Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
5 Precipitation Chemistry Gary J. Stensland, Douglas M. Whelpdale, and Gary Oehlert INTRODUCTION The Role of Precipitation Composition in This Study The chemical composition of wet deposition has been central to the question of the effects of acid deposition and the long-range transport of air pollution since the issue of "acid rain" rose to prominence in Europe two decades ago. Early evidence linked acidic substances in precipitation to harmful effects on surface waters and fisheries in Scandinavia, and as a consequence interest has since focused on precipitation acidity and acid deposition rather than on the more general case of depo- sition of air pollutants. This focus can be attributed to several factors: Changes in precipitation composition could actually be detected, and techniques were available to make measurements; this was not true for dry deposi- tion. In addition, precipitation composition was starting to be used as an indicator of atmospheric quality, so researchers were aware of the value of such measurements. Finally, the correlation between adverse effects and wet deposition could be made more directly than that between effects and emissions. The chemistry of precipitation has been studied both in the expectation that it will provide a measure of the consequences of emissions in time and space and also as an indicator of effects on the environment. Unfortu- nately, problems exist both with the quality of past data and with the interpretation of the data. These difficul- ties have stimulated the examination of other historical data in efforts to improve our understanding of the phenomenon of acid deposition and to corroborate the available precipitation-chemistry records. 128
129 Wet and Dry Deposition Atmospheric constituents, and thus pollutant emissions, reach the surface of the Earth by a variety of processes: those in which precipitation is involved contribute to wet deposition; those that (primarily during dry periods) involve sedimentation, turbulent and molecular diffusion, impaction, and interception contribute to dry deposition; and those that include cloud and fog impaction, dew, and frost are sometimes termed occult deposition. Wet deposi- tion and bulk deposition (i.e., the sum of wet, dry, and occult, obtained with the collector always open) have been measured with varying degrees of reliability for decades; dry deposition has been measured sporadically in a research mode but not routinely; and measurements of the third type of deposition are just beginning. Even his- torical air-concentration data, from which dry deposition might have been estimated, are sparse and incomplete. Out of necessity, therefore, we focus on wet deposi- tion data. From the point of view of determining his- torical trends, this is not necessarily a serious limitation on the regional-to-continental scale, where the wet deposition is expected to reflect total deposition to a reasonable extent. location and at a specific time, At a specific however, wet deposition may not be representative of total deposition. In addition, omitting dry and occult deposition is certainly a serious constraint on determining mass balance and interpreting effects. Over eastern North America the total wet and total dry deposition are thought to be of approximately equal magnitude. Dry deposition is, in general, greater than wet near emission Arms ration where ambient concentrations of pollutants are higher. An additional complexity for dry deposition is that the chemical species behave differently: in general, gaseous nitric acid, sulfur dioxide, and ammonia are removed more efficiently than their particulate forms, nitrate, sulfate, and ammonium. In high-elevation ecosystems the direct input from cloud and fog water is likely to exceed that from either of the other pathways. Interpreting observations of surface ecosystems must allow for these other inputs, even though the majority of past deposition data is for wet or bulk deposition. ~ ~ ~4, _,
130 Constituents of Interest We limit this discussion of wet deposition to the major soluble species in precipitation that account for most of the measured conductance of the samples. The species include the following ions: hydrogen (H+), bicarbonate (HCO3), calcium (Ca ), magnesium (Mg ), sodium (Na+), potassium (K+), sulfate (SO4 ), nitrate (NO3), chloride (C1-), and ammonium (Nut). Experience has shown that a pH value can be calculated from measurements of the last eight ions in the list, and the calculated value is usually in good agreement with the measured pH value. The fact that we can often successfully calculate the pH of precipitation samples indicates that the small list of measured ions is probably sufficient for studies of wet deposition that emphasize the acid precipitation phenomenon. However, there are exceptions. For example, samples from remote locations can be strongly affected by organic acids (Galloway et al. 1982). The data in Table 5.1 demonstrate the relative importance of these ions at three monitoring sites of the National Acid Deposition Program (NADP) in the United TABLE 5.1 Median Ion Concentrations for 1979 for Three NADP Sites (~eq/L) GA MN NY Ion (42 Samples)a (37 Samples)b (49 Samples)C SO4- 38.9 45.8 44.8 NO3 11.6 24.2 25.0 Cl- 8.2 4.2 4.2 HCO3 (calculated) 0.3 10.3 0.1 Anions 59.0 84.5 74.1 NH4+ 5 5 37 7 8.3 Ca2+ 5.0 28.9 6.5 Mg2+ 2.4 6.1 1.9 K+ 0.7 2.0 0.4 Na+ 17.6 13.7 4.9 H+ 17.8 0.5 45.7 Cations 49.0 88.9 67.7 Median pH 4.75 6.31 4.34 NOTE: See Georgia, Minnesota, and New York in Figure 5.23 for location of these sites. aThe Georgia Station site in west central Georgia. bThe Lamberton site in southwestern Minnesota. C The Huntington Wildlife site in northeastern New York.
131 States. Concentrations are expressed in microequivalents per liter (peq/L) to permit a direct evaluation of each ion's contribution to the anion or cation sum. If all ions were measured and there were no analytical uncer- tainty, then the anion sum would equal the cation sum. The values for hydrogen ion concentration were calculated from the measured median pH value, and the values for bicarbonate were calculated by assuming that the sample was in equilibrium with atmospheric carbon dioxide. Although the sulfate and nitrate levels shown are similar at the Minnesota and New York sites, the pH differs greatly owing to the higher levels of ammonium, calcium, magnesium, sodium, and potassium ions at the Minnesota site. These ions are frequently associated with basic compounds. The data in Table 5.1 suggest, therefore, that the concentrations of all the major ions must be considered if we are to understand the time and space patterns of pH. The data also indicate that sites in different regions of the United States exhibit large differences in ionic concentrations. Currently, sites in states such as Ohio, New York, Pennsylvania, and West Virginia have the feature shown for the New York site in Table 5.1, for which hydrogen, sulfate, and nitrate are the dominant ions. For the New York site, 98 percent of the acidity coula be accounted for if all the sulfate were sulfuric acid, whereas nitrate, as nitric acid, could have accounted for about 55 percent of the acidity. Bower sax and de Pena (1980) have concluded, by applying multiple linear-regression analysis for a central Pennsylvania site, that on the average, sulfuric acid is the principal contributor to hydrogen-ion concentration in rain, but the acidity in snow is principally from nitric acid. Variability in Acid Deposition: The Role of Meteorology Any discussion of short- and long-term trends in acid deposition cannot ignore the role of meteorological and climatological variability. Even if pollutant emissions were absolutely constant every hour of the day and every day of the year, variations in atmospheric conditions would bring about uneven deposition patterns in both time and space. For example, both variability throughout the course of a year and long-term trends in precipitation records reflect changes in storm tracks and in the fre-
132 quency of precipitation-producing circulation patterns that affect a given region each year. Differences in precipitation amount from year to year affect total wet deposition, but even if the precipitation amount were constant every year, wet deposition of sulfate, nitrate, and other common ions would not necessarily be, because of the dependence of the concentrations of these ions on the size of the precipitation events and the time interval between the events. Atmospheric processes vary continuously across a wide temporal and spatial range; at whatever scale one selects, atmospheric variability plays a fundamental role in the acid deposition process. Unless we understand this role thoroughly, the meaning of trends in acid deposition will be difficult to resolve unequivocally. (See Chapter 3 for a more detailed discussion.) DATA EVALUATION Given that the object of our evaluation is to deter- mine trends in concentrations (or deposition) of chemicals in precipitation, our first task is to identify approxi- mate data sets and determine their quality. Since adequate methods to determine dry deposition rates for most chemical species do not exist, wet-only deposition data collected with automatic samplers are generally the most reliable and useful. However, owing to the sparsity of such data we must also consider data bases in which less desirable sampling methods were used. Such methods include bulk sampling and manual wet-only sampling in which dust leakage into the sampler and water evaporation and snow sublimation from the collection vessels have apparently introduced serious bias into the data sets. The same biases may occur even in an automatic wet-only sampling device when mechanical failure causes exposure of the sample to the open atmosphere. In this chapter we emphasize trends in concentration rather than in depo- sition to avoid the introduction of spatial variability caused by variability in the amount of precipitation. A primary tool in evaluating possible biases in monitoring techniques is comparison of different data sets, such as bulk data with concurrent wet-only data. This approach has not always been used in previous reports but may be one of the most important evaluation techniques. However, since local sources can affect bulk data through high levels of dry deposition, the only
133 satisfactory comparison must involve concurrence in space and time. Unfortunately, only in a few cases are the two sampling methodologies concurrent spatially and temporally, so that such comparisons are usually far from ideal. Several different evaluations are described in this chapter, with emphasis on evaluating those United States data sets quoted most frequently in trend studies. Comparing Three-Month Regional Bulk Data Concurrently with Wet-Only Data Most of the published comparisons between bulk and wet-only ion concentrations are for eastern sites. The difference between these two sampling methods would be expected to be greater in locations where local sources may be influential, for example, where sampling occurs near tilled fields with an abundance of windblown dust. Peters amd Bonelli (1982) have reported the results of a bulk sampling program for the period from about December 1, 1980, to March 1, 1981. They indicated that consider- able effort in siting was made to avoid the occurrence of local contamination of regional deposition. States from Minnesota and Illinois eastward to Maryland and Maine were included in the study. The NADP network had a large number of stations operating in the same area for the same time period. Therefore we are able to present a simple comparison of the results for the wet-only NADP concentration data versus the bulk concentration data and examine geographic variations in the differences between the two methods. The wet-only and bulk concentration data were plotted and examined visually to select five contiguous geo- graphical areas in which there was some degree of uniformity in values. The areas selected are shown in Figure 5.1. The bulk sample at each site was a single time-integrated sample collected with a continuously open cylindrically shaped sampler. The concentration ratios of the, ions (mean of bulk values divided by mean of wet-only values) are shown in Table 5.2 for each area. One may observe that in all cases the value of the ratio is greater than or equal to 1.0 and that there is considerable variation in the ratio by area and by chemical parameter. The ratios for calcium, magnesium, and potassium are greater than 2 (and as high as 30) for all areas, and bulk sulfate concentrations exceeded
134 :___ 1 15 V · 1 }41 ~ 1~ Jo-'' 70 \ 23 ~ . ~ 3: , 1 -_ _` r~ ~ ~~ my FIGURE 5.1 Map of five areas (indicated by roman numerals) used for comparing concentrations of various ions in wet-only deposition with those in bulk samples. The arabic numbers indicate the number of bulk sampling sites within each area. The NADP sites are indicated by circles. Bulk-measuring sites are distributed uniformly about the areas but are not co-located with NADP sites. See also Table 5.2. wet-only concentrations by factors of 2.0 and 8.1 in areas III and V, respectively. All ion concentrations were elevated by more than 90 percent in areas III and V, with a median ratio value of 3.7 for these two areas. Peters and Bonelli (1982) reported both sample volume and an estimated sample volume derived from the data from the nearest National Weather Service precipitation gauge site. Using these data, we calculated the medians of the ratios of the measured divided by estimated volumes to be 0.88, 0.79, 0.81, 0.69, and 0.45 for areas I, II, III, IV, and V, respectively. These ratios suggest that bulk samples may be concentrated up to a factor of 2 by evaporation. Thus, the many ratios greater than 2.0 in Table 5.2 suggest that differences between concentrations of ions in bulk samples and in wet-only samples cannot by fully explained by evaporation. Even when an average bulk to wet-only ion concentration ratio for an area in Table 5.2 has a value close to 1, an individual bulk site in that area may still differ widely in concentration from the wet-only data. Each site needs
135 TABLE 5.2 Ion-Concentration Ratios (Mean of Bulk Values Divided by Mean of Wet-Only Values) for All Sites in Areas I to V for the Period December 1980- February 1981 Ion Area Ca2+ Mg2+ K+ Na+ Cl- NO3 NH4 SO2- NBa NWb I 2.5 5.6 2.7 1.2 1.0 1.1 1.1 1.0 23 3 II 3.7 4.8 5.9 2.9 2.0 1.5 2.0 1.3 70 11 III 7.2 3.5 11.2 2.7 2.1 1.9 3.5 2.0 45 4 IV 2.7 2.3 3.1 1.9 1.6 1.1 1.3 1.3 26 3 vc 30.5 19.2 6.0 2.8 6.6 3.9 2.1 8.1 15 - NOTE: Areas are depicted in Figure 5.1. a Number of bulk sampling sites used to determine the means. b Number of wet-only sampling sites used to determine the means. C Since there were no NADP sites in area V, computer-generated contour maps (for weighted-average concentration) for all data available for the period 1978 - mid-1983 were used to estimate average NADP ion concentrations for this area. SOURCES: NADP data were provided by Illinois State Water Survey. Bulk data were obtained from Peters and Bonelli (1982~. to be evaluated separately for each chemical parameter to determine if bulk concentrations are likely to be similar to wet-only concentrations. Comparing Bulk and Wet-Only Data at Hubbard Brook, New Hampshire A comparison of bulk and wet-only data from the Hubbard Brook site shows good agreement for most major ions, in contrast to other bulk versus wet-only com- parisons (cf. Tables 5.3 and 5.4 (below)). Weighted- average concentrations of the colocated Hubbard Brook bulk samples and NADP wet-only samples were determined for the period from June 1979 to May 1982. Table 5.3 presents the results as 3-year averages except for pH, for which the averages of three 12-month periods are given. Bulk concentrations of sulfate plus nitrate were 4.8 peq/L (7 percent) higher than for the wet-only values. The sum of calcium plus magnesium, an indicator of dust input, was actually lower for the bulk than for the wet-only data. Apparently, dry deposition of sulfate and nitrate to the continuously exposed bulk collector, either through gaseous precursors or in association with dust particles, was not an important factor. Another
136 TABLE 5.3 Comparison of Bulk and Wet-Only Weighted-Average Concentrations (~eq/L) at Hubbard Brook, New Hampshire Hubbard Brook NADP Ion (bulk)a (wet only)b SO2- 49.9 48.5 NOT 27.3 23.9 C1- 6.6 62 NH4 11.0 9.4 Ca2+ 5.6 5.8 Mg2+ 2.2 3.0 K+ 1.0 0.4 Na+ 5.3 8.4 H+ 60.3 47.9 pH 4.24, 4.19, 4.22 4.39, 4.27, 4.32 Weighted-average concentrations for the period June 1979-May 1982. Data provided by John Eaton, Cornell University, personal communication 1985. Weighted-average concentration for wet-only samples from the NADP collector for the period June 1979-May 1982. typical problem with bulk data, concentration effects owing to sample evaporation, was probably not a major factor given the close agreement of bulk ion concen- trations with the wet-only concentrations. This finding suggests that the funnel and bottle design of the Hubbard Brook bulk sampler effectively prevented this problem for long-term averages at this site. The three (12-month) weighted averages of pH for the bulk samples are lower than the corresponding values for the wet-only samples. The 3-year weighted-average concentration of hydrogen ions in bulk samples is 26 percent larger than in wet- only samples. Such a difference is likely to be significant for these data. In addition to different sampling devices and dif- ferent laboratories for sample analyses of the Hubbard Brook data, different screening and sampling procedures were used for bulk data and the wet-only data. More bulk data than wet-only data were deleted from the data base for the 3-year period. Since the two sampling devices were generally serviced on different days of the week, it was not possible to calculate the data as exactly matched pairs. It would be useful to undertake a more detailed study involving a week-by-week comparison--both to learn more about the long-term record at this important site and to help to relate trends in bulk measurements to those in wet-only measurements.
