Springtime influences of Asian outflow and photochemistry on the distributions of diacids, oxoacids and α-dicarbonyls in the aerosols from the western North Pacific Rim

Abstract Total suspended particle (TSP) samples were collected at Cape Hedo, Okinawa on a basis of 24, 12 and 3 h intervals in March to April 2007, when the Asian outflow is enhanced. The filter samples were analysed for dicarboxylic acids, oxoacids, benzoic acid and α-dicarbonyls to better understand the sources, chemical composition and photochemical ageing of organic aerosols during long-range transport. Their molecular distributions showed a predominance of oxalic acid (C2) followed by malonic (C3) and succinic (C4) acids. Day/night samples did not show clear diurnal variations, suggesting that local anthropogenic influence is not important. The highest concentrations of C2 and other diacid species were observed in April 10 when air masses were delivered from north China. Higher concentrations were observed when air masses arrived from East Asia and coastal China. The 3 h samples showed higher concentrations of oxalic acid at 16:30 or at 13:30 during April 14 and 16, whereas higher relative abundance of C2 in total diacids and higher concentration ratios (C2/C3, C2/C4, C3/C4, C2/Gly, and C2/ωC2) were observed at 7:30 during April 14–16. These results suggest that aerosols were accumulated in the atmosphere during night and photochemically modified early in the morning. Oxalic acid–C/OC did not show a prominent peak in 3-h samples. Temporal variations in the concentrations of diacids, oxoacids and α-dicarbonyls, relative abundances and diacid–C/OC ratios together with 5 day back trajectory analyses show that the aerosols from Cape Hedo are strongly influenced by long-range atmospheric transport from East Asia, especially in China. Strong correlations of C2 with nss-SO42− and NO3− in night-time suggest a secondary formation of water-soluble diacids via heterogeneous aqueous phase oxidation during long-range transport.


Introduction
Atmospheric aerosols contain high abundance of water-soluble dicarboxylic acids (Grosjean et al., 1978). The abundant presence of diacids in aerosols increases their hygroscopic properties (Saxena et al., 1995), affecting their cloud condensation nuclei and ice nuclei activities (Novakov and Penner, 1993). Small organic acids may also have an adverse effect on human health (Highwood and Kinnersley, 2006). Chemical analysis of ambient aerosols from Okinawa, an outflow region of Asian dusts (Kunwar and Kawamura, 2014a) and transmission electron microscopy analysis of soot particles from Okinawa (Ueda et al., 2011) showed that aerosols are significantly aged during long-range atmospheric transport and heavily influenced from East Asian outflow. There are various sources of diacids in East Asia, including fossil fuel combustion (Ho et al., 2006), biomass burning of agricultural wastes , meat cooking operations (Fu et al., 2013) and vehicular emission (Wang et al., 2010) as well as secondary photo-induced oxidations of biogenic unsaturated fatty acids (Kawamura and Gagosian, 1987;Pokhrel et al., 2015Pokhrel et al., , 2016. However, the secondary production of diacids overwhelms the primary emission in the atmosphere (Kawamura, Steinberg et al., 1985;Kawamura and Bikkina, 2016).
Cape Hedo of Okinawa Island is situated in the outflow region of East Asia (Kunwar and Kawamura, 2014a) and has been used as a supersite of Atmospheric Brown Clouds (ABC) project to study the atmospheric transport of Chinese aerosols

Chemical analysis
Low molecular weight diacids, oxoacids, and α-dicarbonyls were analysed using methods described in Kawamura et al. (1996). Briefly, known area of filter was extracted with organic-free ultrapure water under ultra-sonication for 10 min (10 mL × 3). The extracts were passed through a glass column (Pasteur pipette) packed with quartz wool to remove insoluble particles and filter debris, and then concentrated near to dryness using a rotary evaporator under vacuum. Carboxyl and aldehyde groups were then derivatized to dibutyl esters and ω,ω-dibutoxy acetals, respectively, using 14% boron trifluoride (BF 3 )/n-butanol at 100 °C for 1 h, whereas keto group of pyruvic acid was converted to dibutoxy acetal. The derivatives were dissolved in n-hexane and quantified using a capillary gas chromatograph (GC; HP 6890). Identifications of the compounds were performed by comparing GC retention times with those of authentic standards and confirmed by mass spectral examination using GC/MS technique (Kawamura and Ikushima, 1993;Kawamura and Bikkina, 2016).
