3.1. Rainwater Chemistry at Mt. Lulin
 Table 1 shows the volume-weighted mean (VWM) species concentrations for the entire sampling period. The pH value over the entire period was 5.12 (volume-weighted H+ concentration = 7.6 μeq l−1), which approaches the pH value of typical natural rainwater of 5.6. Acidic and alkaline species nss-SO42− and NH4+ (VWM concentrations: 15.3 and 12.3 μeq l−1, respectively) were the most abundant anions (24% of the total concentration) and cations (19% of the total concentration), followed by NO3− (VWM concentration: 10.8 μeq l−1; 17% of the total concentration). This more or less explains the low acidity of the rainwater chemistry at the monitoring site, since the strong acidic anions are neutralized by alkaline cations. Sea-salt species concentrations were low, due to the high altitude of the site.
Table 1. Species Concentrations, μeq l−1, Measured at Mt. Lulina
| ||N||Rainfall||pH||Cl−||NO3−||SO42−||nss- SO42−||H+||NH4+||Na+||K+||Mg2+||Ca2+|
|Apr 2003 toMay 2005|
|Seasonal VWM Concentration|
|Spring (Mar to May)||14||52.7||5.11||5.8||13.1||17.5||17.2||7.8||18.9||2.3||1.7||2.2||5.9|
|Summer (Jun to Aug)||15||61.2||5.15||9.5||9.7||15.0||14.5||7.0||9.6||3.9||1.2||2.2||3.8|
|Fall (Sep to Nov)||7||65.7||5.24||5.1||8.3||11.5||11.2||5.7||4.8||2.7||0.9||0.9||1.4|
|Winter (Dec to Feb)||3||91.7||4.91||6.4||12.2||20.1||20.0||12.4||16.1||1.5||1.8||0.0||3.5|
 Despite a relatively high correlation (R = 0.76, p < 0.001, N = 39) between Cl− and Na+ concentrations (Figure 2a), the ratio of Cl−/Na+ of 1.65 (compared to the seawater ratio 1.16) suggests excessive Cl− in most of the samples. Only five of the samples had ratios similar to the sea-salt value. Excessive Cl− was also reported in the snow samples collected at high-altitude sites ( ≥3000 m) in the Alps [Winiwarter et al., 1998] and the Himalayas [Shrestha et al., 1997]. It is worth noting that the Cl−/Na+ ratio reported in several rainwater chemistry studies near the surface level was similar to that of seawater [Wai et al., 2005, and references therein]. It is well-known that HCl is volatilized after the reaction of strong mineral acids with sea-salt aerosols. However, the scavenging of HCl can lead to a Cl−/Na+ ratio similar to that of seawater and cannot satisfactorily explain the excess of Cl− in our samples. A possible explanation to enrich the Cl− concentration in rainwater sample is the reaction of O3 at sea-salt aerosol surfaces, generating Cl2, followed by rapid photochemical conversion of Cl2 to HCl which is then eventually scavenged by cloud and rain droplets [Talbot et al., 1992, and reference therein]. Such conversion of Cl2 to HCl is enhanced by intense photochemical processes aloft in the air rather than at surface level. The emission of HCl from volcanoes [Orlando and Schauffler, 1999] may also contribute to the Cl− enrichment.
 Throughout the entire sampling period there was a high correlation coefficient (R = 0.92, p < 0.001, N = 34) between nss-SO42− and NO3− concentrations. This suggests that they have a common transport mechanism, rather than there being a reaction between their precursors [Lee, 1993; Báez et al., 1997]. All samples have similar ratio of nss-SO42−/NO3− (1.40). This ratio is also comparable to the value in remote Arctic regions (1.16 [Talbot et al., 1992]). In contrast, this ratio was observed to be 2.29 in Taipei, the capital of Taiwan [Lin et al., 1999]. The excessive nss-SO42− was attributed to the long-range transport of sulfur compounds. The species of Mg2+ was correlated (R = 0.77, p < 0.001, N = 39) with Ca2+, suggesting its nonmarine and crustal origin. Only four samples with higher Mg2+ but low Ca2+ concentrations indicated a marine origin.