137 Comparing U.S. Geological Survey Bulk Data with National Acid Deposition Program Wet-Only Data Nine U.S. Geological Survey (USGS) bulk collection sites have operated in New York state and Pennsylvania since 1965 (Barnes et al. 1982), and several investigators have summarized and analyzed these data. (See later discussion.) Here, we compare these data with NADP wet-only data (Table 5.4). The first column of the table gives the mean and the standard deviation of all New York NADP data available through the end of 1982; the seven NADP sites began operation at different times between November 1978 and TABLE 5.4 Ion Concentrations (~eq/L) and pH for All New York NADP Sites and for Two New York USGS Bulk Sites Time-Adjusted Weighted Average Weighted Average Mean and Standard Deviation for Mays Mays Ion Seven NADP Sitesa Hinckleyb Points Hinckley Point Ne 23 24 23 24 Ca2+ 8.4 + 2.0 30.2 74.2 26.3 71.6 Mg2+ 3.5 + 0.8 7.7 17.1 6.7 16.5 K+ 0.8 + 0.6 2.1 2.9 1.8 2.8 Na+ 6.2 + 3.5 5.4 10.4 4.7 10.0 NH4+ 16.5 + 5.5 31.2 22.7 37.3 27.5 Sumf 35.4 76.6 127.3 76.8 128.4 NO3 30.2 + 5.4 41.7 46.5 62.1 54.7 SO2- 64.8 + 13.4 72.5 82.2 65.9 76.4 C1- 5.8 + 3.3 8.9 15.2 8.9 15.2 pH 4.22 (4.15-4.33)g 3.99 3.99 3.90 3.86 H+ 60.4 + 10.3 102.3h 102.3h 126.4h 138.3h aVolume-weighted concentrations for 1978-1982. bHinckley USGS bulk data for 8/31/75 to 9/15/77. CMays Point USGS bulk data for 8/31/75 to 8/31177. Time adjustments made using time-trend results for 1965-1978 period as determined by Kendall's test (Barnes et al. 1982). An adjustment factor for a 4.5-year time shift was multiplied by the USGS bulk concentrations: Ca2+, Mg2+, K+, and Na+ concentrations were multiplied by 0.87 and 0.965 for Hinckley and Mays Point, respectively; SO2- concentration by 0.909 and 0.93; NO3 concentration by 1.489 and 1.176; NH4+ concentration by 1.194 and 1.211; and H+ concentration by 1.236 and 1.352. eN is the number of samples at USGS sites. f"Sum" gives the total microequivalents per liter potentially available for neutralizing acids in precipitation. g Range. h USGS weighted-average H+ concentrations are likely influenced by a few outliers. Median concentrations are ~50% less.
138 June 1980. The sample volume weighted-average concentra- tions were calculated for each NADP site, and the means of these values are shown in Table 5.4. The table also presents weighted-average concentrations of data from Hinckley and May s Point, two of the USGS bulk sampling sites in New York; later in this chapter these two sites are listed with three others for con- sideration of regional time-trend analysis. The Hinckley and May s Point sites were chosen for inclusion in Table 5.4 because these sites tend to represent the extremes in concentrations found at the USGS stations, with the Hinckley site tending to have the lower concentrations and the May s Point site the higher concentrations (Barnes et al. 1982). Because the two data sets do not overlap in time (1978-1982 for NADP and 1975-1977 for USGS) the USGS data are presented in two ways. The two middle columns in Table 5.4 present the data from 1975-1977. The two right hand columns present the USGS data adjusted to account for the difference in time periods. Temporal trends of the concentrations of the ions listed in the table were determined by Barnes et al. (1982) for the period from 1965 to 1978. We estimated the data for the 1978-1982 period by multiplying the USGS data (1975-1977) by a factor derived from the calculated trend from 1965-1978 (see footnote d in Table 5.4 for details) . Regardless of whether the actual or adjusted USGS values are used, large differences exist between the NADP and USGS data sets. Calcium concentrations are three to nine times higher in the bulk data. Magnesium, potassium, ammonium, and nitrate concentrations are also higher in the bulk data (factors of 1.4 to 4.9). Sodium concentra- tions are about the same, and the hydrogen-ion concentra- tion is much higher in the bulk data, although one might anticipate the reverse. The bulk sulfate values are somewhat higher than the wet-only values at May s Point but are about the same as Hinckley values. The overall conclusion is that the USGS bulk concen- tration data are significantly different from NADP wet-only data for many ions. The sum of the basic cations is larger at the bulk sites than at the wet-only sites by more than twofold. -
139 Comparing Two Wet-Only Networks at Three Sites In this section wet-only data from the Multistate Atmospheric Power Production Pollution Study (MAP3S) network (event samples) and the NADP network (weekly samples) are compared for three locations. Data were selected for the same time interval (March or April 1979 to December 1981) for each location in each network. Median concentrations and pH values for Bondville, Illinois, are shown in Table 5.5; the two collectors at this site are separated by about 10 m. The data in Table 5.6 are for sites near State College, Pennsylvania, and the two collectors are separated by about 25 km. The data in Table 5.7 are for sites near Ithaca, New York, where the two collectors are about 40 km apart. The best agreement of data might be expected for the Bondville site since the two collection devices are located so close to each other. At Bondville the NADP sulfate and nitrate data are similar to those from F~P3S; pH agrees within 0.02 pH unit; calcium, magnesium, and sodium are higher in NADP data; and potassium, ammonium, and chloride are higher for MAP3S data. A similar pattern is found for State College even though the two sites are about 25 km apart at this location and thus not exactly the same precipitation events were sampled. The average calcium, magnesium, and sodium NADP and MAP3S values in Table 5.7 for the Ithaca site are less similar than those in Tables 5.5 and 5.6. The NADP values are also less similar than the MAP3S values for sulfate and nitrate. Therefore, we suggest that the environment may be less pristine at the Ithaca NADP site than at the MAP3S site, resulting in higher ion concentra- tions in samples at the former. Despite this, the pH values for the NADP and MAP3S sites are essentially the same for this 32-month period, indicating that pH com- parisons should be done in conjunction with comparisons of the other major ions. These data show that the 32-month median concentra- tions of ions for NADP and nearby MAP3S sites may differ by more than 10 percent for the various ions. On average, however, the pH agreement is excellent, and nitrate and sulfate data agree within 10 percent for two of the locations. It is likely that some precipitation constituents at the Ithaca NADP site are more affected by local sources than those at the MAP3S site. This illustrates another caution to be observed in carrying out a time-trend
140 TABLE 5.5 Median Concentrations (mg/L) in Precipitation at Bondville, Illinois Ion MAP3S NADP NADP . MAP3S MAP3S . NADP Ca2+ 0.23 0.29 1.26 Mg2+ 0.029 0.041 1.41 Na+ 0.061 0.073 1.20 K+ 0.038 0.027 1.41 NH4+ 0.43 0.41 1.05 SO4- 307 3.28 1.07 NO3 1.74 1.85 1.06 C1- 0.19 0.18 1.06 Median pH Na 140 109 4.32 4.30 aN is number of samples measured. TABLE 5.6 Median Concentrations (mg/L) in Precipitation for Two Sites Near State College, Pennsylvania NADP NADP . MAP3S MAP3S . NADP Ion MAP3S Ca2+ 0.22 0.24 1.09 Mg2+ 0.034 0.048 1.41 Na+ 0.090 0.132 1.47 K+ 0.063 0.036 1.75 NH4+ 0.40 0.32 1.25 SO4- 3.75 3.71 1.01 NO3 2.60 2.49 1.04 C1- 0.32 0.24 1.33 Median pH 4.11 4.15 Na 202 109 aN is number of samples measured. TABLE 5.7 Median Concentrations (mg/L) in Precipitation for Two Sites Near Ithaca, New York Ion MAP3S NADP NADP . MAP3S MAP3S . NADP Ca2+ 0.12 0.22 1.83 Mg2+ 0.020 0.044 2.20 Na+ 0.041 0.080 1.95 K+ 0.028 0.023 1.22 NH4+ 0.34 0.44 1.29 SO4- 3.07 4.28 1.39 NO3 2.23 2.55 1.14 C1- 0.19 0.20 1.05 Median pH 4.15 Na 187 4.16 116 aN is number of samples measured.
141 analysis of constituents in precipitation. Changes in time in the local sources may be quite different from the changes in time of the regional sources af fecting a site. The problem would generally be expected to be more severe if bulk data were being evaluated. Concluding Remarks The paucity of wet-only precipitation data has made essential the use of bulk data for trend analysis. We recognize that bulk concentration data can be affected by evaporation and by dry deposition from distant and local sources. Few data sets are available with which to examine precisely the differences between bulk and wet-only data and the influences of these factors on deposition. Even when bulk sampling sites are located carefully, bulk concentrations usually exceed wet-only values. The differences clearly depend on sampling location, time period, and chemical species. The USGS network data show higher concentrations than do data from wet-only sampling stations in the same region. Thus, the data have limitations for trend analysis despite the length of the record available from this network. In at least one case, the site at Hubbard Brook, New Hampshire, the sulfate and nitrate concentrations in bulk and wet-only samples agree well, and the difference in the sums of the basic cations is relatively small. (See Table 5 e 3.) Thus one may have reasonable confidence in applying the bulk data for trend analysis at this site. The sampling procedure at Hubbard Brook eliminated the evaporation problem, and evidently siting has minimized dry deposition effects, probably because of the long distance of this site from major sources of SOX emissions. Thus, Hubbard Brook seems to be the exception regarding the general lack of compatibility between bulk and wet-only data. Comparing NADP and MAP3S data for three locations suggests that even the wet-only data may differ because of local effects. Differences in data from these two networks at nearby (i.e., 40-km) stations in New York are probably caused by differences in local emission sources or perhaps by different sampling protocols. (See Tables 5.5 to 5.7.)