Recovery and reproducibility tests were performed before real sample analysis. The recovery for oxalic acid (C 2 ) was 85%, more than 90% for C 3 , C 4 , C 5 and C 6 (Kunwar and Kawamura, 2014b) and 90% for glyoxal (Kunwar and Kawamura, 2014b). Kawamura and Yasui (2005) reported that recovery of ωC 2 is and pollutants and their chemical transformation (Takami et al., 2007). The consumption and emission of anthropogenic sources are increasing day by day in East Asia (Ohara et al., 2007). Previously, we reported one-year observation on molecular distributions of diacids and related compounds and carbonaceous components in the Okinawa aerosols to better understand the seasonal variations of those water-soluble organics in the western North Pacific Rim Kawamura, 2014a, 2014b). Although there are few studies of diacids and related compounds in Okinawa and its surroundings (Kunwar and Kawamura, 2014a;Kundu et al., 2010;Deshmukh et al., 2016), there is no time series (e.g. 12 and 3 h) study to decipher the source and photochemical formation of organic acids. The lack of such studies motivates us for collecting time-resolved samples in this region. We identified a time-specific photochemical process of diacids and related compounds, which provides very useful information for future aerosol studies.
In this study, we report molecular distributions of diacids, oxoacids, pyruvic acid, benzoic acid and α-dicarbonyls in ambient aerosols from Cape Hedo, Okinawa on a basis of daily (24 h) and diurnal intervals (day/night and 3 h) to better understand the sources of organic aerosols and in situ photochemical production of water-soluble organic species during long-range atmospheric transport. Based on principal component analysis (PCA), back trajectory analysis, correlation analysis with some specific tracers, we discuss the sources and compositional changes of water-soluble organics in the time-resolved samples, and the photochemical aging of organic aerosols. 88%. The analytical errors in the triplicate analysis were less than 5% for C 2 , C 3 , C 4 , C 5 and C 6 . We also analysed field blanks. Concentrations of all the species reported here are corrected for the field blanks, which are <5% of real samples.

Site description
Organic carbon (OC) and elemental carbon (EC) were measured using a Sunset Laboratory carbon analyser following the IMPROVE temperature protocol (Wang et al., 2005). The method produced data for four OC fractions (OC1, OC2, OC3 and OC4 in a helium atmosphere at 140, 280, 480 and 580 °C, respectively), a pyrolyzed carbon fraction (OP, determined when reflected laser light attained its original intensity after oxygen was added to the combustion atmosphere), and three EC fractions (EC1, EC2 and EC3 in a 2% oxygen/98% helium atmosphere at 580, 740 and 840 °C, respectively). The IMPROVE protocol defined OC as OC1 + OC2 + OC3 + OC4 + OP and EC as EC1 + EC2 + EC3 − OP. The analytical errors in duplicate analysis of the filter sample were less than 8% for OC. OC concentrations reported here are corrected for the field blanks, which are less than 10% of real samples.

Backward air mass trajectory analysis
Five-day backward trajectory analyses were performed at 500 metres above the ground with height for 24, 12 and 3 h samples to decipher the source regions of air masses over Cape Hedo, using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT4) model (https://www.arl.noaa.gov/ ready/hysplit4.html), NOAA Air Resources Laboratory, Silver Spring, Maryland, United States (Draxler and Hess, 1998). Figure 2 shows the air mass trajectories calculated for day and night samples and every 3 h samples. During March 30 (day and night) air masses travel the coastal area of south China, Taiwan and the Pacific. In contrast, during March 23, 24, 31 and April 1 air masses came from the Pacific Ocean. Except for the above-mentioned dates, all the air masses originated from East Asia. For each sample, we have calculated one back trajectory. We performed back trajectory (BT) analysis using geopotential height, time and model vertical velocity. Sampling collection time is the starting time for BT analysis.

Principal component analysis
Principal component analysis (PCA) is a mathematical algorithm that reduces dimensionality of the data while variation is mostly retaining in the data-set. It accomplishes reduction by identifying directions, called principal components, along which the variation in the data is maximal. Using a few components, each sample can be represented by relatively few groups instead of thousands of variables. PCA is a useful statistical tool for data analysis to assess the sources of aerosols. This method has been applied especially when sources of aerosols are different (Hopke, 1985;Kawamura and Sakaguchi, 1999). In this study, the data-sets of diacids from Okinawa (n = 55) were subjected to PCA analysis to investigate their sources and possible formation pathways. The term 'loading' in PCA represents the correlation coefficient between the component and individual species.
