Sulphur dioxide precipitation scavenging

Sulphur dioxide precipitation scavenging

Atmospherrc Enoironmenf Vol. 17, No. 4. pp. 797-805, Oc+4981,83mO797 1983 SULPHUR DIOXIDE -09 $03.00,0 c: 1983 PergamonPress Ltd Pnntedm Gr...

971KB Sizes 0 Downloads 47 Views




17, No. 4. pp. 797-805,







c: 1983 PergamonPress Ltd

Pnntedm Great Britam.





School of Environmental Sciences, University of East Anglia, Norwich, U.K (First received 1 June 1982 and

for publication 9


September 1982)

Attract-Sulphur dioxide washout ratios (rainborne/airborne concentrations) calculated from hourly observations in a heavily industrialized area show a reasonable agreement with predicted values (Bake, 1981, Atmospheric Enoironment 15, 3141) although they are also Iower. Because the equilibrium scavenging assumption may not be valid in a heavily industrialized area, an attempt was made to assess the effect of the atmospheric sulphur dioxide vertical gradient by measuring airborne and rainborne sulphur dioxide at two sites in a steep valley, 100 m vertical distance apart. Although the correlation between the observed washout ratios for the air layer and the predicted ratios was good, the observed ratios were lower (arithmetic mean of the observed washout ratios was about half that of the predicted ratios, although the median values were much closer) and did not appear to be related to vertical atmospheric sulphur dioxide gradients. Because the rainborne and airborne SO2 levels were measured at the entrance and exit of the air layer, it was possible to calculate scavenging coefficients. Sulphur dioxide washout was also studied throughout pr~ipitation events in a rural area with sampling durations from one minute to several hours, Generalization of the fine structure of sulphur dioxide during events is difficult, but the observed washout ratios are still less than those predicted by the adopted model.

INTBODUCTION Dissolved SO2 in rainfall was first measured about ten years ago by Hales et al. (1971) under a power station plume in the U.S.A., and by Davies (1976) in a heavily industrialized area in the U.K. Rainborne SO* levels in regional precipitation were reported by Hales and Dana (1979) and Dana (1980) in the north and east U.S.A. Davies (1979) determined dissolved SOZ and total sulphate in precipitation collected at an urban and rural station in eastern England when the “same” rainfall was collected at each site. In the meantime, theoretical work (e.g. Hales and Sutter, 1973; Barrie, 1978, 1981; Hales and Dana, 1979) predicted the concentrations of rainborne SOZ to be expected, and the observed values of Hales and Dana (1979), Davies (1979) and Dana (1980) broadly agree with the predictions. Davies found that the average contribution of rainborne SO2 to total precipitation sulphur over a period of one year in rural eastern England was 14 7;; in a small city (150,~ population) with little industry, the contribution was 22 7;. In a heavily industrialized area, analysis of a small number of rainfalls indicated that the comparable percentage was around 47 “/ (Davies, 1976). In each of the three locations, however, the range of the contributions on individual occasions was large. Dana (1980) reported similar results in the U.S.A. Davies (1976) attempted to relate hour-to-hour dissolved SO2 levels in pr~ipitation to atmospheric SO2 concentrations and various weather parameters but the continuous and varying emission of SO1 into the atmosphere proved a considerable difficulty. Study of regional rainfall (Davies, 1979) indicated that shortterm variations may be better understood at locations remote from large sources of S02. Here, an attempt is made to relate field observations

to theoretical estimates of SO2 removal by precipitation. In industrial areas with large, continuous emissions of SOZ, there are problems in determining a repr~en~tive atmospheric SO2 concentration which characterizes the conditions pertaining at the time of precipitation. In an attempt to mitigate this problem, rainfall was sampled at the bottom of a heavily industrialized valley, and at the top of the valley-side. Martin (1982) has examined precipitation composition in valley bottoms and adjacent hilltops in mid-Wales. He found that, on average, 15 “/;, more hydrogen ion was deposited in a deep narrow valley than on the adjacent hillside, 250 m above and 600 m away. Martin believed this was associated with occasions when sulphur and nitrogen compounds were held below the valley tops until washout by precipitation. Likens et a/. (1977) also reported significantly higher concentrations of sulphate and nitrate in rainwater from the lower of two locations at 252 and 610m elevation in the Hubbard Brook (USA.). In this study, rainborne SOP content of individ~l pr~ipitation events are examined alongside atmospheric SO, concentrations, in the valley-bottom and atop the valley-side, measured on an hourly basis. The finer structure of SOZ removal by precipitation is difficult to determine in an industrial location, because of the relatively large and varying sources of atmospheric SO,. Consequently, this finer structure is examine by measuring SO2 removal and atmospheric SO2 concentration throughout individual events at a location in rural eastern England. SAMPLING AND


