SEARCH

SEARCH BY CITATION

Keywords:

  • sulfate;
  • nitrate;
  • carbonaceous substances;
  • mineral aerosols;
  • east Asia;
  • long-range transport

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[1] Investigations of chemical properties and transport mechanisms of continental aerosols are necessary for estimating their influences on global radiative budget and on the global material cycle. Intensive measurements of atmospheric aerosols and the associated species on Rishiri Island, near the northern tip of Japan, were conducted from March to May 2001, in order to understand the chemical properties, source regions, transport pathways, and transport patterns of anthropogenic and mineral aerosols over the east Asian Pacific Rim region during the spring. Mean concentrations of nss-SO42−, NO3, NH4+, nss-Ca2+ in aerosols were 2.48, 0.64, 0.72, and 0.17 μg m−3, respectively. Elemental carbon and organic carbon in fine particles (d < 2.5 μm) yielded mean concentrations of 0.25 and 0.80 μg m−3, respectively. The concentrations of these species frequently increased to higher values because of outbreaks of continental polluted air masses, whereas under background conditions, they decreased to lower values similar to those observed over the remote ocean. Our results demonstrate that nss-SO42− and NH4+ coexist in fine particles, that NO3 and nss-Ca2+ coexist in coarse particles, and that each set is transported in an alternate manner. Continentally derived NO3 is transported as coarse particle to the east Asian Pacific Rim region. Anthropogenic pollutants and dust particles are not necessarily transported together. It was often found that anthropogenic fine particles containing abundant nss-SO42− appeared first and were then followed by large mineral particles that had absorbed NO3. Short-term intrusion of the air masses containing abundant particulate carbonaceous compounds, probably due to the influence of biomass burning, also often occurred during the outflow events of continental air masses. Atmospheric behaviors of sulfate, nitrate, and carbonaceous species are different from one another, although they are all derived mainly from combustion processes.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[2] Atmospheric pollutants and dust particles are transported to the remote ocean, and perturb radiation budget and chemical composition in marine atmosphere. In addition, atmospheric inputs of land-derived materials to the ocean may have significant impact on the in-situ marine ecosystems. The western North Pacific is located on the lee side of east Asia, where industrial activities are currently intensifying. Moreover, the region witnesses a large loading of mineral dust called “Kosa,” especially in spring, so that much research has focused on the aerosol chemistry of this oceanic region. Long-term observations of aerosol chemical properties in the east Asian Pacific Rim region have been conducted on several remote islands; for instance, Jeju (also known as Cheju) Island [Carmichael et al., 1997], Oki Islands [Mukai and Suzuki, 1996], and Ogasawara Islands [Matsumoto et al., 1998]. All these researchers have reported remarkable winter-spring enhancement of anthropogenic and mineral aerosols due to the prevailing westerlies.

[3] High-time-resolution intensive measurements of atmospheric aerosols and the associated species during outbreaks of continental air masses are crucial to understand the physical and chemical properties, transport mechanisms, and transformation processes of continental aerosols. Uematsu et al. [1992] reported that the daily concentrations of particulate non-sea salt sulfate and nitrate at Vladivostok and Sapporo, which are located in the east Asian Pacific Rim at a distance of approximately 700 km, covaried during the springtime, but both species showed somewhat different transport patterns. A continuous aerosol sampling at 2-hour intervals during a Kosa event in Nagasaki, Japan, demonstrated that fine sulfur particles behaved differently from large mineral particles and both components did not exist as internally mixed particles [Uematsu et al., 2002]. They concluded that it takes a certain time to mix the different air masses containing various types of aerosols along their transport paths from the Asian continent.

[4] In order to predict the spatial and/or temporal distributions of aerosol species and their depositional flux, and to understand how aerosol chemical species influence on global radiation budget and on global material cycles, it is necessary to clarify the physical and chemical properties, source regions, transport mechanisms, and internal and/or external mixing processes of anthropogenic and mineral aerosols. As a contribution to the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia), we conducted intensive measurements of atmospheric aerosols and the associated species on Rishiri Island, near the northern tip of Japan, in the spring of 2001. During this spring, continental polluted air masses are frequently transported to the northern region of Japan [Matsumoto et al., 2003b]. This study focused on the (1) temporal variations in the concentrations; (2) chemical properties; (3) source regions and transport pathways; and (4) transport patterns of anthropogenic and mineral aerosols over the east Asian Pacific Rim region during the springtime from March to May 2001.

2. Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

2.1. Observation Site and Period

[5] Figure 1 shows the location of the observation site (45.07°N, 141.12°E) on Rishiri Island. The observatory is located at the foot of Mt. Rishiri (1721 m), about 35 m above sea level, and at a distance of 800 m from the nearest coastline. Although there is a residential area and a fishing port about 1 km from the observatory, they are very small, so that the influence of local contamination on our observation has rarely happened. The effects of local contamination can be assessed from the spike-like sudden enhancement of particle number densities, and then the data affected by them are removed in the following discussion.

image

Figure 1. Location of the observation site on Rishiri Island.

Download figure to PowerPoint

[6] In this study, we conducted continuous measurements of particle number densities, the concentrations of carbonaceous substances in aerosols and carbon monoxide (CO) from 1 March to 31 May 2001. To analyze the chemical composition of aerosols, air sampling on filters were also carried out from 29 March to 18 May 2001.

2.2. Measurements of Atmospheric Aerosols and the Associated Species

[7] Ambient aerosols segregated into two size fractions (d < 2.5 μm and d > 2.5 μm) were collected every 24 hours on a PTFE fiber filter (Toyo Roshi Kaisha, Ltd., PF040) using a high-volume dichotomous virtual impactor air sampler equipped with an automatic filter change system (Kimoto Electric Co., Ltd., Model ACS21). A 50 mm inside diameter straight pipe (about 3 m length) was used to transport ambient air into the instrument. The filter samples were ultrasonically extracted with ultrapure water (specific resistivity >18 MΩcm), and then analyzed for major ionic species by ion chromatography (Dionex, DX-120). Anion species (Cl, NO3, and SO42−) were analyzed by an AS4A-SC separate column, and cation species (Na+, NH4+, K+, Mg2+, and Ca2+) were analyzed by a CS12A separate column. The eluent was 1.8 mM Na2CO3/1.7 mM NaHCO3 for anions, and 18mM methanesulfonic acid for cations.