142 GEOGRAPHICAL DISTRIBUTION Current Concentration and Deposition Patterns The geographical distribution of major ionic con- stituents of precipitation and of their wet deposition over North America is generally well known. The many pH, concentration, and deposition maps prepared using various combinations of the 1978-1980 data from several networks show, essentially, the same features; maps for the year 1980 are reproduced below from Barrie and Hales (1984). These data were prepared during the Canada-USA Memorandum of Intent work on the basis of observations from the Canadian Network for Sampling Precipitation (CANSAP), the Air and Precipitation Network (APN), and the MAP3S and NADP networks and are from the first year when a com- patible merger of data from these four federal networks was possible. Figures 5.2 to 5.6 show the annual precipitation amount weighted averages for hydrogen-ion concentration, pH, and concentrations of sulfate, nitrate, and ammonium. Figures 5.2 and 5.3 show hydrogen-ion concentrations up to 60-80 umol/L (pH 4.2-4.1) over a region of some 106 km2 stretching from Illinois and Kentucky north- eastward through New York and southern Ontario. Figure 5.4 shows a background sulfate concentration of less than 10 umol/L in northern Canada and 3-15 umol/L in the western United States, and values reaching greater than 40 nmol/L in Ohio and southern Ontario. Nitrate in Figure 5.5 has a background level of 2-6 nmol/L and rises to 40-50 umol/L over the lower Great Lakes. The patterns of hydrogen, sulfate, and nitrate concentration in precipitation are similar. In contrast, Figure 5.6 shows that ammonium concentration reaches a maximum of greater than 40 umol/L in and downwind of the northern plains states of the United States. Annual average calcium concentrations were not reported by Barrie and Hales (1984) but are shown for U.S. sites in a report of the U.S. Environmental Protection Agency (1984). Values are lowest for sites in states along the East Coast and in the Northwest, with concentrations generally below 5 ~mol/L. Sites in the Upper Great Plains in an area from eastern Wyoming to Lake Michigan and in the Southwest from western Texas to southern New Mexico and Arizona had the highest values, about 12 to 25 vmol/L. The patterns for ammonium and calcium ions identify those areas with the greatest acid-neutralizing potential.
143 Wet deposition is determined as the product of precipitation-amount-weighted concentration (Figures 5.2 to 5.6) and annual precipitation amount. Precipitation- amount fields have greater spatial variability than the concentration fields and in addition may be a source of large year-to-year differences. Figure 5.7 shows that precipitation during 1980 was not highly anomalous in the Northeast, and it indicates the high variability of the field. The wet deposition fields, shown in Figures 5.8 ~ . to 5.11 for hydrogen, sulfate, nitrate, ana ammonium, are similar to the regional patterns of the concentration fields. Wet deposition of hydrogen ion reaches 80 mmol/m2, sulfate and nitrate 40 mmol/m2, and ammonium 30 mmol/m2 in the eastern portion of the continent. Barrie and Hales (1984) present 5 years of observations from three MAP3S stations to demonstrate how representa- tive the year 1980 was of the 5-year normal for the high-deposition region in the Northeast. Table 5.8 shows the deviations (D) of the 1980 values for precipitation amount and concentration from 5-year (1977-1981) means. The 1980 precipitation amount was 8-16 percent below the (1977-1981) average, similar to the difference ir 5-year Figure 5.7. - - - Despite this, the deviations of the ionic concentrations Prom the 5-year means were variable, ranging from -9 percent for hydrogen ion at Pennsylvania State University to +25 percent for sulfate and ammonium at Whiteface Mountain, New York. Longer data records are required in order to establish whether such deviations from long-term means are typical and acceptable. Seasonal variations in the concentrations of particular ions, such as sulfate and hydrogen, have been discussed For example, Herman and Gorham (1957) frequently. reported that in Nova Scotia snow sampled in the early 1950s contained lower sulfur and nitrogen concentrations than did rain sampled during the same period. In the late 1960s, Fisher et al. (1968) observed lower concentra- tions of precipitation sulfate in the cold season. Bowersox and de Pena (1980), Pack and Pack (1980), and Pack (1982) reported strong seasonal variations in sulfate in precipitation at MAP3S sites in New York, Pennsylvania, and Virginia. Bowersox and de Pena (1980) found only slightly higher nitrate in precipitation in the winter than they did in other seasons at the MAP3S site in Pennsylvania. Hydrogen ion had a maximum in the warm months, and sulfate was the principal anion affecting acidity. Nitrate, at concen- trations similar to those of sulfate, did not correlate (text continues on page 157)
144 / : r`~;_ ~ ~ ,,,''''~ - 'A . , _ OH I | ^~ ___ f r \~25 `` ~ l__l______ .'_ ~ ~ ~ 2 I All ; · 09 ~ I I `` r-' ~~ ~~ ~~ C-___ em ~ I r - - - - ~N '; i I V-_. I I \ ~ ~ '_/ '~;~' \ ~ _ 'it 20 FIGURE 5.2 Spatial distribution of the precipitation- amount-weighted annual mean hydrogen-ion concentration (micromoles per liter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Barrie and Hales (1984).
145 ' ~ W~ ~ .,: {/ , .s,o,' (50',,' ·sl7~55\ _~ ~ :~:~:-~.J~ , ' . A.... , ~ ~~,s~ W.,,',<;./' ~ FIGURE 5.3 Spatial distribution of the precipitation- amount-weighted annual mean hydrogen-ion concentration (expressed as pH) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Barrie and Hales (l984).
146 ,. ~~ ~~\ , v, I 16 , . I 17 ./ I 14 1 1 it'- 40 40 / 3~-~- ,`;_ ,' '` 3.4 1' '` ,_-I_ - _ / ~ I (~6.6'~, ,' I 11 · S ·.';2 ! ~ .14 . ~ ~ ~ , ~ '< ~ _' ,:20 , it' - _!,-.,\---. i\ \,>0_ -- (~¢ A, \, ,'_ . , 2 0 ~_~3\ \,(' <0~, '1 6. FIGURE 5.4 Spatial distribution of the precipitation- amount-weighted annual mean sulfate-ion concentration (micromoles per liter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Barrie and Hales (1984).
147 ,~ ~ I. ~ ~ i_. , ~ .; , ~ ~ ~~ ,-<; ~ dI>10\ 'arks .x ~ ~ -' FIGURE 5.5 Spatial distribution of the precipitation- amount-weighted annual mean nitrate-ion concentration (micromoles per liter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Barrie and Hales (1984).
148 / / ~^ ' l V~ . ~ --~2~ ;063 9 ~ Nos ·7020~¢ O I-- I- - , , ; ~ ma 94; 1 it's 1'-'~;-;-_ ~ ~/~ ~ .~, FIGURE 5.6 Spatial distribution of the precipitation- amount-weighted annual mean ammonium-ion concentration (micromoles per liter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Bar rie and Hales (1984).
149 .,i A, , . 1 , , `, - .~L /~ With FIGURE 5.7 Precipitation amount in 1980 expressed as a percentage of the long-term mean amount of precipitation in North America. SOURCE: Barrie and Hales (1984).
150 ,,.~ '_,"' ~ - 24 , _ ·~d ~ , _ ~ _ \' ~ ^~. 05 .~. ~ two FIGURE 5.8 Spatial distribution of annual wet deposition of hydrogen ions (millimoles per square meter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Barrie and Hales (1984).
151 ,~ -iS ~ - - -t ~ r a\\ ·~ 4.6 ~ ~ I ~ . ! \~ _. _i, _.;___. aim\ \ 25 - ,,""- ~10~/ ~_20 :;17 .~ 23~ ~~ ~2 · 1 ~ ~ ~ - 24 ' ~ FIGURE 5.9 Spatial distribution of annual wet deposition of sulfate ion (millimoles per square meter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Barrie and Hales (1984).
152 , ~ . ~05 , .0 9, ,,. ~~~\ ~ 00~ ·70';;"~'-2"""' ;~ 1;~ ~ ^ i-35 1 ( '' -96 .^,'' '' it\ in' ~\ .50 '1, =_s,0 'a ~ .. . . . , . .75 ', :~ ~ ' I ·` . 40 `` ~ ~ ' i_i,'-'<.-;----l' \ 3 9 . , ~_' ·~9 '',( 19 ~ . ~ ~1\ ~ ~ '~ FIGURE 5.10 Spatial distribution of annual wet deposition of nitrate ion (millimoles per square meter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles) and APN (squares); United States, NADP (circles) and MAP3S (squares). SOURCE: Barrie and Hales (1984).
153 ,,.' is ' / al L , , . , , 1 ,, 1 ~ . 3 3_^ \ 54~ U-' 7~ - 6.8 ' I,. ~~ _ _` Z] .66 10~/~° 'a, ~ ~ EMU- _, 27'_-;~ ~/~ i' 16~;~'_"'~ ______- I-- too_' ;2 I · ,. _.-~--'' '` ~ ' 5 1~-~1o '',,,~8.8~~, ~ \^ '%.] ~ 68 FIGURE 5.11 Spatial distribution of annual wet deposition of ammonium ion (millimoles per square meter) in North America in 1980. Data from four networks are plotted: Canada, CANSAP (circles ) and APN (squares); United States, NAPD (circles) and MAP3S (squares). SOURCE: Barrie and Hales (1984).
154 TABLE 5.8 Annual Mean Precipitation Amount (P) and Concentration of Major Ions (~eq/L) at Three Sites in the MAP3S Network in the Northeastern United States from 1977 to 1981 Ion p Year (cm) H+ SO2- NO3 NH4 Site: Whiteface Mountain, New York (73°51'W, 44°23'N) 1977 1 16 46 22 22 1 1 1978 80 60 29 26 17 1979 107 50 20 19 1 1 1980 95 61 27 26 17 1981 124 37 21 18 12 D(%)a 9 20 25 17 25 Site: Ithaca, New York (76°43'W, 42°23'N) 1977 99 69 30 29 13 1978 78 84 34 31 16 1979 102 70 28 25 15 1980 91 67 30 30 17 1981 124 70 37 31 17 D(%) - 8 - 7 - 6 3 9 Site: Pennsylvania State University (77°57'W, 40°47'N) 1977 105 73 33 3 1 15 1978 94 78 21 31 13 1979 132 58 26 26 15 1980 88 75 33 3 1 16 1981 104 59 35 30 25 D(%) - 16 - 9 4 -4 aD is the deviation of 1980 observations from the 5-year mean. SOURCE: Barrie and Hales (1984).
155 soor ,0.0 60.0 50.0 40.0 30.0 20.0 0.0 o.o 36.0 J 30.0 5 3 26.0 z ~ 20.0 1 - 16.0 0.0 no o.o oo.o 70.0 - ~ oo.o Z so.o o ~ ·o.o Z 30.0 o 20.0 Cat 0. o. H. L ~ I I l SC PF KA Al BD 1 WH IT PS CV IL -I -~m SC PF KA Al BD NOT - ~ m: SO4 WH IT PS CV IL 1 l SC PF KA Al BD WH IT PS CV IL REMOTE AREAS EASTERN UNITED STATES FIGURE 5.12 Comparison of the volume-weighted-mean concentration and the upper and lower error estimates among five sites in remote areas and five sites in the eastern United States. Remote sites: San Carlos, Venezuela (SC); Poker Flat, Alaska (PF); Katherine, Australia (KA); Amsterdam Island, Indian Ocean (AI); and Bermuda, Atlantic Ocean (BD). Eastern United States sites: Whiteface Mountain, New York (WH); Ithaca, New York (IT); State College, Pennsylvania (PS); Charlottesville, Virginia (CV); and Champaign, Illinois (IL). SOURCE: Galloway et al. (1984).
/ 156 / / -I 2 ~ ''. .., : .... :' ' .... Coo '" ' ''''''''''''''l ___ -''"'''I '.'...:'....''''12 i',, ~ '"'".':::'::: '''' ' '.,_. , . . ~ flow FIGURE 5.13 Enrichment factors for sulfate concentra- tions in North American precipitation. Numbers signify the ratio of sulfate concentration in precipitation in eastern North America to sulfate concentration in precipitation in remote areas of the world. SOURCE: Galloway et al. (1984).
