian winter monsoon and the 129I-poor East Asian summer monsoon. The mean 129I/127I number ratio of (92.7±124)×10−10 provides an atmospheric background level for the purpose of nuclear environmental safety monitoring. This study suggests that locally discharged stable127I and externally input 129I are likely involved in fine particles formation in urban air, which provides insights into the long-range transport of air pollutants and iodine’s role in particulate formation in urban atmosphere.


Introduction
Iodine is one of the active halogen elements and is involved in plenty of atmospheric chemical reactions (i.e. ozone depletion and new particle formation from condensable iodinecontaining vapours), drawing increasing attention in not only atmospheric science but also environmental fields in recent years (Saiz-Lopez et al., 2012). A number of studies on atmospheric iodine just focus on the processes and mechanisms in marine boundary layer since over 99.8 % of iodine is derived from ocean (McFiggans et al., 2000). Other sources of iodine in air are comprised of volatile iodine and resuspended particles from soil, as well as combustion of fossil fuel (Fuge and Johnson, 1986). Whitehead (1984) estimated annual release of iodine from fossil fuel combustion is about 400 t, accounting for only 0.1 % of total iodine in air. In contrast, anthropogenic iodine in Chinese megacities is believed to be significantly underestimated due to coal combustion (Wu et al., 2014). A few studies have shown high iodine concentrations in air and particles in China (Gao et al., 2010;Xu et al., 2010). Although marine atmospheric iodine has been proven to form fine particles, little is known about terrestrial atmospheric iodine, particularly in urban sites with severe air pollution.
Along with atmospheric circulation of stable 127 I, longlived radioactive 129 I with a half-life of 15.7 million years is also of importance to global transport since it is a major Published by Copernicus Publications on behalf of the European Geosciences Union.
fission product with a yield of 0.7 % from nuclear industry. China is in a transition phase of energy structure in order to solve environmental pollution issues and has put great emphasis on developing nuclear power (World Nuclear Association, 2017). Nuclear waste reprocessing is also in the process of construction in China, which may be a key source of 129 I in the future. Investigation of the level, sources and temporal changes is extremely necessary for nuclear environmental safety assessment and nuclear emergency preparedness. Environmental 129 I/ 127 I number ratios have been increased from the natural 129 I level of 10 −12 to an anthropogenic level beyond 10 −10 in the modern environment due to atmospheric nuclear weapon testing, nuclear accidents, nuclear fuel reprocessing process, etc (Snyder et al., 2010). More than 95 % of the environmental 129 I was discharged by the two European nuclear fuel reprocessing plants (NFRPs), Sellafield in United Kingdom and La Hague in France, to the seas and air in liquid and gaseous forms, respectively. As a consequence of 129 I releases from NFRPs, nuclear accidents and nuclear weapon testing sites, the global distribution of 129 I is rather uneven (Snyder et al., 2010). Atmospheric 129 I investigations have been conducted in Europe, Japan, the United States and Canada, but aerosol 129 I studies are still rare, and no aerosol 129 I data are available in China at present (Hasegawa et al., 2017;Hou et al., 2009;Jabbar et al., 2013;Moran et al., 1999;Toyama et al., 2013;Xu et al., 2013). Furthermore, those previous studies present time series of 129 I in aerosols in monthly resolution for the purpose of nuclear environmental monitoring. Such a low time resolution is not sufficient to understand the sources, transport and temporal variation pattern and how these factors influence 129 I.
Here, we present a day-resolution temporal variation in 129 I and 127 I in aerosols during 2017/18 from a typical monsoonal zone, Xi'an city in the Guanzhong basin of northwest China, to make an attempt to investigate the level, sources and temporal change characteristics of 127 I and 129 I. This study will help to establish a background value of the 129 I/ 127 I number ratio, thereby serving nuclear environmental safety monitoring. The possible factors influencing temporal variation in iodine isotopes are also explored, including meteorological parameters, the East Asian monsoon (EAM) and heavy haze events.

Materials and methods
The aerosol samples were collected using a high-volume sampler on the roof of Xi'an AMS Center (34 • 13 25 N, 109 • 0 0 E) with an elevation of 440 m above mean sea level (Fig. 1). Xi'an, located in the Guanzhong basin, is the largest city in north-west China with a population of 9.9 million. The basin is nestled between the Qinling Mountains to the south and the Loess Plateau to the north, and it is a warm temperate zone with semi-humid continental monsoon climate (Fig. 1b).