The rainfall examined in the industrial city of Sheffield was collected from a site within the heavily 797



industrialized Don Valley which is dominated by the iron and steel industry, although no large buildings lay within 350 m of the sampler. The rainfall was collected on an hourly basis by an automatic sampler; details of the sampler and the site are given in Davies (1976). Twenty-six individ~i rainfall events in Sheffield were sampled at two sites over a 9-month period. One site was in the location described above, at an altitude of 30m, whilst the other was in an area of residential housing overlooking the same valley at an altitude of 130m and about 900m horizontal distance from the lower site. At the upper site the precipitation was collected in amber-glass bottles which held PTFEcoated funnels. The funnel was sluiced with doubledistilled water just before precipitation collection, and the collecting funnel of the valley-bottom samples was automatically sluiced (Davies, 1974). The bottles in the automatic samples and the bottles used at the upper site contained sodium tetrachloromercurate solution (TCM) and corrections for the direct (dry) impingement contribution of atmospheric SOZ were made by the method described in Davies (1976) for the lower site and by using bottles protected from pr~ipitation at the upper site. Atmospheric SOZ concentrations at the two sites in Sheffield were measured automatically on an hourly basis. At the lower site, the instrument was a Nashpattern recorder (Gas Chromatography Ltd.); the accuracy of the recorder is about lO”i, with decreasing accuracy in the lower con~ntration ranges ( < 3Opg m-“), and a detection limit of _ 2Opg m-3. At the upper site a Wljsthoff U3S model was used with slightly better accuracy, especially at lower concentrations, and a detection limit of _ 10pgmv3. The rainfall for the intensive studies in eastern England was collected at a rural station in Norfolk (11 km to the west of Norwich) which represented a “regional” rainfall area (Davies, 1979). Rainborne SO1 Ievels, rainfall sulphate and pH were measured during sampling periods which varied from 1 min to several hours. The period of sampling depended on precipitation intensity; with the smaller sampling periods a large PTFE-coated funnel (40cm diameter) was used. Rainborne SO, was measured by the TCM method (West and Gaeke, 1956) within 48 h of collection. The sensitivity of the method is of the order of 0.005mg (S)/-‘. Where rainfatl sulphate was measured, the pr~ipitation was collected in aseparate bottle and then stored in specimen tubes in a refrigerator before analysis by the Persson (1966) method. Sensitivity of this method is of the order of O.O3mg(S)tC-‘. During the intensive studies in rural Norfolk, atmospheric SOZ levels were determined on a 15min basis by bubbling air through TCM solution at around 301’min- ’ (sensitivity of analysis given above). The atmospheric sampling period during these intensive studies was not the same as the precipitation sampling period, because of the impossibility of conducting a large number of operations simultaneously.


Chamberlain (1960) calculated washout coefficients of SGZ by using Frosting’s equation to calculate the transfer of SO2 to droplets, assuming irreversible absorption. This parameter is more normally called the scavenging coefficient, R (Barrie, 1981) and may be cafcutated from field data by employing the expression: h






where S is the concentration of rainborne SO,; C is the atmospheric SO2 concentration; I is the rainfall rate and h is the depth through which the rain falls. Chamberlain calculated L values (proportion of SO2 removed s-l) to increase from low4 to 3 x 10-4s-’ as the rainfall rate increased from 1.0 to 1Omm h- ‘. More recent theoretical work, allowing for the acidity of the rain, and reversible washout where appropriate (e.g. Billingsley et al., 1976; Hales and Dana, 1979) show that the washout coefficient is an order of magnitude less than Cham~rlain’s estimates. If SO, washout is the only sink, and there are no sources operating, then during the rain event 1 = _ din(C) dt’


Maul (1978), using hourly measurements of atmospheric SOz. and rainfall in the English East Midlands (about 1OOkm to the west of the rural site in Norfolk considered here), made estimates of the apparent 1 values for SO*. He found the results were consistent with 1 = al x 10-5s-’ (a is in the region of 3 if the rainfall rate is measured in mm h- I). Assumptions about the vertical distribution of atmospheric SO2 and the choice of a suitable h value in (1) and assumption about sinks and sources of SO2 in (2) limit the usefulness of 1 for field studies. More appropriate for use with field observations is the dimensionless washout ratio Wso, (Shnn et al., 1978). S (3) wso, = T’ The scavenging coefficient, & is related to Wso, by the following equation: i_!$!..


