Estimation of wet, dry and bulk deposition of ... - Semantic Scholar
Estimation of wet, dry and bulk deposition of atmospheric nitrogen in Connecticut Farhad Nadim,*a Michael M. Trahiotis,a Snieguole Stapcinskaite,b Christopher Perkins,a Robert J. Carley,a George E. Hoaga and Xiusheng Yangc a
University of Connecticut, The Environmental Research Institute, Longley Building, U-5210, 270 Middle Turnpike (Rte. 44), Storrs, CT 06269, USA. E-mail: [email protected]; Fax: (860) 486 5488; Tel: (860) 486 4015 b Institute of Physics, A. Gostauto 12, 2600 Vilnius, Lithuania c University of Connecticut, Natural Resources Management and Engineering, 1376 Storrs, Road, U-87, Storrs, CT 06269, USA Received 1st August 2001, Accepted 5th October 2001 First published as an Advance Article on the web 8th November 2001
Atmospheric nitrogen species including NO32, NH4z and total nitrogen in air and precipitation samples were collected with low-volume filter packs and wet deposition collectors from March 1999 through the end of December 2000 in seven sampling locations in Connecticut. Three sampling locations were chosen along the shores of Long Island Sound and four were chosen in interior sections of Connecticut. Sampling sites were chosen to represent both rural and urban sectors. Wet deposition flux of nitrogen species was calculated using wet concentrations, the volume of collected precipitation and the opening surface area of the Aerochemetrics wet deposition collector. The dry deposition flux of nitrogen species was estimated with the application of the dry deposition inferential model (DDIM). Bulk deposition of nitrogen was collected with the aid of a device based on the Swedish IVL Sampler. The dry deposition fluxes of NO32, NH4z and total nitrogen were found to be significantly higher in urban areas than the rural sampling locations. There was, however, no significant difference between the wet deposition fluxes of different nitrogen species in rural and urban sampling locations. When inland and coastal sites were compared, the dry deposition fluxes of NH4z and total nitrogen were significantly higher in inland locations and there was no significant difference between coastal and inland sampling locations for wet deposition fluxes of nitrogen species. No significant difference was observed between the bulk deposition and the sum of the wet and dry deposition fluxes of total nitrogen at rural sampling locations. In urban sampling locations, the bulk deposition flux of total nitrogen was significantly lower than the sum of dry and wet deposition fluxes. There appears to be a similar seasonal trend in wet and dry deposition fluxes of total nitrogen in Connecticut with high and low deposition fluxes occurring in summer and winter periods, respectively.
1. Introduction Overabundance of nitrogen is of concern in areas that have developed nutrient enrichment problems (i.e., eutrophication). In addition to increasing productivity, nutrient enrichment generally alters the normal ratios of nitrogen to phosphorus and to other elements such as silicon.1 This alteration may induce changes in phytoplankton community structure. Species that normally occur in low abundances may be favored and, in some cases, toxic and/or noxious algal blooms may result. The Long Island Sound (LIS) is one of the 17 major estuarine systems of the United States under EPA’s National Estuary Program that needs to be protected.2 The LIS has a poor or stratified circulation pattern and the over production of algae caused by excessive loading of nutrients especially nitrogen tends to sink to the bottom and decay, using all (anoxia) or most (hypoxia) of the available oxygen in the process, causing loss of habitat.3,4 Atmospheric nitrogen originating from anthropogenic sources is estimated to cause more than 15% of the dissolved oxygen deficit observed in LIS. Recent studies have shown the importance of managing atmospheric sources of nitrogen if water quality objectives are to be achieved and maintained in LIS.5,6 Atmospheric nitrogen can be deposited in the sound in two ways: direct and indirect deposition.7 Direct DOI: 10.1039/b107008h
deposition on LIS contributes to 5.9% of the anthropogenic load of nitrogen to the Sound from all sources. Indirect deposition of nitrogen includes discharges from industry and treatment plants (point sources), storm water runoff from coastal areas, with rivers and streams from throughout the drainage basin (non-point sources) and currents moving into the Sound from the Atlantic Ocean and New York Harbor.3,4 Ammonia and oxides of nitrogen are mineralized forms of nitrogen that are readily usable by plants such as phytoplankton. A small amount of organic nitrogen is also found in atmospheric deposition, which has not been fully characterized. To study the spatial and temporal variation of atmospheric deposition of nitrogen in Connecticut, a collaborative study between the Connecticut Department of Environmental Protection (CTDEP) and the Environmental Research Institute (ERI) at the University of Connecticut was conducted. Wet, dry and bulk deposition atmospheric nitrogen samples were collected and analyzed for estimation of atmospheric wet, dry and bulk deposition fluxes of nitrogen species in Connecticut for a period of 94 weeks. The National Atmospheric Deposition Program (NADP)8 and the USEPA Clean Air Status and Trends Network (CASTNet)9 have a site in a rural section of northeastern Connecticut to monitor the atmospheric deposition of pollutants including NO32, NH4z and HNO3. The mean J. Environ. Monit., 2001, 3, 671–680
This journal is # The Royal Society of Chemistry 2001
671
weekly wet and dry deposition fluxes of total nitrogen reported by NADP and CASTNet were compared with the mean fluxes measured at the rural sampling locations of this study.
2. Methods 2.1. Study site selection Seven sites were selected to monitor ambient levels of atmospheric nitrogen in Connecticut. The primary factor in the selection process was to obtain a distribution of sites to reflect adequately the ambient conditions throughout the State. The other factors included in choosing the sites were the availability of access, power, security and the presence of existing nearby meteorological monitoring stations. Bridgeport (40 km northwest of New York City) was chosen to represent atmospheric nitrogen levels of a highly populated urban and industrial coastal area. Waterbury and East Hartford sites were chosen to represent inland urban levels of atmospheric nitrogen. Hammonasset State Park (Madison) and Avery Point (Groton) were chosen as coastal rural and Mohawk Mountain and Voluntown were chosen as rural interior sites for measurement of atmospheric concentration of nitrogen (Fig. 1). Sampling was performed on Fridays of every week at all of the sites mentioned above for a period of 94 weeks. Samples were collected weekly and analyzed for total nitrogen, ammonium, organic nitrogen and oxides of nitrogen (nitrite/nitrate). The goal of this study was to monitor the spatial and temporal variation of atmospheric nitrogen in Connecticut. 2.2. Field procedures Before collection of samples, all equipment and sample containers were rigorously cleaned using detergent, concentrated acids and de-ionized water. Prior to deployment in the field, sampling equipment and containers were demonstrated to be free from contamination. All sampling equipment and containers were constructed of materials intrinsically low in contaminants. Sampling personnel were required to wear clean, talc-free rubber gloves. All samples collected in the field were placed inside transportable coolers, brought to the laboratory and kept in a walk-in cooler pending analysis.
2.2.1. Dry deposition sample collection. Dry deposition samples were collected on filters. A weekly composite sample of ambient air was drawn through a filter pack at each site and analyzed for total nitrogen, ammonium and oxidized nitrogen. Each site contained a two-stage filter pack with Teflon and nylon filters in series housed in a fiber-glass NEMA enclosure. The Teflon filter (2 mm Gelman Laboratory, Zelfluor) was used to collect dry deposition of particulate nitrogen (total nitrogen, ammonium and NO32). The nylon filter (1 mm Gelman Laboratory, Zelfluor) was used to collect dry deposition of nitric acid (HNO3) vapor. Nylon filter material is an absolute scavenger of HNO3 7 Flow through the filter pack was set to 3.0 l min21 and the target sample volume collected over the course of the week was 30 m3. Teflon filters were extracted with 30 ml of de-ionized water. Teflon filter extracts were analyzed for NH4z and oxidized and total nitrogen. Nylon filters were extracted with 30 ml of a dilute solution of sodium hydroxide and hydrogen peroxide. Nylon filter extracts were analyzed for oxidized nitrogen and the results were reported as HNO3. 2.2.2. Wet deposition sample collection. The network of seven air monitoring stations was outfitted to collect precipitation (wet deposition) to be analyzed for total nitrogen, ammonium and oxidized nitrogen. At the initiation of rain events, moisture triggered a tipping bucket lever, which in turn resulted in the lid being removed from the Aerochemetrics wet deposition collector. At the termination of rain events, the lid swung back and re-sealed the wet deposition collector. Each of the wet deposition buckets was lined with a pre-cleaned Teflon bag to minimize contamination. The wet deposition collectors were sampled weekly. A clean Teflon bag was inserted into the collection apparatus when no rain events occurred within the week. 2.2.3. Bulk deposition sample collection. Collections of bulk deposition samples were initiated on March 19, 1999, at seven sampling locations. A sample collection device based upon the Swedish IVL Sampler was used.10 The device is made of a PVC pipe (y100 cm high with 100 mm id). Inside the pipe a cylindrical glass vessel (110 mm od and 90 mm id) was inserted that was open on top and narrowed at the bottom center to a 5 mm id glass tube. The glass tube extended down to a 300 ml
Fig. 1 Atmospheric nitrogen sampling sites in Connecticut.