Sixty-eight aerosol samples were selected for measurement of iodine isotopes using the pyrolysis combined with AgI-AgCl co-precipitation for separation. The sample collection and preparation procedure are described in detail in the Supplement (Sect. S1), as previously reported (Zhang et al., 2018b). Accelerator mass spectrometry (AMS; 3MV, High Voltage Engineering Europa B.V., the Netherlands) and inductively coupled plasma mass spectrometry (ICP-MS; Agilent 8800, USA) were applied for a determination of 129 I/ 127 I number ratios and 127 I concentrations, respectively. 129 I/ 127 I number ratio of iodine carrier is less than 2×10 −13 , and the analytical precisions are within 5 % for all the aerosol samples. The mean values were 6.22 ± 4.48 ng m −3 , (1.97 ± 1.65) × 10 5 atoms m −3 and (92.7±124)×10 −10 for γ ( 127 I), N( 129 I) and the 129 I/ 127 I number ratio, respectively. 127 I and 129 I in aerosols are characterised with apparent monthly and seasonal variations (Figs. 3 and 4). The minimum and maximum monthly concentrations were observed in August and December for 127 I and July and December for 129 I. The average γ ( 127 I) values in November, December and January (11.4-12.7 ng m −3 ) were 2 times higher than in other months (3.12-6.70 ng m −3 ). Distinct from 127 I, monthly variation in 129 I shows the lowest level in June and July, (0.47-0.50) ×10 5 atoms m −3 , about 2 to 6 times lower than the other months. The maximum 129 I/ 127 I number ratio was not observed in winter months but rather in September.

Results
The average γ ( 127 I) values were 5.68 ± 2.24, 3.61 ± 1.49, 6.05±4.52 and 10.6±6.0 ng m −3 in spring, summer, fall and winter, respectively. The level of 127 I in winter was about 2 times higher than spring and fall, 3 times higher than summer. N( 129 I) values were (1.93±1.90)×10 5 , (1.17±1.55)× 10 5 , (1.92 ± 1.62) × 10 5 and (3.12 ± 0.72) × 10 5 atoms m −3 in spring, summer, fall and winter, respectively. The level of 129 I in winter was about 2 times higher than spring and fall and 3.3 times higher than summer. Seasonal variation in the 129 I/ 127 I number ratio was not so obvious as the concentrations of iodine isotopes. The mean 129 I/ 127 I number ratio of (119 ± 185) × 10 −10 in fall was slightly higher than that of (82.2±79.3)×10 −10 in spring, (71.5±89.3)×10 −10 in summer and (89.3 ± 70.5) × 10 −10 in winter. However, the ratios in all four seasons fell in the similar range as that of the whole year.
A weak correlation between 129 I and 127 I was found with a Spearman correlation coefficient of 0.33 (p < 0.01) for the whole year data, while no significant correlation was found between the two iodine isotopes in each season at the level of 0.05 (Table 1 and Fig. S1 in the Supplement). The correlation analysis between iodine isotopes and total suspended particle (TSP) indicate that there was a strong correlation between 127 I and TSP but no correlation between radioactive 129 I and TSP (Fig. S2).

Level and sources of 127 I and 129 I
The results of a weak correlation in the whole year sampling and no significant correlations in each season between the two isotopes indicate that 127 I and 129 I have different sources and influence factors.

127 I
The levels of 127 I concentrations, particularly in winter, are much higher than those in continental sites (below 0.61 ng m −3 at the South Pole and 2.7-3.3 ng m −3 in Mpumalanga -previously named the Eastern Transvaal) and comparable to those in coastal and ocean sites (typically below 20 ng m −3 and up to 24 ng m −3 in tropic marine aerosols) (Saiz-Lopez et al., 2012). A similar range of 127 I in TSP was observed to be 4.5-22 ng m −3 in a coastal urban area, Shanghai, China, showing the lowest values in summer, and an increase occurred in fall and winter (Gao et al., 2010). Iodine associated with PM 10 and PM 2.5 (particulate matter with the aerodynamic diameter less than 10 and 2.5 µm, respectively). was found to be 3-115 and 4-18 ng m −3 , respectively, in urban and island sites of Shanghai, slightly lower than TSP iodine (Cheng et al., 2017;Gao et al., 2010). The marine aerosol iodine offshore China was found to be below 8.6 ng m −3 during the R/V XueLong cruise from July to September 2008 (Xu et al., 2010). The results suggest a relatively high aerosol 127 I level in both inland and coastal urban areas in China.