[8] The concentrations of organic carbon (OC) and elemental carbon (EC) in aerosols were continuously measured using an ambient carbon particulate monitor (Rupprecht & Patashnick Co. Inc., Model 5400) at 4-hour intervals. Only ambient aerosols with diameter <2.5 μm were introduced into the instrument using a PM2.5 cyclone (with a 50% cutoff efficiency for 2.5 μm particles) to eliminate coarse particles, and then collected with impactors (with a 50% effective cut-off diameter of 0.14 μm) at a flow rate of 16.7 L min−1. A 28.8 mm inside diameter straight pipe (about 3.5 m length) was used as the inlet of the instrument. The collection plate of the impactor was heated at 50°C in order to minimize the adsorption of gaseous organic matter during sampling; therefore a large fraction of the particulate carbonaceous substance evolved below 50°C may have been lost from our measurements. The collected samples of carbonaceous substances were volatilized by heating the collection plate and then transformed to CO2 by combustion at 750°C with an afterburner. The concentration of CO2 was then measured by a non-dispersive infrared (NDIR) CO2 sensor. The heating temperatures of the collection plate were set at four stages, 200°C, 250°C, 340°C and 750°C. It has been reported that the temperature split between OC and EC can be clearly seen at temperatures between 300–340°C: 320°C at Uji, Kyoto [Höller et al., 2002], 300°C at Tokyo and 340°C over the Pacific Ocean [Miura et al., 2000]. In this study, the carbonaceous substances evolved at 340°C were defined as OC and those at 750°C was defined as total carbon (TC). The difference between the amounts of TC and OC gives the amount of EC. Detailed explanation of this instrument has been reported in the previous studies [Uematsu et al., 2001; Höller et al., 2002].

[9] Particle number densities were determined by an optical particle counter (Rion Co. Ltd., KC18) at 20-minute intervals, with five size fractions of 0.10–0.15, 0.15–0.20, 0.20–0.30, 0.30–0.50 and >0.50 μm. The counter shared the same sampling pipe as the ambient carbon particulate monitor. A part of sample air stream was divided from the mainstream, and delivered into the optical particle counter. The concentrations of carbon monoxide (CO) were measured by a modified NDIR photometer instrument (Kimoto Electric Co., Ltd., model 541). The air inlet for CO measurement was made of PFA- Teflon tubing, coupled with a glass manifold [Tanimoto et al., 2002].

3. Results and Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

3.1. Temporal Variations in the Concentrations of Aerosol Species

3.1.1. Particle Number Density

[10] Figure 2 shows the temporal variations of the particle number densities measured from March to May 2001. The particle number densities largely fluctuated in short-term periods, corresponding to the outbreaks of continental air masses. The total number densities of the particles larger than 0.1 μm in diameter frequently increased to 1000 cm−3 or more because of the outbreaks, whereas these values decreased to around 100 cm−3 under background conditions. Matsumoto et al. [2001] reported that particle number densities (d > 0.1 μm) were obtained as 100 cm−3 or less over the remote tropical western North Pacific, but these values increased to >1000 cm−3 over the oceanic region close to the Japan's main island.

image

Figure 2. Temporal variations of particle number densities. The arrows show the outflow events of the continental air masses discussed in section 3.3.

Download figure to PowerPoint

3.1.2. Ion Species in Aerosols

[11] The concentrations of water-soluble ionic species in aerosols are summarized in Table 1. Figure 3 shows the temporal variations and frequency of the concentrations of non-sea salt sulfate (nss-SO42−), nitrate (NO3), ammonium (NH4+), and non-sea salt calcium (nss-Ca2+) in aerosols. The concentrations of nss-SO42− and nss-Ca2+ were calculated from the ratios of SO42−/Na+ and Ca2+/Na+ in seawater, respectively. Temporal variations in the concentrations of four chemical species are similar to one another and to those of the particle number densities.

image

Figure 3. Temporal variations and histograms of the concentrations of (a) nss-SO42−, (b) NO3, (c) NH4+, and (d) nss-Ca2+ in aerosol. The solid triangles show the arithmetic mean, and the open triangles show the median value.

Download figure to PowerPoint

Table 1. Concentrations of Ionic and Carbonaceous Compounds in the Aerosols From 29 March to 31 Maya
 MeanMedianMaximumMinimumMean/Median
  • a

    Concentrations are given in μg/m3. Here, bd, below the detection limit (<0.1 μg/m3).

Total
Cl1.550.866.210.031.81
NO30.640.343.410.031.91
nss-SO42−2.482.239.220.361.11
Na+1.170.754.050.041.57
NH4+0.720.662.870.091.10
K+0.120.070.63<0.011.70
Mg2+0.140.100.44<0.011.39
nss-Ca2+0.170.091.23<0.011.99
 
PM2.5
TC1.050.778.480.241.36
EC0.250.161.92bd1.57
OC0.800.616.560.211.30
OCHT0.410.274.74bd1.54

[12] The mean concentration of nss-SO42− was determined to be 2.48 μg m−3, which is close to the values previously reported from the east Asian Pacific Rim region during the winter and/or spring. For instance, Uematsu et al. [1992] reported a mean concentration of nss-SO42− as 2.69 μg m−3 from Sapporo (about 250 km south of Rishiri) in the springtime, and Matsumoto et al. [1998] obtained a mean concentration of 2.44 μg m−3 for nss-SO42− on Hahajima Island over the western North Pacific during outbreaks of continental air masses. On the other hand, higher concentrations of nss-SO42− (with a mean value of 7.90 μg m−3) have been found on Jeju Island, near the Korean Peninsula, during the spring [Carmichael et al., 1997]. In Rishiri, when the site was affected by the outflows of continental air masses, nss-SO42− concentrations intermittently increased to 5 μg m−3 or more. On the contrary, lower concentrations of nss-SO42− ranged within 0.5–1.0 μg m−3 were most frequently observed as shown by the histogram of nss-SO42− (Figure 3a), and these are close to the values previously reported from the remote ocean; for instance, 0.55 μg m−3 in the oceanic high-pressure region of northwestern Pacific during the summer [Matsumoto et al., 1998].

