Evaluation of Ionic Pollutants of Acid Fog and Rain Using a Factor Analysis and Back Trajectories

Transcription

1 ANALYTICAL SCIENCES JANUARY 2001, VOL The Japan Society for Analytical Chemistry 71 Evaluation of Ionic Pollutants of Acid Fog and Rain Using a Factor Analysis and Back Trajectories Tetsuya ADZUHATA,* Junko INOTSUME,* Tomoko OKAMURA,* Ryoei KIKUCHI,* Toru OZEKI,** Masahiro KAJIKAWA,* and Nobuaki OGAWA* *Faculty of Engineering and Resource Science, Akita University, 1-1, Tegata Gakuen-cho, Akita , Japan **Hyogo University of Teacher Education, Yashiro-cho, Kato-gun, Hyogo , Japan. Fog and rain water samples were collected at the same time in the Akita Hachimantai mountain range in northern Japan from June to September in 1998 and The various ion concentrations in these samples were analyzed, and the fog droplet sizes were measured for each fog event. As the fog droplet size increased, the ion concentration decreased. The slope of log log plots of the concentration versus the droplet size differed with the kind of ion. In order to characterize the air pollutant, moreover, these data were quantitatively analyzed by an oblique rotational factor analysis. We found that three factors were extracted as the air pollutant source: (NH 4) 2SO 4, acids (HNO 3 + H 2SO 4) and sea-salt. Combining the factor analysis with the 72 h back-trajectory at 850 hpa level, we found that the contribution of each factor varied with the transport pattern of air masses. (Received September 26, 2000; Accepted October 16, 2000) Recently, acid precipitation is receiving global interest because it affects terrestrial and aquatic ecosystems. 1,2 Our research group has studied 3 7 the acid precipitation of Hyogo and Akita prefectures in Japan, combining a chemical analysis of the ionic substances with meteorological situations, and has analyzed pollutants by a factor analysis. 3 7 Furthermore, investigations for the fog on a mountain ridge showed by a factor analysis that a fog droplet had some soluble pollutants (e.g. (NH 4) 2SO 4, seasalt, H 2SO 4 and HNO 3), and that the uptake mechanism by a fog droplet would be different for each pollutant. It is well-known that fog/cloud water is more acidic and has a higher concentration of pollutants than rain water. 5,9 11 Nevertheless, prior attempts 15,16 to elucidate the mechanisms of fog acidification, especially the behavior of the ionic substances in the fog droplets, have not been quantitatively conclusive. There has been no investigation, moreover, concerning where the air pollutants specified by employing the oblique rotational factor analysis 3,4,6,12 are transferred. Our purpose here was threefold. First, by using an oblique rotational factor analysis, we examined the observed drop size dependence of the chemical composition during fog events at Akita Hachimantai mountain range from June to September of 1998 and Second, we evaluated the transport of some air pollutants in combination with ionic substances which were quantitatively extracted from a factor analysis developed by our research group, 3,4 with a 72 h back trajectory at the 850 hpa level and from the point of view of a synoptic weather system. Third, from the feature of chemical composition of fog or rain, which was non-ion-balanced, we tried to estimate unknown soluble chemical species. Experimental Fog water was collected on a mountainside (39 56 N, E, 1465 m, a.s.l.) of Mt. Mokkodake (1578 m, a.s.l.) in the Akita Hachimantai mountain range from June to September of 1998 and 1999 (Fig. 1). A passive fog sampler with a cylindrical To whom correspondence should be addressed. azuhata@z2.zzz.or.jp Fig. 1 Sampling sites in northern Japan.

