Importance of Long-Range Nitrate Transport Based on Long-Term Observation and Modeling of Dust and Pollutants over East Asia

Long-term synergetic fine and coarse mode aerosol observations were analyzed at 1-h intervals at Fukuoka, Japan, from January to June 2015. The GEOS-Chem chemical transport model, including dust and sea-salt acid uptake processes, was used for detailed analysis of observation data. Several Asian dust events and long-range anthropogenic aerosol transport events were observed during our analysis period, and the numerical model generally explained the observed time variation for both fine and coarse mode aerosols. We found that (i) the majority of fine mode NO3 can be considered as long-range transport (LRT) outside of Japan during the cold season, and (ii) the peak timing of fine mode NO3 coincided with that of SO4, indicating that both aerosols are controlled by LRT. Also, an observed mass concentration ratio of NO3/SO4 > 0.9 occurred during the cold season, indicating the importance of NO3 as a major contributor to the PM2.5 mass fraction. Finally, we clearly showed that large-scale dust-nitrate outflow from China to Fukuoka was confirmed in all cases of dust events, indicating that the anthropogenic NOx is converted to dust-nitrate and transported to Japan with dust. These results demonstrate the importance of anthropogenic NO3 LRT during the cold season and dust-nitrate LRT for all dust events (even in June).


INTRODUCTION
Understanding the behavior of long-range trans-boundary transport of dust and pollutants in East Asia is an important environmental issue because of the frequent outflow of heavy PM 2.5 pollution from China (e.g., Wang and Shi, 2010).Ground-base and airborne episodic aerosol observations have been studied to determine the physics and chemistry of high-concentration events (e.g., Carmichael et al., 1997;Huebert et al., 2003;Jacob et al., 2003), but the period of these observation campaigns are typically less than one month, which is insufficient to study seasonal variations.EANET (2014), AD-Net Lidar (Sugimoto et al., 2008), and other monitoring data have been accumulating observation data for more than 10 years, but most of these data do not include detailed hourly aerosol compositions.Seasonal changes in Asian monsoons play an important role in the pattern and frequency of long-range transport of pollutants (e.g., Jeong and Park, 2017); therefore, long-term aerosol observation including aerosol compositions with high resolution is required, but such a detailed observation has not yet been well-performed.
Acid deposition problems have been studied since 1990 in Japan (e.g., Fujita et al., 2000) because SO 2 emission increased in China due to the rapid economic development (Ohara et al., 2007, Kurokawa et al., 2013).These studies have used atmospheric sulfate has as an important aerosol tracer, and nitrate has not been well-studied because the atmospheric lifetime of HNO 3 and NO 3 -is shorter than SO 4 2and relatively difficult to measure without artifact.Because NO x emission in East Asia has been rapidly increasing over the last decade, the importance of long-range nitrate transport is becoming increasingly important, and the continuous measurement of both fine and coarse mode aerosol compositions, including NO 3 -, is necessary.We have conducted long-term synergetic observations to capture the behaviors of aerosols around the Chikushi Campus of Kyushu University, located in the suburbs of Fukuoka City (33.52°N,130.47°E),since October 2013 (Pan et al., 2015;Osada et al., 2016;Pan et al., 2106;Uno et al., 2016;Itahashi et al., 2017).We use a state-of-the-art aerosol observation instrument to measure both fine and coarse mode aerosol, measuring single particles (0.5 µm < Dp < 10 µm) using a Polarization Optical Particle Counter (POPC), and multiple-wave length Mie-Raman lidar from the surface to a 10-km height continuously.
During the six-month observation period of January-June 2015, several yellow sand and heavy pollutant transport episodes were observed.This article reports the major characteristics of anthropogenic aerosols and long-range dust transport based on the observation and chemical transport model analysis during this period.We also examined seasonal variations in fine and coarse mode nitrate longrange transport, which has not been reported previously.
This paper is constructed as follows.Section 2 documents the observation dataset and Section 3 describes the detailed chemical transport model simulation.Section 4 discusses the results with respect to temporal variations at observation sites and the model results.Finally, Section 5 provides a summary and conclusions.