However, the concentrations of Okinawa aerosols are 1.7 times higher than those (200 ng m −3 ) from the western North Pacific (>140°E) (Mochida et al., 2003) and 8 times higher than remote marine aerosols (40 ng m −3 ) from the central Pacific (Kawamura and Sakaguchi, 1999). These comparisons suggest that Okinawa aerosols are significantly enriched with diacids, oxoacids and α-dicarbonyls compared to remote marine aerosols, but less enriched compared to the East Asian aerosols in source regions. Figure S1 shows the molecular distribution of diacids, oxoacids and α-dicarbonyls during day and night-time samples. C 2 is the dominant diacid species followed by C 3 , C 4 and produced from photo-degradation of aromatic compounds (e.g. toluene) released from traffic exhausts (Suh et al., 2003). Duan et al. (2008) suggested that oxidation of toluene is one of the significant sources of benzoic acid in the air. As described in experimental section, local anthropogenic activities are not important, thus, abundant presence of aromatic monoacids in the ωC 2 in daytime, whereas benzoic acid is the dominant species after C 4 during night-time. The higher concentration of benzoic acid in night-time suggests more influence from anthropogenic activities via long-range atmospheric transport. Indeed, benzoic acid has been identified as a direct pollutant from the traffic emissions (Kawamura, Ng et al., 1985) and indirect pollutant masses were originated from the Pacific Ocean and travelled over Taiwan before reaching to Okinawa. Higher concentrations were also observed during March 16, 25, 29 (night-time) and April 10. Similar peaks were observed for oxalic, malonic and succinic acids ( Fig. 3b-d). Concentration trends of total oxoacids and α-dicarbonyls are similar to those of total diacids, suggesting a similar source or similar formation mechanism. Based upon the backward trajectories, four source regions of diacids and related compounds were found as (a) East Asia, (b) coastal region, (c) the East China Sea and (d) the Pacific Ocean. study site suggests a significant influence from anthropogenic activities related to traffic emissions in the source regions via long-range atmospheric transport.  diacids (M, mM) show similar trends with higher peaks on 25 April ( Fig. 3h-j). Large amounts of Ph and tPh were found in the Chinese atmosphere (Ho et al., 2006;Wang et al., 2009). In contrast, fumaric acid (F) showed very high peak on April 11 (day), when air masses were delivered from northeastern China (Fig. 3k). The concentration trends of oxoacids (glyoxylic, 3-oxopropanoic, 4-oxobutanoic and 9-oxononoic acids) and α-dicarbonyls (MeGly and Gly) are similar each other, suggesting similar sources of these compounds (Fig. 4).

Temporal variation of diacids, oxoacids and α-dicarbonyls in springtime Okinawa aerosols
On the other hand, pyruvic acid showed a peak on April 6 ( Fig. 4e) when air masses were delivered from East Asia. Pyruvic acid is formed by the oxidation of aromatic hydrocarbons such as toluene and benzene in the atmosphere (Kawamura et al., 1996) and its formation is enhanced in the presence of cloud Lim et al., 2005;Legrand et al., 2007). We did not find any clear diurnal trends in 12 h samples for diacids, oxoacids and dicarbonyls (Figs. 3 and 4), suggesting that local anthropogenic contribution is not important for these species. Pavuluri et al. (2010) reported a clear day and night-time trend in Chennai, India and Mangshan, China, where local sources and in situ formation of diacids and related compounds are both important. He and Kawamura (2010) also reported similar results. Figure 5 shows temporal (3 h intervals) variations of selected diacids and oxoacids as well as pyruvic acid, α-dicarbonyls and benzoic acid. ωC 2 and α-dicarbonyls show similar temporal variations (Fig. 5). Enhanced concentrations of normal diacids (C 2 -C 9 , except for C 6 ) and aromatic diacids (Ph, tPh) were observed at 13:30 pm on 14 April and at 16:30 pm on 16 April. For both dates, air mass trajectories demonstrate the influence from East Asia. Lower concentrations of aromatic diacids (Ph and tPh) were observed on 15 April ( Fig. 5h and i) due to the rainfall from 8 am to noon on 15 April. In contrast, higher concentration of fumaric acid (F) was observed on the same date (Fig. 5g). The peak of F suggests an importance of photo-isomerization of maleic acid (M) under the strong sunlight immediately after the end of rain. The different temporal trend of F from other diacids suggests that source and/or formation mechanism of F is different. Atmospheric NO 3 − is emitted from anthropogenic sources (Deshmukh et al., 2016). Similar trends of diacids and NO 3 − in 3 h samples suggest that diacids are delivered from East Asia to Okinawa. Individual back trajectories suggest a typical advection time of 2-4 days from the source area.