157 well with hydrogen ion in liquid precipitation but did in snow and frozen precipitation. Bowersox and Stensland (1981) analyzed NADP data for seasonal variations in sulfate, nitrate, and hydrogen-ion concentrations. The sampling sites were grouped into five regions in the eastern United States, each contain- ing from two to seven sites. The data for each region were averaged for the cold season (November to March) and the warm season (May to September). The resulting warm- to-cold period ratios for sulfate varied from a maximum of about 2.0 in the New England region to a minimum of 1.25 in the Illinois and Michigan regions. In terms of seasonal (three-month interval) trends, maximum sulfate concentrations occurred in the summer. The investigators noted that aerosol sulfate has a similar seasonal varia- tion but that sulfur oxide emissions for the Northeast have a relatively small seasonal variation. For nitrate, Bowersox and Stensland (1981) found a maximum warm-to- cold period ratio of 1.5 for the region in the Southeast and a minimum ratio of 0.9 for the Michigan region. Thus, the nitrate and sulfate seasonality ratios had different geographical patterns. The free acidity of the precipitation was greater in the warm period for all the regions and reflected the mixture of the patterns for sulfate and nitrate. The free-acidity ratio of warm-to- cold season values was a maximum or '.~ ror one Pennsylvania region and a minimum of 1.2 for the Illinois region. The seasonal pattern of precipitation sulfate concen- tration is different for western Europe from that for the eastern United States. Granat (1978) averaged the data for many European sites and reported a maximum sulfate concentration in the spring, that was 1.6 times greater than the minimum value observed in the fall. The sulfur emissions in the region are at maximum in the winter (Otter 1978). Comparing Global Background and Eastern North American Precipitation Composition Comparing precipitation composition and wet deposition in heavily populated or industrialized regions with those in remote regions of the world shows clearly the higher levels caused by anthropogenic emissions. The elevated concentrations of acid-related species in precipitation in eastern North America, Europe, and Japan attest to this.
158 The results of Barrie and Hales (1984) show that within North America the concentrations of hydrogen, sulfate, and nitrate ions in the Northeast may exceed the con- tinental background levels found in the West and the North by some 20, 5, and 10 times, respectively. On the global scale, Galloway et al. (1984) have compared data from five sites in remote areas of the world--San Carlos, Venezuela; Poker Flat, Alaska; Katherine, Australia; Amsterdam Island, Indian Ocean; Bermuda, Atlantic Ocean-- with those from five wet-only sites in the eastern United States--Whiteface Mountain, New York; Ithaca, New York; State College, Pennsylvania; Charlottesville, Virginia; and Champaign, Illinois (Figure 5.12). Values are precipitation-weighted means of 14 to 125 samples from remote locations and 186 to 373 samples from eastern North America. Figure 5.12 shows that hydrogen-ion concentrations at remote locations are significantly lower than those in eastern North America and that remote sulfate and nitrate concentrations are consistently lower by more than a factor of 5. To illustrate the enrichment of North American concentrations of sulfate over global background, Figure 5.13 (Galloway et al. 1984) shows the ratio of 1980 concentrations in North America to global background concentrations. Although up to a fivefold enrichment of hydrogen ion occurs, Galloway et al. (1984) point out that such a comparison is somewhat misleading because the hydrogen- ion concentration in remote areas may be determined by an entirely different mix of substances, primarily organic acids, from that in eastern North America, where strong mineral acids are the main contributors. For nitrate and sulfate, variability from the different remote sites is small, and the North American levels are from 2 to 10 and 2 to 16 times greater, respectively. From the beginning of the acid rain debate, precipita- tion pH has been used as the central variable, or indicator, of the phenomenon. AS Galloway et al. (1984) showed, this has not been a particularly useful choice. At remote continental locations, organic acids may lower the pH to levels below 5.0 (Galloway et al. 1982). In addition, Charlson and Rodhe (1982) have shown that "clean" rain may contain naturally occurring acids (notably sulfuric acid in the natural portion of the sulfur cycle) in high enough concentrations, in the absence of basic materials, to produce pH values in the 4.5 to 5.6 range. Thus, pH values less than 5.0 may be found in remote areas in the absence of anthropogenic
159 sulfur and nitrogen. In such circumstances, the concentrations of sulfur and nitrogen are much lower than in industrialized areas. Comparing precipitation concentrations of acid species in remote and industrialized regions of the world shows the effect of anthropogenic emissions in the latter. In the absence of lengthy time-trend records, concentrations in remote areas serve to establish baseline values in regions currently affected. T IMETREND I NFORMAT ION Data Available The optimum data set for analyzing trends would consist of a long unbroken series of measurements, made using the same methods, at sites known to be represen- tative of the phenomenon being examined. Unfortunately, no precipitation chemistry data set exists in North America that has all these attributes. The desired record of data would be long enough to cover the time period in which the phenomenon of interest occurred and extend some time before and after. These extended time periods would help to keep erroneous conclusions from being drawn based on shorter-term trends or drifts in the data. Changes in protocols or methods of data collection and analysis introduce uncertainties into the record, unless the changes have been thoroughly documented and there has been a period of overlap when both old and new procedures have been used. With continuous records available from so few locations, it is tempting to assume that trends in such locations are indicative of what has happened over a larger area. This may or may not be the case. Hidy et al. (1984) discuss the problem of site representativeness in this context in their analysis of the USGS data. (See the section below on trends for Hubbard Brook and USGS bulk data.) Figure 5.14 summarizes the types of information available on which an evaluation of trends must be made. Before Junge's measurements in 1955-56, only sporadic measurements were made, beginning about 1910. These have been reviewed by Hidy et al. (1984) and Cogbill et al. (1984) and are discussed below. It is possible to compare the various "snapshots in time" provided by measurements by Jung e in the 1950s and
160 1920 1940 1 960 1980 a) Pre-1955 Dataa b) Junge, PHS/NCAR "Now" Snapshots c) 1964-Present Continuous Recordsb'C Hubbard Brook USGS WMO/BAPMoN d) Recent Datab'C CANSAP/APN NADP MAP3S EPR l-UAPSP aDashed line indicates period of sporadic measurements bDotted lines indicate that measurements are continuing. CSolid lines indicate time period over which data were available to authors of this report. FIGURE 5.14 Summary of available precipitation chemistry data bases. by the Public Health Service/National Center for Atmospheric Research (PHS/NCAR) in the 1960s with present-day measurements. Beginning in 1963, a continuous and valuable series of measurements has been made at Hubbard Brook, New Hampshire. Soon afterward the USGS network was established, and in the early 1970s the World Meteorological Organization (WMO) network was established in cooperation with the governments of the United States and Canada as part of a more extensive global program. (This network is sometimes referred to as WHO/NOAA/EPA in the United States after the National Oceanic and Atmospheric Administration and the U.S. Environmental Protection Agency, the two agencies responsible for administering the monitoring program). More recently, since approximately 1978, several extensive sampling networks have come into operation. NADP and CANSAP/APN (Canadian Network for Sampling Precipitation/ Air and Precipitation Monitoring Network) are discussed in subsequent sections. A detailed description of the MAP3S and EPRI-UAPSP (Electric Power Research Institute- Utility Acid Precipitation Sampling Program) is presented in Wisniewski and Kinsman (1982). They also discuss several other monitoring activities in North America not covered in this chapter.
161 Pre-1955 Data Most precipitation chemistry data available before approximately 1955 were obtained by bulk sampling. In most cases, measurements were taken to characterize nutrient input to agricultural soils for purposes of soil management. The reliability of many of these data is questionable because not only were they based on bulk samples but there were substantial differences in collector materials, sampler designs, chemical analysis methods, and site location. Hidy et al. (1984) have reviewed many of these historical data and have tabulated times, locations, and methods of collection. Figure 5.15, adapted from Hidy et al. (1984), shows sulfate concentrations in New York and Connecticut derived from bulk samples for the period 1910 to 1954. The data were from stations that used stainless steel, aluminum, or glass samplers located outside towns or cities or that reported the lowest deposition in an area for a given year. (Hidy et al. assumed that areas of lowest deposition provided the cleanest samples since they were less likely to be contaminated by dust or other ~mpurz~zes.) These data provide a qualitative picture of the levels encountered during the first half of the present century. In the Northeast, ion concentration and deposition are relatively constant over the period: annual concentration averages for sulfate are in the range of 10 to 40 mg/L, and deposition ranges from 20 to 80 kg ha~1 yr~l. The few available data for the Midwest are comparable to those for the Northeast, whereas those for the Southeast show lower concentrations and depositions. AS we will show below, concentration and deposition values after 1955 are generally much lower. We cannot draw conclusions about the cause of the higher values before 1950; they may be the result of sampling and analysis practices, siting, higher ambient concentrations, or some combination thereof. These data are not indicative of regional wet-only concentration patterns and cannot be compared with modern wet-only data to establish time trends. Trends from Mid-1950s to Present As shown in Figure 5.14, data from four relatively brief periods of time have been used to investigate changes in precipitation composition, primarily pH, in
162 cat o 4 - _ 0 ~ ° m 30 20 10 o 80 60 40 20 o 60 40 20 o 40 20 o 40 20 O ~~ ~ _ _ ~ i. I T111TRI1 1 1 1 1 L ,~ 15 20 25 30 35 40 45 50 55 YEAR FIGURE 5.15 Sulfate concentration (milligrams per liter) in bulk precipitation samples for locations in the northeastern United States. The width of the bar designates the period of sampling. Adapted from Hidy et al. (1984). Data sources are listed in Table 1 of Hidy et al.