Aerosol Compositions
Mass concentrations of PM, sulfate, nitrate, optical black carbon, and water-soluble organic compounds (WSOC) in both fine (Dp < 2.5 µm) and coarse mode (2.5 µm < Dp < 10 µm) were measured using a Continuous Dichotomous Aerosol Chemical Speciation Analyzer (ACSA-12, KIMOTO Electric Co., Ltd) at a time interval of 1 hour (Kimoto et al., 2013) at the rooftop (4F) of the Fukuoka Institute of Health and Environmental Science (130.48°E;33.51°N).The mass concentrations of PM and SO 4 2-were determined using the beta-ray absorption method and the BaSO 4 -based turbidimetric method, respectively, the latter with the addition of BaCl 2 dissolved in polyvinylpyrrolidone solution.The mass concentrations of NO 3 -and WSOC were determined using an ultraviolet absorption-photometric method.It should be noted that ACSA-12 measures both fine and coarse mode aerosols simultaneously and is an important tool for mass budget study and chemical transport model evaluation.Details of the ACSA-12 instrument have been published previously (Kimoto et al., 2013).
The concentration of aerosol NH 4 + was measured using a semi-continuous microflow analytical system (Kimoto Electric Co. Ltd., MF-NH 3 A, Osada et al., 2011) at the Chikushi Campus of Kyushu University.The atmospheric NH x (NH 3 + NH 4 + ) was dissolved in ultrapure water with a continuous air-water droplet sampler and quantified based on fluorescence (excitation, 360 nm; emission, 420 nm) of the o-phtalaldehyde-sulfite-NH 3 reaction product (Genfa et al., 1989).Two inlet lines were used to differentiate the total amounts of NHx and particulate NH 4 + after gaseous NH 3 was removed using a phosphoric acid-coated denuder from the sample air stream.The cut-off diameter of the inlet impactor was about 2 µm (which is smaller than the ACSA PM 2.5 cut-off).In our study, the secondary inorganic aerosols (SO 4 2-, NO 3 -, and NH 4 + ; SIA) were fully observed using our synergetic monitoring system.The horizontal distance from the ACSA-12 site to the Kyushu University site is about 5 km.Both sites are located in a sub-urban area of Fukuoka City.The anthropogenic activity was very limited at both sites, and the air quality showed similar patterns.

Multi-Wavelength Mie-Raman Lidar (MMRL)
The Asian Dust and Aerosol Lidar Observation Network (AD-Net) (Sugimoto et al., 2008) measures both anthropogenic and natural aerosols continuously.This network provides vertical profiles of aerosol backscattering at 532/1064 nm and the depolarization ratio at 532 nm with high spatial and temporal resolution.The automated AD-Net lidar retrieval data for total aerosol extinction coefficient (Shimizu et al., 2010) from the most recent multi-wavelength Mie-Raman Lidar (MMRL) at Fukuoka (Hara et al., 2017;Nishizawa et al., 2017) were used for comparison with the modeled aerosol variations.

Polarization Optical Particle Counter
The atmospheric aerosol particles (0.5 µm < Dp < 10 µm) were measured using a newly developed polarization optical particle counter (POPC) (YGK Corp., Yamanashi, Japan) at Kyushu University, Fukuoka, Japan.The POPC uses a 780nm linearly polarized laser source, and measures both forward scattering and backward scattering intensities.Particle size is determined from the forward scattering intensity, and two polarized compounds (s-polarized/p-polarized, polarization direction perpendicular and/or parallel to the plane of the scattering angle) of backward scattering are measured simultaneously.The depolarization ratio (DR, the fraction of s-polarized to the total backward scattering, S/(S + P)) is a good indicator of particle sphericity because the direction of polarization of scattered light for spherical particles is identical to that of the incident light, while this is not the case for nonspherical particles (e.g., dust).The mass concentration is calculated by assuming the appropriate specific particle weight.The overall measurement uncertainty for number density of super micron particle is less than 15%.Details for the POPC have been reported previously (Kobayashi et al., 2014).