[13] Mean concentration of NO3 was found to be 0.64 μg m−3, which is close to the value reported from Hahajima Island during outbreaks of continental air masses [Matsumoto et al., 1998]. The samples with the concentrations below 0.2 μg m−3, however, were most frequently observed, and these lies within the range of values measured in the remote marine background atmosphere; for instance, 0.18–0.35 μg m−3 [Prospero et al., 1985] and 0.11–0.36 μg m−3 [Savoie et al., 1989] over the remote Pacific.

[14] Mean concentration of NH4+ was 0.72 μg m−3. Previous observations in the east Asian Pacific Rim region found the mean concentrations of NH4+ to be 0.37 μg m−3 on Hahajima Island during outbreaks of continental air masses [Matsumoto et al., 1998], and 1.32 μg m−3 on Jeju Island in the springtime [Carmichael et al., 1997].

[15] Figure 4 shows the temporal variation in the ratio of Cl/Na+ in aerosols. This ratio was found to decrease when the site was hit by continental polluted air masses. Chlorine depletion in aerosols is caused by the reactions of acid gases such as sulfur dioxide and nitric acid with sea-salt particles [e.g., Zhuang et al., 1999]. The outflow of continental air masses containing abundant acid gases may cause chlorine-loss reactions in aerosols, thus leading to depletion of Cl in sea-salt particles and generation of hydrochloric acid gas.

image

Figure 4. Temporal variation in the ratio of Cl/Na+ in aerosol. The dashed line shows the ratio in seawater.

Download figure to PowerPoint

[16] The concentrations of particulate species showed large fluctuations during the observation period. The air quality on Rishiri showed drastic alternations according to air mass origins; the concentrations of particulate species were frequently higher because of outbreaks of continental polluted air masses, whereas under background conditions, they decreased to lower values similar to those observed over the remote ocean. As shown in Table 1, arithmetic mean values of the concentrations of all compounds are larger than the median values, which reflect the intermittent influence from the outflows of continental air masses. The ratios of mean/median values are higher especially for NO3 and nss-Ca2+, but are lower for nss-SO42− and NH4+. More pronounced enhancement in the outflows was found in the concentrations of NO3 and nss-Ca2+ than those of nss-SO42− and NH4+. The histograms and temporal variations for nss-SO42− and NH4+ show similar patterns, and those for NO3 and nss-Ca2+ also resembles each other. It is inferred from these results that the transportation of these compounds is done in the following order: nss-SO42− together with NH4+, and NO3 together with nss-Ca2+.

[17] Table 2 summarizes mean concentrations of ionic compounds in the fine- and coarse-particle ranges. Most of nss-SO42− (92%) and NH4+ (100%) exist as fine particles (d < 2.5 μm), which is consistent with many previous studies. Homogeneous nucleation and subsequent condensation of sulfuric acid derived from the photo-oxidation of sulfur dioxide, and aqueous oxidation reaction of sulfur dioxide absorbed in cloud droplets are the main generation processes of nss-SO42− in aerosols [McHenry and Dennis, 1994], which subsequently react with gaseous ammonia and transform to ammonium salt [Yamato and Tanaka, 1994]. Absorption of sulfur dioxide on sea-salt particles and subsequent oxidation are also a significant generation process of nss-SO42− in marine aerosols [Sievering et al., 1992, 1999; Yvon and Saltzman, 1996]. The size distributions of nss-SO42− and NH4+ obtained in this study can be explained on account of these generation processes. NO3 is mainly included in coarse particle range (d > 2.5 μm), as well as nss-Ca2+. Mineral dust particles act as an important carrier for NO3 in this region, as will be explained later.

Table 2. Mean Concentrations of Ionic Species in Fine and Coarse Particlesa
 Coarse (d > 2.5 μm)Fine (d < 2.5 μm)
MeanRatio to TotalMeanRatio to Total
  • a

    Concentrations are given in μg/m3. Here, nd, not detectable.

Cl1.130.720.440.28
NO30.370.570.280.43
nss-SO42−0.190.082.290.92
Na+0.730.620.440.38
NH4+nd0.000.721.00
K+0.040.290.100.71
Mg2+0.090.600.060.40
nss-Ca2+0.110.550.090.45
3.1.3. Particulate Carbonaceous Species

[18] Figures 5a and 5b show the temporal variations and frequency of EC and OC, respectively. Unfortunately, because of the failure of the thermal control, the measurement of OC evolved below 200°C was unsuccessful until 28 March; therefore the data of OC during this period are not available. The results for carbonaceous species measurements from 29 March to 31 May are also summarized in Table 1. The mean concentrations of TC, EC, and OC are 1.05, 0.25, and 0.80 μg m−3, respectively.

image

Figure 5. Temporal variations and histograms of the concentrations of (a) EC, (b) OC, and (c) OCHT in fine particles (d < 2.5 μm). The solid triangles show the arithmetic mean, and the open triangles show the median.

Download figure to PowerPoint

[19] During the outflow events of continental air masses, the concentrations of carbonaceous species drastically increased to higher values: around 1 μg m−3 and 2 μg m−3 or more for EC and OC, respectively. On the contrary, under background conditions, the carbonaceous species showed lower values, that is < 0.1 μg m−3 for EC and 0.3–0.9 μg m−3 for OC. These temporal variations in the concentrations of particulate carbonaceous species clearly show the alternation of air quality on Rishiri according to the origin of air masses. This pattern was also true for the ionic species and the particle number densities.

[20] Recently, the ambient carbon particulate monitor has been used for continuous measurement of particulate OC and EC in urban [Höller et al., 2002] and remote marine atmosphere [Uematsu et al., 2001; Matsumoto et al., 2001]. Höller et al. [2002] determined the annual mean concentrations of TC and EC in fine particles (d < 2.5 μm) in Uji, Japan, as 4.76 and 2.73 μg m−3, respectively. Matsumoto et al. [2001] reported that the concentrations of TC and EC in fine particles (d < 2.5 μm) were about 0.70 and 0.08 μg m−3, respectively, over the remote tropical North Pacific, whereas they increased to 1.14 and 0.34 μg m−3, respectively, over the oceanic region close to the Japan's main island.