163 the period since the mid-1950s. The major sources of these data are Junge and Werby (1958) for the mid-1950s, Lodge et al. (1968) (also referred to as the PHS/NCAR data base) for the early 1960s, a variety of sources for the mid-1970s (see Likens and Butler 1981), and the NADP , and CANSAP networks In recent years. Analysls ot these data sets and the conclusions drawn from them have given rise to considerable debate in the published scientific literature over the past few years. On the basis of their analyses Likens and Butler (1981) and Cogbill et al. (1984) concluded that a large spread and probable intensification of acid precipitation (pH < 5.6) in eastern North America have occurred during the past 25 years and that acid precipitation may not have been widespread in the eastern United States before 1930. More specifically, they concluded that precipitation has become more acidic, especially in the Midwest and Southeast, between the 1950s and the present and that many areas of the East show an increase in acidity of greater than 22 peq/L during the past three decades. The major points of concern about using these data for trend analysis involve (1) the amount of uncertainty and bias introduced by differences in sampling, analytical, and calculational methods and (2) the influence of differing climatological conditions, particularly drought, during the periods of data collection. Hansen and Hidy (1982) reviewed the available post- 1955 data and assessed the biases and uncertainties associated with (1) use of different sampling, analytical, and calculational methods to deduce precipitation acidity, (2) use of data from different times and conditions, (3) failure to collect brief or light rainfall and initial precipitation in an event, (4) incomplete chemical analyses, and (5) selection of data for analysis. The outcome of this assessment was to determine a series of uncertainty values in pH by decade and geographical region. Cogbill et al. (1984) dispute several of these estimates. Hansen and Hidy (1982) conclude as follows: " m e results show that the historical data are of insufficient quality and quantity to support any long- term trends in precipitation acidity change in the eastern United States. n Somewhat contradictorily, ~~w~v=~ r Hey u~ ~ ~ Mew c'~u ~~ . . . it now seems that the precipitation acidity on the average, was lower in the mid-1950s than in the mid-1960s and thereafter in parts of the eastern United States"; they also conclude that ~ ~ ~ ~ ~ _ _ ~ 1_ _ ~ ll
164 despite limited observations in the Southeast it appears that a region of elevated acidity (pa ~ 5.0) in rainfall from Alabama through Florida exists that was not evident in the mid-1950s. Stensland and Semonin (1982) evaluated the potential role of elevated levels of calcium and magnesium from crustal sources in the composition of precipitation in the mid-1950s. Drought conditions in the mid-1950s are likely to have caused increased concentrations of wind- raised soil components in precipitation samples, possibly resulting in calculated acidity values lower than would have been typical of the period. Stensland and Semonin (1984) also point out that data from the 1960s (PHS/NCAR) and the 1970s (WMO/NOAA/EPA) show generally higher ion concentrations than do current NADP data. Figures 5.16 and 5.17 show a comparison between older and more recent data for calcium and sulfate ions. The analysis suggests that the earlier measurements--1950s, 1960s, 1970s--were perhaps affected by evaporation and dry deposition (dust leakage) and may have been more like bulk than wet-only samples in many cases. On the basis of a simple pH model Stensland and Semonin (1982) recalculated the 1955-1956 pH distribution and found a smaller pH change between the mid-1950s and the present than do the calculations of Likens and Butler (1981) (see Chapter 3, Figure 3.4). Stensland and Semonin (1982) point out the crucial importance of (1) measuring for all major ions in samples to be used for trend analysis and (2 ) analyzing con- currently meteorological/ climatological factors when investigating large-scale temporal and spatial changes. Investigators have not reached a consensus on the subject of precipitation acidity changes since the 1950s; the quality and the quantity of the earlier data make a quantitative evaluation of the change difficult. Never- theless, workers agree that precipitation acidity has increased since the 1950s in parts of the eastern United States despite disagreement on the amount of the change and the mechanism. More specifically, there is agreement that the acidity of precipitation has increased in the Southeast (e.g., Barrie and Hales 1984) since the 1950s. However the reason for the change has not been thoroughly explored. A comparison with the Jung e data shows that the concentrations of sulfate in the southeastern states have generally increased since the 1950s (see Figure 5.18). A similar trend is indicated for nitrate and hydrogen ions (Barrie and Hales 1984). However, concentrations of calcium and calculated magnesium have
65 0.39 kit to. 1 ~56.34\) ~-15 i- . :89 two. ~_0.55 SEX o.~O.4~) ^2~'~ ~0~ ~ °QI A\7 Ca(mgL~l) I~ {~ . \ ( it_ 0.38 .L6\~ ~~K'~ · NADP O PHS/NCA R /\ WMO/NOAA/EPA 3.72 ~ ~ ~ -1 ~~ O ' ~ '0.56 5~. \ A o ~.52: o ~ FIGURE 5.16 Weighted average calcium concentration in precipitation (milligrams per liter). The computer- generated contours are for the NADP wet deposition data for 1978-1981 for sites with at least 20 valid samples. The other sites on the map with numerical data show the calcium concentrations determined in earlier years by PHS/NCAR (1960-1966) (bold numbers) and WMO/NOAA/EPA (1972-1973) (italic numbers). SOURCE: Stensland and Semonin (1984). decreased significantly. We do not know the relative importance of strong acids contributing hydrogen ions versus decreased soil components resulting in less neutralization. For the northeastern states, Figure 5.18 indicates that sulfate concentrations have risen in some areas (Massachusetts, New York, Ohio, and Indiana) although generally not to the same degree as the Southeast. The other northeastern states either exhibit no change (Pennsylvania and Michigan) or show decreases (Maine, Massachusetts, Vermont, and Maryland). The very large negative differences in sulfate concentration (Figure 5.18) at many sites in the Great Plains and the West raise the question of comparability of the 1955-1956 sulfate data with the recent wet-only
166 A JO ~ ~ 35 ~ ~ 447.73 ~y4051~ \' ~,,0~>,,~ '2.43 51~35 ( ~ ~~ ~0 ~ / ·30; jal it W~0~ 3.47 4. 71 .*,+-0 1- ~ / ~ ^.72 7:j 0~ SO4 ( mg L~ 1 ) · NADP O PHS/NCAR WMO/NOAA/EPA `- ~ ~ ..w V ~ _ ~ ~ 1 1 7~ ~~ ~~ . 2.29 FIGURE 5.17 Weighted-average sulfate concentration in precipitation (milligrams per liter). The computer- generated contours are for the NADP wet deposition data for 1978-1981 for sites with at least 20 valid samples. The other sites on the map with numerical data show the sulfate concentrations determined in earlier years by PHS/NCAR (1960-1966) (bold numbers) and WMO/NOAA/EPA (1972-1973) (italic numbers). SOURCE: Stensland and Semonin (1984). data. It seems likely that dry deposition and evapora- tion may have biased the 1950s sulfate data upward for many western sites. The need to have a continuing data base with which to evaluate trends is clearly evident. Although the need for quality-assured data is widespread, such data are crucial in the case of trend analysis since the signal-to- noise ratio may be small. In addition to the particular species of concern in trend analysis of acid deposition, we need to examine other atmospheric chemical components as well as meteorological/climatological parameters.
167 ~\~.3 -9.0 ~0 FIGURE 5.18 Sulfate differences (milligrams per liter) determined by subtracting 1955/1956 precipitation sulfate concentrations (dung e 1958) from NADP sulfate concentrations. Trends for Hubbard Brook and USGS Bulk Data The Hubbard Brook Data In this section we describe a trend analysis undertaken on the Hubbard Brook weekly bulk-precipitation chemistry data (Oehlert 1984a). The data consist of analyses of the more than 500 bulk samples collected between 1964 and 1979. The trend analysis was done for the following ions: hydrogen, sulfate, nitrate, chloride, ammonium, calcium, magnesium, potassium, and sodium; chloride data were not available until June 1967. Data were made available through the courtesy of G. Likens and J. Eaton (Cornell University, personal communication, 1984). Monthly volume-weighted average concentrations and monthly depositions were calculated from the raw data. We analyzed the data for chloride, calcium, magnesium, potassium, and sodium on the logarithmic scale. Seasonal influences were eliminated from the variables by subtracting from each monthly value the mean of the time series for that month.
168 TABLE 5.9 Precipitation-Chemistry Trends at Hubbard Brook Observed Data Smoothed Dataa Ion Scale Trend sob tc Trend seb tc Calicium percent/year - 9.49 1.06 - 8.95 - 9.28 1.07 - 8.67 Magnesium percent/year 6.96 1.94 - 3.59 - 7.13 1.84 - 3.87 Potassium percent/year - 7.01 1.87 - 3.75 - 6.48 1.61 - 4.02 Sodium percent/year -6.25 1.39 - 4.50 -5.82 1.18 -4.93 Chloride percent/year 9.19 4.20 - 2.19 - 8.95 3.34 - 2.68 Sulfate ppm year 0.0383 0.0159 -2.41 -0.0467 0.0147 -3.18 Ammonium ppm/year 0.00554 0.00236 - 2.35 - 0.00651 0.00247 - 2.63 Nitrate ppm year 0.0381 0.0161 2.37 0.0367 0.0152 2.41 pH pH/year 0.00143 0.00259 0.55 0.00204 0.00256 0.80 aTo prevent undue weighting of outliers on this analysis, a rough outlier resistant method was used. Data were first smoothed using a nonlinear data smoother, which smoothed single outliers more than the rest of the data. Smoothed data are then used in the trend regression. If smoothed and unsmoothed results are similar, the unsmoothed result is preferable; if they are dissimilar, the smoothed result is preferable. bse is the standard error. ct is the trend divided by the standard error. SOURCE: Oehlert (1984a). The statistical methods for the trend analysis are those described in Oehlert (1984b). Basically, trends are determined by ordinary least-squares regression, but the standard errors of the trends are adjusted for the presence of autocorrelation in the residuals. Nominally, a t value greater than 2 in absolute value would be considered statistically significant. Here the cut-off value of 2 is only approximate. The linear trends observed in the data over the period of record are shown in Table 5.9. The results show that some strong trends exist in the Hubbard Brook data. In particular, concentrations of all ions but hydrogen and nitrate are decreasing. pH has remained fairly constant. Nitrate has increased significantly over the period 1964 to 1979, although the result is somewhat dependent on statistical technique and the time period of analysis. Inclusion of data more recent than 1979 tends to indicate no change in nitrate concentrations over the entire time span. Figure 5.19 shows the deseasonalized monthly values for pH, sulfate, nitrate, and calcium. One feature of the sulfate trend is that it varies seasonally. With summer defined as June, July, and August, and the other seasons defined accordingly, the trends for summer, fall, winter, and spring are -0.025, -0.023 -0.061 and -0.055 ppm/yr. Thus, spring and
169 . . . . . . ~ ~ ~ Cal Add _ ~ ~ . . . . . . a, Hd o. ~ a, o a, i o <: ~ LU U) Cal o. U) o lo, INCA O [ 1 {V) N c.' Wdd _ o , ~ ~: o r°~ ~ , - o o U] o .,' ( - o 'S N ul ·,1 , - o U] a U] O ~, m U~ Q .Q H ~ :~: ^ £ · - ~, - o ·,, o .,, · - u, o ,' U] a .rl o ~n - ~r - S _ aJ Q O - . :t ~ _ ~ ~ O U) o ·,, o
170 TABLE 5.10 Trends in Ratios of Sulfate, Nitrogen Species, and Total Nitrogen at Hubbard Brook Observed Data Smoothed Data Ion Scale Trend se t Trend se t SO2-/NO3 percent/year6.19 1.62 - 3.83 - 5.81 1.42 - 4.10 NH4 /NO3 percent/year7.52 1.41 - 5.34 - 7.52 1.34 - 5.62 Total N ppm year 0.00431 0.00473 0.91 0.00301 0.00467 0.64 SO2-/total N ppm year 3.24 1.23 - 2.64 - 3.12 1.05 - 2.98 SOURCE: Oehlert (1984a). winter concentrations have been falling more rapidly than the summer and fall concentrations. Ratios of species' concentrations in the data have been used to analyze atmospheric processes (National Research Council 1983) and trends. The data indicate that the ratio of sulfate to nitrate concentrations in Hubbard Brook data has been declining at about 6 percent per year, while the ammonium to nitrate ratio has been declining by about 7.5 percent per year. There has been, however, no overall trend in total nitrogen concentration (nitrate plus ammonium). There is still a decreasing trend in the ratio of sulfate to total nitrogen caused by the decreasing trend in sulfate. Table 5.10 summarizes these trends. We are aware of no plausible explanation for the decreasing trends in calcium-, magnesium-, potassium-, sodium-, and chloride-ion concentrations. The trends are not related to systematic bias introduced by changes in sampling procedures since these methods have not changed over the period of record (G. Likens, Institute of Ecosystem Studies, New York Botanical Garden, personal communication, 1985). That the ions exhibit this change would suggest changes in local source strengths since these substances are usually found in particulate matter that is not transported large distances from sources. As one possible scenario, the decrease in sulfate (about 20 peq/L for the 18-year period) and the decrease in calcium plus magnesium (about 18 peq/L) could result, for example, from a decrease in the input of calcium sulfate, either as dry deposition (sulfur dioxide plus calcium on the funnel surface) or as wet deposition (calcium sulfate in precipitation). Then the portion of the sulfate transported from distant sources would be much less. We have no way to interpret the sulfate trend unequivocally.
171 Several authors using different statistical methods have examined the Hubbard Brook precipitation chemistry data for trends. Table 5.11 summarizes the methods and results from recent publications on trend analysis of the Hubbard Brook data. A summary of the trends since 1964 follows: Hydrogen ion: Record is variable with no significant trend; slight tendency to lower concentrations for parts of the period. Sulfate: Significant decrease of approximately 2~/yr superimposed upon periods of drift; decrease in winter and spring is greater than for summer and fall. Nitrate: Trend over entire record is variously described as "not significant" (1963-1982) and "increasing n (1964-1979 ); an apparent change in slope in about 1971 may be from chance. Ammonium: Three authors describe the trend as "no change n and two others say "decreasing (1.9 and 6.4~/yr)" Sodium, chloride, calcium, magnesium, potassium aluminum: Agreement among three authors of a strong decrease in all these constituents. ratios: Sulfate/nitrate, ammonium/nitrate, and sulfate/hydrogen show a consistent, significant decrease; nitrate/hydrogen shows a significant increase . Several features of this review stand out. Results from the various authors are generally consistent for hydrogen; sulfate; the group including sodium, chloride, calcium, magnesium, potassium; and aluminum and their ratios. For nitrate and aluminum there is qualitative agreement on the behavior of the time-series segments but some disagreement on the trend in the overall record. The different statistical techniques may introduce some differences in the results, but the length of record used in different studies is more important. The end point of the time series varies among studies between 1977 and 1982. In addition, several authors have noted the differences in trends that may occur with an analysis of selected portions of the time series. Finally, the differences in estimated rates of change obtained in the various analyses point to the need for caution if quantitative comparisons are to be made between trends in precipitation composition and, for example, emissions.