672
J. Environ. Monit., 2001, 3, 671–680
BOD bottle that was inserted in the bottom of the PVC pipe. An inverted watch-glass was placed inside the glass beaker to prevent clogging of the tube from debris. During precipitation events, water was directed to the 300 ml BOD bottle. The edge of the opening section of the glass vessel had a convex curve towards the interior part to guide all precipitation towards the bottom of the vessel. At the time of sample collection, the surface of the collecting beaker was rinsed with y100-ml of deionized (DI) water and the rinsate was collected in a precleaned 300 BOD bottle, capped and taken to the laboratory together with the original sample for analysis. After the sample and the rinsate had been taken, the interior of the glass beaker was rinsed with y100 ml of HCl solution followed by 100 ml of DI water and collected in a waste bottle. 2.3. Field quality control 2.3.1. Field blanks. Each sampling crew was equipped with a field blank each week to detect contamination that may have occurred in the sample transport process. This consisted of both a filter pack and a bottle of DI water placed in a plastic bag. The trip blanks were placed in the transport coolers, packaged with the actual sampling media and remained closed for the entire trip. The blanks were returned to ERI with the samples and analyzed in the same manner. 2.3.2. Bag blank. One Teflon bag used in the wet deposition collectors was analyzed for any residual nitrogen contamination that could have originated from the laboratory. After the cleaning process, one bag was rinsed an additional time with 500 ml of DI water. This rinse was poured into a sample bottle and analyzed in the same manner as the wet deposition samplers. One bag blank was prepared weekly for all eight sampling locations. 2.3.3. Co-located (i.e., ‘paired’ samples) sampler. Each week, one station on a rotating basis, had a co-located sampler to determine precision of the sampling technique. The co-located sampler was placed adjacent to the regular sampler and was set up in the same way (i.e., flow rates, equipment, sampling media). The co-located samples were returned to ERI with the regular samples and analyzed in the same manner. 2.4. Analytical procedures 2.4.1. Ammonium (NH4z). Wet and dry samples for ammonium were analyzed by an automated procedure, on a Lachat Quikchem Autoanalyzer, utilizing the Berthelot reaction, in which the formation of a blue compound, believed to be closely related to indophenol, occurs when the solution of an ammonium salt is added to sodium phenoxide, followed by the addition of sodium hypochlorite. A solution of EDTA was added to the sample stream to eliminate the precipitation of the hydroxides of calcium and magnesium. Sodium nitroprusside was added to intensify the blue color. The Lachat Quikchem Autoanalyzer was calibrated with a five-point curve at the time of analysis. The calibration curve was then verified by an external quality control sample from Environmental Resource Associates, Arvada, CO (Waste Water Quality Control Standards). This initial calibration check and initial calibration blank demonstrated that the instrument performance was acceptable at the beginning of the sample analysis. In order to ensure continually acceptable performance, a continuing calibration check and continuing calibration blank were run every tenth sample. For every sample delivery group (SDG, ¡20 samples), a laboratory spike analysis and a laboratory duplicate analysis were performed. 2.4.2. Nitrate z nitrite (oxidized nitrogen) (Teflon and nylon filters). Samples for nitrate and nitrite were analyzed by an
automated procedure, on a Lachat Quikchem Autoanalyzer, whereby nitrate was reduced to nitrite by passage of the sample through a copperized cadmium column. The nitrate reduced to nitrite plus any free nitrite present react under acidic conditions with sulfanilamide to form a diazo compound which coupled with N-1-naphthylethylenediamine dihydrochloride to form a reddish–purple azo dye that was measured at 520 nm. The Lachat Quikchem Auto Analyzer was calibrated with a fivepoint curve at the time of analysis. The calibration curve was then verified by an external quality control sample from Environmental Resource Associates. This second source calibration check and initial calibration blank demonstrated that the instrument was capable of acceptable performance at the beginning of the sample analysis. In order to ensure continually acceptable performance, a continuing calibration check and continuing calibration blank were run every tenth sample. For every sample delivery group (SDG, ¡20 samples), a laboratory spike analysis and a laboratory duplicate analysis were performed. 2.4.3. Total nitrogen (TN). Sample digestion was completed within 14 d from arrival at the laboratory. The sample was placed in screw-capped test-tubes with an oxidizing reagent (potassium persulfate–sodium hydroxide). The tubes were placed in a pressure cooker at 100 uC (3–4 psi) for 60 min. After the samples had cooled to room temperature, hydrochloric acid was added to each sample. Through the processes mentioned above all nitrogen species will be converted into nitrate. The samples were then placed on a vortex mixer to dissolve any precipitates. A boric acid–sodium hydroxide buffer was then added to bring the pH of the sample in the range 7–8. The sample was then ready for the determination of nitrate z nitrite as described in Section 2.4.2. 2.4.4. Organic nitrogen (ON). Organic nitrogen species are typically referred to as a subset of reactive nitrogen. These chemical species result from direct emissions of organic nitrogen compounds or from the interaction between nitrogen and biogenic or anthropogenic and natural sources, speciation of organic nitrogen in the atmosphere is poorly understood. Organic nitrogen flux was quantified by subtracting the oxidized nitrogen and ammonium (NH4z) fluxes from the total nitrogen flux. The total nitrogen flux was derived using the product of total atmospheric concentration of nitrogen and the deposition velocity calculated for particulate nitrogen with the DDIM model. The organic nitrogen flux may also be referred to as the particulate organic nitrogen flux. It should be noted that the organic nitrogen fractions of the total wet and dry depositions were not qualitatively specified by laboratory analysis. 2.5. Laboratory quality assurance/quality control Initial and continuing calibration verification (ICV and CCV) were performed to ensure that the instrument was calibrated properly and remained so during the analytical run. Initially every 10 samples and upon completion of analysis a standard were analyzed. The acceptance criteria for ICV and CCV analyses were 90–110% recoveries. 2.5.1. Preparation, initial and continuing calibration blank. The assessment of blank analysis is to determine the existence and magnitude of contamination problems. A preparation blank was prepared and analyzed for every 20 samples that were digested. A matrix-matched DI water sample was analyzed after the initial calibration, every 10 samples and upon completion of the analytical run. J. Environ. Monit., 2001, 3, 671–680
673
2.5.2. Laboratory duplicate. Duplicate analyses are an indicator of laboratory precision. For every 20 samples a laboratory duplicate analysis was analyzed. The acceptance criteria for the laboratory duplicate analyses were 20% relative percentage difference. 2.5.3. Matrix spike sample analysis. Matrix spike samples are used to monitor, assess and control bias resulting from sample matrix on sample digestion and analysis procedures. For every 20 samples a matrix spike sample analysis was performed. The acceptance criteria for the matrix spike sample analyses were 80–120% recoveries.
3. Data analysis Data were grouped by collection year, location, season, coastal/inland and rural/urban locations. To compare the population means obtained in this study, Tukey’s multiple comparison test method was used. Tukey’s pairwise comparison test was conducted for two mean values obtained for the same sampling period. Tukey’s pairwise comparison test method has been fully described by Berthouex and Brown,11 Kleinbaum et al.12 and Lyman et al.13 Tukey’s pairwise comparison test method makes use of the Studentized range distribution and keeps the family error rate a at a fixed level. This method sets up a (1 2 a)100% confidence interval for the true difference between the two or more means. Instead of testing the null hypothesis that states the difference in the means of two samples is zero, the confidence interval of the difference was tested. If this confidence interval included zero, then it was concluded that there was no statistically significant difference between the two sample means. The a value used throughout this study was 0.05. Data presented graphically were average values calculated from the appropriate dataset (e.g., year, inland, coastal, urban and rural). The Minitab software version 13.3 was used for all statistical analyses. All nitrogen species fluxes were reported as nitrogen. For statistical analysis of the seasonal distribution of atmospheric nitrogen during the course of the study, winter was considered as 1 December–28 February, spring was 1 March–31 May, summer was 1 June–31 August and fall (autumn) was 1 September–30 November. To assess the spatial and temporal association strength of two data set results, the sample correlation coefficient r was used. When r w 0 there was a positive correlation and when r v 0 there was a negative correlation between the two population samples. The association strength increased as r A ¡1.
3.1. Estimation of dry deposition flux of nitrogen Dry deposition is defined as the gases and particles that are deposited to the ground and vegetated surfaces. Previous studies have indicated that acid deposition causes adverse effects on the ecosystem. Studies on dry deposition have shown that it can contribute a nearly equal flux of acid deposition as acid rain.14,15 Measurements of dry deposition fluxes of nitrogen involve the estimation of the deposition velocities of nitrogenous gas and particles. The deposition velocity can be defined as the rate at which each pollutant species is transported towards the Earth’s surface. Weekly concentrations are multiplied by weekly averaged deposition velocities to determine the deposition fluxes of nitrogen. Dry deposition fluxes of ammonia (NH3) and nitrogen dioxide (NO2) were not measured in this study because their measurements required additional equipment that was beyond the objectives outlined by CTDEP. However, based on existing literature, the contribution of the above two gaseous inorganic nitrogen to the dry flux of nitrogen in Connecticut can be 674
J. Environ. Monit., 2001, 3, 671–680
approximated. However, Lear16 states that the contribution of NH3 to total deposition of reduced nitrogen in the United States ranges from 5 to 20% and the contribution of NO2 to total deposition of oxidized nitrogen is in the range 2–15%. Smith et al.17 used a ‘big-leaf’ resistance analogy model for estimation of dry deposition of nitrogen dioxide, ammonia and nitric acid in the United Kingdom. They stated that NO2 dry deposition accounted for 15% of the total oxidized nitrogen deposition in the UK. It could roughly be estimated that the dry deposition flux of NO2 in Connecticut would account for 10–15% of the total oxidized nitrogen flux. Anderson and Hovmand18 conducted a study comparing the measurements of gaseous ammonia (NH3) and particulate ammonium (NH4z) with denuders and filter packs at five different locations in Denmark. They found that the contribution of NH3 to the total concentration of NH3 and NH4z ranged from 18 to 27% and from 14 to 31% for denuder and filter pack collection methods, respectively. The lowest range of NH3 contribution was observed at one of their sampling sites (Anholt Island), which had a distance of at least 50 km to the nearest emission sources of NH3 Considering Connecticut’s landscape and land use, it is likely that the contribution of ammonia to the total deposition of reduced nitrogen could be in the 15–20% range. However, conducting a statistical comparison at local and regional scales requires real measurements of atmospheric ammonia concentrations in Connecticut.
