Overview of the Mount Tai Experiment ( MTX 2006 ) in Central East China in June 2006 : studies of significant regional air pollution

Introduction Conclusions References


Introduction
Asian regional air pollution is a major element of global air pollution (Akimoto, 2003) and is of great interest from the viewpoints of urban and rural air quality, trans-boundary air pollution, hemispherical air pollution, and climate impacts. Central East China (CEC), covering the North China Plain (NCP) and the Yangtze River delta region, stands out as the most highly polluted area in this region, as evidenced by satellite observations of NO 2 , which is regarded as representative of air pollutants (Beirle et al., 2004;Martin et al., 2006), and aerosols (Chin et al., 2004). It has also been observed that the tropospheric column density of NO 2 has been increasing rapidly since 2000 (Richter et al., 2005). Measurements of the aerosol optical depth (AOD) at 550 nm, performed Introduction  (Ohara et al., 2007). The data show that Mt. Tai is in the middle of the highest emission area, Mt. Hua is at the western edge, and Mt. Huang is just outside the heavy emission area. We therefore selected Mt. Tai as our campaign site and chose June as the study period, to represent the most serious regional air pollution in CEC. 5 Wang et al. (2006) simulated a regional high O 3 episode observed at two mountain sites (Mt. Tai and Mt. Huang) in May 2004, using the nested air quality prediction modeling system (NAQPMS) and found that it was the transport of O 3 and its precursors from the Yangtze delta that caused the high O 3 episode at the two sites, with contributions of 20-50 % during the episode. However, the lack of measurements of O 3 10 precursors prevented detailed studies. Gao et al. (2005) reported O 3 and CO observations at the summit of Mt. Tai during the period June-November 2003, showing average concentrations of 58 ± 16 ppbv for O 3 and 393 ± 223 ppbv for CO during the study period. Another paper reported precipitation chemistry at Mt. Tai . Observations at Shangdianzi (a rural 15 site in northern CEC) showed that the average maximum hourly O 3 concentration in June exceeded 120 ppbv . Xu et al. (2008) summarized surface O 3 data from the Lin'an Regional Background Station, Zhejiang Province in the Yangtze delta region, during six periods between August 1991 and July 2006 (Luo et al., 2000;Wang et al., , 2004). The seasonal cycle shows two peaks at around 40-50 ppbv, 20 with the primary peak in May and the secondary peak in October. A maximum hourly mean O 3 concentration of 156 ppbv occurred in June 2006. Intensive field campaigns including measurements of O 3 , CO, NO y , SO 2 , and VOCs were performed at Lin'an, in 1999-2000and in spring 2001(Wang et al., 2002 and key emission factors (e.g. CO/NO x ) were studied. 25 To our knowledge, there have been no other papers reporting air quality with respect to O 3 and its precursors in 2006 or earlier at regionally representative sites in NCP, which can be used for comparisons with results from regional or global chemical transport models. MTX2006 is therefore the first intensive campaign to provide Introduction Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | a comprehensive set of near-surface air pollutant concentrations at a regionally representative site in NCP. The MTX2006 study is a component of the O 3 and BC project sponsored by the Global Environmental Research Fund of the Ministry of the Environment, Japan. This paper presents an overview of the campaign and the obtained results, and a synthesis 5 of the individual papers presented in this special issue on MTX2006, encompassing from field observations to modeling, remote sensing, and laboratory studies. Also, we newly explore comparisons of ∆CO/∆NO y and ∆O 3 /∆NO z ratios between observations and model simulations, correlations between VOCs and levoglucosan, dependence of OC mass concentrations on airmass age, and fractions of OC which were molecularly 10 identified.