Natural iodine in air is from marine emission through sea spray and gaseous emissions from seas, weathering of base rock and continental release through vegetation and suspended soil particles (Carpenter et al., 2013;Fuge and Johnson, 1986). Due to the influence of south-easterly EASM, Table 1. Spearman correlation coefficients between iodine isotopes and atmospheric pollutants and weather conditions * .  Orange bands indicate five heavy haze episodes corresponding with five 127 I peaks. Three dark and two light grey shades in (b, c) demonstrate the high-level and low-level periods, HLP and LLP, respectively, for 129 I and the 129 I/ 127 I number ratio, alternatively dominated by the EAWM and EASM, respectively. moisture from the Pacific Ocean and the Chinese seas might bring marine iodine. Whereas, the mean γ ( 127 I) in summer aerosol is 3.61±1.49 ng m −3 , about 3-fold lower than that in winter. The sampling location, Xi'an, is an inland city about 900 km away from the nearest coastline. The contribution of marine iodine to terrestrial surface system in winter is considered to be negligible when the site is over 400 km away from the ocean (Cohen, 1985). Taking sodium and calcium as reference elements for sea spray and direct volatilisation of iodine from the ocean and weathering of soil and rock, respectively, He (2012) has estimated that less than 0.04 % and 5.2 % of iodine was from direct marine contribution and weathering of soil and rock, respectively, in the precipitation in Zhouzhi County, Xi'an city. Despite being likely underestimated, marine iodine contribution in precipitation samples showed a declining trend with increasing distances of 20 to 1252 km from the sea. And no significant change in marine contribution could be found over 100 km from the sea (He, 2012).
Iodine is also emitted from volatility of terrestrial soil and respiration of vegetation, which was estimated to be 2.27 µg m −2 d −1 in the form of CH 3 I, on a global basis, over an active season of 240 d, together with biome areas for temperate forest and wood lands (28.5 × 10 12 m 2 ) and temperate grasslands (31.9 × 10 12 m 2 ) (Sive et al., 2007). Dry deposition flux of iodine, however, can be calculated to be 5.83-40.7 µg m −2 d −1 based on aerosol 127 I mass concentrations in TSP (13.3-92.5 µg g −1 TSP) multiplying an annual average dustfall flux of 13.2 t (km −2 30 d −1 ) (Xi'an Bureau of Statistics, 2018). The uncertainty for the calculation is about 32 % mainly due the large uncertainty in dustfall flux of about 31 %. Because of different land coverage between urban and forest-grassland in reference of Sive et al. (2007), terrestrial emission of iodine in the sampling site should be even lower than 2.27 µg m −2 d −1 . The dry deposition flux of iodine in Xi'an was therefore far beyond terrestrial sources of soil and vegetation, indicating they might be important iodine sources in summer but not in winter.
The significant increase in 127 I from summer to winter suggests that anthropogenic discharge of iodine is the domi- nant source of 127 I in Xi'an aerosol samples, mainly including combustion of biomass and fossil fuel (Wu et al., 2014). Biomass combustion generally occurs in summer harvest time, normally in later May and early June. In order to improve air quality, Xi'an government has banned biomass combustion since 2009. Additionally, no obvious change in 127 I concentrations was found in May and June, indicating the biomass combustion is not the major source.
A recent study has confirmed that particulate iodine around two coal plants in Nanchang city, China, was greatly increased up to 36 ng m −3 , and iodine concentrations within 9 km from the coal plants were much higher than that in non-coal sites (Duan, 2018). Coal is dominant in energy consumption structure. Coal consumption accounted for 72.7 % of total energy consumption in Shaanxi Province in 2013. In 2017, the coal consumption in Guanzhong basin was 67.4 million t (Shaanxi Provincial Bureau of Statistics, 2018). The 127 I concentration in coal produced in Shaanxi Province ranges from 0.39 to 6.53 µg g −1 with a mean value of 1.47 µg g −1 (Wu et al., 2014). An atmospheric iodine emission factor that equals to the ratio of the iodine released into the atmospheric from the coal is from 78.8 % to 99.4 %, depending on the coal combustion technology and emission control devices (Wu et al., 2014). If simply assuming anthropogenic iodine is solely from combustion of coal in the study area and the atmospheric iodine emission factor is 92 %, about 91 t of 127 I could have been released into the atmosphere in the Guanzhong basin in 2017. Xi'an, a northern city in China, consumes more coal in the heating period from 15 November to 15 March, which aggravates iodine release from coal combustion. Thus, we suggest that coal combustion is the major source of 127 I in Xi'an urban aerosols in particular during the heating period of winter. This also suggests that 127 I was regionally or locally input and can be treated as internal release.

129 I
The aerosol 129 I levels reported in the previous studies and this work can be categorised into three groups (Fig.5).