3.1.1. DDIM model for deposition velocity calculations. The model used for the determination of dry deposition velocities of gaseous and particulate nitrogen is referred to as the dry deposition inferential model or DDIM. This model is used by the National Dry Deposition Network. Details of DDIM can be found in Hicks et al.,19 Clarke and Edgerton20 and Yang et al.21 For the purpose of this study, two velocities were calculated for each site. The velocity for particles was used for ions (NO32, NH4z and total particulate nitrogen) and the gaseous velocity was used for nitric acid vapor. The urban sampling sites were set up in areas surrounded by grass that was mowed in regular intervals. The rural sampling locations were set up in areas surrounded by tall grass. The leaf area index of tall grass was used as an input parameter for calculation of dry deposition velocities at all sampling locations. Geigert22 conducted a study for the estimation of dry deposition of some inorganic gaseous and particulate sulfur and nitrogen to a forested area in north central Connecticut in 1990. He used the DDIM model for the estimation of dry deposition velocities of gaseous and particulate nitrogen and conducted a sensitivity analysis on the DDIM model by arbitrarily varying the input parameters. He stated that the leaf area index (LAI) only affected the SO2 deposition velocity when the vegetation cover was changed from hardwood forest to tall grass and no significant effect on the deposition velocities of nitric acid vapor and particulate nitrate (NO32) was observed,7,22 Hardwood forests cover more than 60% of Connecticut. Trees include ash, beech, birch, elm, hemlock, hickory, maple, sugar maple and oak. The dry deposition fluxes of gaseous HNO3 and particulate NO32 and NH4z based on tall grass LAI could be applied to surfaces covered with hardwood forests without causing significant changes in the fluxes of these nitrogen species.
3.1.2. Mass flux density. The dry deposition flux density, F (mg m22 s21), for each wind class was calculated by F ~ CV where C (mg m23), was the weekly concentration of a chemical
Table 1 Mean weekly deposition flux of nitrogen species, March 19, 1999–December 31, 2000 Rural
Urban
Mohawk Mountain
Hammonasset
Avery Point
Voluntown
Bridgeport
Waterbury
East Hartford
Atmospheric concentration/mg m — 0.743 NH4z 0.207 NO32 0.541 HNO3 Organic nitrogen 0.087 Total nitrogen 1.037
0.780 0.297 0.348 0.105 1.181
0.784 0.327 0.406 0.107 1.218
0.760 0.258 0.303 0.084 1.103
1.149 0.499 0.513 0.138 1.806
1.030 0.358 0.402 0.117 1.505
1.052 0.331 0.376 0.119 1.501
Dry deposition flux/kg ha21 week21— 0.047 NH4z 0.013 NO32 0.031 HNO3 Organic nitrogen 0.005 Total nitrogen 0.095
0.038 0.015 0.019 0.005 0.077
0.031 0.013 0.022 0.004 0.071
0.056 0.021 0.019 0.006 0.102
0.065 0.028 0.033 0.007 0.134
0.072 0.023 0.027 0.008 0.129
0.061 0.018 0.027 0.006 0.112
Wet deposition flux/kg ha21 week21— 0.038 NH4z 0.092 NO32 Organic nitrogen 0.037 Total nitrogen 0.167
0.045 0.077 0.013 0.135
0.043 0.085 0.019 0.146
0.056 0.087 0.018 0.160
0.053 0.090 0.031 0.174
0.049 0.083 0.024 0.156
0.053 0.077 0.028 0.158
Bulk deposition flux/kg ha21 week21— 0.056 NH4z 0.116 NO32 Organic nitrogen 0.054 Total nitrogen 0.225
0.053 0.076 0.030 0.159
0.074 0.097 0.026 0.197
0.194 0.070 0.036 0.299
0.067 0.099 0.026 0.192
0.077 0.073 0.030 0.180
0.063 0.074 0.023 0.160
23
species and V (m s21) was the mean weekly dry deposition velocity of the corresponding species, both for the same wind class. All fluxes were converted to kg ha21 week21, using the appropriate conversion factors.
4. Results and discussion 4.1. Nitrogen flux distribution in seven sampling locations Average weekly dry, wet and bulk deposition fluxes of nitrogen for the sampling period March 19, 1999–December 31, 2000 are given in Table 1.
3.2. Estimation of wet deposition flux of nitrogen To estimate the wet deposition flux of nitrogen species, the measured concentration of each nitrogen species was multiplied by the volume of collected water and the resultant was divided by the opening area of the Aerochemetrics collecting device. Using proper physical units, the mean wet deposition flux of nitrogen species were estimated in kg ha21 week21. The weekly wet deposition samples could have been subject to concentration losses due to biological reactions. Butler and Likens23 conducted a study to compare the weekly aggregated MAP3S (1981–89) and AIRMoN (1992–95) daily precipitation chemistry record with the NADP/NTN weekly data at four collocated sites in the eastern Unites States. They found that the weekly and daily network concentrations of NO32 were very comparable for both time periods. However, data for ammonium showed a statistically significant bias for both time periods with the daily record having wet concentrations y14% higher than the weekly values. They stated that the most probable cause of the higher ammonium concentrations in the daily samples was the loss of biologically active ammonium in the weekly samples, due to extended time in the field. In order to investigate the possible losses of inorganic nitrogen species in the wet samples, an additional bulk deposition collector was placed at our East Hartford sampling site during the first week of April 2001. To the present time, bulk deposition samples are being collected on a daily basis. Laboratory QA/QC allowed the use of 4 weeks (May 11–June 15, 2001, in which precipitation took place) of precipitation concentration data for NH4z and NO32 for data comparison. Wet daily concentration of NH4z showed 7.2–8.8% (mean 8.2%) higher concentrations than the weekly samples. Wet daily concentration of NO32 were z2.5% to 28.5% (mean 2 4.4%) different from the weekly concentrations. However, more data are needed to determine a correction factor for weekly wet concentrations of NH4z and NO32.
4.2. Dry deposition flux Deposition fluxes of total nitrogen, nitrate (NO32), nitric acid vapor (HNO3), ammonium (NH4z) and organic nitrogen were calculated for the sampling period of March 19, 1999– December 31, 2000, using weekly measured atmospheric concentrations and calculated deposition velocities. Overall, out of a potential 646 samples, 590 valid dry deposition samples were collected and analyzed for dry deposition data, leading to a 92% capture rate. The maximum and minimum mean weekly dry deposition fluxes of total nitrogen were measured in Bridgeport and Avery Point at 0.134 ¡ 0.017 and 0.071 ¡ 0.010 kg ha21 week21, respectively. The mean weekly dry deposition fluxes of NH4z and total nitrogen were significantly higher in inland sampling locations than the coastal areas (Table 2). This is similar to the findings of Carley et al.6 when they reported the 1997–99 atmospheric deposition fluxes of nitrogen in Connecticut. For all measured dry deposition fluxes of nitrogen species, urban sampling locations were significantly higher than the rural areas (Table 3). However, Carley et al.6 did not find a significant difference Table 2 Mean weekly dry deposition flux (kg ha21 week21) of nitrogen at coastal and inland sites in Connecticut, March 19, 1999–December 31, 2000 Coastal
Inland
Parameter
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
NH4z NO32 HNO3 Organic nitrogen Total nitrogen
0.044 0.018 0.025 0.005 0.092
0.033 0.021 0.028 0.006 0.066
0.059 0.019 0.026 0.006 0.109
0.048 0.035 0.027 0.007 0.096
Inland w coastal No difference No difference No difference Inland w coastal
a
a ~ 0.05; valid sampling events ~ 340.
J. Environ. Monit., 2001, 3, 671–680
675
Table 3 Mean weekly dry deposition flux of nitrogen at urban and rural sites in Connecticut, March 19, 1999–December 31, 2000 Urban Parameter NH4z NO32
Mean
Rural SD
0.066 0.041 0.023 0.021 0.029 0.031 HNO3 Organic nitrogen 0.007 0.008 Total nitrogen 0.124 0.075 a a ~ 0.05; valid sampling events ~
Mean 0.043 0.016 0.022 0.005 0.086 342.
SD 0.042 0.035 0.024 0.006 0.089
Tukey’s pairwise comparison test resultsa Urban w rural Urban w rural Urban w rural Urban w rural Urban w rural
between the dry deposition flux of HNO3 between urban and rural sampling locations. It should be noted that they had an additional urban sampling location (Old Greenwich) included in their study. Owing to site restrictions, the bulk deposition collector could not be placed in this sampling location and, therefore, the results of the dry and wet sampling analysis of this site were not included in this paper. Dry deposition fluxes of total nitrogen indicated a moderate positive correlation among all sampling locations. The dry deposition fluxes of total nitrogen in urban sites in general had higher correaltions among each other than the rural areas. The highest correlation was found between Waterbury and East Hartford (r ~ 0.68) followed by Waterbury and Bridgeport (r ~ 0.51). There was a positive correlation between the urban and rural sites with the highest correlation found between Waterbury and Mohawk Mountain (r ~ 0.51) followed by Waterbury and Hammonasset (r ~ 0.46). The mean monthly dry deposition flux of total nitrogen indicated that the deposition fluxes peaked during summer months and reached their lowest values in late fall and early winter (Fig. 2). Some previous studies of filter artifacts have described losses of particulate nitrate from the first filter and a corresponding excess on the second, due to either evaporation of ammonium nitrate or its conversion to nitric acid.24,25 Ammonium nitrate (NH4NO3) is an atmospheric aerosol that is formed by reaction of gaseous HNO3 and NH3. Harrison and Msibi25 used three different sampling methods (filter pack, luminol-based chemiluminescence continuous analyzer and annular denuder) for the collection of nitric acid vapor at a suburban site south of Birmingham in the UK. They did not find a statistically significant difference between the results from all the three techniques. However, they found a small constant bias between the chemiluminescence continuous method and the annular denuder system, which they related to possible loss or gain of
Table 4 Mean weekly wet deposition flux of nitrogen at coastal and inland sites in Connecticut, March 19, 1999–December 31, 2000 Coastal
Inland
Parameter
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
z
0.047 0.084 0.022 0.151
0.059 0.074 0.030 0.140
0.048 0.085 0.027 0.159
0.054 0.062 0.063 0.135
No No No No
NH4 NO32 Organic nitrogen Total nitrogen
difference difference difference difference
a
a ~ 0.05; valid sampling events ~ 271.