2 72 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17 Table 1 Arithmetic mean values of the concentration of various ions in fog and rain water Type Site Year n [H + ]/µeq L 1 [Na + ] [Cl 2 (ph) ] [nss-so 4 ] [NO 3 ] [NH 4+ ] All ± 49.0 (4.45 ± 0.14) 17.2 ± ± ± ± ± Fog Mokkodake ± 56.3 (4.38 ± 0.13) 14.4 ± ± ± ± ± ± 26.9 (4.64 ± 0.07) 22.6 ± ± ± ± ± All ± 8.4 (5.16 ±0.14) 4.7 ± ± ± ± ± 30.7 Mokkodake ± 4.0 (5.26 ± 0.05) 5.1 ± ± ± ± ± ± 8.4 (5.12 ± ± ± ± ± ± 30.7 Rain Laboratory ± 4.9 (5.20 ± 0.11) 4.1 ± ± ± ± ± 17.4 All ± 8.6 (5.14 ± 0.09) 6.3 ± ± ± ± ± 29.0 Onumaso Inn ± 9.5 (5.12 ± 0.10) 7.6 ± ± ± ± ± ± 8.1 (5.16 ± 0.09) 4.9 ± ± ± ± ± 33.9 The numbers of data used for the factor analysis were 59 in the total samples of 95 and 45 in 135 rain samples, respectively. Fig. 2 Log log plots of concentrations of various ion ([ion]) vs. mean volume diameter ( D ) for all fog samples. The ion-balanced samples are shown by the marker ( ), with the cation excess ( ), and the anion excess ( ). Solid lines show the fitted curve for ion-balanced ( ) fog samples. The broken lines show a slope (b) of negative three, which indicates a simple attenuation mechanism in fog droplet growth. Teflon wire screen (Model FWP-500, Usui Kogyo Kenkyusho Inc.) was placed there during fog events. Rain water was also collected by a bulk sampler at the same sampling site during the same periods as fog occurrence. Moreover, two locations at different heights, Onumaso Inn (39 59 N, E, 960 m, a.s.l) and Akita University Snow Science Laboratory (39 58 N, E, 1200 m, a.s.l.), were added for rain sampling. The concentrations of various ions (H +, Cl 2, SO 4, NO 3, Na +, NH 4+, K +, Ca 2+, Mg 2+ ) in the fog and rain samples were analyzed using a ph meter and an ion chromatography system 5 8 after filtering each sample through a 0.45-µm membrane filter. The sizes of the fog droplets were estimated by the impaction method of drops on cedar oil-coated glass slides. 13 Furthermore, 72 h back trajectories 20 on a weather chart at 850 hpa level were traced every 12 h in order to determine the transport pattern of air masses. To obtain some factors which characterize several chemical components as the air pollutants in fog water, an oblique rotational factor analysis was used on a matrix made from the chemical data, which meant the absolute equivalent of ions in cloud droplets (the product of the measured concentration and the droplet volume calculated from the mean volume diameter, D), because the absolute amount of pollutants in the droplet contributes to the scavenging mechanism. In the factor analysis

3 ANALYTICAL SCIENCES JANUARY 2001, VOL Fig. 3 (a) Ion composition of fog water of each factor (A, B, C) obtained by the results of factor analysis; (b) the contribution of the factors of fog event; (c) re-sorted (b) in the order of mean volume diameter of fog droplets. Horizontal broken lines (b) show the monthly mean contribution. for the rain water, however, the matrix was made by weighting the precipitation amount to every ion concentration. Only the chemical data with an ion-balance (= [cation]/ [anion]) from 0.83 (1/1.2) to 1.20 were selected. The numbers of data used for the factor analysis were 59 in the total fog samples of 95 and 45 in 135 rain samples, respectively. Results and Discussion Ion concentration and ph in fog and rain water Table 1 represents the arithmetic mean concentrations of the major ion species in fog and rain water (ion-balance: ). A key distinguishing feature of fog water is its considerably higher acidity than the rain water collected at the same site. The dominant ion species of fog and rain water were of three species: non-sea-salt 19 (nss-) SO 2 4, NO 3 and NH 4+, except for the composition of sea salts. In the fog samples, especially, both [nss-so 2 4 ] and [NH 4+ ] were more than half of the total ion concentration. This is caused by an influence of (NH 4) 2SO 4 as the cloud condensation nuclei (CCN) upon the occurrence of fog droplets. When the concentrations of [nss- SO 2 4 ] and [NO 3 ] were high, the ph of sample was low. The rain water at Onumaso Inn had a higher concentration of ions than at other sites. It can be considered that the rain at Onumaso Inn suffered an effect of subcloud