Barrie (1981) has shown that, under equilibrium conditions: logt,wSo, = logro(K,&,)+~H


where K, and K, are equiIibrium constants and Kr I(, = 6.22 x lOmEexp (4755.517)


over the temperature (r) range 273.1-303.1 K. . . . Barrie indicates that Wso, values in the atmosphere are likely to fall in the range 103-106.

Sulphur dioxide precipitation scavenging


It is possible to use data on rainborne SO2 levels and mean daily atmosphere SO2 concentrations from Davies (1979), along with observations of cloud height from synoptic charts, to arrive at a similar range of 1 values to those apparent washout coefficients calculated by Maul (1978) on the basis of hourly measurements of atmospheric SO2 and rainfall amount. However, because little is known about cloud thickness or cloud SO1 content, there are considerable problems in attempting to define 2 values from measurements at the ground. Using paired rainfallevents for rural Norfolk and the small city of Norwich (Davies, 1979) to calculate observed washout ratios (W, = S/C) and air temperature and rainfall pH to determine the theoretical washout ratio W,, from Equations (5) and (6), gives the results in Table 1. The scatter of W, and W, was large and so the median values give a better idea of the “typical” W, and W, values. Most of the extremely high values of W, are associated with low values of W,, yielding no statistically significant correlation between the two, either for city rainfall or for rural rainfall (although the correlation coefficient for the rural site is higher). These poor relationships between W, and W, reflect the difficulties of using mean daily atmospheric SO2 levels for the C value in Equation (3), a problem which is seen to be exacerbated in the city since the SO2 concentration within an urban airshed is more variable than in “regional” air. So, the use of relatively longperiod mean atmospheric SOZ values precludes detailed studies of washout ratios in Norfolk, especially when the complications of different air masses associated with frontal rain are considered. Atmospheric SO2 monitoring on a time-scale more pertinent to the precipitation event is necessary. The measurement of atmospheric and rainborne SO* at the entrance and exit zones of a “layer” of the atmosphere would enable meaningful estimates of 1 to be made. In addition, it may be possible to test the assumption that equilibrium scavenging prevails in the industrial city. Hales (1978) suggests that this condition should prevail about 20 stack-heights downwind of an SO2 plume from a tall stack (in other words, non-equilibrium scavenging may be expected to occur under the condition of large vertical gradients in atmospheric SOZ). SO2 REMOVED BY PRECIPITATION

(i) Washout in an industrial city (Shefield) Rainfall was collected on an hourly basis during every event for the observation period September 1969-August 1970 in industrial Sheffield (Davies, 1976). The overall precipitation-weighted rainborne SO1 concentration was 0.84 mg (S) / - I. Atmospheric SO2 concentrations at the same site were also determined on an hourly basis, although pH measurements were not made as frequently. Therefore, it was not possible to calculate W,, but the distribution of the



(ii) SO1 removed industrial



,.‘, 5





Fig. 1. RelationshIp between observed washout ratio and air temperature in a heavily industrialized area (Sheffield). Precipitation was collected hourly over a period of one year. The solid line is the regression line (r = 0.2). The dotted line is Barrie’s (1981) theoretical curve for a pH of 4.5 (the mean precipitation-amount weighted pH of Sheffield precipitation during the observation period).