 Species with high correlations also included NH4+ and nss-SO42− (R = 0.87, p < 0.001, N = 39; NH4+/nss-SO42− = 1.06), suggesting that NH4+ and nss-SO42− existed in the form of (NH4)2SO4, due to complete neutralization of H2SO4 and NH3. The NH4+-NO3− pair also had a high correlation between them (R = 0.86, p < 0.001, N = 39), indicating that NH4NO3 could coexist with (NH4)2SO4. The species K+ was correlated with NH4+ (R = 0.77, p < 0.001, N = 31) and NO3− (R = 0.74, p < 0.001, N = 31), suggesting that K+ mainly came from the biomass burning. NH3 emissions from biomass burning were reported [Hegg et al., 1988]. Ma et al.  reported that K+ highly correlated (r > 0.95) with NH4+ and NO3− in the plumes dominated by biomass burning in the nearby region (20°N, 120–130°E; >2000 m AMSL) during the TRACE-P campaign.
 Factor analysis of Varimax rotation was further carried out to assess the possible sources of rainwater contamination. Table 2 shows that three factors are identified and the loadings exceeding 0.7 are highlighted. Factors I and II accounted for 63 and 14% of total variance, respectively. The former is a mixing factor including the anthropogenic species (nss-SO42−, NO3−), natural decomposition/biomass burning species (NH4+ and K+) and soil/dust species (Mg2+ and Ca2+) [Momin et al., 2005]. The latter loads on remarkable H+ (0.97) but insignificant nss-SO42− (0.25), are indicative of an acidic factor. The third factor shows the loadings for Na+ and Cl−, in the form of sea-salt, which only accounts for 7.4% of the total variance.
Table 2. Principal Factor Loadings of Varimax Rotation Obtained From Rainwater Samples at Mt. Lulin During 2003–2005a
|Chemical Ions||Factor I||Factor II||Factor III|
|Percent of variance||63.41||14.33||7.43|
3.2. Seasonal Variations of VWM Concentrations at Mt. Lulin
 In order to analyze the data set in more detail, it was further divided based on different seasons, and seasonal VWM species concentrations were tabulated (Table 1, with the definition of the seasons). Five-day backward trajectories at Mt. Lulin, classified by different seasons are shown in Figure 3. The starting heights of the trajectories were set at 2860 and 4000 m AMSL in order to investigate the air motion aloft the site. Whenever rainfall was recorded on the corresponding day, the trajectory sets at two mentioned levels were computed.
Figure 3. Five-day back trajectories at Mt. Lulin in (a) spring, (b) summer, (c) fall, and (d) winter at 2860 and 4000 m AMSL.
Download figure to PowerPoint
 In the spring the VWM pH was 5.11. The concentrations for most of the species (except Na+ and Cl−) were elevated. It should be pointed out that spring is the season when Asian exports of natural and anthropogenic pollution aerosols to the Pacific are at its maximum, due to biomass burning peaks during this time in Southeast Asia. In addition, the cold fronts are responsible for the strong seasonal outflow to the Pacific [Jordan et al., 2003; Jacob et al., 2003, and references therein]. The plots (Figure 4) of the species concentrations show that abnormally high concentrations of all crustal and anthropogenic species appeared mainly in the spring months. However, the high concentrations of such samples are mostly due to exceptionally low amounts of rainfall.
Figure 4. Plots of (a) rainfall amount (mm; bar) and pH value (solid line); concentrations (μeq l−1) of (b) Cl− (green), Na+ (dark blue), K+ (light blue), Mg2+ (violet), and Ca2+ (yellow); and (c) NO3− (red), nss-SO42− (green) and NH4+ (dark blue) versus sampling date at Mt. Lulin.