During 15, 16 and 17 March, 5-day back trajectories show that the air masses are originated from East Asia, whereas on 14 March air masses are delivered from the Pacific passing over the southern part of Japan. On 16 April and 14, the 3-h variations of normal diacids (C 2 -C 8 ), aromatic diacids (Ph and tPh), oxoacids (ωC 2 ), pyruvic acid, MeGly, Gly and benzoic acid all show that the concentrations increased from 10:30 am and peaked at 16:30 pm. The higher concentration of diacids and related compounds at 16:30 is due to increased anthropogenic activities in East Asia via long-range atmospheric transport.
We found higher concentrations of those compounds when air masses arrived from coastal region of China and East Asia.
The highest concentration of C 2 was observed on April 10 when air masses were delivered from coastal region of China. However, the highest concentrations of C 3 , C 4 , pimelic acid (C 7 ) were observed on March 30 (night), when air masses were originated from the Pacific Ocean and travelled over Taiwan before arriving at Okinawa. Highest concentrations of C 6 and C 9 were observed on March 25 when air masses were delivered from East Asia. Cyclic olifines emitted from vehicles are the source of C 6 diacid in the atmosphere (Grosjean et al., 1978). The concentration trends of normal chain diacids are similar to each other ( Fig. 3b-g).
Phthalic acid (Ph) is a tracer of anthropogenic emissions ( vehicle exhaust) (Kawamura and Ikushima, 1993). Terephthalic (tPh) acid is derived from industrial emission and used as major raw material for polyester (polyethylene terephthalate) fibre and plastic bottles (Ding and Li, 2003) and is emitted to the atmosphere by plastic burning (Simoneit et al., 2005;Kawamura and Pavuluri, 2010). Ph, tPh and aliphatic unsaturated pathway Legrand et al., 2007), as summarized in Fig. 7. Relative abundance of individual species is used to evaluate the extent of chemical ageing (He and Kawamura, 2010). An increased C 2 /total diacid ratio (60-70%) has been reported in the marine aerosols from the Pacific during longrange atmospheric transport (Kawamura and Sakaguchi, 1999). The relative abundance of C 2 from this study (67%) is comparable to those in marine aerosols from the western Pacific (65%) (Wang et al., 2006) but less than that from the central equatorial Pacific (>70%) (Kawamura and Sakaguchi, 1999). For daily and day/night samples, the highest relative abundance of C 2 is observed on 24 March (Fig. 8a) when air masses arrived from East Asia. The average relative abundances of C 2 However, we did not get any differences in diurnal variation on March 17, although the airmass source regions are East Asia. Figure 6 shows the average relative abundance (%) of diacid species in total diacids. The average relative abundance of C 2 (67%) in this study is much higher than those of malonic (11%) and succinic acid (4%). Very high percentage of C 2 indicates a significant photochemical processing of long-chain diacids as well as aqueous phase oxidation of biogenic and anthropogenic volatile organic carbons (VOCs) via Gly, MeGly, Pyr and ωC 2 gas phase during day and night-time (Kamens et al., 1982;Fan and Zhang, 2004;Warneck et al., 2004). O 3 and OH radicals are important oxidants in daytime (Kamens et al., 1982;Fan and Zhang, 2004), whereas NO 3 and H 2 O 2 (in the presence of H 2 SO 4 ) are main oxidants in night-time (Hermann et al., 2000;Claeys et al., 2004). NO 3 can react with volatile and semivolatile organic compounds such as hydrated Gly and MeGly in during daytime (70%) and night-time (70%) are similar. There is no sunlight in night-time and the relative humidity is quite high, we suppose that C 2 was produced by aqueous phase oxidation of precursors in darkness. Aqueous phase production of C 2 in cloud/fog/aerosols is important in the atmosphere (Warneck, 2003;Sorooshian et al., 2006). Biogenic and anthropogenic VOCs can be oxidized to MeGly and Gly in the of C 2 is lower, further suggesting the formation of C 3 and C 4 and/or degradation of C 2 .

Photochemical processing of diacids and related compounds
In night-time, oxalic acid is also formed by the degradation of long-chain diacids during long-range atmospheric transport. Figure 9 shows anti-correlations between relative abundances (%) of oxalic acid (C 2 ) in total diacids and concentrations of malonic (C 3 ), succinic (C 4 ), glutaric (C 5 ), adipic (C 6 ) and azelaic (C 9 ) acids. C 2 can be produced from long-chain diacids and related compounds (Kawamura and Sakaguchi, 1999;Matsunaga et al., 1999;Warneck, 2003;Carlton et al., 2006;Legrand et al., 2007, Kundu et al., 2010. Previous studies have shown that concentrations of diacids and related compounds inversely correlated with relative abundance of C 2 (%) for urban Tokyo (Kawamura and Yasui, 2005) and Gosan aerosols (Kundu et al., 2010) in summer. In this study, we found that lower concentrations of long-chain diacids are associated with higher relative abundances of C 2 (%) (Fig. 9). These results suggest that longchain diacids can be decomposed to C 2 . Thus, it can be concluded that secondary production and degradation of organic aerosols are more intensified in night-time due to abundant presence of oxidants.