NUMERICAL MODELING
We used the GEOS-Chem 3-D chemical transport model (version 09-02) (Bey et al., 2001;Park et al., 2004) to simulate the formation of aerosols, including mineral dust and secondary inorganic aerosols, over East Asia with onenesting into the global model.For this study, the reactive uptake of HNO 3 and SO 2 on dust limited by dust alkalinity, and the uptake of gas-phase H 2 SO 4 limited by competition with other aerosol surfaces (Fairlie et al., 2010) were used.Dust-nitrate (mainly Ca(NO 3 ) 2 ) was simulated for each dust bin separately, based on the heterogeneous reaction between dust and nitric acid; the reactive uptake constant is a function of RH%, dust diameter, and HNO 3 gas concentration, and is taken from Fairlie et al. (2010).More details of dustnitrate formation from a heterogeneous reaction can be found in Fairlie et al. (2007Fairlie et al. ( , 2010)).
Dust in GEOS-Chem is distributed into four size bins (radii 0.1-1.0, 1.0-1.8, 1.8-3.0, and 3.0-6.0 µm).The smallest size bin is further divided into 4 bins (radii 0.1-0.18,0.18-0.3,0.3-0.6,0.6-1.0µm) for optical properties and heterogeneous chemistry.Several modifications for seasonal changes in dust source function, dust emission fraction within dust-bins, and the change of wet scavenging efficiency by Yumimoto et al. (2017) were used in this study.General performance of updated GEOS-Chem dust simulation was recently reported by Yumimoto et al. (2017) and Uno et al. (2017).GEOS-Chem also simulated sea salt distributed in two size bins (dry radii; 0.01-0.5 and 0.5-8 µm).Sea-salt emission was simulated by Jaeglé et al. (2011), and similar heterogeneous reactions with HNO 3 and SO 2 were also considered (Fairlie et al., 2010).The model calculations for dust and sea-salt NO 3 -and SO 4 2include two bins (fine and coarse modes), and those for other anthropogenic aerosols include one bin (accumulation mode only).Based on the analysis of PM concentration, PM 2.5 and PM 10 were calculated based on the summation of individual aerosol and dust components of the model.The model used the assimilated meteorological fields from the Goddard Earth Observing System (GEOS) of the NASA Global Modeling and Assimilation Office (GMAO) for the year 2015.The model has a horizontal resolution of 2° × 2.5° for a global run and 0.5° × 0.667° for Asian oneway nesting runs (70-150°E, 11°S-55°N), both with 47 vertical levels from the surface to 0.01 hPa.The lowest model layer thickness was about 130 m.We used the anthropogenic emissions of EDGAR for the global domain and REAS Ver.2.1 for the East Asia domain, as reported by Kurokawa et al. (2013).The NH 3 emission of REAS was modified to include seasonal variations for Asia based on recommendations by Huang et al. (2012) and Xu et al. (2015).The model simulation was conducted from the beginning of December 2014 to the end of June 2015, and the first month results were used as model spin-up.Other basic numerical settings were the same as reported in Uno et al. (2014Uno et al. ( , 2016)).
To investigate whether the effect of Japanese domestic or transboundary air pollution is dominant, we also performed a sensitivity simulation.Because the quantity of emissions from China was larger than that from Japan, to avoid large nonlinearities in the atmospheric concentration response to emissions variation (e.g., Itahashi et al., 2015), the sensitivity simulation was designed to include a 20% reduction in anthropogenic emissions from Japan (defined as JOFF20%).Based on differences between the base case simulation (CNTL) and the JOFF20% simulation, the domestic contribution from Japan was estimated using a multiple of 5 for differences in CNTL and JOFF20% experiments.