(1) Compared to other investigating sites, aerosol N( 129 I) was less than 10 6 atoms m −3 in Xi'an, north-west China. This low level is also found at those sites that are remote from the nuclear facilities in southern and central Europe, as well as Japan before the Fukushima accident (Hasegawa et al., 2017;Jabbar et al., 2013;Santos et al., 2005). The lowest N( 129 I) (< 0.1×10 5 atoms m −3 ) in aerosols have been found at two high-altitude sites of the Alps (about 3000 m above the sea level). (2) The high values beyond 10 8 atoms m −3 have been reported at the sites directly contaminated either by nuclear reprocessing plants, such as Hanford, Sellafield and the reprocessing plant (WAK) at Karlsruhe, or by the Fukushima nuclear accident in 2011 (Brauer et al., 1973;Jackson et al., 2002;Wershofen and Aumann, 1989;Xu et al., 2015).
(3) In between, aerosol N( 129 I) within the range from 10 6 to 10 8 atoms m −3 is mainly found in the sites and periods with global fallout from atmospheric nuclear weapon testing and from indirect contamination from nuclear fuel reprocessing plants (Brauer et al., 1973;Englund et al., 2010;Kadowaki et al., 2018;Tsukada et al., 1991;Zhang et al., 2016).
The source of 129 I is crucial for spatial and temporal distributions of 129 I in global scale. 129 I/ 127 I number ratios in the Xi'an aerosols range from 10.6 × 10 −10 to 743 × 10 −10 , at least 3 orders of magnitude higher than the naturally produced 129 I level (1.5×10 −12 ) (Fehn et al., 2005). This clearly indicates human nuclear activities are a dominant contributor to the increase in 129 I level in the environment. The level and source of 129 I in soil, vegetation, rain and river water samples have been previously investigated in the Xi'an region, where 129 I/ 127 I varied from 1.1×10 −10 to 43.5×10 −10 with a mean value of 20.6 × 10 −10 . Aerosol 129 I/ 127 I number ratios were about 1 order of magnitude higher than those in other environmental media, indicating 129 I in Xi'an aerosols was not released by local soil suspension and vegetation release. Weathering of bedrock is not a major source of airborne 129 I, since weathering just contributes 5 % of stable iodine, and 129 I in bedrock can be considered even lower than the nature-produced 129 I level because of the continuous decay. Coal combustion contributes a large proportion of stable 127 I in winter, while the amount of 129 I in coal is almost negligible, because coal was formed in the Tertiary (2.58-66 million years) at the latest so that 129 I has been decayed out or in an extremely low value of 10 −13 -10 −10 for 129 I/ 127 I. Thus, coal combustion is not a major source of atmospheric 129 I.
Nuclear activities including the historic nuclear weapon testing sites, nuclear reactors, NFRPs in China and Europe, and underground nuclear weapon testing are considered. Two nuclear weapon testing sites, Semipalatinsk and Lop Nor, located upwind, may input 129 I into Xi'an region through soil resuspension and gaseous re-emission. However, evidence from the 129 I distribution in surface soils from upwind regions reveals that the two nuclear weapon testing sites have a limit impact on the atmospheric 129 I level in the remote regions farther than 1000 km from these test sites (Fan, 2013). This is also supported by the back-trajectory analysis that 129 I concentration did not significantly rise when abundant air masses from Xinjiang passed through the Lop Nor test site on 28 December 2017 (Fig. S3g). Five nuclear power plants are in operation along the south-east coastal areas in China. 129 I data from seawater collected within 10 km from a Chinese nuclear power plant suggest that normal operation of reactors does not cause a significant increase in 129 I concentrations . Although information on the gaseous release of 129 I from these reactors is unknown, the low 129 I/ 127 I (about 7 × 10 −10 ) in the surface soil of southern China (Guangxi, Jiangxi and Fujian provinces) close to the reactors can confirm that there is no marked deposition from the gaseous release (Fan, 2013). Toyama et al. (2013) have shown a direct nearby influence of a pilot plant in Tokaimura (Ibaraki Prefecture), Japan on the 129 I deposition in Tokyo. Similarly, a pilot nuclear spent fuel reprocessing plant (NFRP) has been established and operated in Gansu Province, China, since 2010. This NFRP is located in an upwind area and about 1200 km north-east of Xi'an. During the sampling period in 2017/18, no abnormally high 129 I was observed; however this contribution cannot be neglected in the future operation and should be continuously monitored. In addition, the possible influence of the sixth underground nuclear weapon test conducted by North Korea on 3 September 2017 has been excluded based on the back and forward trajectories and the nuclear environmental monitoring around the Chinese north-east border by the government (Ministry of Environmental Protection of the People's Republic of China, 2017).