172 _ o ._ _ Ct o Ct~ An, _ ~ =~ o Ct ~ s UO C) ~ m S: o ._ ._ U. o o I: o ._ Cot ._ ._ o He m Ct I: ._ U. o V) m . . o V: ~ ° ~3 ~ b ~ ·Q (: (,) <0 en o . ~ 3 ~ ~ . ~ .- - a, a c ~ C ~ ~ , ~ ~ o ° C C ~ ~ C C . , ~ Hi. , is ~ .-. /C ~ ~ ~ ~ ~ · ~ a. · · · ~ · - · · · ~ · - - -
173 .. .. O O O ~ v: v: 0 ~ 1 1 I t~ ~ ~ u E c E _ Y 3 E E A C + we E ° I E u ~ ~ e _ a e ~ E ~ ° D 'O ~ Z 3 CQ ~ ~ ° O ON ~ ad ~ ~ 5 en: ~~ c.> By 0 iO TIC c ~ + V of c of 0 ~ e e Z e e ~ e e e e _Q e e e
174 o ._ Ct - ~: C) Ct ~ At Cal o Ct ~ Ct ~ m :S ._ it: o m as via oo ~ . . . . ~ - == 1 1 1 1 .... .... += 1 + + ~ _ ~ ~ Zen o O to C) =: S: Cal ._ ~ ~ do C.) 3 .C~ ~ o .^ so On ~ Ct V ~ 3 to Ct. =- C C,) so ~ ~ _ c ° 3 ° 'a ~ ~ C,: += 1 U3 Z ~ ~ ~ to O~ C;, Ct Cd o o oo ~ U' _ ~ ~ ~ ~ o ~ o c~ ~ . (', 1 1 1 1 1 1 1 1 .. .. .. .. .. .. .. .. .. 1 m+= 1 + + + + 1 ~ 1 ~ OZ Z ~ Z ~ ~Ce mb0Z O 1 += o ~ v) bO U3 .^ .^ ~ ~ . ~ ~ ~ ~ ~ ~ ~ ~ li~ l~ ~= i~ -o c,3 ~- i ~ c c om+~= ~ + + +~d m=o== · · · · · · · · ~
175 Changes in sulfate deposition at a site are caused by changes in natural and anthropogenic emissions of sulfur compounds and by changes in the transport, conversion, and deposition processes that deliver the sulfur to the receptor. As a rule, a source close to a receptor will affect the receptor more strongly than a source of the same size farther away. Thus, changes in local source strengths are a potential concern when trying to infer regional trends from site data. Recently, Coffey (1984) speculated about the possible influence of a local paper mill on the Hubbard Brook precipitation-sulfate data, although data on actual emissions from the plant were lacking. John Pinkerton (National Council of the Paper Industry of Air and Stream Improvement, personal communication, 1984) estimated sulfur dioxide emissions from the paper mill from 1963 to 1980 based on available information on strikes and closings at the plant, pulp and paper production rates, energy consumption, and knowledge of the boilers and processes involved. No direct measures of emissions were made at the plant. The sulfur dioxide emissions from the plant were roughly constant at about 4200 tons/yr from 1963 until a shutdown of the plant in August of 1970. After that, the plant operated sporadically with emissions during operational periods at about one third of the pre-1970 rate (Figure 5.20). The longest period of total shutdown was from July 1976 to November 1978, during which there were no sulfur dioxide emissions from the mill. No large step changes of the type illustrated in Figure 5.20 are evident in the monthly data for sulfate concentration in precipitation at Hubbard Brook. Quali- tatively, the Hubbard Brook sulfate data show short-term increases and decreases superimposed upon a linear decrease from 1964 through 1980. The large step decrease in emissions in 1970 occurs at a time when the Hubbard Brook sulfate record is at the end of a short-term (4-year) decrease and the beginning of a short-term (3-year) increase. Quantitatively, linear regression of the deseasonalized monthly precipitation sulfate data on the deseasonalized emissions data from the mill is not statistically significant, having a t value of 1.22, while the regression of deseasonalized precipitation sulfate on a time trend has a t value of -2.41. Thus, we would accept the null hypothesis of no association between sulfate in precipitation at Hubbard Brook and sulfur dioxide emission from the paper mill, while we
176 \,: L il 1 1 1 o o o HV3A ~ ~d OS J0 SN01 _ oo _ oo O ~ ~ -- C~ cs) I_ , , ~ ~Q 6 ~ O U] U2 ,! a s" C9 ~ U] ~5 a ~ 3 ~q E~ CO CD ~3 z H _ CD O ~ S~ a) a Sh~ ~ U2 a) ~_ . d. ~0 O .,' ~ a) ~ 0 O ) ~, - Q,l 0- ~ O Z _1 O O ~ U] S~ $_~ S~ 0 ~ a y: ·,1 ~ . a. a · O H o S" _ S~ U) 1 S~ O .,1 ~ S · U] Q. $ 3 Z; . U] o ·,, U2 U] .,, ·,' a) Q 3 o o U2 o ·,, a) U2 s~ ~l o s~ u] u] H ~1
177 would reject the null hypothesis of no linear trend. A linear time trend is a better predictor than the estimated emissions from the paper mill. In the general case, the deposition at Hubbard Brook is likely to be the result of a combination of regional and local influences, so it is reasonable to predict the sulfate concentrations using linear regression on both time and emissions from the mill. Using both variables, the t value for a linear effect over time is -3.62 and the t value for the emissions effect is -2.17. Thus, allowing for some regional linear trend, any effect of the shape of the paper mill emissions curve is statistically estimated to be significantly negative. From this nonphysical result we conclude that the trend present in the Hubbard Brook sulfate data is essentially uninfluenced by the paper mill emissions. The U.S. Geological Survey Data The USGS network was established in 1965 to provide data and continuous records on atmospheric contributions for use in hydrologic and geochemical studies (Peters et al. 1982). The network comprises nine stations, eight in New York and one in northern Pennsylvania; six sites are inland rural, two are urban, and two are coastal (one urban/coastal). Samples are monthly bulk samples. Because this data set provides one of the longest continuous records of precipitation composition and deposition in North America, it has been examined by several authors. Virtually all these authors have commented on problems associated with the data (in particular, see Miller and Everett 1981, Peters et al. 1982, Miles and Yost 1982, Bilonick and Nichols 1983, Hidy et al. 1984, Bilonick 1984, Cogbill et al. 1984, Peters 1984). Potential problems include changes in sites, collectors, analytical methods, and laboratories. Authors differ in their assessments of the importance and implications of these factors for trend analysis; however, the various analyses have been instructive and have produced some valuable trend information. A time-trend regression analysis by Miller and Everett (1981) for pH data showed no significant trend for any of the nine sites but a slight downward trend for the pooled data. Analysis of the nitrate data showed slight increases for individual stations and a greater increase for pooled data.
178 TABLE 5.12 Time Trends for Five New York Sites in the USGS Bulk Data Set %per year Ion Site a b c SO2- Canton - 0.7 - 1 .6Ci - 1 .2 Hinckley - 1.80 - 3.4e - 2.6f Mays Point 0.8 - 2.6e - 2.0 Salamanca - 0.8 - 2.7Ci 0.8 Rock Hill 0.4 - 2.00 1. l NO3 Canton 2.3 7.60 9.4 (after April, 1969) Hinckley 4.5e lO.9f 13.9f Mays Point 2.4 6.3e 5.0f Salamanca - 3.7 - 1.7 - 2.5 Rock Hill 2.5 3.0 1 .5 NH4 Canton 2.3f 2.2 4.6f Hinckley 5.2e 3.6e 5 Se Mays Point 7.6e 1.0 6.0e Salamanca 2.9 - 3.7e 0.2 Rock Hill 6.1'l 3.4~ 4.4 Sum of Ca2+, Canton 2.90 2.2 3.70 Mg2+, Na+, and K+ Hinckley 4.7f 3.6e - 3.70 Mays Point 0.8 - 1.0 -1.0 Salamanca - 2.7 - 3.7e - 2.3 Rock Hill 6.5f 3.4~ 6.1f H+ Canton 1.2 - 6. 8e _ 0.5 Hinckley 3.4f 8.4e 6.7f Mays Point 4.5e 4.3 1O.Of Salamanca 8.6f 10.2e 8.7f Rock Hill 2.6 8.9f 6.3f aTrend by simple linear regression between measured ion concentration and date of collection. bTrend by linear regression between monthly normalized ion concentrations and date of collection. CTrend by seasonal Kendall's test. ~p <0.10. ep < 0.05. fp < 0.01. SOURCE: Ba~nes et al. (1982). Barnes et al. (1982) analyzed the USGS data for time trends at all nine sites with monthly bulk data since 1965. Results for five of the sites are shown in Table 5.12; data for the other four sites are not included as they are viewed as less likely to be regionally repre- sentative; i.e., urban sites (Albany and Mineola, New
179 York), coastal sites (Upton and Mineola, New York), and sites affected by local sources. The site at Athens, Pennsylvania, was characterized by high levels of nitrate and ammonium, which were perhaps due to a nearby dairy farm. The time-trend values were calculated with three different methods as described in Table 5.12. The time trend by simple linear regression is commonly used, while the seasonal Kendall test is described as the most robust of the three. The data in Table 5.12 show that the mag- nitude of a time trend varies by as much as threefold depending on which of the three methods is used (for example, note the figures on nitrate and hydrogen ions at Hinckley, New York, as extreme cases). The time trend can change sign as the method is changed (for example, . . . ~ . . note those tor ammonium at H,nckley and hydrogen ion at Rock Hill). Also, the magnitude of the time trend can change by more than a factor of 2 among the five sites for the same method (for example, note the change in sulfate for method (b)). The data in Table 5.12 suggest that quantitative time-trend results from the USGS bulk data are subject to imprecision. Qualitatively, the most consistent result from Table 5.12 is that sulfate has decreased by about 1 to 3 percent/yr at the five nonurban, noncoastal sites. Nitrate increases by 4 to 14 percent/yr for significant values (p < 0.10) and decreases by up to 4 percent/yr for some nonsignificant values. The significant values of ammonium-ion trends vary from -4 to +8 percent. Similarly, values of cation- sum (calcium plus magnesium plus sodium plus potassium) trends vary from -5 to +7 percent, and the significant hydrogen ion trends vary from -9 to +10 percent. The large variability in trends observed at the five USGS bulk sampling sites may be caused by local effects masking the regional patterns. Bilonick and Nichols (1983) applied the Box-Jenkins time series analysis method to "cleaned," pooled, monthly deposition data for 1965-1979. They found that the long-term mean levels of hydrogen, sulfate, nitrate, and calcium ions in bulk samples were essentially constant. Hidy et al. (1984) attempted to establish the reliabil- ity of USGS bulk data by comparing time series from the Hinckley and Mays Point stations with the event data from the MAP3S stations at Whiteface Mountain, New York and Ithaca, New York, respectively. They concluded that although the USGS data displayed a systematic positive bias in sulfate concentrations on a monthly basis, these data composed a self-consistent, reliable long-term record
180 of sulfate deposition that was adequate to reflect drifts and trends. However, the long-term patterns of hydrogen, nitrate, and ammonium in the USGS data were less reliable and difficult to interpret for long-term trends. Hidy et al. (1984) also examined the spatial repre- sentativeness of a number of stations in the Northeast. From an examination of intersite correlations for sulfate among the Hinckley, Mays Point, Canton, Salamanca, and Athens USGS stations they found that only the first three tended to have similar and simultaneous month-to-month variations in sulfate. This points out the difficulty of expecting that a single station or a grouping of a few stations will necessarily provide data that will represent regional or subregional patterns without having estab- lished the fact from supplementary information beforehand. Figure 5.21 shows, as an example, box plots of the analysis of Hidy et al. (1984) for the Hinckley, New York, site. Median sulfate concentrations showed an overall downward trend, while median nitrate concentra- tions increased in the 1960s and fluctuated thereafter. Median ammonium concentrations appear to have increased after 1975, and median pH showed a weak downward trend. Results obtained from the box plot analyses of the five USGS stations and analysis of the data from Hubbard Brook, New Hampshire, are summarized in Table 5.13. (We note that there are some contradictions between Tables 5.12 and 5.13, presumably a result of different statis- tical analyses.) Two central New York sites, Hinckley and Canton, and Hubbard Brook show a downward trend in sulfate concentration. Nearby but farther west, Mays Point, Salamanca, and Athens do not. Two of the central New York sites, Hinckley and Mays Point, along with Athens, Pennsylvania, show a positive nitrate trend and the site at Hubbard Brook shows a slightly positive trend; Canton and Salamanca show no trend. Ammonium appears to have increased steadily at all USGS sites except Salamanca, and the site at Hubbard Brook shows a decreased trend. pH trends, although less reliable, are either neutral or downward over the period of record. Hidy et al. (1984) also applied an autoregressive integrated moving-average model to the sulfate data from Hinckley and Mays Point, the two sites where the data record was of sufficient length. Results showed a statistically significant decrease in sulfate concentra- tion of about 2 percent/yr between 1965 and 1980 at both sites.