Specific objectives of MTX2006
The general objective of MTX2006 was to quantify air quality in the middle of the NCP. To attain this objective, the specific aims were to answer the following questions.
1. What are the concentrations of surface O 3 , aerosols (including BC), their precur- 15 sors, and related species at the top of Mt. Tai, located in the middle of the regional pollution area of CEC, in the highest pollution season of June? 2. Why does the surface O 3 concentration peak sharply in June, as opposed to showing a broad plateau in the summer, as observed in Europe, another continental source region? Introduction . So, although the air quality of Mt. Tai would be affected by regional emissions when the boundary layer builds up, covering the top of the mountain, in the afternoon, as will be described later, the site 15 is believed to be free from local anthropogenic pollution, and is expected to provide regionally representative data. conditions are summarized in Fig. 2. On 9, 15, and 29 June, low-pressure systems passed over the region. During the periods of passage, the wind direction was north. During the period 22-25 June, the main wind direction was east, as a result of a highpressure system located in the east, near Japan. The wind direction in the other periods was normally south or southwest. Precipitation was recorded on 7, 13,14,21,22,26,15 28, and 29 June. The average daytime maxima of J(NO 2 ) and J(O 1 D) (2π sr) were 7.4 × 10 −3 and 2.8 × 10 −5 s −1 , respectively. The temperature and relative humidity (RH) ranges were 9-25 • C and 20-100 %, with averages of 17 • C and 67 %, respectively.

Modeling activities
Modeling activities were included as part of MTX2006 as post-mission analysis. Ta-20 ble 2 lists the modeling systems used, targeted subjects, and principal investigator and affiliation. A one-box model was used to study photochemical production and loss of O 3 by fixing the key species and parameters to observed values (Kanaya et al., 2009). Two different three-dimensional chemical transport models (3-D CTM) were used; one is NAQPMS, developed by Z. Wang et al. (2006), and the other is Community Multi- 25 scale Air Quality Modeling System (CMAQ), developed by Byun and Ching (1999 (Yamaji et al., 2010), and to study the impact of the Asian monsoon on O 3 and precipitation acidity in Eastern and Central China (He et al., 2008;Ge et al., 2011).
6 Overview of MTX2006 findings 5 Figure 3 shows the overall temporal variations in O 3 , CO, benzene, CO 2 , NO x , NO y , BC, OC, sulfate, and nitrate concentrations during the campaign period. Table 3 lists their average, maximum, and minimum concentrations. In Fig. 3, two highconcentration episodes are clearly seen for most gases and aerosols during the peri-10 ods 6-7 and 12-13 June. These episodes are caused by the strong impact of agricultural waste burning plumes, as will be discussed later. A regular diurnal pattern with a daytime maximum and nighttime minimum can be seen particularly for NO y , BC, and OC in the latter half of the campaign period. This pattern is the result of a polluted air mass being transported to the mountain top during the daytime; this is associated with concentrations occurred in June, the same as at Mt. Tai. The monthly average O 3 mixing ratio at Mt. Tai is therefore even higher than in the outskirts of Beijing, and the maximum concentration is almost the same as that at Miyun. The monthly mean mixing ratio of CO at Mt. Tai, 560 ppbv, is about the same as the 600 ppbv and 600-700 ppbv levels reported at Miyun and Changping, respectively, but 5 the maximum level of 1950 ppbv at Mt. Tai is much higher than that in the outskirts of Beijing (∼ 1500 ppbv). In contrast, the mixing ratios of NO y at Mt. Tai, with average and maximum values of 6.7 and 21 ppbv, were much lower than those at Changping, i.e. 15 and 50 ppbv, respectively (T. Wang et al., 2006). The ∆CO/∆NO y ratio at Mt. Tai, around 50 ppbv/ppbv, shown as the slope of Fig. 4a, is 1.4 times that at Lin'an  40, Wang et al., 2002Wang et al., , 2004, about twice that at Changping (20-30, T. Wang et al., 2006) and in a plume from Mexico City (20.6, Kleinman et al., 2008), and about five times that in US vehicular emissions (∼10, Parrish et al., 2002). The large ∆CO/∆NO y ratio at Mt. Tai could be attributed to open biomass burning. However, the ratio was similar for the latter half of the campaign, where the influence from biomass burning 15 was minimal. Figure 4d shows that the high ∆CO/∆NO y ratio is well reproduced by the CMAQ model simulation (Yamaji et al., 2010). Figure 4a and d also indicate that high O 3 concentrations are associated with relatively high CO and NO y concentrations.