It is well documented that gaseous and liquid discharges from the NFRPs in Sellafield, United Kingdom, and La Hague, France, as well as the secondary emission from the contaminated seas and land, are the predominant source of 129 I in the modern atmosphere, particularly in the European environment (Jabbar et al., 2013). The two NFRPs are located from 50 to 55 • N, the westerly belt. The prevailing westerly winds throughout the year in the midlatitude act as a crucial pathway of 129 I transport from its source to the whole midlatitude regions of the Northern Hemisphere, as observed in the sediment core from Jiaozhou Bay on the east coast of China (Fan et al., 2016). The 60-year record of 129 I in a lacustrine sediment from the Philippines further shows that the EAWM plays an important role in transporting the midlatitude 129 I to the low-latitude regions (Zhang et al., 2018a). The feature of 129 I variation also shows that 129 I was at a high level in spring and winter when the EAWM is prevailing and at a low level in summer when the EASM is prevailing, supporting that 129 I is dominantly sourced from the long-range transport of European NFRPs discharges. In this case, 129 I is externally input, in contrast to the locally released stable 127 I.

Factors influencing temporal variation in iodine isotopes
As discussed above, even though variation patterns of 127 I and 129 I were similar, they were considerately influenced by many factors owing to their different sources. In this work, meteorological factors including precipitation, wind speed, temperature and dust storm events including, atmospheric circulation (in particular the EAM) and heavy air pollution periods are discussed.

Meteorological factors
Precipitation and wind speed. As discussed in Sect. S2, the influences of precipitation and wind speed on temporal changes in iodine isotopes are not significant ( Fig. 2e and f). However, the winter days with an absence of wet precipita- tion and lower wind speed corresponded well to the heavy haze episodes when iodine concentrations, in particular stable 127 I, were greatly increased, indicative of less dispersion. The details about haze influence on iodine will also be discussed in the following section.
Temperature. Temperature and its associated physiochemical processes and biological release of iodine from source regions might be reasons for the variation patterns. In summer, the temperature is about 20-40 • C in the North Hemisphere, which is favourable for direct volatilisation of iodine from the surfaces of land and seas. Ozone in the air-sea boundary layer is suggested to act as an oxidant to transform iodide in seawater into volatile molecular iodine that enters into the air, which is believed more significant than the biological process (Carpenter et al., 2013). Ozone concentrations in summer are around 30 nmol mol −1 , roughly 2 times higher than winter (Ayers et al., 1996), which may increase the re-emission rate of iodine from the ocean and 129 I-contaminated sea surface into the air. Additionally, the bloom of phytoplankton and algae in summer, can release biogenic organic iodine into the air through a mechanism of anti-oxidation (Küpper et al., 2008). The temperature, ozone concentration and marine biomass greatly reduces in winter, which results in less iodine released from the source regions and this can be used to explain the relatively weak peaks in winter than in summer. As discussed above, 127 I and 129 I in Xi'an aerosols were mainly derived from coal combustion and long-range transport from Europe, respectively. The change in the release amount of 127 I and 129 I at the source regions is obviously not the determining factor for the changes in iodine isotopes since Xi'an is far from the oceans and the 129 I source regions. Furthermore, the seasonal variation in 127 I and 129 I with low levels in summer can also easily exclude the possibility of temperature influence.
In addition to atmospheric stability reflected by precipitation, wind and temperature, atmospheric boundary layer height determines the vertical dispersion scale of air pollution by thermal turbulent mixing, which might be a factor for variation in iodine isotopes. Since 127 I is locally input and 129 I is remotely transported from Europe, the influence of boundary layer height might be different for the two iodine isotopes. It will be further explored with longer temporal variation in iodine isotopes in the future.
Dust storms. Two severe dust storm events occurred in Xi'an in 17-18 April and 4-6 May 2017, as indicated by the peaks of air quality index (AQI) of 268 and 306, respectively. A 127 I peak, 11.0 ng m −3 , was observed on 18 April 2017, while 127 I levels in other samples were almost below 6 ng m −3 in spring and summertime. Dust storms frequently occur in winter and spring in northern China and normally originate from the arid and semi-arid desert regions mainly located in Mongolia and north-west China. The first dust storm arrived the Guanzhong basin on 17 April 2017 and lasted until 19 April (China Meteorological Administration, 2018). The small peak of 127 I is likely attributed to the suspended particulate matter from the soil surface in the dust storm source. In contrast, variation in the 129 I level did not reflect the dust storm influence. The fact that 129 I was not correlated with particulate concentrations (Fig. S4), indicates that the extrinsic 129 I is not related to the heavy particulate events, since the major dust source areas include Taklimakan desert, the Gobi Desert in Inner Mongolia and the Loess Plateau, where the 129 I/ 127 I number ratios in surface soil fell below 60 × 10 −10 , apparently much lower than those in aerosols . Furthermore, the back-trajectory anal-ysis also showed that the low 129 I level on 18 April can be partially attributed to an 129 I-poor low-altitude air mass (< 900 m) (Fig. S3a). This is because either the low-altitude air mass might be formed in 129 I-poor inland areas, not from the 129 I-rich European area, or else the long-range-transported 129 I in low-altitude air mass could be easily lost by the topographic countercheck (Dong et al., 2018).