Time-Series Variation of PM and Aerosol Compositions
Figs. 1(a)-1(c) show the time variation (1-hour average) of PM 2.5 and PM 10 by ACSA-12 and particle volume concentration between 3.6 µm < Dp < 6.0 µm observed by POPC (POPC3, which is the same range of GEOS-Chem dust 3 bin) at Fukuoka.POPC3 can be considered dust particles and were labeled A-G for major dust events.The figures also include the corresponding model values (2hour average).Fig. 1(d) shows the aerosol total extinction coefficient from MMRL lidar at 532 nm averaged from the height 400-700 m.It is clear that the observed PM 2.5 exhibited frequent intermittent peaks between January and March, but the frequency decreased after April, which is related to the frequency of cold air outbreak from Asian continent.The modeled PM 2.5 reproduced most of the observed variation.The modeled PM 2.5 value corresponded to approximately 55% of the observed PM 2.5 (PM 2.5_model = 0.55PM 2.5_obs + 0.64, unit = µg m -3 , r = 0.77) because the observation included un-modeled aerosols (e.g., secondary organic aerosol (SOA), many trace metals and others).The model results were systematically under-estimated during the A event, which was caused by SO 4 2-under-estimation, as reported for the same event by Wang et al. (2016).Similar under-estimation was also reported by Wang et al. (2014) and He et al. (2014).The sharp peak in the C event (see Fig. 5) occurred during the strong dust onset period due to the cold front passage, and it was difficult to capture these sharp increases with a horizontal grid size of 50 km and 3-hour interval meteorological data.
The observed PM 10 included coarse aerosols, and its peaks corresponded to POPC3 peaks when major dust was observed.The modeled DST3 (dust bin 3) was in good agreement with the observed POPC3 and reproduced most of the dust episodes, which indicated that the data assimilation for monthly dust source function by Yumimoto et al. (2017) is accurate.The model failed for the D event because the modeled west wind was not strong enough for the transport of dust to Fukuoka (see Fig. 7(c)).The modeled PM 10 was in good agreement with observations, but constantly underestimated for the same reasons as PM 2.5 (there are many unmodeled coarse aerosols in observations) and showed some uncertainty in the large dust concentration.It is important to note that dust concentration is highest during the B event (higher than PM 2.5 ), while PM 2.5 is higher for the C event (because the dust concentration in C is smaller than B).Based on Fig. 2, the daily SO 4 2-and NH 4 + are reasonably well-predicted by the model.The sensitivity analysis of JOFF20% shows that the contribution from Japan is small and the majority of SO 4 2-can be considered as LRT from outside of Japan (Japan domestic contribution is small at 6.3% for JFM and 11.5% for MAM).For NH 4 + , we can see the contribution of LRT is large until April, after which the Japanese domestic contribution coming from NH 3 emission (11.1% for JFM and 29.3% for AMJ) increases as NH 3 emission increases in warm seasons.
The modeled fine NO 3 -reproduced major time variations, but could slightly over-predict during cold seasons (especially in January and February).These over-predictions can be seen when the NH 4 + is high because the formation of NH 4 NO 3 is sensitive to NH 3 .The NO 3 -concentration is low during the warm season due to the thermal equilibrium between NO 3 -and HNO 3 shifts to a gas phase under warm conditions.The sensitivity analysis of JOFF20% also confirmed that the majority of fNO 3 -can be considered as LRT from outside of Japan during the cold season (in JFM,

85% of fNO 3
-originates from long-range transport).After the middle of May, the Japanese contribution becomes dominant.The peak timing of fine mode NO 3 -coincided with that of SO 4 2-, indicating that aerosol is controlled by LRT, especially during the cold season.
The coarse mode NO 3 -showed a distinct peak value during the dust event.Modeled sea-salt nitrate was constantly in the order of 0.5-1.0 µg m -3 as the baseline of cNO 3 -.The model results showed that the dust-nitrate concentration increased during the dust episode (Japanese domestic contribution is only 2.5%).During the cold season (average of ABC events), the ratio of modeled cNO 3 -and observed cNO 3 -was 0.48, and in the warm season (average of EFG events), it was 0.80.This indicates that the under-estimation of dust-NO 3 -was not due to the assumption of the dust acid uptake coefficient, and supports the importance of further studies on the cNO 3 -formation mechanism, including other processes (such as coagulation of fine particles onto coarse particles) that are not considered in the current model.concentration is lower, as shown in Fig. 6) was used to determine which aerosol contributes more significantly to the mass concentration of PM 2.5 .From this figure, the red circle (NO 3 -/SO 4 2-> 0.9, diameter of circle is PM 2.5 concentration) indicates that the NO 3 -becomes important when PM 2.5 concentration increases, which commonly occurs during the cold season (January and February).After May, NO 3 concentration decreases and the ratio of fNO 3 -/fSO 4 2-becomes less than 0.4, indicating that fSO 4 2-becomes more important for aerosol LRT.
Total aerosol ion is important for aerosol acidity, and we evaluated the mass concentration to total equivalent concentration ratio.The total ion amount is the sum of fine and coarse aerosols.If the mass ratio of NO 3 -/SO 4 2-exceeds 1.3 (which corresponds to an equivalent concentration ratio of 1), then NO 3 -becomes as important as SO 4 2- . We observed 10 cases exceeding this criterion within the 6-month period.This result suggests that dry and wet deposition of NO 3 -cannot be ignored for acid deposition problems.