HNO3. The filter pack method produced results that were slightly higher than those from the annular denuder system, which could be attributed to the evaporation of ammonium nitrate from the Teflon filter in the filter pack method and the prefilters in the chemiluminescense continuous method. In this work, however, only filter packs were used for collection of particulate nitrate and nitric acid vapor and measurement of volatilization losses of ammonium nitrate and its conversion to nitric acid was not feasible. Based on the work of other researchers, it appears, however, that the volatilization losses of NH4NO3 would not significantly increase the HNO3 loading on the nylon filter. 4.3. Wet deposition flux Wet deposition fluxes of total nitrogen, nitrate (NO32), ammonium (NH4z) and organic nitrogen were calculated for the sampling period March 19, 1999–December 31, 2000, using weekly measured wet concentrations and the volume of water collected in precipitation collectors. From a total of 515 wet deposition samples, the data for 486 valid samples were analyzed and reported, resulting in a 94% capture rate. The maximum and minimum mean weekly wet deposition fluxes of total nitrogen were measured in Bridgeport and Hammonasset at 0.174 ¡ 0.032 and 0.135 ¡ 0.035 kg ha21 week21, respectively. There was no significant difference between the coastal and inland sampling locations for mean weekly wet precipitation fluxes of different nitrogen species (Table 4). This is similar to the findings of Carley et al.6 There was also no significant difference between the urban and rural sampling locations for mean weekly wet deposition fluxes of different nitrogen species (Table 5). Carley et al. reported significantly higher wet deposition flux of ammonium in urban sampling locations. There was a moderate positive correlation for total wet deposition flux measurements among all sampling sites. The highest correlation in rural sampling locations was found
Fig. 2 Montlhy comparison of wet, dry and bulk deposition fluxes of total nitrogen in Connecticut, March 19, 1999–December 31, 2000.
676
J. Environ. Monit., 2001, 3, 671–680
Table 5 Mean wet deposition flux of nitrogen at urban and rural sites in Connecticut, March 19, 1999–December 31, 2000 Urban Parameter NH4z NO32 Organic nitrogen Total nitrogen
Mean 0.052 0.083 0.028 0.162
Rural SD 0.053 0.061 0.062 0.144
Mean 0.045 0.085 0.022 0.150
SD 0.058 0.072 0.043 0.132
Tukey’s pairwise comparison test resultsa No No No No
difference difference difference difference
Table 6 Bulk deposition flux of nitrogen at coastal and inland sites in Connecticut, March 19, 1999–December 31, 2000 Coastal
Inland
Parameter
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
z
0.065 0.091 0.027 0.181
0.112 0.057 0.046 0.173
0.0977 0.083 0.036 0.207
0.339 0.062 0.056 0.288
No No No No
NH4 NO32 Organic nitrogen Total nitrogen
difference difference difference difference
a
a
between Hammonasset and Avery Point (r ~ 0.749) followed by Voluntown and Avery Point (r ~ 0.645). The highest correlation in urban sampling locations was found between Waterbury and Bridgeport (r ~ 0.616) followed by East Hartford and Bridgeport (r ~ 0.474). There was also a moderate positive correlation between urban and rural sites with the highest correlation found between Waterbury and Hammonasset (r ~ 0.681). The highest mean monthly wet deposition flux of total nitrogen occurred in the months of May and March of 2000. Similar to the dry deposition flux, the wet deposition flux of total nitrogen was lower during winter and fall than the summer and spring sampling periods (Fig. 2). The mean weekly concentration of total nitrogen in the seven sampling locations was plotted against the total weekly precipitation depth on a log-scale graph (Fig. 3). The scattered data points indicate that higher precipitation concentrations of total nitrogen occurred at total weekly precipitation depths of v1 cm. The sample correlation between the log-transformed precipitation data and the mean weekly concentration of total nitrogen was r ~ –0.61, implying that the total weekly precipitation depths were inversely correlated to total nitrogen concentrations. The analysis of variance rejected the null hypothesis H0: b1 ~ 0 (no significant linear association between the mean weekly precipitation concentration of total nitrogen and precipitation depth data sets), because using the logtransformed precipitation data the F-value was 203, which was much higher than the critical F(1, 509, 0.95), P v 0.001 ~ 3.86.
Table 7 Bulk deposition flux of nitrogen at urban and rural sampling sites in Connecticut, March 19, 1999–December 31, 2000
a ~ 0.05; valid sampling events ~ 264.
4.4. Bulk deposition flux Bulk deposition fluxes of total nitrogen, nitrate (NO32), ammonium (NH4z) and organic nitrogen were calculated for the sampling period March 19, 1999–December 31, 2000, using the Swedish IVL Sampler discussed in Section 2.2.3. Overall, out of a potential 658 (7 sites 6 94 weeks) bulk deposition samples, the data for 459 valid samples were analyzed and
a ~ 0.05; valid sampling events ~ 260.
Urban Parameter
Mean
Rural SD
NH4z 0.069 0.071 0.081 0.048 NO32 Organic nitrogen 0.026 0.041 Total nitrogen 0.177 0.111 a a ~ 0.05; valid sampling events ~
Mean
SD
Tukey’s pairwise comparison test resultsa
0.095 0.090 0.037 0.211 263.
0.352 0.067 0.059 0.312
No difference No difference Rural w Urban No difference
reported, leading to a 70% capture rate. The maximum and minimum mean weekly wet deposition fluxes of total nitrogen were measured in Voluntown and Hammonasset at 0.299 ¡ 0.125 and 0.159 ¡ 0.025 kg ha21 week21, respectively. There was no significant difference between the coastal and inland sampling locations for mean weekly bulk fluxes of different nitrogen species (Table 6). The bulk deposition of organic nitrogen in rural locations was significantly higher than the urban sites (Table 7). The sample correlation coefficient for bulk deposition of total nitrogen between sites varied from site to site. The highest positive correlation was found between Mohawk Mountain and East Hartford (r ~ 0.720) followed by Waterbury and East Hartford (r ~ 0.482). The lowest negative correlation was found between Hammonasset and Bridgeport (r ~ 20.100) followed by Voluntown and Mohawk Mountain (r ~ 20.093). When urban and rural sites were separately grouped together, a moderate positive correlation was found between the urban and rural sampling locations (r ~ 0.343). The mean monthly bulk deposition of total nitrogen followed the same pattern as the wet and dry deposition fluxes. The bulk deposition fluxes peaked in late spring and summer and reached their lowest levels in late fall and winter. The highest
Fig. 3 Total weekly precipitation depth versus mean weekly concentration of total nitrogen in precipitation in Connecticut, March 19, 1999– December 31, 2000.
J. Environ. Monit., 2001, 3, 671–680
677
mean monthly bulk deposition flux of total nitrogen occurred in May 1999, and July 2000 (Fig. 2). 4.5. Comparison of bulk deposition fluxes with the sum of wet and dry deposition fluxes of nitrogen In order to assess the accuracy of the IVL bulk deposition sampler, a comparison between the mean weekly deposition fluxes of bulk total nitrogen with the sum of wet and dry deposition fluxes of total nitrogen at each sampling location was made. There was no significant difference between the bulk deposition flux of total nitrogen and the sum of wet and dry fluxes of total nitrogen in the four rural sampling locations. Significant difference was observed between the two fluxes for the urban sampling locations (Table 8) (Fig. 4). The comparison between the monthly averages indicated that the bulk deposition and the sum of wet and dry deposition fluxes of total nitrogen corresponded well during the months of June and November of 1999 (Fig. 2). 4.6. NADP and CASTNet atmospheric monitoring sites in Connecticut The National Atmospheric Deposition Program (NADP) has set up a sampling station in Abington, Connecticut to measure the weekly precipitation concentration of various chemicals including inorganic nitrogen species. The USEPA Clean Air Status and Trends Network (CASTNet) has a dry deposition sampling station in the same location (Fig. 1). Abington is a rural area located in northeastern Connecticut in Windham
County. The NADP and CASTNet methods of wet and dry deposition collections in Abington were similar to the methods used in this study. For calculation of dry deposition velocities of gaseous and particulate nitrogenous compounds, however, CASTNet used a multilayer DDIM model,26 whereas in this study a singlelayer DDIM model was used. The mean annual concentrations of dry and wet deposition fluxes of nitrogen estimated for Abington in 1999 were compared with the mean values obtained for rural sites (January 1–December 31, 1999) in this study. The mean weekly wet deposition flux of nitrogen (0.101 ¡ 0.027 kg ha21 year21) measured by NADP in Abington in 1999 was comparable to the mean weekly wet deposition flux of nitrogen (0.125 ¡ 0.022 kg ha21 year21) measured in rural areas in this study. The mean weekly dry deposition flux of nitrogen (0.046 ¡ 0.033 kg ha21 year21) measured by CASTNet in Abington was comparable to the total dry deposition flux of nitrogen (0.071 ¡ 0.008 kg ha21 year21) measured in rural areas in this study. The total nitrogen fluxes reported by NADP and CASTNet were 35 and 20% lower, respectively, than the mean deposition fluxes reported in rural areas of this study. Various reasons may be responsible for this phenomenon. For wet deposition flux measurements of nitrogen, NADP reported NH4z and NO32. For dry deposition, CASTNet reported NH4z, NO32 and HNO3. Neither CASTNet nor NADP reported organic nitrogen that was reported in this study. Abington is located in a rural area northeast of Connecticut, where no major towns, industrial activities or heavy traffic that could be a major source of oxidized nitrogen exist within a 40 km radius of this site.