Download figure to PowerPoint
 Most of the trajectories originate, same as in the winter as discussed later, from Southeast or south Asia (Figure 3a). This is especially true for the trajectories that started at 4000 m AMSL, and suggests that in the spring strong westerlies prevail in the middle troposphere over east Asia. Therefore as expected, the high concentrations of K+, NH4+, and NO3− can be explained by the peak activities of biomass burning in Southeast and south Asia. Some of the trajectories point to India, where the emission of the precursor gas NH3 is very high [Streets et al., 2003]. The high emission of NH3 in southern China may also have contributed to the elevated NH4+ concentration of our rainwater samples. The concentrations of nss-SO42− and NO3− were half of those measured at our remote monitoring site at Peng Jia Yu (25.65°N, 122.07°E, 100 m AMSL) near the surface layer. This can be explained by the fact that the low-level trajectories from Peng Jia Yu originate from northeastern China where the emissions of precursor gases SO2 and NOx are intense [Streets et al., 2003]. However, the high-level trajectories traverse over southern China and then over Southeast or south Asia. The paths of these trajectories suggest that the impacts of Asian dust storms on our samples are insignificant. Therefore the high Ca2+ content (5 out of 14 samples with concentrations >10 μeq l−1 and one sample has a Ca2+ concentration of 99.9 μeq l−1) in samples taken in the spring is not due to the impacts of Asian dust storms on northwestern China. Ma et al.  reported little influences of mineral dust or sea salt on the plumes dominated by biomass burning in the nearby region mentioned at 3000 m AMSL during the TRACE-P campaign (9 March 2001). On the contrary, TOMS images (http://toms.gsfc.nasa.gov/) show high values of aerosol index (AI) from the Sahara/Arabian Deserts to India/Southeast Asia corresponding to most samples with high Ca2+ concentrations. High dust concentrations are also evident from the global aerosol simulation SPRINTARS (http://cfors.riam.kyushu-u.ac.jp/∼toshi/SPRINTARS/) from Africa to India. Momin et al.  attributed the high SO42− and Ca2+ concentration in their rainwater samples in India (similar to our cases) to the transport from the African continent. In our case the trajectory air masses associated with two samples with high Ca2+ concentrations intercepted the areas of high AI over the South China Sea/Philippine Sea. The trajectories for the rest of these samples originated from the Indian desert Thar (about 17–27°N, 70–75°E) with their starting height being at 4000 m AMSL in general. This implies the significant influence of the chemical signatures of free tropospheric air on our samples, at least in the spring time. Owing to the impact of the Thar Desert, the frequency of dust storms from northern and northwestern India is maximum during the premonsoon season (March to May) [Dey et al., 2004, and reference therein].
 In summer the VWM concentrations for all species are similar (except H+ concentration) to those of the entire monitoring period. It is worth noting that concentrations for these species are not the lowest, contrary to the findings of the site as mentioned in Peng Jia Yu and despite the dilution effects due to the highest amount of rainfall per sample. More than 95% of trajectories come from the clean marine atmosphere and from Southeast Asia where low pollution emissions are found during the summer (Figure 3b). Therefore emissions from Taiwan are the dominant source of pollution. The increase in the thickness of the planetary boundary layer during summer months, due to the intense heating up of the ground, leads to the frequent impact of boundary layer pollution on our site. The highest rate of conversion of HCl from Cl2 by intense incoming solar radiation as mentioned in the previous section is evident from the highest ratio of Cl−/Na+ (2.8) during the summer months. A small amount of HCl is believed to come from incineration processes. The highest seasonal VWM pH value of 5.15 approaches the value of typical natural rainwater.
 The rest of our data set for the summer and fall months, when TCs (tropical cyclones) appear frequently over the western Pacific Ocean, has been checked for the possibility of the mentioned impacts of the TC. However, the TCs located far from Taiwan and their influences are not significant in some cases. The trajectories show their origins as being marine or they traverse from regions with a recorded noticeable rainfall. This implies that the impact of high pollution loadings associated with the TC on our samples is infrequent.
 Another sample that started on 5 August 2003 also shows peaks of species concentrations (Figure 4). Rainfall was recorded as 10 mm per week on 5–11 August 2003. Unstable air occurs frequently in the summer months due to the intense solar heating of the surface. The infrared satellite images for these days show local clouds (not shown here) appearing over Taiwan in the afternoon (but not in the morning). This indicates that a deep convection developed during these days, which is the main engine for transporting the boundary layer pollutants to the upper level air and resulted in elevated species concentrations in this sample.