observed washout ratios (W,) with air temperature on an hourly basis is shown in Fig. 1. Although there is considerable scatter, the regression line (r = 0.2, significant at 1 ‘;,, level) shows a quite close relationship with W, calculated by Barrie (1981). The mean (precipitation-amount weighted) pH of Sheffield rainfall during the observation period (pH measured from samples which were kept in a refrigerator and bulked to provide a “weekly” value) was 4.55, and the equivalent curve from Barrie (1981) for that pH is also shown in Fig. 1. The agreement is relatively close because of the strong dependence of W, on pH [Equations (5) and (6)] with a ten-fold increase in W, for unit increase in pH. There are, of course, problems in examining precipitation scavenging in a heavily industrialized area. The hourly mean atmospheric SO2 values fluctuate considerably because of the large sources (Davies, 1976) and precipitation does not commence and terminate exactly on the hour. Whilst precipitation is removing SO*, because of large emission rates, atmospheric SO2 concentrations can increase during rainfalls. In the previous Sheffield study, out of the 206 individual rainfalls which were sampled on an hourly basis, only on 54”. of the occasions did the atmospheric SO1 concentration in the hour before rainfall exceed the concentration in the last hour of rainfall (Davies, 1976). However, whilst recognizing these problems, study of SOZ washout through a layer of air will enable estimates to be made of the scavenging coefficient (2) and could provide a more realistic C-value than a single (valley bottom) SO, concentration; besides permitting an examination of the equilibrium scavenging assumption.

by precipitafion

at two heights

in an


Table 2 shows the results relating to the 26 individual precipitation events where SO2 content of rainwater and the air was measured at two heights in the industrialized Don Valley in Sheffield. The assumption is that these data give a reasonable picture of conditions through a 100-m depth of air. There are difficulties associated with this assumption; on three occasions the average atmospheric SO, concentration throughout the precipitation event at the upper station was greater than, or equal to, that at the valley-bottom station. No obvious meteorological reasons were apparent. The differences between the atmospheric SO2 concentrations at the two sites were small in these three cases (0-10ngm-3) and it is assumed that they were due to measurement error. The observations were used to calculate the washout ratio from the difference between the rainborne SOZ concentrations at the two sites and the difference between the average atmospheric SOZ concentrations (throughout the rainfall) at each site, i.e. (S, - S,)/(C, -C,) (refer to Table 2). The washout ratio, thus calculated, pertains strictly to the 100-m layer of air between the two sites. It assumes that at the top of the 100-m layer, a raindrop which contains no rainborne SO, is surrounded by air with zero SOZ concentration. It should be pointed out that strictly, by definition, this is not a washout ratio, although it clearly has physical significance. Thus, although it may be regarded as a washout ratio in this study, to avoid confusion, it shall be referred to as a layer-washout ratio (W,). It is not possible to calculate W, for the three cases referred to in the previous paragraph, where C, 3 C,. There is no alternative but to ignore these three cases, although it may be argued that this is introducing bias into the calculations. There is no statistically significant relationship evident between W, and temperature, but W, and pH show the expected positive relationship (Fig. 2) with a correlation coefficient of 0.67-for log (W,) and pH (significant at the 0.1 7; level). The relationship is similar to that predicted by Barrie (1981) who pointed out that the washout ratio will change much more over the range of pH values encountered in rainfall than it would over the range of temperatures encountered in the atmosphere. Figure 3 shows the relationship between W, and W,. The correlation coefficient between log (W,) and log (IV,) is 0.76 (significant at 0.1 “/level). W, is less than Wr in all but six of the plotted 23 cases. Previously, it was suggested that apparently depressed washout ratio values may be caused by overestimates of the atmospheric SO2 concentration (C) because of continuous emission of SO2 within the vicinity of the rainfall sampler. This is less likely to be a pertinent factor here because the difference between two atmospheric levels is incorporated into the calculation. This is confirmed, to some extent, by the behaviour of the values of the

0.32 2.0 0.27 2.37 0.07 0.04 1.28 0.2 0.57 0.14 0.40 0.49 0.46 0.39 0.37 0.21 0.04 1.12 0.49 1.58 0.38 1.21 1.34 I .98

4 Nov 9 Nov 9 Nov 16 Nov 20 Nov 27 Nov IDec 12 Dee 15 Dee 20 Dee 8 Jan 11 Jan 18 Jan 9 Feb 21 Mar IO Apr 25 Apr 27 Jun 8 Jul 19 Jul 24 Jul 21 Jul 16 Aug 20 Aug

Mean Standard deviation

0.2 1.3

19 act 24 Ott

Rainfall intensity (mm h-‘)