Table 8 Comparison between the bulk deposition flux of nitrogen with the sum of wet and dry fluxes, March 19, 1999–December 31, 2000 Bulk deposition flux
Sum of wet and dry deposition flux
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
Bridgeport (urban coastal)— NH4z NO32 Organic nitrogen Total nitrogen
0.067 0.099 0.026 0.192
0.040 0.052 0.049 0.095
0.099 0.124 0.030 0.262
0.060 0.082 0.041 0.155
Valid sampling events ~ 87 Wet and dry w bulk Wet and dry w bulk No difference Wet and dry w bulk
Waterbury (urban inland)— NH4z NO32 Organic nitrogen Total nitrogen
0.077 0.073 0.030 0.180
0.106 0.044 0.042 0.143
0.103 0.105 0.024 0.232
0.609 0.076 0.027 0.140
Valid sampling events ~ 92 No difference Wet and dry w bulk No difference No difference
East Hartford (urban inland)— NH4z NO32 Organic nitrogen Total nitrogen
0.063 0.074 0.023 0.160
0.041 0.046 0.030 0.082
0.097 0.101 0.026 0.225
0.078 0.068 0.082 0.189
Valid sampling events ~ 91 Wet and dry w bulk Wet and dry w bulk No difference Wet and dry w bulk
Mohawk Mountain (rural inland)— 0.056 NH4z 0.116 NO32 Organic nitrogen 0.054 Total nitrogen 0.224
0.061 0.086 0.072 0.153
0.074 0.114 0.033 0.221
0.051 0.076 0.067 0.142
Valid sampling events ~ 90 No difference No difference No difference No difference
Hammonasset (rural coastal)— NH4z NO32 Organic nitrogen Total nitrogen
0.053 0.076 0.030 0.157
0.050 0.049 0.044 0.105
0.069 0.089 0.013 0.169
0.070 0.080 0.015 0.150
Valid sampling events ~ 88 No difference No difference Bulk w wet and dry No difference
Avery Point (rural coastal)— NH4z NO32 Organic nitrogen Total nitrogen
0.074 0.097 0.026 0.194
0.181 0.067 0.045 0.261
0.0584 0.091 0.016 0.164
0.059 0.086 0.016 0.150
Valid sampling events ~ 90 No difference No difference Bulk w wet and dry No difference
Voluntown (rural inland)— NH4z NO32 Organic nitrogen Total nitrogen
0.193 0.069 0.036 0.262
0.657 0.049 0.066 0.526
0.092 0.098 0.018 0.208
0.086 0.098 0.020 0.183
Valid sampling events ~ 91 No difference Wet and dry w bulk Bulk w wet and dry No difference
a
a ~ 0.05.
678
J. Environ. Monit., 2001, 3, 671–680
Fig. 4 Comparison of mean total deposition of nitrogen with wet and dry deposition fluxes in Connecticut, March 19, 1999–December 31, 2000.
5. Conclusions Wet, dry and bulk deposition fluxes of atmospheric nitrogen were measured in seven sampling locations in Connecticut for a period of 94 weeks. The overall mean weekly wet, dry and bulk deposition fluxes of nitrogen in Connecticut were 0.157, 0.103 and 0.202 kg ha21 week21, respectively. The data collected in this study indicated that there is no discernible spatial gradient for wet, dry and bulk deposition fluxes of total nitrogen in Connecticut. The dry deposition fluxes of nitrogen species were higher in urban areas than the rural sampling locations. No significant difference was observed between the wet deposition fluxes of total nitrogen between rural and urban sampling locations. When inland and coastal sites were compared, the dry deposition fluxes of NH4z and total nitrogen were significantly higher in inland locations and no significant difference was observed between the coastal and inland sampling locations for wet deposition fluxes of nitrogen species. The highest wet and dry deposition fluxes of total nitrogen were measured in May 2000 and July 1999, respectively. Wet and dry deposition fluxes of total nitrogen peaked in summer and reached their lowest values during late fall and winter sampling periods. Many industrial activities take place along the shores of Hudson River north of New York metropolitan area. In the past few years, the population of New York City and its suburbs has had an increasing trend resulting into more congested traffic and more vehicular emissions. Mohawk Mountain is a rural site located in northwestern part of Connecticut. This site is approximately 50 km east of Hudson River and 100 km northeast of the New York City area. Therefore, high dry, wet and bulk deposition fluxes of nitrogen observed at this sampling location maybe due to the proximity of the site to areas with high emission sources. Voluntown is another rural site located in eastern part of Connecticut about 35 km west of Providence (capital of Rhode Island) that is a center of many industrial activities. Rhode Island and southeastern Connecticut are major tourist attraction areas all year around. Heavy traffic and industrial emissions could have been the main causes of high deposition fluxes of nitrogen observed at Voluntown sampling location. The lifetime of NH3 ranges from y0.5 h to 5 d. This short lifetime is attributed to the rapid gas-to-particle conversion of
NH3 to NH4z and deposition of NH3 to natural surfaces. The lifetime of particulate NH4z is 5–10 days. If the conversion of NH3 to NH4z proceeds slowly, most of the NH3 emissions will be deposited locally and less NH4z will be made available for long-range transport.27 It maybe possible that a major portion of the ammonia gas transported to Connecticut as long-range transport will be converted to ammonium particles. In this study, NH3 and NO2 were not quantitatively captured, which may have led to low bias for nitrogen dry and total deposition calculations. Agricultural and livestock activities are not very common in Connecticut and forests cover more than 60% of the land surface. To identify the spatial distribution of ammonia and nitrogen dioxide in Connecticut, the authors suggest that measurements of atmospheric concentrations of NH3 and NO2 in at least one rural and one urban sampling location in Connecticut be added to the program schedule.
Acknowledgements We acknowledge the financial support of the Connecticut Department of Environmental Protection (CTDEP) and the Environmental Research Institute (ERI) for this project. We also acknowledge the collaborative efforts and feedback from Carmine DiBattista, Thomas Morrissey, Robert Smith, Teresa Gutowski and Paul Stacey of CTDEP and Hugo Thomas and David Miller of the University of Connecticut. This project could not have been completed without the sampling help from CTDEP Air Toxics Monitoring Group personnel. Finally, we acknowledge the efforts of the ERI personnel who conducted sampling and analytical duties for this study: Janet E. Heitert, Steve Chirdon, Mark Groszek, Stephan Nowakowski, Mike Morrison, Dan Tubbs, Gary Ulatowski and Todd Wheeler.
References 1 Connecticut Department of Environmental Protection, The Atmosphere and Hypoxia in Long Island Sound, Hartford, Connecticut, http://www.dep.state.ct.us/wtr/lis/atmoslis.htm, 2000. 2 H. L. Hu, H. M. Chen, N. P. Nikolasdis, D. R. Miller and X. Yang, Water Air Soil Pollut., 1998, 105, 521. 3 Connecticut Department of Environmental Protection, Water Quality of Long Island, Hartford, Connecticut, http://www.dep. state.ct.us/wtr/lis/wtrqual.htm, 2000. 4 L. J. Puckett, Nonpoint and Point Sources of Nitrogen in Major
J. Environ. Monit., 2001, 3, 671–680
679
5 6
7 8 9 10 11 12 13 14
680
Watersheds of the United States, Water Resources Investigations Report 94-001, US Geological Survey, Reston, VA, 1994. Connecticut Department of Environmental Protection, Long Island Sound: Fact Sheet, DEP-FS-08, Connecticut Department of Environmental Protection, Hartford, CT, 1999. R. J. Carley and C. Perkins, Nitrogen Deposition Monitoring in Connecticut: Final Report, submitted to Connecticut Department of Environmental Protection, Hartford, Connecticut, www.eri. uconn.edu, 2001. M. A. Geigert, N. P. Nikolaidis, D. R. Miller and J. Heitert, Atmos. Environ., 1994, 28, 1689. National Atmospheric Deposition Program (NADP), NADP/NTN Sites in Connecticut, CT15, Abington, http://nadp.sws.uiuc.edu/, 1999. USEPA Clean Air Status and Trends Network (CASTNet), Sites/ Abington, Connecticut, Monitoring Station ABT147, http://www. epa.gov/castnet/, 1999. A. Iverfeldt, Water Air Soil Pollut., 1991, 56, 553. P. M. Berthouex and L. C. Brown, Statistics for Environmental Engineers, Lewis Publishers, CRC Press, Boca Raton, FL, 1994. D. G. Kleinbaum, L. L. Kupper and K. E. Muller, Applied Regression Analysis and Other Multivariable Methods, PWS-Kent Publishing, Boston, 1988. W. J. Lyman, W. F. Reehl and D. H. Rosenblatt, Handbook of Chemical Property Estimation Methods, American Chemical Society, Washington, DC, McGraw Hill, New York, 1990. E. S. Edgerton, T. F. Lavery and H. S. Prentice, Project Summary:
J. Environ. Monit., 2001, 3, 671–680
15 16 17 18 19 20
21 22 23 24 25 26 27
Nation Dry Deposition Network Third Annual Progress Report, ESE, Gainesville, FL, 1990. T. P. Meyers, B. B. Hicks, J. P. Hosker Jr., J. D. Womack and L. C. Satterfield, Atmos. Environ., 1991, 25, 2361. G. Lear, U.S. Environmental Protection Agency, Clean Air Markets Division, Washington, DC, personal communications. R. I. Smith, D. Fowler, M. A. Sutton, C. Flechard and M. Coyle, Atmos. Environ., 2000, 34, 3757. H. V. Anderson and M. F. Hovmand, Atmos. Environ., 1994, 28, 3495. B. B. Hicks, D. D. Baldocchi, T. P. Meyers, R. P. Hosker and D. R. Matt, Water Air Soil Pollut., 1987, 36, 311. J. F. Clarke and E. S. Edgerton, Dry Deposition Flux Calculations for the National Dry Deposition Network, Atmospheric Research and Exposure Assessment Laboratory, USEPA, Research Triangle Park, NC, EPA/600/R-93/065, 1993. X. Yang, D. R. Miller, X. Xu, L. H. Yang, H. M. Chen and N. P. Nikolaidis, Atmos. Environ., 1996, 30, 3801. M. A. Geigert, Masters Thesis, University of Connecticut, Storrs, CT 1991. T. J. Butler and G. E. Likens, Atmos. Environ., 1998, 32, 3749. A. G. Allen and R. M. Harrison, Atmos. Environ., 1989, 23, 1591. R. M. Harrison and M. I. Msibi, Atmos. Environ., 1994, 28, 255. T. P. Meyers, P. Finkelstein, J. Clarke, T. G. Ellestad and P. F. Sims, J. Geophys. Res., 1998, 103, 22645. J. T. Walker, P. V. Aneja and D. A. Dickey, Atmos. Environ., 2000, 34, 3407.