High concentrations of O 3 and aerosols at Mt. Tai and related features
The median NO x /NO y ratio of 0.15, and the 15 and 85 percentile values of 0.07 and 0.27 at Mt. Tai, are much lower than the values at Miyun, predicted by a model 20 to be 0.4-0.7 . Our range is even lower than that of the NO x /NO y ratio reported by Kleinman et al. (2008) over the Mexico City plateau of between 0.11 and 0.83. From these findings, it was concluded that air at Mt. Tai is affected less by emissions of NO x but more by emissions of CO, and is photochemically more aged, than is the case for air in studied regions surrounding the megacities of Beijing and 25 Mexico City. It should be noted that CEC is not simply a source region; photochemical aging has great importance there.
The mean concentrations of BC (PM 1 ), EC (PM 1 , NIOSH temperature protocol, 2-20 June 2006), and OC (the same as EC) at Mt. Tai were 3.4, 2.6, and 9.2 µg C m −3 , ACPD 13,2013 Overview of the Mount Tai   Tai show episodically very high-concentration events, as seen in Fig. 3. Two prominent events during the periods 6-7 and 12-13 June gave peak concentrations of 29 (22) and 77 µg C m −3 for BC (EC) and OC, respectively; these can be ascribed to the plumes of agricultural waste burning, as will be described in Sect. 6.2. The mean concentrations of PM 2.5 and total suspended particles (TSP) during the campaign period were 123 and 135 µg m −3 , respectively. These values are even higher Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | autumn, and low in summer (Fig. 5b). The BC concentration levels were higher at Mt. Tai than at Mt. Huang throughout the year. The BC level at Mt. Tai in June 2006 was one of the highest monthly averages. Figure 6 shows the average diurnal variations of selected species. Ozone showed a broad peak in the late afternoon (14:00-19:00 LT, LT = UTC + 8 h) as high as 92-5 94 ppbv. NO y and CO had earlier peaks in the afternoon, coinciding with the transport of pollutants from lower altitudes with respect to the buildup of the boundary layer. BC showed high concentrations in the early morning (04:00-06:00 LT) as well as a peak in the afternoon. OC showed relatively flat diurnal variations in the early half-period (until 20 June, when the NIOSH temperature protocol was used with PM 1 sampling), but 10 showed regular diurnal variations with early afternoon maxima in the latter half-period, when an IMPROVE-like protocol was used with PM 2.5 sampling. CO 2 showed a regular daytime decrease, mainly in response to the source/sink provided by vegetation. Figure 7 shows three photographs looking westward from the observatory at the summit of Mt. Tai. In the upper panel, the outlines of distant mountains are clearly visible, 15 when aerosols are not abundantly present. Once regional pollution occurs, as shown in the middle panel, the top boundary of the polluted layer is clearly visible below the mountain altitude level in the morning. The pollution layer develops in the later hours, as shown in the bottom panel, and then the mountain top is incorporated in the polluted layer, at around noontime.