scavenging, while the rain at Mt. Mokkodake and Snow Science Lab. collected in clouds and near the cloud base. In comparison with the rain in 1999, the rain in 1998 had low acidity but high concentrations of ions. This variance is explained in the following section. Relationship between the concentrations of various ions and fog droplet size Figure 2 shows log log plots of the concentration of various ions ([ion]) versus mean volume diameter ( D). The fitted curves are for fog samples which were ion-balanced. The concentration of each ion decreases with the droplet size, and the curves have different slopes. If a CCN is diluted only by ambient water vapor in the growth process of droplets, this slope would be negative three. When it becomes more than negative three, it should be considered that the air pollutants were uptaken to the fog droplets after the activation of CCN. 2 Comparing these slopes, nss-so 4 (Fig. 2f) and NH 4+ (Fig. 2e) have a smaller slope than other species, and Na + (Fig. 2b) and Cl (Fig. 2c) have a larger slope than nss-so 2 4 and NH 4+. It suggests that even after the activation of CCN, more sea-salt as interstitial aerosol are uptaken by fog droplets than (NH 4) 2SO 4, relatively. The slope of NO 3 (Fig. 2d) is also strongly related to the growth process of fog droplets after the activation of the component of CCN. There were many fog samples of cation excess in the range of larger drop sizes, particularly for all ions showing a lower concentration of ions than the fitted curve, except for the case of NH 4+. In other words, the fog water had a tendency to exceed the cation when the total concentration was low and the ph was high. On the other hand, as the samples of the cation excess in the case of NH 4+ show a higher concentration than its fitted

4 74 ANALYTICAL SCIENCES JANUARY 2001, VOL. 17 Fig. 4 (a) Ion composition of rain water of each factor (A, B, C) obtained by the results of a factor analysis; (b) the contribution of the factors of rain event. The horizontal broken lines (b) show the monthly mean contribution. Table 2 Contributions of each factor for fog water in various groups of synoptic weather system Synoptic weather system n( ) Factor A, % Factor B, % Factor C, % Total, % Stationary front Early summmer 22( ) 64.1 ± ± ± autumn 11( ) 37.0 ± ± ± Low on sea of Japan 16 (8 + 8) 56.0 ± ± ± High on the Japan islands 7 (7 + 0) 38.7 ± ± ± curve. That is to say, these fog water samples should be neutralized by ammonia, and then the ph of the fog would increase. Therefore, these higher ph samples would contain a hydrogencarbonate (and/or an organic acid anion) because of an insufficiency of the anion. For samples of anion excess, there were no evident characteristics. However, the fog samples in excess of anion might contain some heavy metal ions, such as iron and so on. 17 Oblique rotational factor analysis A data matrix of fog for a factor analysis was prepared from samples from both 1998 and Figure 3 shows the result of an oblique rotational factor analysis 8 which evaluated the three factors in ion-balanced fog water and comprised 86.4% of the entire data. The pollutants which had the highest contribution (Factor A: 58.8%) consisted of (NH 4) 2SO 4 (Fig. 3a). There was a tendency toward a decreased contribution for each fog sample in the autumn (Fig. 3b). Factor B (18.1%), which had a composition containing such acids as H 2SO 4 or HNO 3 with ammonium, tended to have a higher contribution in the summer. It appears that this factor is air pollutants affecting the acidification of fog water, since it has high [H + ]. The third factor (Factor C: 9.5%) was mainly seasalt. Its composition was of Cl, Na +, K + and Mg 2+, and it had a tendency toward an increased contribution in autumn. Since the soluble salts accounting for these extracted components are well-known as CCN, 18 our analysis shows that the pollutants in fog water are quantitatively separated into some appropriate factors. Upon re-sorting these contributions of each sample in order of drop size (Fig. 3c), Factor A showed a tendency to be the highest contribution in the range of smaller sizes. On the other hand, Factor B had two peaks in both smaller and larger drop sizes, and Factor C showed a trend that the contribution