University of Connecticut, The Environmental Research Institute, Longley Building, U-5210, 270 Middle Turnpike (Rte. 44), Storrs, CT 06269, USA. E-mail: [email protected]; Fax: (860) 486 5488; Tel: (860) 486 4015 b Institute of Physics, A. Gostauto 12, 2600 Vilnius, Lithuania c University of Connecticut, Natural Resources Management and Engineering, 1376 Storrs, Road, U-87, Storrs, CT 06269, USA Received 1st August 2001, Accepted 5th October 2001 First published as an Advance Article on the web 8th November 2001
Atmospheric nitrogen species including NO32, NH4z and total nitrogen in air and precipitation samples were collected with low-volume filter packs and wet deposition collectors from March 1999 through the end of December 2000 in seven sampling locations in Connecticut. Three sampling locations were chosen along the shores of Long Island Sound and four were chosen in interior sections of Connecticut. Sampling sites were chosen to represent both rural and urban sectors. Wet deposition flux of nitrogen species was calculated using wet concentrations, the volume of collected precipitation and the opening surface area of the Aerochemetrics wet deposition collector. The dry deposition flux of nitrogen species was estimated with the application of the dry deposition inferential model (DDIM). Bulk deposition of nitrogen was collected with the aid of a device based on the Swedish IVL Sampler. The dry deposition fluxes of NO32, NH4z and total nitrogen were found to be significantly higher in urban areas than the rural sampling locations. There was, however, no significant difference between the wet deposition fluxes of different nitrogen species in rural and urban sampling locations. When inland and coastal sites were compared, the dry deposition fluxes of NH4z and total nitrogen were significantly higher in inland locations and there was no significant difference between coastal and inland sampling locations for wet deposition fluxes of nitrogen species. No significant difference was observed between the bulk deposition and the sum of the wet and dry deposition fluxes of total nitrogen at rural sampling locations. In urban sampling locations, the bulk deposition flux of total nitrogen was significantly lower than the sum of dry and wet deposition fluxes. There appears to be a similar seasonal trend in wet and dry deposition fluxes of total nitrogen in Connecticut with high and low deposition fluxes occurring in summer and winter periods, respectively.
1. Introduction Overabundance of nitrogen is of concern in areas that have developed nutrient enrichment problems (i.e., eutrophication). In addition to increasing productivity, nutrient enrichment generally alters the normal ratios of nitrogen to phosphorus and to other elements such as silicon.1 This alteration may induce changes in phytoplankton community structure. Species that normally occur in low abundances may be favored and, in some cases, toxic and/or noxious algal blooms may result. The Long Island Sound (LIS) is one of the 17 major estuarine systems of the United States under EPA’s National Estuary Program that needs to be protected.2 The LIS has a poor or stratified circulation pattern and the over production of algae caused by excessive loading of nutrients especially nitrogen tends to sink to the bottom and decay, using all (anoxia) or most (hypoxia) of the available oxygen in the process, causing loss of habitat.3,4 Atmospheric nitrogen originating from anthropogenic sources is estimated to cause more than 15% of the dissolved oxygen deficit observed in LIS. Recent studies have shown the importance of managing atmospheric sources of nitrogen if water quality objectives are to be achieved and maintained in LIS.5,6 Atmospheric nitrogen can be deposited in the sound in two ways: direct and indirect deposition.7 Direct DOI: 10.1039/b107008h
deposition on LIS contributes to 5.9% of the anthropogenic load of nitrogen to the Sound from all sources. Indirect deposition of nitrogen includes discharges from industry and treatment plants (point sources), storm water runoff from coastal areas, with rivers and streams from throughout the drainage basin (non-point sources) and currents moving into the Sound from the Atlantic Ocean and New York Harbor.3,4 Ammonia and oxides of nitrogen are mineralized forms of nitrogen that are readily usable by plants such as phytoplankton. A small amount of organic nitrogen is also found in atmospheric deposition, which has not been fully characterized. To study the spatial and temporal variation of atmospheric deposition of nitrogen in Connecticut, a collaborative study between the Connecticut Department of Environmental Protection (CTDEP) and the Environmental Research Institute (ERI) at the University of Connecticut was conducted. Wet, dry and bulk deposition atmospheric nitrogen samples were collected and analyzed for estimation of atmospheric wet, dry and bulk deposition fluxes of nitrogen species in Connecticut for a period of 94 weeks. The National Atmospheric Deposition Program (NADP)8 and the USEPA Clean Air Status and Trends Network (CASTNet)9 have a site in a rural section of northeastern Connecticut to monitor the atmospheric deposition of pollutants including NO32, NH4z and HNO3. The mean J. Environ. Monit., 2001, 3, 671–680
This journal is # The Royal Society of Chemistry 2001
671
weekly wet and dry deposition fluxes of total nitrogen reported by NADP and CASTNet were compared with the mean fluxes measured at the rural sampling locations of this study.
2. Methods 2.1. Study site selection Seven sites were selected to monitor ambient levels of atmospheric nitrogen in Connecticut. The primary factor in the selection process was to obtain a distribution of sites to reflect adequately the ambient conditions throughout the State. The other factors included in choosing the sites were the availability of access, power, security and the presence of existing nearby meteorological monitoring stations. Bridgeport (40 km northwest of New York City) was chosen to represent atmospheric nitrogen levels of a highly populated urban and industrial coastal area. Waterbury and East Hartford sites were chosen to represent inland urban levels of atmospheric nitrogen. Hammonasset State Park (Madison) and Avery Point (Groton) were chosen as coastal rural and Mohawk Mountain and Voluntown were chosen as rural interior sites for measurement of atmospheric concentration of nitrogen (Fig. 1). Sampling was performed on Fridays of every week at all of the sites mentioned above for a period of 94 weeks. Samples were collected weekly and analyzed for total nitrogen, ammonium, organic nitrogen and oxides of nitrogen (nitrite/nitrate). The goal of this study was to monitor the spatial and temporal variation of atmospheric nitrogen in Connecticut. 2.2. Field procedures Before collection of samples, all equipment and sample containers were rigorously cleaned using detergent, concentrated acids and de-ionized water. Prior to deployment in the field, sampling equipment and containers were demonstrated to be free from contamination. All sampling equipment and containers were constructed of materials intrinsically low in contaminants. Sampling personnel were required to wear clean, talc-free rubber gloves. All samples collected in the field were placed inside transportable coolers, brought to the laboratory and kept in a walk-in cooler pending analysis.
2.2.1. Dry deposition sample collection. Dry deposition samples were collected on filters. A weekly composite sample of ambient air was drawn through a filter pack at each site and analyzed for total nitrogen, ammonium and oxidized nitrogen. Each site contained a two-stage filter pack with Teflon and nylon filters in series housed in a fiber-glass NEMA enclosure. The Teflon filter (2 mm Gelman Laboratory, Zelfluor) was used to collect dry deposition of particulate nitrogen (total nitrogen, ammonium and NO32). The nylon filter (1 mm Gelman Laboratory, Zelfluor) was used to collect dry deposition of nitric acid (HNO3) vapor. Nylon filter material is an absolute scavenger of HNO3 7 Flow through the filter pack was set to 3.0 l min21 and the target sample volume collected over the course of the week was 30 m3. Teflon filters were extracted with 30 ml of de-ionized water. Teflon filter extracts were analyzed for NH4z and oxidized and total nitrogen. Nylon filters were extracted with 30 ml of a dilute solution of sodium hydroxide and hydrogen peroxide. Nylon filter extracts were analyzed for oxidized nitrogen and the results were reported as HNO3. 2.2.2. Wet deposition sample collection. The network of seven air monitoring stations was outfitted to collect precipitation (wet deposition) to be analyzed for total nitrogen, ammonium and oxidized nitrogen. At the initiation of rain events, moisture triggered a tipping bucket lever, which in turn resulted in the lid being removed from the Aerochemetrics wet deposition collector. At the termination of rain events, the lid swung back and re-sealed the wet deposition collector. Each of the wet deposition buckets was lined with a pre-cleaned Teflon bag to minimize contamination. The wet deposition collectors were sampled weekly. A clean Teflon bag was inserted into the collection apparatus when no rain events occurred within the week. 2.2.3. Bulk deposition sample collection. Collections of bulk deposition samples were initiated on March 19, 1999, at seven sampling locations. A sample collection device based upon the Swedish IVL Sampler was used.10 The device is made of a PVC pipe (y100 cm high with 100 mm id). Inside the pipe a cylindrical glass vessel (110 mm od and 90 mm id) was inserted that was open on top and narrowed at the bottom center to a 5 mm id glass tube. The glass tube extended down to a 300 ml
Fig. 1 Atmospheric nitrogen sampling sites in Connecticut.