Monthly Variation
Fig. 6 shows the observed and modeled temporal variation of monthly averaged (a) PM 2.5 , (b) fine and coarse mode SO 4 2-, and (c) fine and coarse mode NO 3 -between January and June at Fukuoka.SO 4 2-concentration peaks during the warm season rather than winter.The variation in the modeled SO 4 2-explains the observed results, but the model under-estimated observed results during January and February.Similar under-estimation can be found in other CTM applications (Wang et al., 2016;Itahashi et al., 2017), and the exact reason remains unclear.
Coarse mode SO 4 2-ranged from 1/10 to 1/6 of fine mode SO 4 2-.Modeled cSO 4 2-is very small compared with observation and is not shown in the figure .NO 3 -levels are higher during the winter and lower during the summer, and exhibit clear seasonal changes.The monthly average of fine NO 3 -for June was 0.90 ± 0.50 -from the trapped filter surface is small (e.g., Osada et al., 2016).
The ratio of observed fNO 3 -: cNO 3 -is approximately 2:1 in January and February.The monthly averaged cNO 3 ranged from about 1.3-1.4µg m -3 until April, even when the impact of yellow sand was different each month.The monthly average of modeled fNO 3 -was in better agreement with observations.For cNO 3 -, the model results were underestimated until March.After April, modeled fNO 3 -and cNO 3 -were nearly in the same order and were in good agreement with observations.It is important to note that NH 4 NO 3 is unstable during the warm season and moves to a gas phase (HNO 3 + NH 3 ), while dust-NO 3 -and seasalt-NO 3 -is stable once formed.Thus, higher cNO 3 -during the warm season is important for HNO 3 -NO 3 -mass budget and acid deposition studies because coarse mode aerosol is easily deposited on the surface of the ground due to its large gravitational velocity (and will supply excess N to the ground eco-system).

Outflow of SO 4
2-and NO 3 -from the Asian Continent to Western Japan Fig. 7 shows the time-longitude variation of daily averaged (a) fSO 4 2-, (b) fNO 3 -, and (c) total dust-nitrate from GEOS-Chem model simulations at 33.5°N (as shown in Fig. 3).Lines in the figure represent the concentration variation at Fukuoka, as shown in Fig. 2 (in (b), we also plot HNO 3 gas, which is not in shown in Fig. 2).The longitude of 119°E corresponds to north of Shanghai, 131°E of Fukuoka.Because the major wind direction is from the west (due to westerly), the vertical high concentration lines in the figure indicate a large-scale outflow from the Asian continent to western Japan.
From this figure, it is clear that the high concentration of SO 4 2-in China occurs intermittently during the winter, and after April SO 4 2-reaches a higher level in China.Outflow from China can be seen during all periods (intermittently during the winter time), but the concentration is higher and more frequent outflow is observed during the warm season.On the other hand, NO 3 -concentration is higher in winter in China, and this air mass outflow intermittently reaches Fukuoka with SO 4 2-.During the warm season, NO 3 -level decreases and HNO 3 becomes important.It is important that we can see large-scale dust-nitrate outflow from China to Fukuoka during all dust events (all A-G dust episodes).This indicates that the anthropogenic NO x is converted to dust-nitrate and transported to Japan with dust.The results shown in Fig. 6 demonstrate the importance of anthropogenic NO 3 -LRT during the cold season, and dust-nitrate LRT for all cases (even in June).This contributes a major fraction of aerosol in PM 2.5 and PM 10 and is important for supplying N to the surface of the ground.

Averaged Horizontal Distribution of Dust-Nitrate
Fig. 8 shows the horizontal distribution (at z = 65 m level) of 6-months averages for (a) total dust, (b) total dust-nitrate, (c) ratio of total dust-nitrate/total dust, and (d) total dust-nitrate/fine NO 3 -.The ratio of dust NO 3 /dust_all increased from the dust source region to the central China plain because of rapid loss of coarse dust as it moved away from the dust sources and generated dust-nitrate.This ratio becomes 8-10% over the Japan area, and fractions of dust-NO 3 -increased compared with dust concentrations (heterogeneous reaction progressed).The ratio of dust-NO 3 -/fNO 3 -reached 18% over the Yellow Sea, 25% in Fukuoka, 60% over the Pacific Ocean just east of Tokyo, and 50% over the northwest Pacific Ocean.Over the warm ocean, thermodynamic partitioning of total nitrate would be favored for the gas phase, and this may be the reason for the increased ratio of dust-NO 3 -/fNO 3 .Thus, dust-NO 3 -becomes increasingly important compared with anthropogenic fNO 3 -.Figs.8(e) and 8(f) show the downwind changes of these four parameters along the two paths (Fig. 8(e) for the dust source from Beijing -north Korea -Hokkaido -north of the northwestern Pacific Ocean, and Fig. 8(f) shows the changes for the dust source from Shandong -Fukuoka -south part of the northwestern Pacific Ocean).These two results clearly show the spatial changes and demonstrate that the chemical transport model, which does not include acid uptake processes over coarse aerosol, may over-estimate fine-mode NO 3 -and HNO 3 concentrations.