 In the fall the trajectory plots starting at 2862 m AMSL indicated that most of the trajectories traverse over a marine atmosphere and the coastal areas of China and Korea, even though very few of the trajectories originated from Southeast Asia (Figure 3c). At 4000 m AMSL, the trajectories originated from Southeast and south Asia, and westerly flows reappeared at the middle troposphere. The species concentrations for this season were the lowest. This is explained by the facts that (1) most of the air masses are of oceanic origin; (2) the direct influence of boundary layer pollution is much less in the fall compared to the summer months because of less surface heating; (3) export of boundary layer pollutants from deep convective (other than TC) and frontal lifting processes are less common compared to those in summer and winter/spring months, respectively; (4) the average surface wind speed is higher in the fall than in the summer because of the smaller influence of the subtropical high-pressure system on Taiwan, resulting in a higher ventilation of local pollution in the fall within the boundary layer in general (The average ventilation coefficients on the upwind side of the Mt. Lulin for the summer and fall of 2004 were estimated to be 1430 and 2403 m2 s−1, respectively.) The VWM pH value in the fall was 5.24, which is less than that in the summer. This indicates that rainwater acidity is controlled by the balancing effects of acidic and alkaline species concentrations but not by the absolute concentrations of these species or by the emission strengths of their precursors [Wai et al., 2005].
 In winter the VWM concentrations for most of the species (except Mg2+) are relatively high. The volume weighted pH value (4.91) is the lowest among all seasons. The average rainfall amount per sample (91.7 mm per week) is the highest. Generally, the trajectory paths (Figure 3d) are to the west of our site pointing to Southeast or south Asia, indicating the prevailing westerly flows in the middle troposphere in winter. However, only three weekly samples were taken in winter and the data set should be analyzed with caution. No further discussion is therefore made.
3.3. Rainwater Chemistry of Mt. Fuji
 Rainwater chemistry studies have been conducted and reported at the summit of Mt. Fuji (35.35°N, 138.73°E, 3776 m AMSL) in Japan since 1990 [Dokiya et al., 1995; Hayashi et al., 2001]. Only recent data sets from 2001 to 2004 are included for discussion here, as well as that of 2002 adopted by Watanabe et al. . The sampling durations were mostly 4 h (ranging from 1.5 to 14 h depending on whether the rainwater amount was enough for chemical analysis) for these studies, compared to the weekly samplings of our work at Mt. Lulin. Therefore it is possible that several samples were collected on the same day, depending upon the actual meteorological conditions of that day.
 About 75% of data were retained after the same data quality check was performed as mentioned in section 2. The rejected data were likely due to possible contamination of the samples during the sampling procedure [Dokiya et al., 2001] under extreme conditions at the summit. The rainfall amount measured by standard rain gauge is not available since the precipitation comes not only from aloft but also from every direction at the summit [Dokiya et al., 2001]. Therefore VWM species concentrations cannot be calculated instead average concentrations at Mt. Fuji are presented here.
 Table 3 summarizes the rainwater chemistry of Mt Fuji from 2001 to 2004. A smaller number of samples is available for species K+, Ca2+, and Mg2+ (26, 30, and 21 samples, respectively). The average pH value for the whole period was 4.66. The acidic species nss-SO42− and the H+ ions (mean concentrations: 23.8 and 22.1 μeq l−1, respectively) were the most abundant anions and cations (19% of the total concentration for each), followed by the species Na+ and Cl− (15% of the total concentration for each).
Table 3. Species Concentrations, μeq l−1, Measured at Mt. Fuji Compared With Those at Mt. Lulina
| ||Station||N||pH||Cl−||NO3−||SO42−||nss- SO42−||H+||NH4+||Na+||K+||Mg2+||Ca2+|
|Seasonal Mean Concentration|
|Spring (Mar to May)||Mt. Fuji||15||4.37||9.7/3.7||8.3/5.7||41.8/16.3||40.8/15.6||42.9/11.2||10.9/6.6||8.0/5.9||4.1/3.6||3.0/0.6||7.9/1.4|
|Summer (Jun to Aug)|| ||11||4.88||0.5||3.0||7.8||7.7||13.1||2.5||1.2||0.7||0.8||3.3|
|Fall (Sep to Nov)|| ||13||5.54||45.2||13.6||22.7||17.0||2.9||18.3||47.4||16.5||1.5||8.2|
|Winter (Dec to Feb)|| ||2||4.70||0.0||8.8||30.1||29.8||20.0||18.1||2.8||2.7||-||-|
|29–30 May 2001|| ||2||3.7||48.7||25.1||207.5||204.9||249.1||38.7||22.0||5.6||4.2||21.8|
|6 Jun 2001|| ||6||4.9||-||2.5||5.3||5.1||12.6||1.7||1.3||-||1.2||1.9|
| ||Mt. Fuji||-||5.01||10.6||5.3||16.8||16.0||9.8||7.3||7.6||6.1||1.0||5.0|
| ||Mt. Lulin||-||5.23||7.3||12.4||15.9||15.5||5.9||15.4||2.8||2.2||0.8||5.3|
 The species Cl− and Na+ were highly correlated (R = 0.87, p < 0.001, N = 39) with a slope of about one (Figure 2b). This indicates minimal chloride depletion in the rainwater samples and a sea-salt origin of these species. It is worth noting that there is a very high correlation between K+ and these species (R > 0.82), with a K+/Na+ equivalent ratio of about 0.34. This ratio is about 15 times higher than that of sea water, and the excess K+ is thought to originate from biofuel consumption as will be discussed later. The Varimax rotated factor analysis also indicates very high loadings (>0.92) of K+, Na+, and Cl− species with eigenvalue 5.59. This suggests that these loadings are brought together through long-range transport by the westerly flows (refer to the trajectory analysis in the following section) from east Asia.