The second dust storm started from the south central Mongolia and west-central Inner Mongolia autonomous regions on 3 May, arrived at Xi'an on 5 May and retreated on 6 May. It is a pity that no sample was analysed in this event, but a significant 129 I peak with value of 7.53 × 10 5 atoms m −3 was found after 3 d of this event (Fig. 2b). The back-trajectory analysis suggests the 129 I peak on 8 May 2017 is found to relate to the downdraft originating from high altitude (2000-6000 m) to low altitude (500 m) (Fig. S3b). This elevation of 129 I after the dust storm events is likely attributed to the intensified winter monsoon and strong cold high pressure transporting greater 129 I from Europe to China.

Heavy haze episodes during 2017/18 winter
A significantly positive correlation between 127 I and air quality index (AQI) was found with a high Spearman correlation coefficient of 0.72 (p 0.05) for the whole-year sampling period and an increased coefficient of 0.87 in winter ( Table 1). The 127 I concentration in winter can reach up to 10 times as much as in summer (Fig. 2a). Furthermore, five 127 I peaks from 12.8 to 21.4 ng m −3 were clearly identified on 15 and 29 November, 14 and 28 December, and 16 January, respectively, which coincided well with the heavy haze episodes, with AQI mostly over 200, namely heavily polluted air (Fig. 2e). As discussed in Sect. 4.1, the irrelevance between 127 I and 129 I in aerosols for each season attributed to their different sources, also demonstrates that locally discharged iodine and externally input iodine are not contemporaneously subjected to formation of iodine-containing particles.
Further analysis showed close relationship between 127 I and six air pollutants, including PM 10 , PM 2.5 , CO, SO 2 , NO 2 and O 3 (Table 1 and Fig. S4). In spring and summer, the high correlation between 127 I and AQI can be attributed to the high correlation of 127 I with PM 10 and PM 2.5 . In fall and winter, 127 I, is significantly positively correlated with PM 10 , PM 2.5 , CO, SO 2 and NO 2 and negatively correlated with O 3 . In contrast, there is no such good agreement between 129 I and these gaseous pollutants. Despite that, three 129 I peaks were found on 15 November 2017, 14 December 2017 and 16 January 2018, which corresponded well with high 127 I concentrations ( Fig. 2a and b) during the haze episodes. This reflects that the formation mechanism of iodine-containing aerosols might be seasonally different. However, the three peaks of 129 I in aerosols during the heavy haze episodes suggest that local and external iodine are likely subjected to subsequent growth of particles and capture by particles due to a longer residence time in stagnant weather conditions.

Impact of the EAM for long-range transport of 129 I
Increasing evidence has suggested that the prevailing westerly and the EAM system act as crucial driving forces and pathways for transport of the 129 I derived from European NFRPs from Europe to East Asia and even low-latitude south-east Asia (Fan et al., 2016;Zhang et al., 2018a). Monthly variations in atmospheric 129 I in Japan also showed a clear pattern of low 129 I deposition in summer and high 129 I deposition in winter, which is also attributed to the impact of the EAM (Hasegawa et al., 2017;Kadowaki et al., 2018;Toyama et al., 2013). In this work, seasonal variation in 129 I was identical to the observation in the previous studies (Toyama et al., 2013). However, the day-resolution variation patterns of 129 I and 129 I/ 127 I in Xi'an, distinct from monthly variation in Japan, showed three periods with high levels and two periods with low levels, indicating a more complex influence of the EAM in the typically continental monsoon climate city, Xi'an. The whole-year time series can be divided into five periods with three high-level periods (HLP) from (a) late March to early May (HLP 1), (b) middle August to early September (HLP 2) and (c) middle November 2017 to late February 2018 (HLP 3), as well as two low-level periods (LLP) from (d) early May to middle August (LLP 1) and (e) middle September to early November 2017 (LLP 2) ( Fig. 2b and c). 129 I levels in the three HLPs fell within the range of (1.98-2.41) ×10 5 atoms m −3 , which is 3-5 times higher than those during the two LLPs with (0.49-0.66) ×10 5 atoms m −3 (Table S1). The relative standard deviation shows much higher variability during HLP 1 and 2 from 91 % to 109 % in contrast to the variability in other clusters of less than 60 %. The significant difference between the HLPs and LLPs suggests the transportation process of 129 I is obviously distinct. The westerly is a crucial driving force of 129 I from the NFRP point sources and their contaminated seas, which are labelled as having a high 129 I level, up to 10 −6 for the 129 I/ 127 I number ratio (Michel et al., 2012; (Fig. 1a). Due to interplay between the westerly and EAWM , EAWM inherits the high 129 I feature of 10 −7 -10 −9 for a 129 I/ 127 I number ratio in the long-distance transport process. Therefore, the HLP 1 and 3 were strongly affected by the EAWM prevailing from early September to early May in 2017. Compared to the violent fluctuation of 129 I in spring (HLP 1), the weak fluctuations of HLP 3 in winter might be attributed to a relatively stable interaction process between the strengthened westerly and the EAWM. In addition, the 129 I level in March 2018 was much less than that in March 2017, seeming to be a consequence of a weaker EAWM strength in March 2018 compared to in March 2017. This is in good agreement with the EAWM in-dices of 2.04 in 2017 and −1.86 in the 2018 MODES forecast monitor (National Climate Center of China Meteorological Administration, 2019). HLP 2 was a different case from HLPs 1 and 3, since the period should be under control of the EASM.