Table 2. Rainborne

I 8 4 13 2 3 4 13 8 9 7 I 7 7 5 4 2 I 2 4 16 4 7 IO

15 3


Duration of event

6.5 3.0 3.0 3.0 5.0 5.5 2.5 5.0 5.0 1.5 0.5 5.5 5.0 0 8.5 2.0 7.0 19.0 22.0 15.0 14.0 15.0 11.0 13.0

13.0 5.0

0.78 0.51

0.36 1.12 1.68 0.64 0.80 0.64 0.12 1.60 1.44 1.60 1.20 1.20 0.12 0.56 2.0 0.56 0.16 0.32 0.28 0.48 0.32 0.16 0.24 0.80

0.12 0.40

top (S,) mge-’


SO, before, during

Average air temperature throughout event (“C)

SO,; atmospheric


1.60 1.38

0.41 2.10 6.50 1.60 4.00 1.70 1.24 2.00 1.94 I.75 2.81 1.90 0.94 3.20 2.40 0.63 0.30 0.34 0.56 1.00 0.75 0.80 0.88 0.88

0.28 0.66


I60 70 80 70 60 110 180 170 150 370 90 210 170 110 160 II5 140 110 120 40 50 35 40 50

70 120


215 130 145 100 135 150 215 370 150 430 165 285 175 190 165 195 165 90 220 45 15 40 50 55

130 125


115 70 250 80 125 85 120 150 180 240 140 340 160 285 165 165 170 180 160 80 I80 30 55 65 40 50

50 80 120 60 60 55 40 60 140 130 140 300 80 170 150 70 140 100 110 90 80 35 45 30 35 50

50 20 220 25 105 10 110 150 110 0 130 50 150 280 155 110 170 I65 155 70 145 20 30 40 30 50

40 50 80 40 50 40 30 40 70 80 100 120 70 90 130 50 110 70 80 70 70 10 35 IS 10 40


bottom Top (CL

Average atmospheric SO1 concentration throughout the rainfall event pgm-s

with 26 individual


Atmospheric SO, concentration in last hour of rainfall event flgmm3

and pH value associated




Atmospheric SO1 concentration in hour before rain fell


event; temperature

Rainborne SO2 bottom (S,) mg/-t

and after rainfall

’ (C,)

0.07 0.24 0.75 0.07 0.24 0.10 0.40 0.13 0.66 0.13 0.35 0.20 0.10 0.14 0.13 0.06 0.12 0.05 0.1 3.61 I.55 0.58 4.60 I .03

0.005 0.49 0.74 0.32 0.39 0.12 0.13 0.03 3.7 4.1 4.6 3.6 4.15 3.8 4.3 3.9 4.6 3.8 4.2 4.1 3.8 3.8 4.0 3.5 3.9 3.8 4.2 5.6 5.2 4.8 5.6 5.0

0.03 1.70 0.43 0.18 1.28 0.40

0.03 0.2 0.06 0.15 0.28 0.13 0.01 0.03

0.16 0.42

Calculated washout


Observed washout ratio (W,)

from Equations

4.2 4.4

pH of rain collected throughout event at.the bottom site

events [ wr has been calculated





L-----l 3.5






Fig. 2. Relationship between observed washout ratio (calculated for the lowest 100-m layer of the atmosphere in industrialized Sheffield) and pH at the bottom of the layer. Correlation coefficient for log (W,) against pH is 0.67 (significant at 0.1 “/, level).






w xlos 01


t 0001





sources (e.g. 27 November, 21 March, 27 July). Stable atmospheric conditions did not exist during the sampling periods (25 of the events were frontal rain). Another possible explanation of apparently depressed washout ratio values is that the equilibrium scavenging assumption is invalid. Non-equilibrium scavenging might be expected to occur in the presence of large vertical gradients of atmospheric SOZ (Hales, 1978). However, in this study, there is no clear-cut relationship between W,/W, and C, -C, (the vertical SOZ gradient). Neither is there a statistically significant relationship between W,/W, and precipitation intensity (which is closely related to the size distribution of raindrops). Smaller drops may be expected to equilibrate more quickly with the surrounding air (Barrie, 1978). The calculations W, = S,/C,, and W = S,/C, yield the washout ratios at the valley bottom and valley top, respectively (as opposed to W, = (S, - S,)/(C, -C,), which is the washout ratio which characterises the 100-m depth layer between the two sites). The pH of the rainfall was not measured at the top site, so no attempt was made to estimate the theoretical washout ratio at that location. However, W, was compared with Wrcalculated for the 1OOm layer (Fig. 4); the same 23 events which were shown in Fig. 2 were used. The relationship is similar to that between W, and W, except that the correlation coefficient is smaller (r = 0.48, significant at the 5 “, level) and the gradient is marginally further from unity. So the theoretical washout ratios (W,) do exhibit a better correlation with washout ratios determined from field observations which characterize a layer of air near the ground than with washout ratios determined from S and C values at the ground.