increased with the droplet size. These results support the suggestion by Pruppacher and Klett (1997) 18 that smaller CCN (e.g. ammonium sulfate) form smaller droplets and larger ones, such as sea-salt, form larger droplets. Rain samples collected at three sites (Mt. Mokkodake, Snow Science Lab., and Onuma Inn) in the Hachimantai mountain range were also analyzed by our factor analysis as well as fog water. Two years of rain samples were combined in order to clarify the results. Three factors specified by the factor analysis have common compositions for the results of all fog samples (Fig. 4a), though the contribution of sea-salt (Factor B) was larger than acids (Factor C), compared with the fog results. Moreover, Factor C consisting of acids mainly contained more NO 3 than the fog water. These results appear to be influenced by the uptake rate by rain or fog drop being different from each ion. The ion balances of the factor concerning the acid pollutants (fog: Factor B, rain: Factor C) were evaluated to be lower (0.83, 0.75, respectively). These results suggest an insufficiency of heavy metal ions which could not be measured in this study. From the point of view of the sampling season (Fig. 4b for rain), there was a tendency for the factors to have a similar spectrum to the results of fog water, as shown in Fig. 3b (by a comparison between Factor A of fog and Factor A of rain, Factor B of fog and Factor C of rain, and Factor C of fog and Factor B of rain, respectively). Hence, it seems that the composition of rain water reflected most of the fog s result. In other words, it seems that few air pollutants were scavenged by rain drops present in the sampling area. In addition, there were

5 ANALYTICAL SCIENCES JANUARY 2001, VOL Fig. 5 Comparison of the 72 h back trajectory at the 850 hpa level with the results of the factor analysis for fog events under a westerly wind; trajectories with higher and lower contributions for the Factor A (a-1,2), for Factor B (b-1,2), and for Factor C (c-1,2). Thick solid lines: trajectories with higher contributions for each factor. The dotted lines are trajectories with lower contributions for each factor. no differences for contributions during the same time and each rain event between sites. Relationship between the factor analysis and synoptic weather systems Table 2 presents the average contributions of each factor for fog in various groups of synoptic weather systems. These contributions were calculated on the assumption that the sum of three factors can account for the entire composition of a fog sample, and thereby the total contribution is 86.4%. The contribution of Factor A has a tendency to increase under the weather system of a stationary front type (Bain front type) in early summer. It seems reasonable that a westerly wind was prevailing based on the intensification of the front; thus, air pollutants with rich (NH 4) 2SO 4 were present. For Factors B and C, the contributions rise in the types of stationary front in the autumn and high pressure on the Japan islands. Since in these types the wind direction of the sampling sites vary frequently, it can be considered that the air pollutants arrived from various directions. Relationship between the factor analysis for fog water and the 72 h back trajectory Figure 5 shows that 72 h back trajectories corresponding to the fog samples had higher and lower contributions (approximately a fourth of the total ion-balanced samples, respectively) in the factor analysis. Factor A (1998), consisting of (NH 4) 2SO 4, had a strong contribution in fog that occurred in the summer, and the pollutants were transferred from the area of northeastern China (Fig. 5a-1). In contrast with Factor A, the contribution of Factor C increased in the autumn and in a large number of trajectories which passed along the sea of Japan in There was no obvious relationship between the results of the trajectory analysis and the contribution of the factor analysis in 1999 in comparison with the results of The reason for this can be thought of as the result of differences in the synoptic weather system, which can be characterized by the absence of prevailing westerly winds in Thus, these results show that the course of transport of the air pollutants