20
The high concentrations of BC must have important implications for the regional climate in CEC. Here the RF is roughly estimated. First we simply calculate its column concentration ( Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | radiative budget. More detailed studies using radiative transfer models and chemistryclimate coupled models are needed 6.2 Importance of agricultural waste burning and regional transport  2012), on this issue, similarly report the impact of burning using backward trajectories that passed over hotspots. On 8-10 June, the wind was from the north and thus the influence was small, although hotspots were continuously found south of Mt. Tai. Pan et al. (2012)  to high CO and BC concentrations was elevated concentrations of levoglucosan and other anhydrosugars (Fu et al., 2008), K + , and oxalic acid in PM 2.5 (Xu et al., 2009;Deng et al., 2011), CH 3 Cl (Suthawaree et al., 2010), CH 3 CN (Inomata et al., 2010), and lowered stable carbon isotope ratios (δ 13 C) of organic aerosol particles (Fu et al., 2012). 5 The impact of OCRB on O 3 , CO, BC, and OC concentrations over CEC during MTX2006 was evaluated using a regional chemical transport model, CMAQ, employing estimated daily emissions from OCRB, based on hotspot data obtained by MODIS (Yamaji et al., 2010). The impact of OCRB on regional emissions during the period 6-9 June amounted to 31-44 % (NO x ), 47-61 % (CO), 50-63 % (EC) and 56-69 % (primary organic aerosols) of the total sources in CEC. After 17 June, the impact of OCRB was almost zero. Figure 6 in Yamaji et al. (2010) demonstrates that the daily emissions from OCRB conveyed by southerly winds were an essential factor in reproducing atmospheric concentrations of pollutants during MTX2006. These emissions have a large impact not only on primary pollutants but also on secondary pollutants, 15 such as O 3 , in the first half of June. The run with OCRB emissions (06DS) clearly captured the observed variations in daily average O 3 concentrations, with a correlation coefficient R of 0.61 between the model and the observations, which is much better than that obtained without burning (NOCRB; R = 0.34). The monthly O 3 concentration simulated using 06DS (80.8 ppbv) was much closer than that obtained using NOCRB 20 (73.9 ppbv) to the observed monthly O 3 concentration. The average impacts of OCRB emissions contributed 6 % of O 3 , 20 % of CO, 43 % of EC and 53 % of OC concentrations over CEC for the whole month of June; for the episodic period 6-9 June, the impacts increased to 12 % for O 3 , 35 % for CO, 56 % for BC daily emissions, and 80 % for OC over CEC. The real cause of such elevated O 3 concentrations in June at Mt. Tai, 25 which had been a mystery, was therefore identified as biomass burning, by conducting a comprehensive field campaign covering the precursors of O 3 and related species.  (Li et al., 2008a). During this time period, CEC was covered with air masses with high O 3 concentrations (65-85 ppbv) in the model simulation Mt. Tai was at the center of a high O 3 region with the prevailing wind from the south. The source regions were divided into CEC and outer regions (see Fig. 1  Source attributions from the regional scale down to subprovince levels were therefore made. 15 In general, regional photochemical production of O 3 within CEC was the most important mechanism, showing a contribution of 51.4 ppbv (60.2 %) to a monthly mean O 3 mixing ratio of 85.4 ppbv. Specifically, the largest fractions of O 3 were formed in SSD, AH, and JS (32.4 ppbv or 37.9 %), influenced by high emission rates (including those from biomass burning) and favorable meteorological conditions in these areas.

20
They concluded that local photochemical production (9.8 ppbv or 11.5 %) is a minor contribution, suggesting that emissions from local cities (e.g. Tai'an) could be of minor importance. The calculated in-situ O 3 production rate was in good agreement with the results of Kanaya et al. (2009), which will be described in Sect 6.4. Li et al. (2008b) showed that the same NAQPMS model was able to reproduce the observed diurnal 25 variations in O 3 at Mt. Tai, and studied the roles of chemical production and transport. They suggested that the photochemistry in the surrounding region is more dominant than transport, giving diurnal variations in O 3 mixing ratios, with a maximum in of 9 % in highly polluted regions).