Fig. 3. Relationship between observed washout ratio (W,) and washout ratio predicted from Equations (5) and (6) (Wr) for the lowest 100-m layer of the atmosphere in industrial Sheffield. The correlation coefficient for log (W,) and log (Wr) is 0.76 (significant at 0.1 “/, level).


WT x 105

difference in atmospheric SOZ concentrations between the bottom and top sites before rain (AS,) and the same values in the last hour of rainfall (AS,). The relationship is AS, = 0.82 (AS,) + 7.6 with a correlation coefficient of 0.56 which is significant at the 1 “/, level (excluding the observations for 12 December, where the two values show an inexplicably large difference). Only under very stable atmospheric conditions would changes in atmospheric SO2 concentrations at the lower site not be reflected by changes at the upper site (Huddy, 1964); although during precipitation scavenging the valley-bottom atmospheric SO* levels might be less depressed because of closer proximity to the


0 01



Wb x 105

Fig. 4. Relationship between t’he observed washout ratio at the ground (IV,, calculated from ground-level rainborne and atmospheric SO, concentrations) and the predicted washout ratio (Wr). The correlation coefficient for log (W,) and log (Wr) is 0.48 (significant at the 5 “/, level).


Sulphur dioxide precipitation scavenging









Cb-Cl (p9 w3 i

Fig. 5. Relationship between the ratio of the washout ratio for the bottom site in the valley (W,) to the washout ratio for the top site (IQ, i.e. k&,/y:, and the difference in atmospheric SO, concentrations at the two sites (C, -C,). The correlation coefficient is - 0.44 ~signifi~nt at the 2 T0level).

Figure 5 shows the relationship between the ratio of the washout ratio for the bottom site (W,) to the washout ratio for the top site (JQ i.e. Q/w:, and the difference between the atmospheric SO2 concentration at the two sites (i.e. C, - C,). The correlation coefficient is - 0.44 (significant at the 2 “i;,level). Generally, W,/w is greater than 1.0, but it approaches unity with higher C, -C, values. There is no statistically significant relationship between W, and C, -C, or between wand C, -C,. If non-equilibrium scavenging occurs in an industrial air-shed, it might be expected that a negative relationship would exist between W, and C, -C,. The fact that W, decreases {relative to w) as C, -C, increases could be regarded as an indication of nonequilibrium scavenging (with large vertical gradients of atmospheric S02), but a difficulty is the fact that W, is generally greater than w. This problem may have been resolved had the pH values of precipitation been determined at the upper site. It is possible that the removal of other constituents of the urban atmosphere in the lowest IOOm elevates the pH value of the falling raindrop which then leads to relatively greater removal of SOI. Since the depth of the air-layer is known, it is possible to calculate the washout coefficient, 1, from Equation (I), assuming that there is a linear change of concentration of atmospheric SO, from the bottom site to the top site. For the 26 rainfall events there was a large range of L, from 0.01 to 8.56 x 10e5 s-l, with a mean value of 1.7 x 10-5s-1 (standard deviation 2.6 x lo- 5, and a median value of 2.6 x lo- ’ s- t . These values are similar to those found by Maul (1978) for similar rainfall rates in a less industrialized area. It is not possible to employ Equation (2) to calculate I., since most of the precipitation events were frontal and large sources of SOz were present in the valleybottom.