[21] The measurements of particulate carbonaceous species by our method have significant uncertainties as with many previous studies. Matsumoto et al. [2003a] reported that a positive artifact probably caused by the adsorption of organic gases on the collectors of the ambient carbon particulate monitor could significantly lead to an overestimation of OC, especially the OC evolved at temperatures below 200°C. Similar results have also been reported by the measurements using quartz fiber filters, which have traditionally been used to collect and analyze particulate carbon [Novakov et al., 1997]. In marine atmosphere, however, the OC evolved at high temperatures between 200 and 340°C (OCHT) and EC may be not largely affected by the positive artifact in our method [Matsumoto et al., 2003a]. The concentrations of OCHT are also given in Table 1. The ratio of OCHT/OC was about 0.51, indicating that 49% of OC could be affected by positive artifact. The temporal variation and frequency of the OCHT concentrations are shown in Figure 5c, which are similar to those of OC. The concentrations of OCHT measured in this study should be considered as lower limited values, since they may be subject to underestimation because of a negative artifact induced by the evaporation of collected particulate organic matter during sample collection. Besides, in the thermal analysis used in our method, it is possible that the pyrolysis of some portion of OC also result in the underestimation of OC [Novakov et al., 2000a; Chow et al., 2001].

[22] Another problem associated with the ambient carbon particulate monitor is that particulate carbonaceous substances smaller than 0.14 μm in diameter are not collected by the impactor, and this may result in the underestimation of carbonaceous substances, especially of EC [Höller et al., 2002]. Although OC is mainly concentrated in the submicron size range, only a small fraction of OC is expected to be smaller than 0.14 μm [McMurry and Zhang, 1989; Höller et al., 2002]. Höller et al. [2002] argued that even though a large fraction of EC exists as particles smaller than 0.14 μm in the urban area, it concentrates in accumulation-mode particles larger than 0.14 μm in the rural sites. The underestimation of EC due to missing the particles smaller than 0.14 μm should be small on Rishiri, since the site is mostly affected by chemically aged air masses.

3.2. Source and Transport of Chemical Species

3.2.1. Relationships Between Air Trajectories and Aerosol Component Concentrations

[23] In this study, the relationship between the air trajectories and the concentrations of aerosol components was analyzed to identify the transport pathway and the source region of each chemical species. The trajectories used in this study are isentropic 7-day backward trajectories, which were calculated using the hybrid single-particle Lagrangian integrated trajectory (HY-SPLIT 4) model (NOAA Air Resources Laboratory, web address http://www.arl.noaa.gov/ready/hysplit4.html). The trajectories were calculated twice daily at 0000 and 1200 GMT. Starting altitudes were set at the top of the boundary layer over the site, which was judged from vertical profiles of potential temperature obtained from the Wakkanai Observatory of the Japan Meteorological Agency located about 50 km east of Rishiri Island.

[24] The trajectories were classified into four types according to their directions as shown in Figure 6a. Figure 6b shows the appearance frequency of each air mass from March to May 2001. Continental air mass from east Siberia (region SIB) was a predominant air mass.

image

Figure 6. (a) Four categories of the air trajectories. (b) Appearance frequency of the air trajectories of the air masses that reached Rishiri during the observation period. The number of trajectories from four regions is shown in respective columns.

Download figure to PowerPoint

[25] Figure 7 shows the average values of normalized concentrations (N) of aerosol components in each air mass, which were calculated by the following equation:

  • equation image

where m is the average concentration of aerosol species for the samples collected in each air mass, and M is the average concentration for all samples. The concentrations of all species show higher values in the air masses from region CHI, whereas they are lower for the air masses coming from regions OHK and PAC, indicating that these species are mainly derived from continental sources. Although the air masses from regions SIB and CHI are both derived from the continent, that from region CHI has higher concentrations of aerosol species, implying that the emission sources of them are mainly located in region CHI.

image

Figure 7. Normalized concentrations classified into the four trajectory pathways shown in Figure 6.

Download figure to PowerPoint

[26] The pattern that higher concentration was found in the air masses from region CHI is more prominent for ionic components, especially for NO3 and nss-Ca2+. This suggests that a mineral dust plume was transported to the site after passing over region CHI, since nss-Ca2+ is an indicator of mineral dust. The concentration of NO3 shows similar distribution to that of nss-Ca2+, which supports the interpretation that mineral particles act as an important carrier of NO3. Compared to the ionic components, EC shows relatively higher concentrations in the air masses from region SIB, suggesting that the source regions of EC were somewhat different from those of the anthropogenic ionic compounds such as nss-SO42−. Particle number density (d > 0.1 μm) shows similar distribution to EC concentration, being relatively higher in the air masses from region SIB. Both EC concentration and particle number density may increase in the air masses directly affected by combustion processes. It is possible that the air masses from region SIB were significantly affected by biomass burning, since biomass smoke particles contain abundant carbonaceous compounds but relatively scarce sulfate [Novakov and Corrigan, 1996]. Recently, it has been recognized that natural forest fires often occur in and around eastern Siberia [Kasischke et al., 1999; Chankina et al., 2001], and they have a profound impact on the regional air pollution [Tanimoto et al., 2000; Kato et al., 2002]. In Rishiri, the episodic high concentrations of CO derived from boreal forest fires in far eastern Siberia were frequently observed during the period from summer to early fall 1998 [Tanimoto et al., 2000]. On the other hand, the OC concentrations do not show appreciable differences among the air trajectory types, probably since OC is derived not only from combustion source but also significantly from natural sources such as the ocean and/or terrestrial vegetation, as will be discussed later.