 The species Mg2+ shows a high correlation (R = 0.79, p < 0.001, N = 21) with Ca2+, suggesting its nonmarine and crustal source origin. When removing one sample of extreme values, nss-SO42− and NO3− are associated with each other (R = 0.81, p < 0.001, N = 40) with a slope of about 2.44.
 If two samples with an exceptionally high nss-SO42− concentration (316.1 and 93.6 μeq l−1) are removed from the data set of Mt. Fuji, NH4+ becomes well correlated with nss-SO42− (r = 0.76, p < 0.001, N = 40) and the average slope decreases from 2.37 to 1.06. This indicates that NH4+ and nss-SO42− existed in the form of (NH4)2SO4. The NH4+ and NO3− correlation was very high (R = 0.87, p < 0.001, N = 40).
 The species K+ was correlated with NH4+ (R = 0.71, p < 0.001, N = 26). A relatively large value of correlation coefficient (0.76) between K+ and NH4+ was also found during ACE-Asia aircraft measurement over northeastern Asia [Lee et al., 2003]. It is believed that NH4+ sampled within such areas originated from the biofuel consumption in China [Lee et al., 2003; Ma et al., 2003].
3.4. Seasonal Differences of Species Concentrations at Mt. Fuji
 The data set of Mt. Fuji is further divided based on different seasons (Table 3). Owing to the 4-h sampling protocol, a larger sample size was obtained although only 9, 6, 6, and 2 sampling days were available in spring, summer, fall, and winter, respectively. Therefore these results should be interpreted with caution and the seasonal differences in species concentrations at Mt. Fuji are only discussed briefly with some case studies, as follows.
 The starting heights of the trajectories were set at 4000 m and 5000 m AMSL, similar to those adopted by Watanabe et al. . The trajectories are computed on a daily (rather than every 4 h) basis according to the sampling days. Many trajectories (Figure 5) originate from the west side of the summit, indicating that the westerly flows prevail over the site. This agrees with the year-round measurements of wind speed and direction reported by Igarashi et al. .
Figure 5. Five-day back trajectories at Mt. Fuji in (a) spring, (b) summer, and (c) fall at 4000 and 5000 m AMSL. Trajectories in winter are not shown here.
Download figure to PowerPoint
 The mean concentrations of NH4+ and nss-SO42− were elevated during the winter months (Table 3). However, the sample size is too small to support further meaningful discussion. The transport pathways of the trajectory in spring (Figure 5a) were more or less similar to those reported by Igarashi et al.  for the middle of February 2003. Although most of them traversed over north/northeastern China with high pollution emissions at the boundary layer as noted earlier, no elevated species concentration were found when two samples with episodic concentrations (as discussed below) were removed. A closer examination of the heights of the trajectories indicates that most of them were above 3000 m AMSL, well above the polluted boundary layer. Lee et al.  also summarized that most of the species (NH4+, SO42−, NO3−, K+, Na+, and Cl−) concentrations in fine aerosols above this level were at least a factor of 5 less than those in the boundary layer. Asian dust storms occur most frequently in springtime and their impact on the upper level air is expected to be significant, particularly for enhancing the crustal Mg2+ and Ca2+. However, the lowest Mg2+ and Ca2+ concentrations among other seasons are believed to be due to the fewer samples available for capturing the dust storm signatures as discussed later in section 3.5.