Photochemical O 3 production processes over CEC in June 2006
An observation-based boxmodel approach was undertaken to estimate concentrations of OH, HO 2 , and RO 2 radicals and the net photochemical production rate of O 3 , i.e. transportation is more important than in-situ photochemistry for high O 3 concentrations. The net production rate of O 3 was estimated to be 6.4 ppbv h −1 as a 6-h average (09:00-15:00 LT), in rough agreement with the insitu production rate (∼ 5 ppbv h −1 ) estimated in a three-dimensional regionalscale modeling study (Li et al., 2008b). These values fall within the range of the rates estimated for the boundary layer over major 20 cities in the United States, 3.5-11.3 ppbv h −1 (Kleinman et al., 2002).
The estimated 6-h averages of F − D(O 3 ) suggest that 58 ± 37 ppbv of O 3 are produced in 1 d. A sensitivity model run was made incorporating heterogeneous loss of HO 2 on aerosol particle surfaces, which could impede photochemical O 3 production. Pure water extracts of aerosol samples collected on quartz filters were used to regenerate aerosol particles in an aerosol flow tube in the laboratory to study the loss kinetics of HO 2 radicals as a result of heterogeneous reactions on the aerosol particles  Taketani et al., 2012). The uptake coefficients (γ values), determined for real atmospheric aerosol particle samples (> 10 samples) for the first time had an average value of 0.25, larger than the values determined for particles with major single components (e.g. ammonium sulfate), suggesting the possibility that minor components of aerosols play important roles. When this heterogeneous loss was included, the daily integral 5 of F − D(O 3 ) fell from 58 to 39 ppbv. This quantity was still larger than the observed daytime increases in the O 3 concentration observed at the mountain top, 23 ppbv on average (Fig. 6) suggesting that the daytime buildup could be explained by in-situ photochemistry. Another sensitivitymodel runs with altered NO x and hydrocarbon concentrations suggested that O 3 production occurred normally under NO x -limited conditions.
10 Figure 4b shows that NO x is almost depleted by photochemical aging (in comparison with Fig. 4a, using NO y as the x-axis), and thus the abundance of NMHCs, correlated with CO and used for the y-axis in Fig. 4b, becomes relatively high, resulting in NO x -limited conditions. This is consistent with the box-model analysis in this study An exception was 10 June, when fresh pollution from the north affected the site and the O 3 15 production was VOC-limited This analysis suggests that the O 3 pollution over the CEC region will become even worse when NO x emissions increase in the future. Figure 4e shows the modeled dependence of O 3 on NO x and CO. Although the modeled dependence is similar to the observations (Fig. 4b), the model tends to underestimate CO and therefore the CO/NO x ratio. Because the O 3 concentration is positively dependent 20 on the CO concentration (see color codes in Fig. 4e, b), an increase in CO in the model to reproduce the observed levels will increase the O 3 concentration.

Figure 4c shows that hourly O 3 concentrations increase with NO z ([NO z ]=[NO y ]−[NO]−[NO 2 ]
) up to about 10 ppbv, but saturation occurs with further increases in NO z . This feature is consistent with the above-mentioned analysis 25 using a box model which suggested general NO x -limited conditions. The slope ∆[O 3 ]/∆[NO z ] was 5.8 ± 0.5 ppbv ppbv −1 when the NO z mixing ratio was below 10 ppbv, implying efficient O 3 production from NO x . This is quite similar to the value of 3-6 ppbv ppbv −1 suggested for selected episodes studied at Changping near Beijing  Wang et al., 2006), and to that of 6.2 ppbv ppbv −1 found for the Pico de Tres Padres (PTP) site near Mexico City (Wood et al., 2009). The observed slope is lower than the modeled slope (Fig. 4f, slope = 8.7 ± 0.5 ppbv ppbv −1 ). This suggests that the O 3 production efficiency per unit NO x molecule oxidation is slightly lower than that predicted by the model. Although not shown, high O 3 concentrations are sometimes 5 associated with aged air masses (ln([NO y ]/[NO x ]) > 2 or [NO x ]/[NO y ] < 0.14).

NO 2 measurements by MAX-DOAS: comparisons with satellite and in-situ observations
During the campaign period, MAX-DOAS measurements of vertical profiles of NO 2 were performed from the top of Mt. Tai as well as from the foothills in the city of Tai'an 10 (Irie et al., 2008). The mean difference between the NO 2 volume mixing ratios measured by MAX-DOAS instruments at Tai'an and Mt. Tai and from insitu data based on chemiluminescence detection coupled with a photolytic converter were as small as −0.01 ± 0.60 ppbv and −0.29 ± 0.65 ppbv, respectively (see Fig. 5 of Irie et al., 2008). The mean NO 2 volume mixing ratio for the 0-1 km altitude layer above the surface 15 derived from the MAX-DOAS measurements at Tai'an was about 4 ppbv. The MAX-DOAS measurements showed complete diurnal variations in the tropospheric NO 2 VCD in the daytime. Typically, the value was highest at ∼ 15 × 10 15 molecules cm −2 in the early morning (6:00-9:00 LT) and dropped to a minimum of ∼ 7 × 10 15 molecules cm −2 at 13:00-15:00 LT. This indicates that the OMI measure-20 ments (at about 13:40 LT) were made when the tropospheric NO 2 VCD was lowest in its typical diurnal cycle, and that the systematic difference between tropospheric NO 2 VCDs derived from OMI and those observed by SCIAMACHY and GOME2 in the morning could be explained by this strong diurnal variation. Critical comparisons between tropospheric NO 2 VCDs derived from MAX-DOAS and OMI were made for Introduction the use of a stricter coincident criterion of 0.1 • resulted in better agreement within (+1.6 ± 0.6) ×10 15 molecules cm −2 (+20 ± 8 %) (OMI minus MAX-DOAS).