(iii) Short period sampling during precipitation


The use of hourly mean atmospheric SO2 concentrations, allied to the large and varying sources in an

industrial area, is less satisfactory than shorter-period sampling in areas relatively remote from large sources. Georgii (1963) and Newall and Eaves (1962) had previously studied the effects of rainfall on atmospheric SO2 concentrations during precipitation events. During the intensive studies at the rural station in Norfolk, air temperature was also recorded so that W, could be calculated

(Table 3). Three individual

are shown in Fig. 6. Generalizations


are difficult to

Table 3. Observed washout ratios (W,) and theoretical washout ratios (Ur) during intensive studies of three precipitation events. The time over which the calculations are made is governed either by the air sampling period (15 min) or the length of the precipitation sample (see Fig. 6). Time (min) from commencement of event

W, x lo5

wrx 105

g-50 S&l 70 17%120

6.20 0.80 0.84

0.28 0.29 0.70

480-530 [email protected] 560-590 59tM30 63u660 66&7lO 710-780

0.45 0.30 0.53 0.40 0.80 0.73 0.48

0.26 0.35 0.41 0.79 2.20 0.50 0.28


tk15 15-30 30-45 45-60 6&?5 75-90

0.23 0.25 0.37 0.23 0.22 0.17

0.55 0.65 0.53 0.48 0.54 0.58


&15 15-30

2.70 0.81

2.70 0.36

60-75 75-90 9ck105 1055120 12&135

0.15 0.10 0.40 0.23 0.46

0.35 0.23 0.23 0.21 0.20


T. D.






80 ttme





Fig. 6. Rainborne SO, concentrations (mg /-I) rainfall pH, rainfall intensity (I, mm mint x 102) total rainfall sulphate (mg /- I). atmospheric SO1 concentrations (pg mm3) throughout three individual precipitation events. The duration of sampling for atmospheric SO, was 15 min. The sampling duration for precipitation depended on intensity and varied from minutes to several hours. Times are zeroed at start of the precipitation event. (a) 19.1.78 (b) 31.1.80 (c) 5.2.80. The first rainfall SO:- determination in (a) was 34.0 mgf-‘.

make from the individual events studied, but the three examples shown indicate that rainborne SO, concentrations decrease from the early stages of the rainfall, except for the later stages of the event shown in Fig. 6(c). Rainborne sulphate, shows a decrease with time, except in the case on 5 February 1980 (Fig. 6c). The second stage of this particular event also exhibits an increase in rainborne SO2 with time. The atmospheric SO, concentrations during this event increased after the interruption (of _ 30min) in precipitation. The precipitation was associated with an occluded front and the wind veered by 60” as the front passed over. The wind direction shift was much more marked during this event than during the other two cases when

it was fairly constant. It appeared quite clear that a different air mass was passing over, with higher ambient SO, concentrations. The relationship between rainborne SO2 and pH value of precipitation is very clear during the first 30 min of the event depicted in Fig. 6(c). Another noticeable feature is that there is an obvious inverse relationship between the sulphate content of precipitation and both SO2 and pH during this initial stage of the event. The precipitation during the event on 31.1.80 was sufficiently heavy to enable samples to be taken every few minutes, Fig. 6(b). During this event, there is a broad correlation between rainborne SO2 and pH and atmospheric SO2 decreases throughout the event.


Sulphur dioxide precipitation scavenging Rainborne sulphate also exhibits a general decrease through the precipitation, except for an increase towards the end of the rainfall. The precipitation event on 19.1.78 comprised of snow in the first phase (before the interruption) and sleet in the later stages. There is no relationship between rainborne SO2 and pH. Both rainborne sulphate and atmospheric sulphur dioxide concentrations show a fairly steady decrease over time, except for the step just after the interruption ( - 5 h) when atmospheric SO2 levels regained their initial values. The mean of the observed washout coefficients (W,) is smaller than the mean of the predicted washout coefficients (W,) in Table 3, although there is no statistically significant difference between these two samples. The mean W, value {excluding the first, very large, value on 19.1.78) is 0.53 x 10’ (standard deviation, 0.54 x 10’) and the mean W, value (also excluding the first value on 19.1.78) is 0.61 x 10’.


In spite of the continuous injection of SO2 into an urban airshed, even within a heavily industrialized city, when washout data and atmospheric SO2 levels are based on a sampling period of one hour, there is a reasonably close relationship between measured washout ratios and predicted washout ratios (when the mean rainfall-amount weighted pH value is used). Because of the possibility of equilibrium scavenging conditions not pertaining due to large vertical gradients ofatmospheric SO2 concentration within the airshed of a heavily industrialized city, washout of SO2 through a 100-m depth of the atmosphere was measured. A good relationship was found between the layer-washout ratio which characterized the 100-m layer of air and theoretically predicted washout ratios. The agreement was greater than when the washout ratio was determined from rainborne SO2 concentrations and atmospheric SO2 levels measured at one site at the ground. The measured washout ratios were less than the predicted ratios, but there was no clear-cut relationship between the disparity and vertical atmospheric SO2 gradients. The fine-scale structure of individual precipitation events for “regional” rainfall appeared to vary considerably, even during a relatively small number, and generalizations are difficult to make. However, it does appear that the closer together are the sampling periods for precipitation and air, the better the agreement between measured and predicted washout ratios.