672
J. Environ. Monit., 2001, 3, 671–680
BOD bottle that was inserted in the bottom of the PVC pipe. An inverted watch-glass was placed inside the glass beaker to prevent clogging of the tube from debris. During precipitation events, water was directed to the 300 ml BOD bottle. The edge of the opening section of the glass vessel had a convex curve towards the interior part to guide all precipitation towards the bottom of the vessel. At the time of sample collection, the surface of the collecting beaker was rinsed with y100-ml of deionized (DI) water and the rinsate was collected in a precleaned 300 BOD bottle, capped and taken to the laboratory together with the original sample for analysis. After the sample and the rinsate had been taken, the interior of the glass beaker was rinsed with y100 ml of HCl solution followed by 100 ml of DI water and collected in a waste bottle. 2.3. Field quality control 2.3.1. Field blanks. Each sampling crew was equipped with a field blank each week to detect contamination that may have occurred in the sample transport process. This consisted of both a filter pack and a bottle of DI water placed in a plastic bag. The trip blanks were placed in the transport coolers, packaged with the actual sampling media and remained closed for the entire trip. The blanks were returned to ERI with the samples and analyzed in the same manner. 2.3.2. Bag blank. One Teflon bag used in the wet deposition collectors was analyzed for any residual nitrogen contamination that could have originated from the laboratory. After the cleaning process, one bag was rinsed an additional time with 500 ml of DI water. This rinse was poured into a sample bottle and analyzed in the same manner as the wet deposition samplers. One bag blank was prepared weekly for all eight sampling locations. 2.3.3. Co-located (i.e., ‘paired’ samples) sampler. Each week, one station on a rotating basis, had a co-located sampler to determine precision of the sampling technique. The co-located sampler was placed adjacent to the regular sampler and was set up in the same way (i.e., flow rates, equipment, sampling media). The co-located samples were returned to ERI with the regular samples and analyzed in the same manner. 2.4. Analytical procedures 2.4.1. Ammonium (NH4z). Wet and dry samples for ammonium were analyzed by an automated procedure, on a Lachat Quikchem Autoanalyzer, utilizing the Berthelot reaction, in which the formation of a blue compound, believed to be closely related to indophenol, occurs when the solution of an ammonium salt is added to sodium phenoxide, followed by the addition of sodium hypochlorite. A solution of EDTA was added to the sample stream to eliminate the precipitation of the hydroxides of calcium and magnesium. Sodium nitroprusside was added to intensify the blue color. The Lachat Quikchem Autoanalyzer was calibrated with a five-point curve at the time of analysis. The calibration curve was then verified by an external quality control sample from Environmental Resource Associates, Arvada, CO (Waste Water Quality Control Standards). This initial calibration check and initial calibration blank demonstrated that the instrument performance was acceptable at the beginning of the sample analysis. In order to ensure continually acceptable performance, a continuing calibration check and continuing calibration blank were run every tenth sample. For every sample delivery group (SDG, ¡20 samples), a laboratory spike analysis and a laboratory duplicate analysis were performed. 2.4.2. Nitrate z nitrite (oxidized nitrogen) (Teflon and nylon filters). Samples for nitrate and nitrite were analyzed by an
automated procedure, on a Lachat Quikchem Autoanalyzer, whereby nitrate was reduced to nitrite by passage of the sample through a copperized cadmium column. The nitrate reduced to nitrite plus any free nitrite present react under acidic conditions with sulfanilamide to form a diazo compound which coupled with N-1-naphthylethylenediamine dihydrochloride to form a reddish–purple azo dye that was measured at 520 nm. The Lachat Quikchem Auto Analyzer was calibrated with a fivepoint curve at the time of analysis. The calibration curve was then verified by an external quality control sample from Environmental Resource Associates. This second source calibration check and initial calibration blank demonstrated that the instrument was capable of acceptable performance at the beginning of the sample analysis. In order to ensure continually acceptable performance, a continuing calibration check and continuing calibration blank were run every tenth sample. For every sample delivery group (SDG, ¡20 samples), a laboratory spike analysis and a laboratory duplicate analysis were performed. 2.4.3. Total nitrogen (TN). Sample digestion was completed within 14 d from arrival at the laboratory. The sample was placed in screw-capped test-tubes with an oxidizing reagent (potassium persulfate–sodium hydroxide). The tubes were placed in a pressure cooker at 100 uC (3–4 psi) for 60 min. After the samples had cooled to room temperature, hydrochloric acid was added to each sample. Through the processes mentioned above all nitrogen species will be converted into nitrate. The samples were then placed on a vortex mixer to dissolve any precipitates. A boric acid–sodium hydroxide buffer was then added to bring the pH of the sample in the range 7–8. The sample was then ready for the determination of nitrate z nitrite as described in Section 2.4.2. 2.4.4. Organic nitrogen (ON). Organic nitrogen species are typically referred to as a subset of reactive nitrogen. These chemical species result from direct emissions of organic nitrogen compounds or from the interaction between nitrogen and biogenic or anthropogenic and natural sources, speciation of organic nitrogen in the atmosphere is poorly understood. Organic nitrogen flux was quantified by subtracting the oxidized nitrogen and ammonium (NH4z) fluxes from the total nitrogen flux. The total nitrogen flux was derived using the product of total atmospheric concentration of nitrogen and the deposition velocity calculated for particulate nitrogen with the DDIM model. The organic nitrogen flux may also be referred to as the particulate organic nitrogen flux. It should be noted that the organic nitrogen fractions of the total wet and dry depositions were not qualitatively specified by laboratory analysis. 2.5. Laboratory quality assurance/quality control Initial and continuing calibration verification (ICV and CCV) were performed to ensure that the instrument was calibrated properly and remained so during the analytical run. Initially every 10 samples and upon completion of analysis a standard were analyzed. The acceptance criteria for ICV and CCV analyses were 90–110% recoveries. 2.5.1. Preparation, initial and continuing calibration blank. The assessment of blank analysis is to determine the existence and magnitude of contamination problems. A preparation blank was prepared and analyzed for every 20 samples that were digested. A matrix-matched DI water sample was analyzed after the initial calibration, every 10 samples and upon completion of the analytical run. J. Environ. Monit., 2001, 3, 671–680
673
2.5.2. Laboratory duplicate. Duplicate analyses are an indicator of laboratory precision. For every 20 samples a laboratory duplicate analysis was analyzed. The acceptance criteria for the laboratory duplicate analyses were 20% relative percentage difference. 2.5.3. Matrix spike sample analysis. Matrix spike samples are used to monitor, assess and control bias resulting from sample matrix on sample digestion and analysis procedures. For every 20 samples a matrix spike sample analysis was performed. The acceptance criteria for the matrix spike sample analyses were 80–120% recoveries.
3. Data analysis Data were grouped by collection year, location, season, coastal/inland and rural/urban locations. To compare the population means obtained in this study, Tukey’s multiple comparison test method was used. Tukey’s pairwise comparison test was conducted for two mean values obtained for the same sampling period. Tukey’s pairwise comparison test method has been fully described by Berthouex and Brown,11 Kleinbaum et al.12 and Lyman et al.13 Tukey’s pairwise comparison test method makes use of the Studentized range distribution and keeps the family error rate a at a fixed level. This method sets up a (1 2 a)100% confidence interval for the true difference between the two or more means. Instead of testing the null hypothesis that states the difference in the means of two samples is zero, the confidence interval of the difference was tested. If this confidence interval included zero, then it was concluded that there was no statistically significant difference between the two sample means. The a value used throughout this study was 0.05. Data presented graphically were average values calculated from the appropriate dataset (e.g., year, inland, coastal, urban and rural). The Minitab software version 13.3 was used for all statistical analyses. All nitrogen species fluxes were reported as nitrogen. For statistical analysis of the seasonal distribution of atmospheric nitrogen during the course of the study, winter was considered as 1 December–28 February, spring was 1 March–31 May, summer was 1 June–31 August and fall (autumn) was 1 September–30 November. To assess the spatial and temporal association strength of two data set results, the sample correlation coefficient r was used. When r w 0 there was a positive correlation and when r v 0 there was a negative correlation between the two population samples. The association strength increased as r A ¡1.
3.1. Estimation of dry deposition flux of nitrogen Dry deposition is defined as the gases and particles that are deposited to the ground and vegetated surfaces. Previous studies have indicated that acid deposition causes adverse effects on the ecosystem. Studies on dry deposition have shown that it can contribute a nearly equal flux of acid deposition as acid rain.14,15 Measurements of dry deposition fluxes of nitrogen involve the estimation of the deposition velocities of nitrogenous gas and particles. The deposition velocity can be defined as the rate at which each pollutant species is transported towards the Earth’s surface. Weekly concentrations are multiplied by weekly averaged deposition velocities to determine the deposition fluxes of nitrogen. Dry deposition fluxes of ammonia (NH3) and nitrogen dioxide (NO2) were not measured in this study because their measurements required additional equipment that was beyond the objectives outlined by CTDEP. However, based on existing literature, the contribution of the above two gaseous inorganic nitrogen to the dry flux of nitrogen in Connecticut can be 674
J. Environ. Monit., 2001, 3, 671–680
approximated. However, Lear16 states that the contribution of NH3 to total deposition of reduced nitrogen in the United States ranges from 5 to 20% and the contribution of NO2 to total deposition of oxidized nitrogen is in the range 2–15%. Smith et al.17 used a ‘big-leaf’ resistance analogy model for estimation of dry deposition of nitrogen dioxide, ammonia and nitric acid in the United Kingdom. They stated that NO2 dry deposition accounted for 15% of the total oxidized nitrogen deposition in the UK. It could roughly be estimated that the dry deposition flux of NO2 in Connecticut would account for 10–15% of the total oxidized nitrogen flux. Anderson and Hovmand18 conducted a study comparing the measurements of gaseous ammonia (NH3) and particulate ammonium (NH4z) with denuders and filter packs at five different locations in Denmark. They found that the contribution of NH3 to the total concentration of NH3 and NH4z ranged from 18 to 27% and from 14 to 31% for denuder and filter pack collection methods, respectively. The lowest range of NH3 contribution was observed at one of their sampling sites (Anholt Island), which had a distance of at least 50 km to the nearest emission sources of NH3 Considering Connecticut’s landscape and land use, it is likely that the contribution of ammonia to the total deposition of reduced nitrogen could be in the 15–20% range. However, conducting a statistical comparison at local and regional scales requires real measurements of atmospheric ammonia concentrations in Connecticut.