can distinguish the effective CCN into two categories, namely ammonium sulfate and sea salt. Considering the trend of Factor B (Fig. 5b-1), the trajectories from the area of northeastern China had a higher contribution, while there were also higher contributions for the trajectories passing over the Japan islands from the south. It should be considered that the air pollutants had been uptaken into the fog droplets by not only the long-range transport from northeastern China, but also short-range transport over the Japan islands or the sea of Japan. Conclusions An analysis of fog observed in the Akita Hachimantai mountain range from June to September of 1998 and 1999 was performed using the oblique rotational factor analysis and the 72 h back trajectory method. The principal results are as follows: (1) Fog water was remarkably acidified and their pollutants were more concentrated compared with rain water. The

6 76 ANALYTICAL SCIENCES JANUARY 2001, VOL common dominant ions in fog and rain water were nss-so 4, NO 3 and NH 4+, except for the composition of sea salts. (2) Comparing each slope of the fitted curves for log log plots of the concentration versus the mean volume diameter of fog 2 droplets, nss-so 4 and NH 4+ have smaller slopes than other ionic species, and Na + and Cl have larger ones. This suggests that much sea-salt as interstitial aerosol is relatively uptaken by droplets, even after the activation of CCN than (NH 4) 2SO 4 is activated. (3) The pollutants extracted from the factor analysis were the well-known CCN. In particular, the factor for (NH 3) 2SO 4 had the highest contribution under a westerly wind system. Moreover, the results of this factor analysis were consistent with that the smaller CCN (e.g. H 2SO 4 mist, (NH 4) 2SO 4) forms smaller droplets and the larger ones such as sea-salt form larger droplets. (4) Combining the factor analysis with the 72 h back trajectory, we found that the factor composed of (NH 4) 2SO 4 and acids had increased the contribution in early summer and mid summer under a westerly wind system, in which pollutants were transferred from the area of northeastern China. These considerations invite a further investigation using a refinement of our current method or alternative methods, synthetically. Acknowledgements We thank Miss Kumi Nakagawara for her association in sampling the rain water at Akita University and analyzing the back trajectories. References 1. J. M. Seinfeld and S. N. Pandis, Atmospheric Chemistry and Physics, 1998, Wiley, New York, A. Singer, Water, Air, Soil Pollut., 1999, 109, T. Ozeki, K. Koike, and T. Kimoto, Environ. Sci. Technol., 1995, 29, T. Ozeki, K. Koike, N. Ogawa, T. Adzuhata, M. Kajikawa, and T. Kimoto, Anal. Sci., 1997, 13, N. Ogawa, R. Kikuchi, H. Goto, M. Kajikawa, and T. Ozeki, Bunseki Kagaku, 1998, 47(8), N. Ogawa, R. Kikuchi, T. Okamura, T. Adzuhata, M. Kajikawa, and T. Ozeki, Atmos. Res., 1999, 51, N. Ogawa, R. Kikuch, T. Okamura, M. Kajikawa, T. Adzuhata, Y. Iwata, and T. Ozeki, Intern. J. Mater. Engin. Resour., 1999, 7, N. Ogawa, R. Kikuchi, T. Okamura, J. Inotsume, T. Adzuhata, T. Ozeki, and M. Kajikawa, Atmos. Res., 2000, 54, J. M. Waldman, J. W. Munger, D. J. Jacob, R. C. Flagan, J. J. Morgan, and M. R. Hoffman, Science [Washington, D.C.], 1982, 218, T. Hosono, H. Okochi, and M. Igawa, Bull. Chem. Soc. Jpn., 1994, 67, K. J. Hoag, J. L. Collet Jr., and S. N. Pandis, Atmos. Environ., 1999, 33, T. Adzuhata, J. Inotsume, T. Okamura, R. Kikuchi, T. Ozeki, M. Kajikawa, and N. Ogawa, Water, Air, Soil Pollut., 2000, in press. 13. T. Okita, Tellus, 1961, 13, N. Ogawa, T. Adzuhata, and M. Kajikawa, Seppyo, 1998, 60, J. L. Collett Jr., A. Bator, X. Rao, and B. Demoz, Geophys. Res. Lett., 1994, 21, S. N. Pandis and J. H. Seinfeld, Atmos. Environ., 1990, 24A(7), J. W. Munger, D. J. Jacob, J. M. Waldman, and M. R. Hoffmann, J. Geophis. Res., 1983, 88(C9), H. R. Pruppacher and J. D. Klett, Microphysics of Clouds and Precipitation, 1997, Chap. 17, Kluwer Academic Publishers. 19. R. F. Mc Allister and E. F. Corcoran, Chemical Oceanography, in Hand Book of Ocean and Underwater Engineering, ed. J. J. Myers, 1969, McGraw-Hill, New York, I S. Petterssen, Weather Analysis and Forecasting, 2nd ed., 1956, Vol. 1, Chap. 2, McGraw-Hill.