3.2.2. Relationships Among the Concentrations of Chemical Species

[27] Figure 8a shows a good correlation between the concentrations of nss-SO42− and NH4+. It is suggested that particulate nss-SO42− on Rishiri mainly exist as ammonium salt, which is also evidenced from similarities in their temporal variations, histograms, and size distributions. The average mole equivalent ratio of NH4+/nss-SO42− was determined as 1.56, which is higher than those reported from Hahajima Island during outbreaks of continental air masses (1.04) [Matsumoto et al., 1998] and from Jeju Island in the spring (0.89) [Carmichael et al., 1997]. Neutralization of sulfate was relatively progressed on Rishiri, since this site is significantly affected by ammonia, which is mainly emitted from continental surfaces such as fertilization and livestock waste [Bouwman et al., 1997], and, moreover, not largely affected by heavy loading of nss-SO42−.

image

Figure 8. Relationships between the concentrations (units, μg m−3) of (a) nss-SO42− and NH4+, (b) nss-SO42− and NO3, and (c) NO3 and nss-Ca2+. The dotted circle shows the samples collected from 9 to 11 April.

Download figure to PowerPoint

[28] Figure 8b shows the relationship between the concentrations of nss-SO42− and NO3. Both of them show a good correlation, since they are both derived from fossil fuel combustions and subsequent photo-oxidations. As discussed above, however, the size distribution of NO3 is different from that of nss-SO42−, and the temporal variations and histograms of NO3 are similar to those of nss-Ca2+. Figure 8c shows the relationship between the concentrations of NO3 and nss-Ca2+. Exceptionally higher concentrations of nss-Ca2+ marked by a dotted circle were obtained in the samples from 9 to 11 April, which will be discussed later. When these samples are excluded, a good correlation is obtained between the concentrations of NO3 and nss-Ca2+. It is inferred from these results that NO3 is mostly absorbed by mineral particles, and that mineral particles act as an important carrier for NO3 in this region, which has been found in the previous observation in east Asia [Carmichael et al., 1997]. This contrasts to the fact that most of NO3 is generated through the reaction with sea-salt particles in remote marine atmosphere [Savoie and Prospero, 1982; Matsumoto et al., 1998].

[29] As shown in Figure 9a, the concentrations of EC and CO show a good correlation, although these species exist in particulate and gaseous phases, respectively. Both EC and CO are primary compounds emitted from combustion processes, and this common origin may explain the good correlation between their concentrations. It is inferred that both were emitted from the same source regions and transported together to the site. The relationship between the concentrations of EC and nss-SO42− is shown in Figure 9b. The regression line between EC and nss-SO42− in the aerosols over the central Pacific obtained by Kaneyasu and Murayama [2000] is also plotted in this figure. Although the measurements of Kaneyasu and Murayama [2000] yielded lower concentration values for both EC and nss-SO42−, the slope of the line is close to that obtained in our study. Exceptionally higher concentrations of EC marked by a dotted circle were obtained in the samples on 17 April and those collected from 26 to 29 April; these will be discussed later. Excluding these samples, the concentrations of EC and nss-SO42− show a good correlation, which can be attributed to the fact that most of them are derived from combustion sources. It is expected, however, that the source region and the generation processes of EC are somewhat different from those of nss-SO42−, as evidenced from the air trajectories depicted in Figure 7. Most of nss-SO42− is derived from fossil fuel combustions and subsequent photo-oxidations, but EC is a primary compound derived from combustion processes including both fossil fuel combustion and biomass burning.

image

Figure 9. Relationships between the concentrations of (a) EC and CO and (b) EC and nss-SO42−. The dotted circle shows the samples collected on 17 April and those collected from 26 to 29 April. The bold line shows the regression line between EC and nss-SO42− in aerosols over the central Pacific obtained by Kaneyasu and Murayama [2000], and the thin line is from this study.

Download figure to PowerPoint

[30] Figure 10a shows the relationship between the concentrations of EC and OC. A good correlation is found between both concentrations. OC may be derived from a wide variety of sources; combustion processes including fossil fuel combustion and biomass burning, and natural sources such as ocean and terrestrial vegetation [Duce et al., 1983]. The good correlation between both concentrations may suggest that main sources of OC are combustion processes as well as EC. As shown by the air trajectories (Figure 7), however, relatively higher concentrations of OC were found in marine air masses. Figure 10b shows the relationship between the ratios of OC/EC and CO concentrations. The ratios of OC/EC are higher in the air masses with lower CO concentrations, indicating that sources other than combustion processes contributed to the OC concentrations under the unpolluted conditions, and that the spatial distributions of OC and EC are not exactly the same.

image

Figure 10. Relationships between (a) the concentrations of EC and OC, (b) CO concentrations and the ratios of OC/EC, and (c) CO concentrations and the ratios of OCHT/EC.

Download figure to PowerPoint

[31] Figure 10c shows the relationship between the ratios of OCHT/EC and CO concentrations, which is similar to the case of OC/EC shown in Figure 10b. Since the measurement of OCHT includes only a small positive artifact as mentioned above, these results suggest that the existence of OC at relatively higher concentrations under the unpolluted condition should not be attributed merely to the positive artifact in the measurement of OC but that it is necessary to consider other sources for OC than combustion processes. Natural OC emissions from ocean and terrestrial vegetation should be considered as significant sources of particulate OC, and should be incorporated in the numerical simulations for global aerosol distribution.

3.3. Chemical and Physical Properties of Aerosols During the Outflow Events

[32] In Rishiri, although the concentrations of anthropogenic and mineral aerosols increased during the outflow events of continental air masses, the behavior of these components were not completely synchronized with one another. During ACE-Asia, the Rishiri site was exposed to outbreaks of continental air masses several times. In order to clarify the transport patterns of anthropogenic and mineral aerosols, the following discussion focuses on three outflow events termed as episodes 1–3 in Figure 2.

3.3.1. Episode 1

[33] Figure 11 shows the temporal variations in the concentrations of nss-SO42−, NO3, NH4+, nss-Ca2+, and carbonaceous species in aerosols, and the number densities of the particles larger than 0.1 μm and 0.5 μm in diameter during the period from 3 to 13 April. It is interesting to note that the concentrations of nss-SO42−, NH4+, and carbonaceous species in aerosols gradually increased from 4 April, reaching its maximum peak on 11 April, whereas the concentrations of NO3 and nss-Ca2+ drastically increased on 9 April and showed higher values until 11 April. The number densities of the particles larger than 0.1 μm and 0.5 μm in diameter were synchronized with the concentrations of nss-SO42− and nss-Ca2+, respectively. These results indicate that the plume reaching Rishiri consisted of two air masses, which were not mixed together. An anthropogenic polluted air mass containing abundant particulate nss-SO42−, NH4+, and carbonaceous species appeared first, and afterward, a dust storm containing mineral particles that had adsorbed NO3 reached the site.

image

Figure 11. Temporal variations of the aerosol species during episode 1 (3–13 April).