Barrie L. A. (1978) An improved model of reversible SO* washout by rain. Atmospheric Environment 12, 407412. Barrie L. A. (1981) The prediction of rain acidity and SO2 scavenging in eastern North America. Atmosphrric Environment 15, 31-41. Biliingsley J., Kallend A. S. and Marsh A. R. W. (1976) Washout of sulphur dioxide by rain. Centrai Electricity Research Laboratories, L~therhead U.K. (unpublished report). Chamberlain A. C. (1960) Aspects of the deposition of radioactive and other gases and particles. Inr. J. Air wur. Pollut. 3, 75-88. Dana M. T. (1980) SO2 versus sulfate wet deposition in the Eastern United States. J. Geophys. Res. 185, 4475-480. Davies T. D. (1974) Determination of the local removal of sulphur dioxide by precipitation. Observation and measurement of atmospheric pollution. WMO-No. 368, 567-578, World Meteorological Organisation, Geneva. Davies T. D. (1976) Precipitation scavenging of sulphur dioxide in an industrial area. Atmospheric ~nri~on~e~r IO, 879-890. Davies T. D. (1979) Dissolved sulphur dioxide and sulphate in urban and rural precipitation (Norfolk, U.K.). Armosphrric Enuironmenr 13, 1275-1285. Georgii H. W. (1963) On the effect of rainfall in the sulphur dioxide concentration in the atmosphere. Int. J. Air Wur. Pollut. 7, 1057-1059. Hales J. M. (1978). Wet removal of sulphur compounds from the atmosphere. Atmospheric Enoironment 12, 389-399. Hales J. M. and Dana M. T. (1979) Regional scale deposition of sulphur dioxide by precipitation scavenging. Armospheric ~~vironme~f 13, 1121-I 132. Hales J. M. and Sutter S. L. (1973) Solubility of suiphur dioxide in water at low concentrations. .~frn~.~~~~~~~~ En~~ro~menz7, 997-1001. Hales J. M., Thorpe J. M. and Wolf M. A. (1971) Final report on field investigation of sulphur dioxide washout from the plume of a large coal-fired plant by natural precipitation. Batelle Me Inst., Richland. Huddy W. (1964). Correlations between atmospheric SO1 concentrations and meteorological variables in Shefield. Sheffield Air Pollution Unit Report. Likens G. E., Bormann F. H., Pierce R. S., Eaton J. S. and Johnson N. M. (1977) Biogeochemisrry elf u Foresred Ecosystem. Springer, New York. Martin A. (1982) A short study on the influence ofa valley on the composition of rainwater. Atmospheric ~nL~ir~~n~i~~r 16, 785-793. Maul P. R. (1978) Preliminary estimates of the washout coefficient for sulphur dioxide using data from an East Midlands ground level monitoring network. ,4rmosphrric Enl;ironmem 12, 2515-2517. Newall H. E. and Eaves A. (1962) The effect of wind speed and rainfall on the concentration of sulphur dioxide in the atmosphere. Int. J. Air Wut. Pollut. 6, 173.-177. Persson G. A. (1966) Automatic calorimetric determination of low concentrations of sulphate for measuring sulphur dioxide in ambient air. Inr. J. Air War.Poitur. 10,845-852. Slinn W. G. N., Hasse L., Hicks B. B., Hogan A. W.. D.. Liss P. S., Munnich K. 0.. Sehmel G. A. and Vittori 0. (1978) Some aspects of the transfer of atmospheric trace constituents past the air-sea interface. ,&mnsphrric Environment 12, 2055-2089.

.4cknowledyement --Some of the results reported here were obtained from a project funded by the Natural Environment Research Council.

West P. W. and Gaeke G. C. (1956) Fixation of sulphur dioxide as disulphitomercurate (II) and subsequent colorimetric estimation. Analyt. Chemist 28, 1816-1819.