181 In summary, despite the uncertainties inherent in the USGS data and the different statistical analysis tech- niques used, some general statements can be made about trends in the data. Although slight downward trends in pH are evident at some stations for some portions of the record, pH generally shows no trend. The sulfate records at Hinckley and May s Point show a downward trend, con- sistent with the record at Hubbard Brook. Because of problems with the nitrate data the various analyses are equivocal--some show increases, some decreases, some no change. Ammonium generally appears to be increasing. As important as the trend estimates themselves are, several points stand out from the published analyses. Large variability is evident between stations and param- eters; even methods of analysis and data preprocessing affect the results. Stations within a limited geo- graphical area may be affected by different processes or climates and therefore may not be assumed to be repre- sentative of a region. The trends in sulfate at the two stations examined in detail by Hidy et al. (1984) appear to be most certain because of their self-consistent, reliable, long-term record. The fact that the trends are similar to that at Hubbard Brook gives some confidence to proposing a regional trend in precipitation sulfate concentration. The World Meteorological Organization/ Background Air Pollution Monitoring Network Data In the early 1970s the United States and Canada established, respectively, ten and eight stations as part of the WMO Background Air Pollution Monitoring Network (BAPMoN) to measure wet-only precipitation composition on a monthly basis. These stations have undergone several changes over the years, including station relocation, changes in collectors, sample handling protocols, analytical methods, and sampling frequency with the result that the data are difficult to interpret. Preliminary analysis of both the U.S. and Canadian data indicates that sample contamination in the early years of the network and the subsequent changes in operation have made the data unsuitable for trend analysis without applying more rigorous data screening and correction criteria than have been used thus far. However, within the last 3 to 4 years, these BAPMoN stations have been incorporated into
182 (a ) 1 5.0 12.5 1nn _ ~ 75 _ E J _ ~ 5.0 _ 2.5 _ O _ -2.5 (b) 2.0 1 .6 - z - cL 0.8 a: 1.2 z 0.4 Ol ~.4 ' SO= 4 o * 1 65 70 75 80 WATER YEAR BEGINNING NO3 o * ._. :+ _ . +. . _ * 1 1 65 70 * WATER YEAR BEGINNING · ~ · +. · ~ FIGURE 5.21 Box plots for precipitation chemistry at the Hinckley, New York, USGS site. (a) sulfate concentration, (b) nitrate concentration, (c) ammonium concentration, (c) pH. Data are from monthly bulk-deposition samples. The box encloses the middle 50 percent of the data for the year. The median value is denoted by a +. The dotted lines extending beyond the box give the range of all data for the year. The * and o symbols denote possible and probable outliers, respectively. SOURCE: Hidy et al. (1984).
183 (c) 2.0 1.6 1 .2 0.8 0.4 o -0.4 (d ) 7.0 5.6 2.8 NH4 * * o o 65 pH n 70 V . 75 WATER YEAR BEGINNING 4F 80 ME 65 FIGURE 5.21 (continued) . 70 WATER YEAR BEGINNING 75 80
184 TABLE 5.13 Summary of Apparent Trendsa Based on Boxplot Analysis of Annual Median Precipitation Chemistry Data From the USGS Sites and the Hubbard Brook Site (1965-1980) Hubbard Parameter Brook Hinckley N.Y. Canton Mays Point Salamanca Athens N.Y. N.Y. N.Y. Pa. Precipitation SO2- NO3 NH4 pH + o o O+ + _c + + oc _ 0 O O O Ob o o+ O + O + _ O O + + o a + is an upward trend,is a downward trend, 0 is no trend, and 0 + is a slightly upward trend. bTime-ser~es analysis shows a downward trend at this site. CFrom Oehlert (1984a). SOURCE: Hidy et al. (1984). the NADP and Canadian Air Pollution Monitoring Network (CAPMoN). Their most recent data are discussed in the following section. The 1979-1982 CANSAP-APN and NADP Data As the importance of deposition monitoring became evident, several longer-term networks began operating in the late 1970s (Figure 5.14). In addition, there are a number of other networks of more limited geographical and jurisdictional scope. All have added stations and made changes in their operations since their inception, but of major importance has been the increased emphasis on quality control/quality assurance and on data analysis. In most cases the period of available high-quality data and increased spatial coverage extends from 1979 or 1980, although MAP3S began providing high-quality data in late 1976. This is too brief a period for an analysis of trends. However, this period may have been characterized by rather large changes in emissions, and it is of great interest to determine whether these changes can be detected in the precipitation data. In what follows, recent results from two networks, CANSAP-APN and NADP, are examined in the context of emission changes over the past 3 to 5 years.
185 a ~ V a W~ ~ 1 i'- ~~: ~ .1. . At, ~ _, ~_/. ( ~ 1 ~ ! ,,>7; ~ ~ | ~~- ~ ~) .it'' ~''.-.~4,1`,,,' /, , fee_ _ ' ,, _ ,7: ~ ~ ~ - W so ~~ H + · 'a FIGURE 5.22 Comparison of the annual precipitation- weighted-mean concentrations of sulfate, hydrogen, nitrate, calcium, and precipitation amount (designated P in top panels in (a) to (c)) for (a) 1980; (b) 1981; (c) 1982 at eastern CANSAP stations. Filled circles denote maximum of three annual values; filled triangles denotes minimum.
186 b ~910 ~ O 60/ ~ . ~ .? ;' · i~\ ~ ·/ so '\A , I, i;- 'aft .~ 40 ~ p 1981 ~ ~ ~ ~ ~ -sol ~ ·> 90 ~ ) ~ waif 70 -A 1981 1 90 ~ ) ANY ~ 70 - 1981 ~ SC NOs 70 - 1~1 ! FIGURE 5.22 (continued). CANSAP-APN Data Data from the CANSAP-APN (now CAPMoN) network were examined with particular interest in the period of reduced emissions from 1980 to 1982. Between the beginning of 1980 and late 1983 sampling was carried out in a monthly composite mode, and these data are considered to be more reliable than earlier data (Barrie and Sirois 1982). Two features 'nave emerged from the preliminary analysis of the data: (1) data frown stations within the same 0~
187 c ~~ art' 4'\i FIGURE 5.22 (continued). ,lr' "-'^1~: 4o P 198Z '- .~1 40 He 1982 For?' ~~ so 1982 ~ ~' ~ Cat geographical region of eastern Canada often showed similar behavior and (2) ions of different origin also tended to show similar behavior within regions. Figures 5.22(a), 5.22(b), and 5.22(c) show these features for the ions sulfate, nitrate, hydrogen, and calcium during the years 1980, 1981, and 1982, respec- tively; precipitation amount (P) is also shown. In 1980, for example, several stations in the Atlantic Provinces and the upper St. Lawrence Valley had the highest annual precipitation-weighted-mean values of the three years for
188 . ~ ~ ~ At, ~ \ ~ \, oH~1 15 .) NH 02 \ 1 1 ~ :] t: FIGURE 5.23 Locations of twelve NADP sites referred to in Figures 5.24 and 5.25. the four ions (filled circles, region designated by dashed lines). Similarly, several stations in the lower St. Lawrence Valley had the lowest values of the three years (filled triangles), except for calcium. As a second example, in 1981 the easternmost stations in the network, those in the Atlantic Provinces and parts of eastern Quebec and Labrador, had the lowest annual precipitation-weighted means of the three years. (See areas enclosed by dashed lines in Figure 5.22(b).) In 1982, the maximum values for the three-year period tended to occur in the lower St. Lawrence Valley and lower Great Lakes region. (See dashed lines in Figure 5.22(c).) Even this descriptive approach does point out the strong role that factors other than emissions have in determining the composition of precipitation at a given station. Sulfate and nitrate tend to behave in a similar way at many stations, but calcium, not primarily an anthropogenic constituent, also shows similar behavior in many instances. The fact that calcium is often similar to sulfate and nitrate suggests that meteorology, both precipitation and wind regime, plays a major role in the spatial pattern of precipitation composition.
189 120 100 - cn 0 80 LLl 60 by o C' Z 40 _: \ so2- +1 \NO3 + Cl -/\ \ so2 _^ _ \\ H ~ 1 ~ _~_ ~ NO3 20 NH4 1 1 1 1 o NY 20 Sow + on ~ ~ SHO 2- _N, {~' NO3 ~ _ / NH4 80 82 NC 41 ~so2~+ _ NO3 + Cl - ~ iso2~ _ _ ,~ Hit 1 / ~ NO3 N H 4 1 1 1 1 80 82 80 82 CALENDAR YEAR ~ Ca2 + | M92+ + 1 Na + K + NH4 FIGURE 5.24 Calendar year median ion concentrations for three NADP sites for 1979-1982. NADP Data Because the data have only recently become available, no studies have been published that examine the time pattern for NADP stations (in particular for the 1979-1982 time period). Here an initial examination is presented. When the NADP began in 1978 most of the stations were located in the northeastern quadrant of the United States, although there were fewer stations in other areas. These stations have compiled relatively complete records for the 1979-1982 period. Samples were collected for each week that had precipitation and problems of contamination or equipment malfunction were infrequent. Results are presented for 12 sites whose locations are shown in Figure 5.23. The median annual concentrations for major ions and combinations of ions are shown in Figure 5.24 for three of the sites (ILll, NY20, and NC41), and volume-weighted
190 ~ N ~ I ~ ~ ~ I ~ ~4 '' ~ ~ ~ ~ ~ N; I ~ I I ° ° 1 1 1 1 1 + ~ N ~ I ~ ~ _ 1 1 1 1 1 1 1 1 O O O O C-l O 00 ~ N ~ , O ~ O I v, I _ z Z ~1 1 1 1 1 ~ ¢ 141 - o l , ~ ~ ~ I _ Oz I C~ / _ g ~ _ O O d. (l/bad) SNOIlVUlN3~:)NOO G31H!313M-3WnlOA c: z LL c, z LL J ~: C~ o N N , + ~ ~ ' ~ Y Z ~ ~ Z ~ Y ~ ~ ~r U) (D N O + O I Z Z t~ N tD 1 1 1 1 14 1 1~ ° N ~ O I i/ ~ ~: LL t~d (D t(~0 Z J :: 1 - N ~ ~ ~ ~ ~ _, O _ Z Z ~, ,,, C`' O O O O O O O C~l O 00 C.C) ~ ~ r~ _ (l/bad) SNOIl~lUlN3ONOD 031HE)13M-3Wnl OA
191 + + - <r' - , om-/ ~ + 0 I `,, , O _ Z _ ~ ~ _ 1 1 1 1 + 1 1 1 1 ~o" ~o~ 2 I O O I --- ~' ~ ~ 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 - "' =, o O _ cn Z (~) _ I I 1 1 + ~ , ~ I O Z Z + DJ I -~?~ ~ /_ 1 1 1 1 1 O O C-l O O O O d. N (l/ba?~) SNOIl~lUlN3~)NOO O31HE)I3M-3Wl)lOA 0 + ~) I , m+ O I Z Z ~_ - - , m+ ~ I _ Z Z ~ <~: (~_ 1 1 1 1 1 , ~ + ~r O I Z Z ~_ o - - , =, O O _ cn Z c) + + 1 ~ ~r O + cn I _ 1 1 1 O O O O C~ 0 00 CO O O ~ (~ (l/bad) SNOIl~lUlN3ONOO G31HE)13M-3VUnlOA c~ oo o oo cr LL c~ ~ oo ~r 0 ~ oo z UJ J 00 0D 1 o U2 a' U] a z 0 00 S~ o C~ o ~, a o 00 ~r UJ 0O o ~ oo z LLJ J C~ 00 o 00 a' S .~1 3 a) o a c) . In H
192 average concentrations (to be referred to as weighted concentrations) are shown for all twelve sites in Figure 5.25. The results shown in Figures 5.24 and 5.25(a) can be compared to evaluate differences in the two data presen- tations. For site ILll the anions and hydrogen ion have similar median and weighted concentration patterns, although hydrogen ion is lowest in 1979 for the weighted concentration and in 1982 for the median concentration. For ammonium and the base cation sum (the sum of calcium, magnesium, potassium, sodium, and ammonium) the year of maximum concentration differs for the median and weighted concentrations. There are also differences in median and weighted concentrations at sites NY20 and NC41. For site NY20 the concentration patterns of ammonium ion and the base-cation sum differ between the two figures, while for site NC41 the sulfate and hydrogen ion concentration patterns differ in the two presentations. To compare the 12 sites, volume-weighted average concentrations are shown in Figure 5.25. From Figure 5.25 (b) for the three sites in Ohio, we see that almost every curve has some site-to-site differences even though the sites are spatially close together. In particular, sulfate has a different pattern for each site. From Figure 5.25(c) we see that the various curves of the first two sites, GA41 and NC25, are quite similar in shape but different from NC03. However, the patterns for site NC03 are quite similar to those for site NC41 in Figure 5.25(a). From Figure 5.23 we see that GA41 and NC25 are adjacent, as are NC03 and NC41, suggesting a spatial consistency for the annual concentration patterns. In summary, the most persistent pattern for all ions and all sites is higher concentrations in 1980 and 1981 and lower concentrations in 1979 and 1982. However, not all sites or all ions follow the general pattern. Even multiple sites within a single state have different time patterns (for example, in North Carolina and Ohio). It is not uncommon to have most of the major conservative ions (sulfate, nitrate, ammonium, and calcium) following the same time pattern (for example, see ILll). This suggests that for the 1979-1982 period meteorology, as opposed to the temporal changes in source emissions, was important in determining time trends of ion concentrations.