CONCLUSIONS
Long-term synergetic fine and coarse mode aerosol observations were analyzed at 1-hour intervals at Fukuoka, Japan, from January to June 2015.The GEOS-Chem chemical transport model including the dust and sea-salt acid uptake processes were used for the analysis of observation data.Our findings from this study can be summarized as: 1) Several Asian dust events and long-range anthropogenic aerosol transport were observed during our analysis period, and a numerical model generally explained the observed time variation for both fine and coarse mode aerosols.2) Sensitivity analysis revealed that the fine mode SO 4 2peak in Fukuoka was controlled by LRT outside of Japan (around 90%), and fine mode NH 4 + was also controlled by LRT during the cold season (89%) and its fraction decreased during late spring (70%).
3) The majority of fine mode NO 3 -can be considered as LRT from outside of Japan during the cold season (in JFM, 85% of fNO 3 -originates from long-range transport).The peak timing of fine mode NO 3 -coincided with that of SO 4 2-, and indicated the both aerosols are controlled with LRT (especially during the cold season).The mass concentration ratio of NO 3 -/SO 4 2-> 0.9 was observed during the cold season, supporting the importance of NO 3 -as a major contributor to the PM 2.5 mass fraction.4) From the time-longitude variation of daily averaged concentration shown in Fig. 7, a high concentration of SO 4 2-in China occurs intermittently during the winter, and after April, SO 4 2-reaches a higher level in China.Outflow from China can be seen during all periods.On the other hand, NO 3 -concentration is higher in winter in China, and this high air mass outflows intermittently and reaches Fukuoka with SO 4 2-.5) Large-scale dust-nitrate outflow from China to Fukuoka was confirmed in all cases of dust events (all A-G dust events).This indicates that anthropogenic NO x is converted to dust-nitrate and transported to Japan with dust.These results support the importance of anthropogenic NO 3 -LRT during the cold season, and dust-nitrate LRT for all dust events (even in June).

Fig
are indicated by red dots; the straight line indicates the model control experiment (CNTL: all emission), and the blue barline is the Japan contribution calculated from CNTL and JOFF20%.The modeled cNO 3 -is the sum of dust-nitrate and sea-salt nitrate.Figs.3 and 4show the daily changes in dust (color) and SO 42-(contour and pink shade > 10 µg m -3 ) horizontal distributions using the model's first vertical level (z = 130 m) from Feb 21-Feb 23 (dust episode B) and March 20-22 (dust episode C), respectively.The symbol "$" indicates that the dust phenomena was observed at the SYNOP report site.The HYSPLIT back trajectory path starting from Fukuoka

Fig
colored circle).The observed SIA concentration explained 60% of the PM 2.5 concentration (SIA_obs = 0.60PM 2.5 _obs + 0.44, unit = µg m -3 , r = 0.92), which is reasonable because the observed PM 2.5 included BC, OC, sea salt, and other trace metal components.The counterpart cation of NO 3 -and SO 4 2-is NH 4 + , and NO 3 and SO 4 2-are considered major mass components of PM 2.5 .

Fig. 4 .
Fig. 4. Labeling is the same as Fig. 3, but for dust event C and the back-trajectory starting from Fukuoka (z = 500 m) on March 22 0900UTC.

Fig. 7 .
Fig. 7. Time-longitude variation of daily averaged (a) fSO 4 2-, (b) fNO 3 -, and (c) total dust-nitrate from GEOS-Chem model simulations at 33.5°N (see Fig. 3(a) for the location of this section).Lines in the figure represent the concentration variation at Fukuoka taken from Fig. 2.

Fig. 8 .
Fig. 8. Horizontal distribution (at z = 65 m level) of 6-month averages for (a) total dust and extraction path lines, (b) total dust-NO 3 -, (c) ratio of total dust-NO 3 -/total dust, and (d) total dust-nitrate/fine NO 3 -, (e) and (f) concentration and ratio changes along the north and south paths shown in (a).