Volatile organic compound and oxygenated volatile organic compound (VOC/OVOC) compositions and concentrations
The relative concentration ratios of individual NMHCs (10 alkanes, six aromatic com-5 pounds, seven alkenes, and acetylene), normalized by ethane, at Mt. Tai, derived from daily canister sampling followed by gas chromatography-flame ionization detector analysis, were compared with those observed in Beijing city center on 20 November 2005 (see Fig. 7 of Suthawaree et al., 2010). Here, the average concentration of ethane at Mt. Tai was 2.4 ppbv, comparable to those in other areas of China (Barletta et al., 2005) 10 reflecting its long lifetime and the absence of local sources. As expected, we found that the ratios of major alkanes and alkenes to ethane are much lower for observations at Mt. Tai than they are in Beijing, reflecting the fact that the air mass at Mt. Tai is isolated from sources and is more aged. The measured benzene/toluene ratio at Mt. Tai was 3.2 ± 1.0, much higher than those (0.6) for other Chinese cities (Barletta et al., 2005) 15 and other parts of the world where vehicular emissions make significant contributions. A high ratio to ethane was also found for acetylene probably reflecting the contribution of biomass/biofuel combustion in the surrounding areas. The analysis of halogenated compounds (C 2 Cl 4 , CH 3 Cl, CH 3 Br) by gas chromatography-mass spectrometry showed low concentrations of C findings is that the observed mixing ratios of acetaldehyde were approximately 1.6 times higher than those of formaldehyde during the whole observation period. This is in contrast to the finding that the mixing ratios of formaldehyde are generally higher than those of acetaldehyde in Beijing ) and other urban, suburban and rural sites in the world, suggesting the possibility of primary emission of acetaldehyde. The 5 concentrations of methanol were high, ∼ 10 ppbv on average, compared with aldehydes and acetone. This result is similar to that observed in Beijing ) and near biomass burning (Karl et al., 2007). The HCHO concentrations derived from PTR-MS were in good agreement with those from MAX-DOAS observations (Inomata et al., 2008). Figure 9 shows that the PTR-MS signal at m/z 97 showed a strong positive cor-10 relation (R = 0.95) with levoglucosan concentrations in the TSP samples. The correlation coefficient was even larger than that for CH 3 CN and levoglucosan (R = 0.89). Karl et al. (2007) studied tropical biomass burning and assigned the peaks at m/z 97 to substituted furans and furfurals. This may be valid for our study. However, our (m/z 97)/(m/z 42) ratio of 1.6 ppbv ppbv −1 was slightly lower than the value of 2.4-3.2 ppbv ppbv −1 re-15 ported by Karl et al. (2007); this suggests that either the furans were removed by fast OH reactions during travel for 10-50 h and the original emission ratio was much higher, or that the species we detected at m/z 97 was different and had more chemical stability, resulting in the high correlation coefficient. Additionally, m/z 85 and m/z 87 also showed strong correlations (R > 0.90) with levoglucosan; the (m/z 85)/(m/z 42) and 20 (m/z 87)/(m/z 42) ratios were 1.2 and 1.6 ppbv ppbv −1 , respectively, assuming a typical ion-molecule reaction rate constant of 2×10 −9 cm 3 molecule −1 s −1 . These analyses were made possible by the comprehensive coverage of gas and aerosol species during this MTX2006 field campaign More studies are needed in the vicinity of biomassburning sources for better characterization.