Scatter plots (Fig. S2) between relative abundance of oxalic acid and its precursor compounds such as C 3 , C 4 , C 5 , C 6 , C 7 , ωC 2 , MeGly and Gly further support that oxalic acid is formed by the degradation of long-chain diacids, oxoacids and α-dicarbonyls in early morning hours. We further checked the correlations among OC, C 2 , C 3 , C 4 , C 6 , Ph, tPh, ωC 2 and 9-oxononanoic acid (ωC 9 ). We found very strong correlations among them, suggesting that morning-time aerosols are influenced by long-range atmospheric transport. Large amounts of unsaturated fatty acids are emitted in the Chinese atmosphere from biogenic sources, meat cooking the aqueous phase. Aqueous phase Gly can be oxidized to ωC 2 , and finally to C 2 (Lim et al., 2005). Similarly, MeGly is further oxidized to pyruvic acid, hydrated glyoxylic acid and finally to C 2 (Legrand et al., 2007).
For the 3-h samples, we clearly observed the formation and destruction of short-chain diacids. Figure 8b shows 3-h variations in relative abundance of oxalic and succinic acids. During the morning hours (centre time 7:30 am) when the relative abundance of C 2 is high, the relative abundances of C 3 (not shown as a figure) and C 4 became low, suggesting that photochemical production of C 2 from C 3 and C 4 is intensified in morning hours. Similarly, when relative abundances of C 3 and C 4 are significantly higher in the afternoon, relative abundance  as VOCs such as isoprene (Limbeck et al., 2001), terpenes (Fick et al., 2004) and xylene (Birdsall and Matthew, 2011). Except for a few samples (25, 29 March, 3 and 4 April), the daily and day/night variations of C 2 and ωC 2 are similar.
Pyruvic acid is formed in the atmosphere by the oxidation of aromatic hydrocarbon (Kawamura et al., 1996;Legrand et al., 2007). Pyr-C/OC ratios (%) are almost constant except for March 29 (night), 30 (day and night) and April 6 ( Fig. S3h). Very high value was observed on 6 April. Gly and MeGly are the oxidation products of anthropogenic VOCs such as naphthalene, alkylnapthalene and toluene as well as biogenic VOCs such as isoprene and terpenes (Kunwar and Kawamura, 2014b and reference therein). These semi-volatile precursors of diacids are further oxidized to result in C 2 . Except for the samples of 17 March and 3 April and 4, Gly shows similar trend with C 2 (Fig. S3i).
For 3-h samples, ratios of C 2 to C 5 , ωC 2 and benzoic acid increased until 13:30 pm on 14 April and then decreased (Fig. S4). However, C 2 peak is not prominent, suggesting that the aerosols from Cape Hedo are more aged. During 15 March, C 2 -C/OC ratio is sharply decreased at 13:30 pm (Fig. S4a) as well as the ratios of other species. This sharp decrease should operation and wood burning and degraded to small diacids during long-range transport to Okinawa. We also found that when relative abundance of C 4 increases, relative abundance of C 2 decreases (Fig. 8b) in the afternoon, suggesting a favourable degradation of C 2 in the presence of Fe 2+ , during which C 2 can be degraded easily via the formation of oxalate-Fe complex to result in formic acid and CO 2 (Pavuluri and Kawamura, 2012;Hoigne 1992, 1994), although we did not measure Fe in this study.