Download figure to PowerPoint

[34] Figure 12 shows the synoptic weather charts and the reproduced spatial distributions of nss-SO42− and dust particles simulated by the chemical weather forecasting system (CFORS), which was developed by Uno et al. [2003], during episode 1. CFORS integrates a regional chemical transport model with a multitracer system built within the RAMS (Regional Atmospheric Modeling System) mesoscale meteorological model, and produces 3-dimentional distributions of particulate and gaseous chemical species and major meteorological parameters. A wide variety of chemical species are included in this system; for instance, sulfur dioxide, sulfate, black carbon, organic carbon, mineral dust and sea salt. Detailed information on CFORS is given by Uno et al. [2003].

image

Figure 12. Representative synoptic surface weather charts (0900 JST) and spatial distributions of nss-SO42− and dust particles reproduced by CFORS (1200 JST) during episode 1. Solid circles in the panels indicate the location of Rishiri.

Download figure to PowerPoint

[35] Weather charts and numerical simulations show that a low-pressure system contributed to sporadic transport of dust storm, which reached Rishiri around 9 April and covered the site until 11 April. On the other hand, particulate nss-SO42− was successively transported to Rishiri from 5 to 11 April, via the northern or western edge of an anticyclone, because of sequential passing of low-pressure systems through the north of Japan. On 12 April, the low-pressure system advancing in the south of Rishiri brought about easterly winds from the ocean, so that the concentration of every aerosol component decreased to the lower values. These numerical simulations sufficiently reproduced the observation results. Both field and numerical studies show that particulate anthropogenic pollutants and dust particles are not transported together during this episode. Anthropogenic fine particles appeared first, and afterward, large mineral particles arrived at Rishiri. This transport pattern might be not exceptional phenomenon, since similar pattern was also observed during a Kosa event in Nagasaki, near the western tip of Japan [Uematsu et al., 2002].

[36] As shown in Figure 8c, extremely higher concentrations of nss-Ca2+ were obtained from 9 to 11 April. It should be noted that a large dust storm developed in the inland of the Asian continent around 5 April, spread rapidly over the east Asian coastal regions, and then traveled across the North Pacific [Seinfeld et al., 2003]. The higher concentrations of nss-Ca2+ observed at Rishiri around 10 April can be attributed to this dust storm. The large dust storm effectively transported mineral particles together with anthropogenic substances from the east Asian continent to the northwestern Pacific.

3.3.2. Episode 2 and Episode 3

[37] The temporal variations in the concentrations of the aerosol species from 12 to 20 April (episode 2) are shown in Figure 13. During this episode, carbonaceous compounds showed different behavior from ionic species. The concentrations of carbonaceous compounds drastically increased on 17 April. Figure 14 shows the synoptic weather charts and isentropic 7-day backward trajectories from 16 to 19 April. The air trajectory on 17 April shows the intrusion of the air masses from eastern Siberia (region SIB in Figure 6a) to Rishiri. The number densities of the particles larger than 0.1 μm in diameter show a trend similar to the concentrations of carbonaceous species. These results imply that, during this period only, Rishiri was affected by the temporary intrusion of the air masses from region SIB, which contained abundant carbonaceous compounds and fine particles. Throughout this episode, temporal variations of other ionic species resemble each other. The transport of anthropogenic and mineral particulate substances from the Asian continent to the observation site during this episode was accelerated by the low-pressure system passing over the north of Japan.

image

Figure 13. Temporal variations of the aerosol species during episode 2 (12–20 April).

Download figure to PowerPoint

image

Figure 14. Synoptic surface weather charts (0900 JST) and 7-day isentropic backward trajectories (0900 JST) from 16 to 19 April.

Download figure to PowerPoint

[38] Episode 3 represents the outbreak of continental polluted air masses from 26 to 30 April. Although the concentrations of all aerosol components were synchronized with one another as shown in Figures 2, 3, and 5, particulate carbonaceous species were relatively enriched during this episode, as shown in Figure 9b with a dotted circle. The number densities of the particles larger than 0.1 μm in diameter show a trend similar to the concentrations of carbonaceous substances. The synoptic weather charts and isentropic 7-day backward trajectory of 28 April are shown in Figure 15. A large traveling anticyclone carried the continental aerosols to the observation site. The air trajectory also shows the intrusion of the air masses from region SIB during episode 3.

image

Figure 15. Synoptic surface weather charts (0900 JST) and 7-day isentropic backward trajectories (0900 JST) on 28 April.

Download figure to PowerPoint

[39] It is possible that the air masses containing enriched carbonaceous compounds, which were obtained in episodes 2 and 3, were significantly affected by biomass burning in region SIB. The ratio of OC/EC in the aerosols, however, did not largely change during these episodes. Several studies have reported that the ratio of OC/EC would be high in biomass smoke particles [Novakov and Corrigan, 1996; Novakov et al., 2000b]. On the other hand, it has been reported that the ratio of OC/EC is quite different among burning conditions (flaming burning or smoldering burning) [Novakov and Corrigan, 1996]. Although we cannot exactly determine the source of the air masses containing abundant particulate carbonaceous substances derived from region SIB, our results suggest that the source regions of carbonaceous compounds are not the same as for the ionic species such as nss-SO42−, which is meaningful for the numerical study of aerosol distributions. In addition, it is inferred from our results that particulate carbonaceous compounds have a large influence on the fine particle number density.