The EASM originates from the Pacific and Indian tropical regions under the role of subtropical highs and transports moisture from the ocean to East Asia starting in early summer. 129 I/ 127 I number ratios in the Pacific Ocean, the East China Sea and the Indian Ocean are as low as 10 −10 Povinec et al., 2011). Even after the Fukushima accident, 129 I/ 127 I number ratios are still less than 40 × 10 −10 in the western Pacific Ocean (Guilderson et al., 2014). Thus, the EASM is poor in 129 I in comparison to the winter monsoon. This is in good agreement with the low 129 I level during the two LLPs (Fig. 2b). The 850 hPa water vapour transmission flow field showed the south-east wind moisture moving northward to the north of 35 • N on 2 May, followed by another two outbreaks on 21 May and 3 June (Fig. S5), indicative of a EAWM retreat and a EASM advance. During this period, 129 I dropped abruptly from 3.45×10 5 atoms m −3 on 27 April to 1.10 × 10 5 atoms m −3 on 2 May, followed by a maximum on 8 May; then there was a sudden decline to 0.64 × 10 5 atoms m −3 on 15 May. The violet fluctuation of 129 I is likely caused by the onset of the EASM, which is quite violent with stepwise northward jumps. This conclusion is fully supported by previous metrological observations (Ding and Chan, 2005). As the EASM moved into the active stage in mid-May, the 129 I level was low and in a relatively stable state, as shown in LLP 1.
After the active stage of the EASM, however, it is unexpected that increased and variable 129 I levels were observed from middle August to early September (HLP 2). The 129 I peak on 6 September 2017 was the highest throughout the sampling year. The back-trajectory model shows that five low-altitude air masses (< 1000 m above ground level) from the Baltic Sea moved fast eastward and arrived at the Guanzhong basin within 5 d (Fig. S3e). The Baltic Sea contains a high 129 I concentration due to the water exchange with the North Sea that receives over 100 kg yr −1 129 I from La Hague and Sellafield NFRPs (Snyder et al., 2010). Therefore, a 129 I peak observed here indicates that there is interplay between the 129 I-enriched westerly and the EASM, the latter of which was retreating to the south. It is reported that Xi'an enters into the EASM break stage during this time based on rainfall data (Ding and Chan, 2005). The intensive interaction between westerly winds and the EASM facilitates the formation of rainfall at their confluence area, resulting in the drastically fluctuating 129 I levels. Therefore, the elevated and variable 129 I levels in HLP 2 can be attributed to the EASM break stage.
After the break stage with significant 129 I fluctuation, the second LLP of 129 I from 21 September to 11 October (LLP 2) occurred when the summer monsoon moved into the revival stage (Fig. 2b). Despite being lower than in the break period, the 129 I level in this period was slightly increased from 0.49 × 10 5 atoms m −3 in the active stage to 0.66 × 10 5 atoms m −3 in the revival stage. After the active-break-revival cycle of summer monsoon, reflected by a low-high-low 129 I level, the 129 I level increased stepwise since mid-October, suggesting the EAWM had taken the place of the EASM in the Guanzhong basin and lasted until March next year.