BC measurement comparisons and ∆BC/∆CO ratio analysis
In addition to the year-round observations of BC using a summit of Mt. Tai, we installed several more instruments, namely a particle soot absorption photometer (PSAP; Radiance Research, Shoreline, WA, USA) and an ECOC semi-continuous analyzer (Sunset Laboratory, Tigard, OR, USA), operated either with IMPROVE-like or NIOSH temperature programs, for the purpose of instrument intercomparison during the campaign period (Kanaya et al., 2008). These instruments were operated either with PM 1 or PM 2.5 cyclones, depending on the time period. An aethalometer (AE-21, Magee Scientific, Berkeley, CA, USA) was also operated for several days during the first part of the campaign. The regression analysis for each pair showed that the correlations were strong and the slopes ranged from 1.03 to 1.54 for the first period (before 20 June) with PM 1 sam-10 pling, and from 1.07 to 1.46 for the second period (after 20 June), using the IMPROVElike temperature program for EC analysis with PM 2.5 sampling (see Fig. 8 of Kanaya et al., 2008). Considering that disagreement up to a factor of 4 (e.g. Jeong et al., 2004) was reported in the United States among similar instruments, the spread at Mt. Tai was thought to be relatively narrow. We therefore obtained BC concentration data suitable 15 for constraining BC emission rates from China, whose uncertainty was thought to be large (e.g. factors of 4-5; Bond et al., 2004;Streets et al., 2003).
We also found that enhanced MAAP BC/EC ratios (e.g. higher than 2) occurred only when low NO x /NO y ratios were recorded, implying that aging is related to large MAAP BC/EC ratios. Given that the tendency to underestimate EC is not straightfor-20 wardly explained in terms of aging, it is reasonable to conclude that the BC particles are coated by transparent materials after aging and that the MAAP results based on absorbance measurements gave overestimates possibly as a result of a lens effect induced by the coating. Pan et al. (2012) studied ∆BC/∆CO ratios in detail, using CO as a conservative 25 tracer, during the biomass-burning events and found that the ratio had a clear tendency to decrease with increasing transport time of the air mass from the biomass-burning source area. Here, the transport time was estimated using the WRF-FLEXPART model. The estimated lifetime of BC was ca. 4.1 d, near the lower limit of the range suggested by past studies.

PM and inorganic ion/element concentrations
The daily mass concentrations of TSP and PM 2.5 and their ionic and element compositions, were measured during MTX2006 and compared with the data obtained in spring

10
This ratio in summer is much higher than those in other cities in China (0.27-0.68), indicating that the air mass at Mt. Tai is chemically very aged consistent with other evidence discussed earlier.
Water-soluble ions contributed 10.8 % of TSP and 24.0 % of PM 2.5 by mass in spring, compared with 40.9 % and 41.1 % during MTX2006. Sulfate, nitrate, and am-15 monium ions were the major water-soluble species in PM 2.5 , accounting for 61.5 % and 72.7 % of the total measured ions in spring and during MTX2006, respectively. Also, higher concentrations of K + a tracer of biomassburning sources, were observed during MTX2006, accounting for 8.3 % of the total ions in TSP with high correlations with BC (R = 0.90) and oxalic acid (R = 0.87).
20 Xu et al. (2009) reported size-segregated water-soluble ions and metals in aerosol particles. Sulfate, ammonium, and K ions, and Pb, Zn, and Ti peaked in the accumulation mode (0.43-1.1 µm) Ca, Mg, Fe, Al, Ba, and Mn had peaks in the coarse mode (4.7-5.8 µm). In contrast, nitrate, Na, and Cl ions, and Co, Ni, Mo, and Cu showed bimodal distributions (0.43-0.65 and 4.7-5.8 µm). was assumed to be 90 ppbv. The increase in the ∆[OC]/∆[CO] ratio with photochemical aging suggests that carbon mass flux from gas-to-particle conversion is present. Robinson et al. (2007) assumed that SVOC and IVOC (semi-volatile and intermediate volatility organic compound) vapors are oxidized via a gas-phase OH reaction with a rate constant of 4×10 −11 cm 3 molecule −1 s −1 to yield nextgeneration compounds that 10 are 10 times less volatile, and calculated larger partitioning from gas-to-particle phases The assumed reaction rate is about four times faster than the oxidation of NO x to NO y .