Oxalic acid-C/OC ratio can be used as a tracer for photochemical aging of organic aerosols (Kunwar and Kawamura, 2014b). Figure S3 presents the contribution (%) of selected diacids to organic carbon. Total diacid-C/OC, oxoacid-C/ OC and α-dicarbonyl-C/OC ratios ranged from 0.92 to 12% (6.2 ± 2.0%), 0.02 to 1.90% (0.70 ± 0.43%), and 0.03 to 0.48% (0.2 ± 0.09%), respectively. Normal chain diacids (C 2 to C 5 and C 7 ) show the similar trend. For day and night samples, we did not find clear trends although we got higher ratio in March 27 (day), 28 (night), 29 (night) and 30 (day and night) samples. A very high ratio was observed on 7 April for C 2 and ωC 2 ( Fig.  S3a and g). ωC 2 is the precursor of C 2 , which is produced in the atmosphere by the oxidation of aromatic hydrocarbons as well Fig. 8. Temporal variations in the relative abundances of (a) C 2 /total diacids and C 4 /total diacids (b) diurnal variations in the relative abundances of C 2 /total diacids and C 4 /total diacids in the ambient aerosols collected at Cape Hedo, Okinawa. Colours indicate similar as other figures. SPRINGTIME INFLUENCES OF ASIAN OUTFLOW AND PHOTOCHEMISTRY C 3 (r = 0.92), C 4 (0.91), and ωC 2 (0.94). MSA − that is produced by the atmospheric oxidation of dimethyl sulphide (DMS) and dimethyl sulphoxide (DMSO), shows a strong correlation with C 2 (0.75). C 2 showed no correlation or weak correlation with NH 4 + (0.07), NO 3 − (0.38), non-sea salt (nss)-SO 4 2− (0.22) and nss-Ca 2+ (0.1). No correlation between of C 2 and nss-SO 4 2− suggest the different sources or different formation mechanism. In night-time samples, C 2 is strongly correlated with its precursors such as C 3 (r = 0.91), C 4 (0.94), C 6 (0.92) and C 7 (0.83). Again, C 2 is strongly correlated with ωC 2 (0.96), Pyr (0.95), Gly (0.86) and MeGly (0.77). Strong correlations of C 2 with longer chain diacids suggest that C 2 is formed by the oxidation of long-chain diacids. Strong correlations among C 2 , Pyr, Gly and MeGly further suggest that aqueous phase reaction is also important in night-time during long-range atmospheric transport (Fig. 7).
Both nss-SO 4 2− and NO 3 − are emitted to the atmosphere from the anthropogenic activities including industrial and vehicular emissions and biomass burning (Wang et al., 2009). Formation of sulphate from SO 2 oxidation is either accomplished in gas phase by OH radicals (homogenous reaction) or through in-cloud/fog processes (heterogeneous pathway) by oxidants. In night-time, we found very high concentration (15 μg m −3 ) of nss-SO 4 2− on 11 April (Fig. S5). We considered this value as an outlier. After the removal of outlier, we found a good correlation (r = 0.86) between C 2 and nss-SO 4 2− in night-time (Fig.  S3). A good correlation between these two components suggests the main formation pathways from gaseous precursors in aqueous phase chemistry in night-time. In contrast, no correlation between C 2 and nss-SO 4 2− in daytime may demonstrate that local biogenic emission of isoprene, monoterpene and fatty acids are the sources of C 2 . We further found strong correlation of OC with C 2 (r = 0.97), C 3 (0.98), C 4 (0.98), C 6 (0.92), Ph (0.88), ωC 2 (0.98), ωC 9 (0.96) and Gly (0.94) in night-time.
Interestingly, we could not find such correlations of OC in daytime, except for ωC 9 and ωC 2 . ωC 9 is an intermediate compound during the oxidation of biogenic oleic acid (C 18:1 ) to C 9 (Kawamura and Gagosian, 1987). In daytime, we found very strong correlation of OC with ωC 9 and ωC 2 . These results further suggest that local biogenic emission is important in daytime during 27 March to 1 April and 11-13 April. Tambunan et al. (2006) and Zhu et al. (2016) reported that there are isoprene emitter trees in Okinawa. NO 3 − shows weak correlation with C 2 in daytime (0.38) and strong correlation (0.70) in night-time. The aqueous phase production of oxalic acid in the presence of NO 3 − may be important in night-time (Harrmann et al., 2000). The positive correlations of C 2 with NO 3 − and nss-SO 4 2− in night-time suggest that aqueous phase production of C 2 in the presence of H 2 O 2 along with H 2 SO 4 . Claeys et al. (2004) reported the aqueous phase chemistry of secondary organic aerosols in the presence of H 2 O 2 and acid.
tPh is a tracer of plastic burning (Simoneit et al., 2005;Pavuluri et al., 2010). We found a fair correlation between C 2 be caused by rainfall, suggesting that particles associated with oxalic acid and related water-soluble organic species act as CCN and thus selectively scavenged from the air. There is heavy rainfall in Okinawa before the noontime on that day.