193 CONCLUSIONS AND RECOMMENDATIONS Conclusions · The eastern half of the United States and south- eastern Canada south of James Bay experience concentra- tions of sulfate and nitrate in precipitation that are, in general, greater by at least a factor of 5 than those in remote areas of the world, indicating that levels have increased by this amount in northeastern North America since sometime before the 1950s. · The northeastern quadrant of the United States, consisting of Region B. the northernmost states of Region C, and most of Region D, is the area most heavily affected by acidic species (the ions of hydrogen, sulfate, and nitrate) in precipitation. · Spatial distributions of annual (precipitation- weighted-mean) concentrations and wet deposition for the major ionic species in precipitation are adequately known on the regional to continental scale in North America to afford comparison with spatial patterns of emissions and other possible indicators of acid deposition, such as visibility and the quality of surface waters. The 1980 concentration and deposition fields are acceptably representative of the broad-scale patterns present in the 7-year period 1977-1983. · Data on the chemistry of precipitation before 1955 should not be used for trend analysis. . Precipitation is currently more acidic in parts of the eastern United States than it was in the mid-19SOs or mid-1960s; however, the amount of change and its mech- anism are in dispute. Changes in natural dust sources, anthropogenic acid sources, and sampling methods are all major factors that need to be considered in interpreting the changes in precipitation. . Precipitation sulfate concentrations and possibly acidity leave increased in the southeastern United States (Region C) since the mid-1950s. · For Hubbard Brook in New England, general agree- ment exists among several reports on the magnitude and direction of trends since 1964 in some species: hydrogen ion shows no overall significant trend; sulfate has decreased at approximately 2 percent/yr; sodium, chloride, calcium, magnesium, and potassium have shown strong decreases with time; nitrate appears to have increased until about 1970-1971 and subsequently leveled off; the analyses for the ammonium record are contradictory.
194 · On the basis of past analyses of the USGS bulk data from New York and Pennsylvania since the mid-1960s a consistent network-wide trend does not emerge: results differ from station to station (with a few exceptions). Thus the USGS data base should be screened carefully when general conclusions about regional-scale trends are drawn. The USGS stations at Mays Point and Hinckley, New York, show sulfate trends based on time-series analysis to be consistent with those at Hubbard Brook, making it more likely that a regional trend exists in this parameter. . On the basis of preliminary analyses of CANSAP- APN and NADP data for the eastern half of the continent, changes in anthropogenic emissions during the period 1979-1982 are not readily apparent; concentration changes from year to year occur concurrently in both anthropogenic and natural constituents and are thought to be more strongly affected by meteorological rather than emissions changes during this period. . Because data from the WMO BAPMoN network have not yet been rigorously quality assured, trend analyses have not yet been carried out. . The uncertainties associated with available time- trend estimates of precipitation chemistry make it unwise to draw quantitative conclusions on the basis of direct comparisons between changes in concentrations or deposi- tion and changes in emissions. In cases in which artifacts may be introduced in a precipitation-chemistry data set, for example, by changes in analytical methods over time, even qualitative conclusions related to emission changes may not be possible. . In general, individual sites or groups of a few neighboring sites cannot be assumed a priori to provide regionally representative information; regional repre- sentativeness must be demonstrated on a site-by-site basis. · General conclusions drawn from merging bulk precipitation data sets with wet-only precipitation data sets are not valid unless detailed site-by-site analyses justify the approach. Similar evaluations should be made when different wet-only precipitation data sets are merged. · Conclusions regarding time trends depend to some extent on the statistical analysis technique chosen (see Table 5.12). They also depend on the period of the record used for analysis, and this point is frequently ignored.
195 · Trends for particular ions in precipitation (including hydrogen) should be interpreted only in relation to trends of all the major ions present. Recommendations · Expand the scope of deposition monitoring to include dry and occult deposition. · Strive for improved understanding of the char- acteristics of all forms of deposition, particularly as a function of altitude and species. . Continue long-term, wet-only national monitoring, preferably under the auspices of a single jurisdiction. · Use measures other than pH or hydrogen-ion concentration as a measure or an indicator of the acid deposition problem. . Investigate the role of organic acids in precipitation acidity. . sites. Discourage bulk sampling except in specific circumstances in which it can be demonstrated that bulk and wet-only samples are providing comparable data. · Use other major ionic species (in addition to hydrogen and sulfate) in future trend analyses; also use ancillary information such as meteorological/ climatological data. · Investigate the role of meteorological and climatological variability in temporal and spatial variations of acid deposition. · Monitor air concentrations in closely located · Take great care in all aspects of selecting and maintaining a station, particularly its siting and operation, if it is to be used in the future for trend analysis. . Devote more effort to the analysis of available data. · Emphasize network design to ensure that networks are capable of detecting changes, particularly those changes that may result from strategic emissions reductions. · Establish a mechanism and a procedure to compile data, to perform quality control and assurance, to analyze routinely, to synthesize results, and to publish multinetwork data.
196 REFERENCES Barnes, C. R., R. A. Schroeder, and N. E. Peters. 1982. Changes in the chemistry of bulk precipitation in New York State, 1965-1978. Northeast. Environ. Sci. 1(3-4):187-197. Barrie, L. A., and J. M. Hales. 1984. The spatial distributions of precipitation acidity and major ion wet deposition in North America during 1980. Tellus 36B:333-355. Barrie, L. A., and A. Sirois. 1982. An analysis and assessment of precipitation chemistry measurements made by CANSAP: 1977-1980. Internal Report AQRB-82-003-T, Atmospheric Environment Service, Downsview, Ontario, 163 pp. Bilonick, R. A. 1984. Discussion of Miles and Yost, 1982. Atmos. Environ. 18:479. Bilonick, R. A., and D. G. Nichols. 1983. Temporal variations in acid precipitation over New York state--what the 1965-1979 USGS data reveal. Atmos. Environ. 17:1063-1072. Bower sax, V. C., and R. G. de Penal 1980. Analysis of precipitation chemistry at a central Pennsylvania site. J. Geophys. Res. 85:5614-5620. Bowersox, V. C., and G. J. Stensland. 1981. Seasonal patterns of sulfate and nitrate in precipitation in the United States. Paper No. 81-6.1 presented at the 74th Annual Meeting, Air Pollution Control Association, Philadelphia, Pa., June 21-26. Charlson, R. J., and H. Rodhe. 1982. Factors controlling the acidity of natural rainwater. Nature 295:683-685. Coffey, P. E. 1984. Source-receptor relationships for acid deposition: pure and simple? Critical Review Discussion Papers. J. Air Pollut. Control ASSOC. 34:905-906. Cogbill, C. V., G. E. Likens, and T. A. Butler. 1984. Uncertainties in historical aspects of acid precipitation: getting it straight. Atmos. Environ. 18:2261-2270. Environmental Protection Agency. 1984. Deposition Monitoring, Chapter A-8 in Critical Assessment Review Papers, 1984. Volume I. EPA-600/8-83-016AF. Fisher, D., A. Gambell, G. Likens, and F. Bormann. 1968. Atmospheric contributions to water quality of streams in the Hubbard Brook Experimental Forest, New Hampshire. Water Resour. Res. 4:1115-1126.
197 Galloway, J. N., G. E. Likens, and M. E. Hawley. 1984. Acid precipitation: natural versus anthropogenic components. Science 226:829-831. Galloway, J. N., G. E. Likens, W. C. Keene, and J. M. Miller. 1982. The composition of precipitation in remote areas of the world. J. Geophys. Res. 87:8771-8776. Galloway, J. N., and G. E. Likens. 1981. Acid precipitation: the importance of nitric acid. Atmos. Environ. 15:1081-1085. Granat, L. 1978. Sulfate in precipitation as observed by the European atmospheric chemistry network. Atmos. Environ. 12:413-424. Hansen, D. A., and G. M. Hidy. 1982. Review of questions regarding rain acidity data. Atmos. Environ. 16:2107-2126. Herman, F., and F. Gorhman. 1957. Total mineral material, acidity, sulfur and nitrogen in rain and snow at Kentville, Nova Scotia. Tellus 9:180-183. Hidy, G. M., D. A. Hansen, R. C. Henry, K. Ganesan, and J. Collins. 1984. Trends in historical and precursor emissions and their airborne and precipitation products. J. Air Pollut. Control Assoc. 31:333-354. Junge, C. E., and R. T. Werby. 1958. The concentration of chloride, sodium, potassium, calcium, and sulfate in rainwater over the United States. J. Meteorol. 15:417-425. Likens, G. E., F. H. Bormann, R. S. Pierce, J. S. Eaton, and R. E. Munn. 1984. Long-term trends in precipitation chemistry at Hubbard Brook, New Hampshire. Atmos. Environ. 18:2641-2647. Likens, G. E., and T. J. Butler. 1981. Recent acidification of precipitation in North America. Atmos. Environ. 15:1103-1109. Likens, G. E., F. H. Bormann, and J. S. Eaton. 1980. Variations in precipitation and streamwater chemistry at the Hubbard Brook Experimental Forest during 1964-1977. In Effects of Acid Precipitation on Terrestrial Ecosystems, T. C. Hutchinson and M. HavaS eds., New York: Plenum Press, pp. 443-464. Lodge, J. P., J. B. Pate, W. Basbergill, G. S. Swanson, K. C. Hill, E. Lorange, and A. L. Lazrus. 1968. Chemistry of United States precipitation. Final report on the National Precipitation Sampling Network. Boulder, Colo., National Center for Atmospheric Research. 66 pp.
198 Miles, L. J., and K. J. Yost. 1982. Quality analysis of USGS precipitation chemistry data for New York. Atmos. Environ. 16:2889-2898. Miller, M. L., and A. G. Everett. 1981. History and trends of atmospheric nitrate deposition in the eastern U.S.A. In Formation and Fate of Atmospheric Nitrates. Workshop Proceedings, EPA-600/9-81-025, 162-178. Environmental Protection Agency, Wasnington, D.C. Munn, R. E., G. E. Likens, B. Weissman, J. Hornbeck, C. W. Martin, F. H. Bormann, G. W. Oehlert, and R. Bloxam. 1982. A climatological analysis of the Hubbard Brook (New Hampshire) precipitation chemistry data. Final report to the U.S. Department of Energy, Grant DE-FGO1-SOEV-10455, 221 pp. National Research Council. 1983. Acid Deposition: Atmospheric Processes in Eastern North America. Washington, D.C.: National Academy Press. 375 pp. Oehlert, G. W. 1984a. A statistical analysis of the trends in the Hubbard Brook bulk precipitation chemistry data base. Technical Report No. 260, Series 2, Department of Statistics, Princeton University, 64 PP ~ Oehlert, G. W. 1984b. Basic statistical methods for trend analysis. Unpublished manuscript, Department of Statistics, Princeton University. Ottar, B. 1978. An assessment of the OECD study on long range transport of air pollutants (LEAP). Atmos. Env 12:445-454. Pack, D. H. 1982. Precipitation chemistry probability-- the shape of things to come. Atmos. Environ. 16:1145-1157. Pack, D. H., and D. W. Pack. 1980. Seasonal and annual behavior of different ions in acidic precipitation. World Meteorological Organization Special Environmental Report No. 14, pp. 303-313. Peters, N. E. 1984. Discussion of Miles and Yost, 1982 Atmos. Environ. 18:1041-1042. . Peters, N. E., and J. E. Bonelli. 1982. Chemical composition of bulk precipitation in the North-Central and Northeastern United States, December 1980 through February 1981. Circular 874, U.S. Geological Survey, 63 pp. Peters, N. E., R. A. Schroeder, and D. E. Troutman. 1982. Temporal trends in the acidity of precipitation and surface waters of New York. Water Supply Paper 2188, U.S. Geological Survey, 35 pp.
199 Stensland, G. J., and R. G. Semonin. 1982. Another interpretation of the pH trend in the United States. Bull. Am. Meteorol. Soc. 63:1277-1284. Stensland, G. J., and R. G. Semonin. 1984. Response to comment of T. S. Butler, C. V. Cogbill, and G. E. Likens. Bull. Am. Meteorol. Soc. 65:640-643. Wisniewski, J., and J. D. Kinsman. 1982. An overview of acid rain monitoring activities in North America. Bull. Am. Meteorol. Soc. 63:598-618.