4. Summary and Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[40] As a contribution to the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia), we conducted intensive measurements of atmospheric aerosols and associated species on Rishiri Island, near the northern tip of Japan, during the spring period from March to May of 2001. The conclusions of this study are summarized as follows:

[41] 1. Mean concentrations of nss-SO42−, NO3, NH4+, and nss-Ca2+ in the aerosols were determined as 2.48, 0.64, 0.72, and 0.17 μg m−3, respectively. EC and OC in fine particles (d < 2.5 μm) showed mean concentrations of 0.25 and 0.80 μg m−3, respectively. The concentrations of every particulate compound showed large fluctuations during the observation period. The air quality on Rishiri showed drastic alternations according to the origin of the incoming air masses; the concentrations of particulate species were frequently higher because of outbreaks of continental polluted air masses, whereas under background conditions, they decreased to lower values similar to those observed over the remote ocean.

[42] 2. More pronounced enhancement in the outflows was found in the concentrations of NO3 and nss-Ca2+ than those of nss-SO42− and NH4+. Nss-SO42− mainly exists as ammonium salt in fine aerosols. NO3 is absorbed on coarse mineral particles, so that mineral particles may play an important carrier for NO3 in this region. It is inferred that nss-SO42− and NH4+ coexist in fine particles, NO3 and nss-Ca2+ coexist in coarse particles, and each pair is respectively transported in an alternate manner. Continentally derived NO3 is transported as coarse particle to the east Asian Pacific Rim region.

[43] 3. The source regions of EC were found to be somewhat different from those of nss-SO42−, although both are mainly derived from combustion processes. It is expected that the air masses from eastern Siberia were significantly affected by biomass burning. Although the main sources of OC are combustion processes as well as EC, natural OC emissions from the ocean and terrestrial vegetation should also be considered as significant sources of particulate OC. Carbonaceous compounds have a large influence on the fine particle number density.

[44] 4. Anthropogenic pollutants and dust particles are not always transported together. It was found that anthropogenic fine particles containing abundant nss-SO42− appeared first, and afterward, large mineral particles that had absorbed NO3 reached Rishiri. This transportation pattern is probably not exceptional for the east Asian Pacific Rim region. The temporary intrusion of the air masses containing abundant particulate carbonaceous compounds often appeared during the outbreaks of continental air masses. Atmospheric behavior of sulfate, nitrate and carbonaceous species are different from each other, although all of them are mainly derived from combustion processes. It takes a certain time to mix the different air masses containing various types of aerosols along their transport pathways.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[45] We express our gratitude to H. Akimoto, Y. Miyata, K. Ohta, and M. Sawaki for their generous support during our observations and to NOAA Air Resource Laboratory for using the HY-SPLIT 4 model. This study was financially supported by the Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Corporation (JST).