To quantitatively characterise the influence of the EAM on the variation in 129 I, Z-normalised 129 I concentrations and 129 I/ 127 I number ratios were used to build a quantitative model during the winter monsoon and different stages of the summer monsoon including onset, active, break, revival and retreat (Fig. 6). The value of Z( 129 I) varies from −1.11 to 3.38 with a median value of −0.29 and Z( 129 I/ 127 I number ratio) from −0.66 to 5.26 with a median value of −0.34. Based on the observation during 2017/18 in the Guanzhong basin, when Z( 129 I) is less than −0.5 and Z( 129 I/ 127 I number ratio) is smaller than 0, this period is in good agreement with the onset, active and revival stages of the EASM. During the stable active stage, Z scores for 129 I and 129 I/ 127 I were minimal, which was followed by the second lowest value during the revival stage. The onset and break stage showed a much larger fluctuation with Z scores changing from −0.8 to −0.3. The break stage of the East Asian summer monsoon is an exception, which exists with alternative influences from both factors in our studied region. The Z( 129 I) values from 1.57 to 1.96 for the break stages were much higher than the period controlled by the East Asian winter monsoon with Z( 129 I) from −0.5 to 1.53. This result clearly confirms that the EAM plays a decisive role on the temporal variation and long-range transport of not only 129 I but also other air pollutants (i.e. persisting organic pollutants or inorganic air pollutants) in Chinese monsoon-affected regions.

Atmospheric background level of 129 I/ 127 I number ratio
For the purpose of nuclear environment safety monitoring, the average 129 I/ 127 I number ratio of (92.7 ± 124) × 10 −10 can be simply regarded as the atmospheric background level of 129 I in north-west China. The previous studies on 129 I environmental baseline have never carefully investigated the influence of weather on time variation in 129 I. Here our dayresolution 129 I dataset in this monsoon climate city showed that time variation in the atmospheric baseline level related to metrological conditions, heavy haze events and atmospheric circulation, has to be carefully considered and used for better evaluation of the impact of possible nuclear incidents in a practical way. Particularly, a pilot nuclear reprocessing plant located upwind to Xi'an, might be extended and will be a source of radionuclides in the future. The baseline established in this work is, therefore, of significance to monitor nuclear environmental safety for the long-term and to sensitively assess the impact of nuclear incidents and apply on environmental process tracing. Figure 6. Two-dimensional graph of Z-score-normalised 129 I concentrations and 129 I/ 127 I number ratios, suggesting the refined features of the East Asian summer (onset, active, break and revival in yellow diamond, green triangle, red circle and pink circle, respectively) and winter monsoons (WM; black dot) (a). The coloured symbols clearly demonstrate a detailed onset-active-break-revival cycle for the summer monsoon with Z129 I ≤ −0.5 and Z Ratio ≤ 0, as illustrated in the blue oval area (b).

Conclusions
The study firstly presents a high-resolution temporal variation in atmospheric 127 I and 129 I in north-west China, showing the vivid seasonal characteristics of iodine isotopes and an 129 I/ 127 I baseline ratio of (92.7 ± 124) × 10 −10 . Variation in 127 I strongly linked with atmospheric pollution and heavy haze episodes, particularly in winter, indicates that 127 I in Xi'an aerosols is mainly derived from combustion of fossil fuel. Aerosol 129 I mainly originates from European nuclear reprocessing plants through long-range transport, and its temporal variation is strongly dominated by the interplay of the East Asian winter and summer monsoon. Previous studies on temporal changes in atmospheric 129 I in other monsoonal regions showed a simple pattern with the lowest level in summer and the highest in winter, while our dayresolution dataset showed that high 129 I levels could be found in summertime due to the break of the East Asian summer monsoon. The locally input 127 I and exogenous 129 I were greatly increased during haze events, reflecting the possible role of iodine in the formation of urban fine particles, therefore, further investigations are expected to focus on the speciation of iodine isotopes for mechanism study of iodine's impact on air pollution.
Data availability. All the data generated during the study appear in the article.
Author contributions. LZ, XH and SX designed and optimised the experiment. LZ and NC performed the experiment, with the help of PC and YF. TF collected the air pollutant data. LZ, TF, PC and YF created the figures. The data analysis and interpretation were carried out by LZ, XH, SX, TF and NC. LZ prepared the paper, with contributions from all coauthors.
Competing interests. The authors declare that they have no conflict of interest.
Acknowledgements. Luyuan Zhang gratefully acknowledges Qi Liu, Miao Fang, Yanyun Wang, Lin Wang and Jie Zhou at IEECAS for their kind help on AMS measurement, sample collection and figure drawing. We acknowledge the two anonymous referees and editor for their constructive comments. Review statement. This paper was edited by Jan Kaiser and reviewed by two anonymous referees.