 The abnormally high concentrations for most of the species on 29–30 May 2001 are believed to be due to the influence of the volcanic plume of Miyake-Jima (34.08°N, 139.53°E) located 170 km southeast of Mt. Fuji. The differences in mean concentrations in spring before and after the removal of these two samples are shown in Table 3. On the basis of the forward and backward trajectory plots (not shown here) the influence of the plumes from Miyake-Jima on Mt. Fuji starts less than 12 h after leaving the volcano. Episodic concentrations of SO2 and aerosol SO42− were reported to be due to the influences of the plumes of Miyake-Jima [Igarashi et al., 2004; Naoe et al., 2003]. However, Igarashi et al.  commented that the volcanic episodes occurred very infrequently and that the local wind direction and the existence of local convection played important roles in such episodes. Lee et al.  reported coemission of NH3, SO2, and aerosol SO42− of Miyake-Jima. The elevated NH4+ and nss-SO42− concentrations (52.3 and 316.1 μeq l−1, respectively) on 30 May 2001 are the highest in the data set of Mt. Fuji.
 On the basis of the trajectory plots in the summer months (Figure 5b), the air masses are mostly resident above 3000 m AMSL. At the same time, some of them originate from the marine boundary layer where clean air masses are expected, and they result in the lowest concentrations for most species except Mg2+ and Ca2+ which can be attributed to a marine origin. Igarashi et al.  also reported the lowest SO2 concentration being measured during the summer months at Mt. Fuji. They argued that a Pacific High covering the Japanese islands can transport clean marine air to the summit.
 The concentrations for all the species were low on 6 June 2001. For instance, the concentrations of Ca2+ and nss-SO42− were 1.9 and 5.1 μeq l−1, respectively. However, the corresponding trajectories pointed to the boundary layer of southern China, where high pollution emissions were expected from the Pearl River Delta (PRD). Closer examination of the surface weather maps from 1 to 6 June (Figure 6) shows that a quasi-stationary (Mei-yu) front was located over southern China which was associated with heavy rainfall, thereby reducing the pollution levels. Meanwhile, no TC was found near the South China Sea during those days and the associated high pollution loadings mentioned in section 3.2 did not appear over the coastal areas of southern China.
Figure 6. (a) Surface pressure map and (b) its corresponding visible satellite image showing the Mei-yu front located over southern China at 0000 UTC, 2 June 2001.
Download figure to PowerPoint
 In the fall the concentrations for most species are the highest among the seasons. The species concentrations in the fall were averaged from the data set available in September 2002 only, which was analyzed in detail by Watanabe et al. . The backward trajectories in the fall are plotted in Figure 5c. They argued that the high concentrations of Na+ and Cl− in the rainwater were due to the rising motion associated with a strong TC (Sinlaku, central pressure 960 hPa) approaching Taiwan and which then struck the southern part of China. After these species were lifted from the boundary layer to the free troposphere, they were transported to the summit through long range transport. It should be pointed out that other anthropogenic and crustal species can also be transported to the summit along with the sea salt species. It is reflected by the elevated concentrations of NH4+, SO42−, NO3−, K+, and Ca2+, which are always associated with the elevated sea salt species concentrations in the samples. The elevated anthropogenic and crustal concentrations are believed to come from the high pollution loadings in the boundary layer over eastern China, which is similar to the case of Mt. Lulin discussed in section 3.2.
 The rest of the data set has been checked for the possible impact of the TC on the middle troposphere rainfall chemistry. This was done by checking for any TC located in the western Pacific region from the surface weather map, starting 10 d prior to each sampling day. However, no TC existed in this region, or if it did its influence was unlikely to affect the rainwater chemistry at Mt. Fuji, as indicated by the paths of backward trajectories.