Interestingly, fumaric (F) acid is formed soon after the rain ended on 14 April (Fig. S4e) and solar radiation came back, showing a peak at 16:30 pm. Except for F, all the diacids and related compounds show lower ratio at this time, suggesting the additional source of fumaric acid. tPh, a tracer of plastic burning (Jung et al., 2010;Pavuluri et al., 2010), shows time specific diurnal variation. Higher values were obtained at 16:30 pm on 14 March and at 13:30 pm and 16:30 pm on 16 April, suggesting the influence from plastic burnings in the afternoon. These results suggest that low molecular weight diacids, oxoacids and α-dicarbonyls are more aged in the Okinawa aerosols during long-range transport from East Asia as illustrated in Fig. 7.

Correlation coefficients among selected diacids and related compounds
Correlation analysis was performed separately for day and night samples and for 3-h samples to better understand the chemical processing in the atmosphere. For daytime samples (12 h), C 2 is correlated with its precursor compounds such as .99, respectively). However, correlation between oxalic acid and benzoic acid is time specific. At 7:30 am, there is negative correlation between C 2 and benzoic acid. At 10:30 am, the correlation is very weak (0.09). During afternoon (at 13:30 pm and 16:30 pm), correlations are fair (0.62 and 0.67, respectively), suggesting the common source of C 2 and benzoic acid in the afternoon because benzoic acid is emitted from the vehicle exhausts (Ho et al., 2010) and local anthropogenic input is less important, hence, we conclude that benzoic acid and its precur-and tPh during night-time (r = 0.61). Plastic burning is significant in East Asia including Mongolia (Jung et al., 2010) and influences the study site via long-range atmospheric transport. However, correlation of C 2 with tPh is rather weak (0.28) in daytime.
MSA − shows a strong correlation with C 2 during daytime (0.75) and night-time (0.88). MSA − can be found in both marine and coastal regions. However, negative correlation of C 2 with ss-SO 4 2− suggests an anthropogenic origin. Dimethyl sulphide (DMS) from terrestrial emissions and dimethyl sulphoxide (DMSO) from industrial wastes could be two possible precursors of MSA − (Hui et al., 2004).
The ratios of C 2 /C 3 , C 2 /C 4 , C 2 /MeGly, C 2 /Pyr and C 2 /ωC 2 are used to decipher photochemical chain reactions in the atmosphere. Interestingly, those ratios are higher at 7:30 am for 3 days (14, 15 and 16 April) (Fig. 10), although we did not find the time specific trend on 17 April. The higher ratios at 7:30 am and strong correlations (r = 0.88-0.99) of OC with C 2 , C 3 , C 4 , C 6 , Ph, ωC 2 , ωC 9 , and Gly in both night-time and morning hours, as well as decreased concentrations of the C 2 precursors with increased relative abundance of C 2 /total diacid (%) (Fig.  S2), suggest the accumulation of the precursor compounds in the atmosphere in night-time and subsequent photochemical sors such as toluene were emitted to the Chinese atmosphere from vehicles and long-range transported to Okinawa.
Although daytime (12 h) samples in March and April did not show any correlation between C 2 and nss-SO 4 2− , we found a strong correlation in 3-h daytime samples during 14-17 April, suggesting more anthropogenic influence on diacid concentrations. We further checked the concentration of nss-SO 4 2− during this period and concentration of nss-SO 4 2− is very high during the same period except for 15 April. During this period significant amounts of nss-SO 4 2− were emitted to the Chinese atmosphere and long-range transported to the western North Pacific, and thus we detected large amounts of nss-SO 4 2− (up to 30 μg m −3 ) in Okinawa during this period. Compared to daytime (12 h) samples during late March to mid April, 3 h daytime samples during 14-17 April are more influenced by anthropogenic activities.

In situ formation of oxalic acid and possible photolysis during long-range atmospheric transport
Malonic acid (C 3 ) can be formed in the atmosphere by the photochemical degradation of succinic acid and thus C 3 /C 4 ratio has been used as indicator of enhanced photochemical aging of organic aerosols (Kawamura and Ikushima, 1993;Ho et al., 2007;Deshmukh et al., 2016). Lower ratios (0.25-0.45, av.  (Hui et al., 2004).
For 24 h samples, we found 3 components with higher loadings of C 5 and C 6 , which are associated with anthropogenic activities because both species are tracers of anthropogenic cyclohexene (Kawamura et al., 1996). Higher loadings of Ph, tPh, ωC 2 and Gly again suggest the photochemical oxidation of aromatic hydrocarbon. Component 1 is associated with secondary formation of small diacids and related compounds from aromatic diacids; components 2 and 3 and are associated with dust influence and marine emission, respectively. The high loading of tPh in component 1 suggests an influence from plastic burning because tPh is a tracer of plastic burning products (Simoneit et al., 2005;Jung et al., 2010).