ACPD
To cover eight generations, corresponding to the eight orders of magnitude range in volatility considered in their model study, a ln([NO y ]/[NO x ]) range between 0 and 2 needs to be studied. The study at Mt. Tai actually covered this range and the data are 15 suitable for the quantitative investigation of possible increases in OC with photochemical aging. Fu et al. (2008) summarized the organic molecular compositions (except dicarboxylic acids) of TSP samples collected on quartz filters using a high-volume air sampler and analyzed their temporal variations. They found that n-alkanes, fatty acids, fatty 20 alcohols, sugars, glycerol polyacids, and phthalate esters are the major species, and that lignin and resin products, sterols, aromatic acids, hopanes, and polycyclic aromatic hydrocarbons (PAHs) are minor species. They showed the importance of plant emissions of waxes and soil resuspension as well as biomass burning, as source processes. From the same TSP samples, Kawamura et al. (2012) reported that water- 25 soluble OC (WSOC)/total carbon (TC) ratio was 0.41 ± 0.09 in average, which is larger than 0.35 or less observed for PM 2.5 samples at Changping or other Chinese cities, Shanghai, Lanzhou, and Guangzhou (Pathak et al., 2011). Kawamura et al. (2012) also  reported that very high concentrations of saturated and unsaturated diacids (C 2 -C 11 up to 61 µg m −3 ) were detected among WSOC (ranged from 1 to 37 µg m −3 ), with a predominance of oxalic (C 2 ) acid followed by malonic (C 3 ) and succinic (C 4 ) acids as well as C 2 -C 9 ω-oxocarboxylic acids, pyruvic acid and α-dicarbonyls (glyoxal and methylglyoxal). The highest concentrations of total diacids (> 6 µg m −3 ) found at Mt. Tai were 5 several times higher than those reported at ground levels in Chinese megacities. The temporal variations in diacids were interpreted to be caused by a combination of direct emissions from field burning of agricultural wastes and secondary photochemical production via oxidation of volatile and semi-volatile organic precursors emitted from field burning of wheat straw during transport to the summit of the mountain. biomass-burning period in early June, the diurnal trends of most of the primary and secondary organic aerosol tracers were characterized by concentration peaks observed at midnight or in the early morning, whereas in late June (without the influence of intense biomass burning) most of the organic species peaked in the late afternoon. A strong anti-correlation was found between levoglucosan and δ 13 C values, again suggesting 20 the importance of crop residue burning. Altogether, 165 molecularly identified organic compounds including diacids, oxo/ketocarboxylic acids, dicarbonyls, n-alkanes, fatty acids, fatty alcohols, sugars, phthalates, polyols polyacids, aromatic acids, lignin and resin products, sterols, hopanes, PAHs, and biogenic secondary organic aerosol tracers (Fu et al., 2012;Kawamura et al., 2012) accounted for 14±6 % of the total OC mass 25 in the TSP. The OC mass concentrations in the TSP were 2.6 ± 0.8 times larger than those in PM 1 (before 20 June, using the NIOSH protocol, n = 46) and 2.4 ± 0.7 times larger than those in PM 2.5 (after 20 June, IMPROVE-like protocol, n = 22) analyzed using a semi-continuous ECOC analyzer (Fig. 11). This suggests the presence of coarse organic aerosol particles, and the presence of molecularly unidentified organics, even in PM 1 . Wang et al. (2009) investigated the size distributions of n-alkanes, PAHs, and hopanes collected using an Andersen eight-stage air sampler during the period 22-29 June 2006. The size distributions were compared with those sampled in winter at 5 the summit of Mt. Tai and those sampled in the marine boundary layer and urban atmosphere. During MTX2006, bimodal distributions with peak sizes of 0.7-1.1 µm and 4.7-5.8 µm were recorded for both plant-wax and anthropogenic n-alkanes. All the PAHs had maximums in the range 0.7-1.1 µm. Hopanes showed bimodal distributions with peaks in the ranges 0.7-1.1 µm and > 3.3 µm.