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information
  • Bouwman, A. F., D. S. Lee, W. A. H. Asman, F. J. Dentener, K. W. Van Der Hoek, and G. J. Olivier, A global high-resolution emission inventory for ammonia, Global Biogeochem. Cycles, 11, 561587, 1997.
  • Carmichael, G. R., M.-S. Hong, H. Ueda, L.-L. Chen, K. Murano, J. K. Park, H. Lee, Y. Kim, C. Kang, and S. Shim, Aerosol composition at Cheju Island, Korea, J. Geophys. Res., 102, 60476061, 1997.
  • Chankina, O. V., T. V. Churkina, A. V. Ivanov, V. A. Ivanov, G. A. Ivanpva, K. P. Koutsenogii, and G. A. Kovalskaya, Multielement composition of the aerosols of the forest fires of boreal forests upon burning of forest combustibles, Nucl. Instrum. Methods Phys. Res., Sect. A, 470, 444447, 2001.
  • Chow, J. C., J. G. Watson, D. Crow, D. H. Lowenthal, and T. Merrifield, Comparison of IMPROVE and NIOSH carbon measurements, Aerosol Sci. Technol., 34, 2334, 2001.
  • Duce, R. A., V. A. Mohnen, P. R. Zimmerman, D. Grosjean, W. Cautreels, R. Chatfield, R. Jaenicke, J. A. Ogren, E. D. Pellizzari, and G. T. Wallace, Organic material in the global troposphere, Rev. Geophys., 21, 921952, 1983.
  • Höller, R., S. Tohno, M. Kasahara, and R. Hitzenberger, Long-term characterization of carbonaceous aerosol in Uji, Japan, Atmos. Environ., 36, 12671275, 2002.
  • Kaneyasu, N., and S. Murayama, High concentrations of black carbon over middle latitudes in the North Pacific Ocean, J. Geophys. Res., 105, 19,88119,890, 2000.
  • Kasischke, E. S., K. Bergen, R. Fennimore, F. Sotelo, G. Stephens, A. Janetos, and H. H. Shugart, Satellite imagery gives a clear picture of Russia's boreal forest fires, Eos Trans. AGU, 80, 141147, 1999.
  • Kato, S., P. Pochanart, J. Hirokawa, Y. Kajii, H. Akimoto, Y. Ozaki, K. Obi, T. Katsuno, D. G. Streets, and N. P. Minko, The influence of Siberian forest fires on carbon monoxide concentrations at Happo, Japan, Atmos. Environ., 36, 385390, 2002.
  • Matsumoto, K., I. Nagao, H. Tanaka, H. Miyaji, T. Iida, and Y. Ikebe, Seasonal characteristics of organic and inorganic species and their size distributions in atmospheric aerosols over the northwestern Pacific Ocean, Atmos. Environ., 32, 19311944, 1998.
  • Matsumoto, K., M. Uematsu, K. Ohta, M. Sasakawa, and K. Miura, Chemical composition of atmospheric aerosols over the western North Pacific, paper presented at 11th Symposium on Atmospheric Chemistry, Sol.-Terr. Environ. Lab., Nagoya Univ., Toyohashi, Japan, 11–12 Jan. 2001.
  • Matsumoto, K., M. Uematsu, and T. Hayano, Measurement of particulate carbonaceous substances by an ambient carbon particulate monitor, in Programs and Abstracts of the 8th International Conference on Atmospheric Sciences and Applications to Air Quality (ASAAQ2003), p. 239, Natl. Inst. of Adv. Ind. Sci. and Technol., Tsukuba, 2003a.
  • Matsumoto, K., M. Uematsu, T. Hayano, K. Yoshioka, H. Tanimoto, and T. Iida, Simultaneous measurements of particulate elemental carbon on the ground observation network over the western North Pacific during the ACE-Asia campaign, J. Geophys. Res., 108(D23), 8635, doi:10.1029/2002JD002744, in press, 2003b.
  • McHenry, J. N., and R. L. Dennis, The relative importance of oxidation pathways and clouds to atmospheric ambient sulfate production as predicted by the regional acid deposition model, J. Appl. Meteorol., 33, 890905, 1994.
  • McMurry, P. H., and X. Q. Zhang, Size distributions of ambient organic and elemental carbon, Aerosol Sci. Technol., 10, 430437, 1989.
  • Miura, K., H. Miyashita, K. Suzuki, and S. Nakae, Measurements of carbonaceous particles in urban and marine atmosphere, paper presented at 17th Symposium on Aerosol Science and Technology, Japan Soc. of Aerosol Sci. and Technol., Hiroshima, 2–4 Aug. 2000.
  • Mukai, H., and M. Suzuki, Using air trajectories to analyze the seasonal variation of aerosols transported to the Oki Islands, Atmos. Environ., 30, 39173934, 1996.
  • Novakov, T., and C. E. Corrigan, Cloud condensation nucleus activity of the organic component of biomass smoke particles, Geophys. Res. Lett., 23, 21412144, 1996.
  • Novakov, T., C. E. Corrigan, J. E. Penner, C. C. Chuang, O. Rosario, and O. L. Mayol Bracero, Organic aerosols in the Caribbean trade winds: A natural source? J. Geophys. Res., 102, 21,30721,313, 1997.
  • Novakov, T., S. Timothy, S. Bates, and P. K. Quinn, Shipboard measurements of concentrations and properties of carbonaceous aerosols during ACE-2, Tellus, Ser. B, 52, 228238, 2000a.
  • Novakov, T., M. O. Andreae, R. Gabriel, T. W. Kirchstetter, O. L. Mayol-Bracero, and V. Ramanathan, Origin of carbonaceous aerosols over the tropical Indian Ocean: Biomass burning or fossil fuels? Geophys. Res. Lett., 27, 40614064, 2000b.
  • Prospero, J. M., D. L. Savoie, R. T. Nees, R. A. Duce, and J. Merrill, Particulate sulfate and nitrate in the boundary layer over the North Pacific Ocean, J. Geophys. Res., 90, 10,58610,596, 1985.
  • Savoie, D. L., and J. M. Prospero, Particle size distribution of nitrate and sulfate in the marine atmosphere, Geophys. Res. Lett., 9, 12071210, 1982.
  • Savoie, D. L., J. M. Prospero, and E. S. Saltzman, Nitrate, non-sea sulfate and methanesulfonate over the Pacific Ocean, in Chemical Oceanography, vol. 10, edited by J. P. Riley, R. Chester, and R. A. Duce, pp. 219250, Academic, San Diego, Calif., 1989.
  • Seinfeld, J. H., et al., Regional climatic and atmospheric chemical effects of Asian dust and pollution, Bull. Am. Meteorol. Soc., in press, 2003.
  • Sievering, H., J. Boatman, E. Gorman, Y. Kim, L. Anderson, G. Ennis, M. Luria, and S. Pandis, Removal of sulfur from the marine boundary layer by ozone oxidation in sea-salt aerosol, Nature, 360, 571574, 1992.
  • Sievering, H., B. Lerner, J. Slavich, J. Anderson, M. Posfai, and J. Cainey, O3 oxidation of SO2 in sea-salt aerosol water: Size distribution of non-sea-salt sulfate during the First Aerosol Characterization Experiment (ACE-1), J. Geophys. Res., 104, 21,70721,717, 1999.
  • Tanimoto, H., Y. Kajii, J. Hirokawa, and H. Akimoto, The atmospheric impact of boreal forest fires in far eastern Siberia on the seasonal variation of carbon monoxide: Observations at Rishiri, a northern remote island in Japan, Geophys. Res. Lett., 27, 40734076, 2000.
  • Tanimoto, H., H. Furutani, S. Kato, J. Matsumoto, Y. Makide, and H. Akimoto, Seasonal cycles of ozone and oxidized nitrogen species in northeast Asia: 1. Impact of regional climatology and photochemistry observed during RISOTTO 1999–2000, J. Geophys. Res., 107(D24), 4747, doi:10.1029/2001JD001496, 2002.
  • Uematsu, M., T. Sugita, V. V. Anikiev, and A. N. Medvedev, Large-scale transport of pollution aerosol over the east coast of Asia, Geophys. Res. Lett., 19, 22192221, 1992.
  • Uemastu, M., K. Ohta, K. Matsumoto, and I. Uno, Short term variation of marine organic aerosols under the Northwestern Pacific high pressure region in the summer of 1999, Geochem. J., 35, 4957, 2001.
  • Uematsu, M., A. Yoshikawa, H. Muraki, K. Arao, and I. Uno, Transport of mineral and anthropogenic aerosols during a Kosa event over east Asia, J. Geophys. Res., 107(D7), 4059, doi:10.1029/2001JD000333, 2002.
  • Uno, I., et al., Regional chemical weather forecasting system CFORS: Model descriptions and analysis of surface observations at Japanese island stations during the ACE-Asia experiment, J. Geophys. Res., 108(D23), 8668, doi:10.1029/2002JD002845, 2003.
  • Yamato, M., and H. Tanaka, Aircraft observations of aerosols in the free marine troposphere over the North Pacific Ocean: Particle chemistry in relation to air mass origin, J. Geophys. Res., 99, 53535377, 1994.
  • Yvon, S. A., and E. S. Saltzman, Atmospheric sulfur cycling in the tropical Pacific marine boundary layer (12°S, 135°W): A comparison of field data and model results: 2. Sulfur dioxide, J. Geophys. Res., 101, 69116918, 1996.
  • Zhuang, H., C. K. Chan, M. Fang, and A. S. Wexler, Formation of nitrate and non-sea-salt sulfate on coarse particles, Atmos. Environ., 33, 42234233, 1999.

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Methods
  5. 3. Results and Discussion
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.