Summary and conclusions
Daily (24 h), day/night (12 h) and time-resolved (3 h) samples were collected from Cape Hedo, Okinawa to better understand the source and in situ formation of organic acids in the western North Pacific Rim. Oxalic acid (C 2 ) is found as a dominant species followed by malonic (C 3 ) and succinic (C 4 ) acid. We found higher concentration ratios (C 2 /C 4 , C 2 /C 3 , C 3 /C 4 , and C 2 /ωC 2 ) at 7:30 am, suggesting that organic precursors are accumulated in night-time and then photo-oxidized early in the morning in the presence of sunlight. No clear trends of diacids and related compounds were found in day and night-time, suggesting that there are no local anthropogenic emissions of diacids and related compounds. The strong correlations of C 2 with precursor compounds such as C 3 , C 4 , C 6 , Ph, C 9 , Gly, MeGly and Pyr suggest the formation of low molecular weight diacids via the aqueous phase oxidation of long-chain diacids and other precursors. Further, the average relative abundance of C 2 during day and night-time is similar (70%), suggesting that Okinawa aerosols are already aged during long-range atmospheric transport from the Asian Continent. Strong correlation (0.86) between nss-SO 4 2− and oxalic acid during night-time suggests the similar source and/or formation mechanism. However, no correlation between nss-SO 4 2− and oxalic acid and strong correlation of OC with ωC 2 (0.81) and ωC 9 (0.73) further suggest the local biogenic emission during daytime. For 3-h samples, oxalic acid shows good correlations with secondary atmospheric tracers (NO 3 − , SO 4 2− , NH 4 + , MSA − ), again supporting the secondary photochemical production of C 2 during long-range atmospheric transport. Compared with daytime (12 h) samples on 27 March-1 April and 11-13 April, we found that 3-h daytime samples on 14-17 April are more influenced by anthropogenic activities in China via long-range transport. production of oxalic acid immediately after sunrise. Similar trends are also observed for C 2 /Pyr and C 2 /ωC 2 although peaks are less prominent for C 2 /Pyr ( Fig. 10f and g). In contrast, C 2 / Gly ratios showed higher values at 16:30 pm of 14 April and 7:30 am of 16 April. Similar trend was observed for C 2 /MeGly ratios. These results suggest that photochemical production of LMW diacids is important early in the morning, although morning aerosols can also be influenced by long-range transport.
Relative abundances of C 2 for 24 h samples show negative correlations with the relative abundance of C 3 , C 4 , C 5 and C 6 (figure is not presented). This result again suggests a possible formation of oxalic acid from longer chain diacids such as C 3 and C 4 .

Principal component analysis (PCA) for dicarboxylic acids and related compounds
We performed time series PCA analysis. First, we normalized the data of chemical species using Statistical Package for the Social Sciences (SPSS) software. Table 2 presents a result of PCA for day and night-time and 24 h samples. The number of PCs to extract is determined based on eigenvalue. The component having more than 1 eigenvalue is extracted here. We used a varimax rotation with Kaisher normalization method. We trimmed the missing or <LOD data which is less than 1%, so, there is not a significant change in result.
We found four components in daytime samples, whereas three components were found in night-time and 24 h samples ( Table 2). High loadings of C 2 , C 3 , C 4 , C 6 , C 7 , Ph, ωC 2 , Pyr, Gly and MeGly were found in Component 1 in daytime, suggesting an influence from anthropogenic emission with photo-oxidation because these diacids are associated with anthropogenic sources, where Gly, MeGly, Pyr and ωC 2 are associated with photochemical oxidation in the atmosphere, and Ph is formed by the oxidation of polycyclic aromatic hydrocarbons (PAHs) including naphthalenes (Ho et al., 2010 and references therein). Higher loadings of Na + , Cl − and ss-SO 4 2− in component 2 are associated with marine source. In addition, higher loading of nss-Ca 2+ in component 2 is beyond our explanation. Higher loadings of NH 4 + and nss-SO 4 2− in component 3 suggest an industrial influence and biomass burning of animal excreta. Based on the PCA analyses, we emphasize that anthropogenic and industrial sources and biomass burning of animal dung are important as well as marine inputs over the study site via longrange transport from the Asian Continent in daytime.
In night-time, we detected very strong loadings of C 2 -C 9 , Pyr, ωC 2 , MeGly and MSA − in component 1, which are associated with secondary oxidation. Very high loadings of Na + , Cl − and ss-SO 4 2− in component 2 indicate the marine influence. High loadings of nss-Ca 2+ and nss-K + in component 2 in night-time suggest the mixed sources from marine and continent have