3.5. Comparison Between Sampling Sites and the Aerosol Chemistry During TRACE-P and ACE-Asia
 Apart from the intercomparison of the rainwater chemistry between the two sites, the data sets of the aerosol chemistry (in the fine mode) obtained from the TRACE-P and ACE-Asia campaigns are included in this study to examine if there is consistency in the findings between aerosol and rainwater chemistry as mentioned in section 1. The insights and relationship between aerosol and rainwater chemistry are then to be investigated. The findings from Ma et al.  regarding the measurements of the plumes from the biomass burning in Southeast Asia by TRACE-P P-3B aircraft and the findings of Lee et al.  regarding the measurements over the East China Sea and around Japan and the Korean Peninsula by ACE-Asia C130 aircraft are mostly used in the present study. Detailed flight tracks and sampling procedures of these experiments can be found in the two literatures and references therein. However, the comparison should be made with caution since these campaigns were conducted in late winter/spring only and the chemical properties of the aerosols may vary from season to season.
 Table 3 shows that the acidity of the rainwater at Mt. Lulin (5.12) is lower than that at Mt. Fuji (4.66). While the species concentrations for nss-SO42−, Ca2+, and Mg2+ at Mt. Lulin and Mt. Fuji were relatively comparable, the sea-salt species Na+ and Cl− and the K+ concentrations were more than double at Mt. Fuji, but they were about a factor of two less for NH4+ and NO3−. Only three samples had a ratio of Cl−/Na+ larger than 1.6 at Mt. Fuji. No excess chloride was found for most of the samples, which differs from the finding at Mt. Lulin. The reasons for that might be (1) the photochemical reactions were not rigorously compared to those at the lower latitude where Mt. Lulin is located; (2) Cl− emitted from incineration processes at the boundary layer is unlikely to reach the summit of Mt. Fuji which is located higher than our site by about 1000 m.
 The average ratio of nss-SO42−/NO3− was 1.38 at Mt. Lulin, which is within the typical range obtained from the fine aerosol measurements during TRACE-P [Ma et al., 2003]. This may imply that the contribution of trace gases SO2 and HNO3 to nss-SO42− and NO3−, respectively, in rainwater are insignificant, compared to the contributions of aerosol nss-SO42− and NO3−. The higher ratio (2.44) at Mt. Fuji indicates a higher contribution of nss-SO42− to the rainwater acidity than that of NO3−, which may be due to the influence of the pollution plumes (enriched by SO2) from north/northeast China and the volcanic plumes from Japan.
 The ratio of Mg2+/Ca2+ was 0.15 at Mt. Lulin, which is in excellent agreement with the value 0.16 measured by the TRACE-P aircraft P-3B when the high Mg2+ samples of marine origin were removed [Lee et al., 2003]. The flight tracks of the aircraft covered from 8°N to 40°N, taking some measurements near Taiwan. The agreement may suggest that these species were scavenged directly by cloud or rain droplets from the aerosol phase to the aqueous phase. The ratio was 0.20 at Mt. Fuji as compared to the value of 0.12 measured by aircraft NCAR C130 with flight tracks mainly around Japan, eastern China, and Korea [Lee et al., 2003]. They also pointed out that the high correlation between these species was due to the crustal origin of Mg2+ for most of the samples.
 The species NH4+ and nss-SO42− appeared as (NH4)2SO4 at both sites. Lee et al.  concluded that in most conditions NH3 was emitted in quantities comparable to or greater than H2SO4 and its precursor SO2 throughout the flight tracks of the TRACE-P and ACE-Asia campaigns. This more or less explains the complete neutralization of H2SO4 and NH3. The molar ratio of NH4+/NO3− obtained in Mt. Lulin (1.39) and in Mt. Fuji (1.49) showed good agreement. The larger-than-one ratio may be explained by the higher emission quantities of NH3 than NOx over China and Southeast/south Asia [Streets et al., 2003]. The molar ratio of NH4+/(NO3−+ SO42−) showed an altitude dependence in the TRACE-P and ACE-Asia campaigns. This ratio decreased with the altitude from 2000 to 5000 m in both campaigns. A similar variation is found in the present study when comparing the data sets of Mt. Lulin (median value of 0.58) with those of Mt. Fuji (median value of 0.20) in the spring.
 Mineral particles were reported to be important carriers and sinks for NO3− [Matsumoto et al., 2004; Dentner et al., 1996] and thus a high correlation should exist between Ca2+ and NO3−. However, the correlation was only moderate (r = 0.62) at Mt. Lulin and Mt. Fuji (r = 0.63).
3.6. Comparison of Baseline Concentrations
 In general, the baseline concentration of a specific pollutant is determined on the basis of large volume of data set, while only 3-year data are available for Mt. Lulin. Instead, the comprehensive statistic approaches and comparisons with other sites are adopted. The baseline species concentrations (Table 3) are defined as the average concentrations which are free from the effects of episodic influences from the source regions, and their concentrations are temporally more uniform [Al-Momani et al., 1998]. While the definition of episodic concentration was not mentioned clearly by these authors, it was defined as two standard deviations plus the average concentration [Wai and Tanner, 2005; Chung et al., 1999]. However, too few data can be removed from the data sets in our case, and therefore such criteria is relaxed to one standard deviation plus the average concentration here, resulting in at most 20% of data being removed for both sites. While it is worth noting that Ca2+, Mg2+, nss-SO42− concentrations and pH agree so well at both sites, the NH4+ and NO3− concentrations at Mt. Lulin are about double. On the other hand, the K+ and Na+ concentrations at Mt. Fuji are higher than those at Mt. Lulin by a factor of two. Therefore the baseline species concentrations of Mt. Lulin are found to be comparable in magnitude to those of Mt. Fuji. The summit of Mt. Fuji was reported as an excellent platform for atmospheric background monitoring [Igarashi et al., 2004, and references therein]. In view of the above, it is suggested that the measurements of Mt. Lulin are representative of the background characteristics in rainwater in east Asia, after removal of the influencing episodic periods. However, there are drawbacks in such a comparison due to the limited samples mentioned and the short sampling durations at Mt. Fuji, which may influence the reliability of the findings here. Therefore the Mt. Lulin data set is also compared with those of other remote sites at high elevations as follows.
3.7. Comparison With Data Sets Measured in Other Areas
 Rainwater chemistry data sets from other remote or background monitoring sites located at high altitudes and remote islands as well as those in the arctic region are compared here (Table 4). It should be noted that a few measurements were conducted more than 10 years ago, so the chemical properties in the rainwater may have already been changed. Nevertheless, these data sets can act as some reference information regarding the rainwater chemistry of our site. The comparison is also complicated by the fact that the distributions of pollution sources and their strengths are highly inhomogeneous in a global sense. Continental-scale pollution transport is another complicated factor. Therefore Asian measurements are more important for our comparison here.
Table 4. A Comparison of VWM Species Concentrations, μeq l−1, at Our Site With Those at Other Areasa
|Locations||Period||Elevation, m||Cl−||NO3−||SO42−||nss- SO42−||H+||NH4+||Na+||K+||Mg2+||Ca2+|
|Mt. Fuji, Japan||2001–2004||3780||18.9||8.6||26.1||23.8||22.1||11.6||18.5||9.8||1.5||6.4|
|Mt. Rocky, US||2004||2920||1.0||9.5||6.1||6.0||4.8||6.3||1.0||0.3||0.0||5.8|
 A comparison of the data sets (Table 4) obtained by sites at high altitude (>1000 m AMSL) indicates that sea salt species and K+ in our site ranks at the low side, in the sense of Asian measurements or on a global comparison. Sea salt concentrations are even higher in Lhasa and Mt. Fuji than those in sites of lower altitude in Asia. A similar K+ concentration, as a biomass burning tracer, was measured at Malaysia and at our site. The concentrations of anthropogenic species such as nss-SO42−, NO3−, and NH4+ tend toward the high side when they are compared with those of Asian measurements, but they rank at the middle in a global sense. On the other hand, the opposite holds true for the observation on H+ concentration. The H+ concentration is less than 1μeq l−1 in Lhasa, due to the large amount of alkaline species Ca2+ existing in the rainwater samples. A low Ca2+ concentration was measured in our site, based on Asian or global measurements. When comparing the results at our site with those at remote islands and the arctic region, the sea salt species concentrations are comparable to the arctic values, while the concentrations on Cape Hedo (Okinawa Island) and Ogasawara Island are expected to be very high because the monitoring sites are close to the ocean. The anthropogenic species concentrations mentioned at our site rank at the high side. However, H+, Ca2+, and K+ concentrations at our site are low. To summarize, the concentrations of all species measured at Mt. Lulin rank within the range of concentrations recorded at other high-altitude or remote sites. Combined with previous discussions, we suggest that Mt. Lulin can be used as a regional site in Southeast Asia for the study of background rainwater chemistry and long-range transport of atmospheric pollutants.