3.1.1. DDIM model for deposition velocity calculations. The model used for the determination of dry deposition velocities of gaseous and particulate nitrogen is referred to as the dry deposition inferential model or DDIM. This model is used by the National Dry Deposition Network. Details of DDIM can be found in Hicks et al.,19 Clarke and Edgerton20 and Yang et al.21 For the purpose of this study, two velocities were calculated for each site. The velocity for particles was used for ions (NO32, NH4z and total particulate nitrogen) and the gaseous velocity was used for nitric acid vapor. The urban sampling sites were set up in areas surrounded by grass that was mowed in regular intervals. The rural sampling locations were set up in areas surrounded by tall grass. The leaf area index of tall grass was used as an input parameter for calculation of dry deposition velocities at all sampling locations. Geigert22 conducted a study for the estimation of dry deposition of some inorganic gaseous and particulate sulfur and nitrogen to a forested area in north central Connecticut in 1990. He used the DDIM model for the estimation of dry deposition velocities of gaseous and particulate nitrogen and conducted a sensitivity analysis on the DDIM model by arbitrarily varying the input parameters. He stated that the leaf area index (LAI) only affected the SO2 deposition velocity when the vegetation cover was changed from hardwood forest to tall grass and no significant effect on the deposition velocities of nitric acid vapor and particulate nitrate (NO32) was observed,7,22 Hardwood forests cover more than 60% of Connecticut. Trees include ash, beech, birch, elm, hemlock, hickory, maple, sugar maple and oak. The dry deposition fluxes of gaseous HNO3 and particulate NO32 and NH4z based on tall grass LAI could be applied to surfaces covered with hardwood forests without causing significant changes in the fluxes of these nitrogen species.
3.1.2. Mass flux density. The dry deposition flux density, F (mg m22 s21), for each wind class was calculated by F ~ CV where C (mg m23), was the weekly concentration of a chemical
Table 1 Mean weekly deposition flux of nitrogen species, March 19, 1999–December 31, 2000 Rural
Urban
Mohawk Mountain
Hammonasset
Avery Point
Voluntown
Bridgeport
Waterbury
East Hartford
Atmospheric concentration/mg m — 0.743 NH4z 0.207 NO32 0.541 HNO3 Organic nitrogen 0.087 Total nitrogen 1.037
0.780 0.297 0.348 0.105 1.181
0.784 0.327 0.406 0.107 1.218
0.760 0.258 0.303 0.084 1.103
1.149 0.499 0.513 0.138 1.806
1.030 0.358 0.402 0.117 1.505
1.052 0.331 0.376 0.119 1.501
Dry deposition flux/kg ha21 week21— 0.047 NH4z 0.013 NO32 0.031 HNO3 Organic nitrogen 0.005 Total nitrogen 0.095
0.038 0.015 0.019 0.005 0.077
0.031 0.013 0.022 0.004 0.071
0.056 0.021 0.019 0.006 0.102
0.065 0.028 0.033 0.007 0.134
0.072 0.023 0.027 0.008 0.129
0.061 0.018 0.027 0.006 0.112
Wet deposition flux/kg ha21 week21— 0.038 NH4z 0.092 NO32 Organic nitrogen 0.037 Total nitrogen 0.167
0.045 0.077 0.013 0.135
0.043 0.085 0.019 0.146
0.056 0.087 0.018 0.160
0.053 0.090 0.031 0.174
0.049 0.083 0.024 0.156
0.053 0.077 0.028 0.158
Bulk deposition flux/kg ha21 week21— 0.056 NH4z 0.116 NO32 Organic nitrogen 0.054 Total nitrogen 0.225
0.053 0.076 0.030 0.159
0.074 0.097 0.026 0.197
0.194 0.070 0.036 0.299
0.067 0.099 0.026 0.192
0.077 0.073 0.030 0.180
0.063 0.074 0.023 0.160
23
species and V (m s21) was the mean weekly dry deposition velocity of the corresponding species, both for the same wind class. All fluxes were converted to kg ha21 week21, using the appropriate conversion factors.
4. Results and discussion 4.1. Nitrogen flux distribution in seven sampling locations Average weekly dry, wet and bulk deposition fluxes of nitrogen for the sampling period March 19, 1999–December 31, 2000 are given in Table 1.
3.2. Estimation of wet deposition flux of nitrogen To estimate the wet deposition flux of nitrogen species, the measured concentration of each nitrogen species was multiplied by the volume of collected water and the resultant was divided by the opening area of the Aerochemetrics collecting device. Using proper physical units, the mean wet deposition flux of nitrogen species were estimated in kg ha21 week21. The weekly wet deposition samples could have been subject to concentration losses due to biological reactions. Butler and Likens23 conducted a study to compare the weekly aggregated MAP3S (1981–89) and AIRMoN (1992–95) daily precipitation chemistry record with the NADP/NTN weekly data at four collocated sites in the eastern Unites States. They found that the weekly and daily network concentrations of NO32 were very comparable for both time periods. However, data for ammonium showed a statistically significant bias for both time periods with the daily record having wet concentrations y14% higher than the weekly values. They stated that the most probable cause of the higher ammonium concentrations in the daily samples was the loss of biologically active ammonium in the weekly samples, due to extended time in the field. In order to investigate the possible losses of inorganic nitrogen species in the wet samples, an additional bulk deposition collector was placed at our East Hartford sampling site during the first week of April 2001. To the present time, bulk deposition samples are being collected on a daily basis. Laboratory QA/QC allowed the use of 4 weeks (May 11–June 15, 2001, in which precipitation took place) of precipitation concentration data for NH4z and NO32 for data comparison. Wet daily concentration of NH4z showed 7.2–8.8% (mean 8.2%) higher concentrations than the weekly samples. Wet daily concentration of NO32 were z2.5% to 28.5% (mean 2 4.4%) different from the weekly concentrations. However, more data are needed to determine a correction factor for weekly wet concentrations of NH4z and NO32.
4.2. Dry deposition flux Deposition fluxes of total nitrogen, nitrate (NO32), nitric acid vapor (HNO3), ammonium (NH4z) and organic nitrogen were calculated for the sampling period of March 19, 1999– December 31, 2000, using weekly measured atmospheric concentrations and calculated deposition velocities. Overall, out of a potential 646 samples, 590 valid dry deposition samples were collected and analyzed for dry deposition data, leading to a 92% capture rate. The maximum and minimum mean weekly dry deposition fluxes of total nitrogen were measured in Bridgeport and Avery Point at 0.134 ¡ 0.017 and 0.071 ¡ 0.010 kg ha21 week21, respectively. The mean weekly dry deposition fluxes of NH4z and total nitrogen were significantly higher in inland sampling locations than the coastal areas (Table 2). This is similar to the findings of Carley et al.6 when they reported the 1997–99 atmospheric deposition fluxes of nitrogen in Connecticut. For all measured dry deposition fluxes of nitrogen species, urban sampling locations were significantly higher than the rural areas (Table 3). However, Carley et al.6 did not find a significant difference Table 2 Mean weekly dry deposition flux (kg ha21 week21) of nitrogen at coastal and inland sites in Connecticut, March 19, 1999–December 31, 2000 Coastal
Inland
Parameter
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
NH4z NO32 HNO3 Organic nitrogen Total nitrogen
0.044 0.018 0.025 0.005 0.092
0.033 0.021 0.028 0.006 0.066
0.059 0.019 0.026 0.006 0.109
0.048 0.035 0.027 0.007 0.096
Inland w coastal No difference No difference No difference Inland w coastal
a
a ~ 0.05; valid sampling events ~ 340.
J. Environ. Monit., 2001, 3, 671–680
675
Table 3 Mean weekly dry deposition flux of nitrogen at urban and rural sites in Connecticut, March 19, 1999–December 31, 2000 Urban Parameter NH4z NO32
Mean
Rural SD
0.066 0.041 0.023 0.021 0.029 0.031 HNO3 Organic nitrogen 0.007 0.008 Total nitrogen 0.124 0.075 a a ~ 0.05; valid sampling events ~
Mean 0.043 0.016 0.022 0.005 0.086 342.
SD 0.042 0.035 0.024 0.006 0.089
Tukey’s pairwise comparison test resultsa Urban w rural Urban w rural Urban w rural Urban w rural Urban w rural
between the dry deposition flux of HNO3 between urban and rural sampling locations. It should be noted that they had an additional urban sampling location (Old Greenwich) included in their study. Owing to site restrictions, the bulk deposition collector could not be placed in this sampling location and, therefore, the results of the dry and wet sampling analysis of this site were not included in this paper. Dry deposition fluxes of total nitrogen indicated a moderate positive correlation among all sampling locations. The dry deposition fluxes of total nitrogen in urban sites in general had higher correaltions among each other than the rural areas. The highest correlation was found between Waterbury and East Hartford (r ~ 0.68) followed by Waterbury and Bridgeport (r ~ 0.51). There was a positive correlation between the urban and rural sites with the highest correlation found between Waterbury and Mohawk Mountain (r ~ 0.51) followed by Waterbury and Hammonasset (r ~ 0.46). The mean monthly dry deposition flux of total nitrogen indicated that the deposition fluxes peaked during summer months and reached their lowest values in late fall and early winter (Fig. 2). Some previous studies of filter artifacts have described losses of particulate nitrate from the first filter and a corresponding excess on the second, due to either evaporation of ammonium nitrate or its conversion to nitric acid.24,25 Ammonium nitrate (NH4NO3) is an atmospheric aerosol that is formed by reaction of gaseous HNO3 and NH3. Harrison and Msibi25 used three different sampling methods (filter pack, luminol-based chemiluminescence continuous analyzer and annular denuder) for the collection of nitric acid vapor at a suburban site south of Birmingham in the UK. They did not find a statistically significant difference between the results from all the three techniques. However, they found a small constant bias between the chemiluminescence continuous method and the annular denuder system, which they related to possible loss or gain of
Table 4 Mean weekly wet deposition flux of nitrogen at coastal and inland sites in Connecticut, March 19, 1999–December 31, 2000 Coastal
Inland
Parameter
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
z
0.047 0.084 0.022 0.151
0.059 0.074 0.030 0.140
0.048 0.085 0.027 0.159
0.054 0.062 0.063 0.135
No No No No
NH4 NO32 Organic nitrogen Total nitrogen
difference difference difference difference
a
a ~ 0.05; valid sampling events ~ 271.
HNO3. The filter pack method produced results that were slightly higher than those from the annular denuder system, which could be attributed to the evaporation of ammonium nitrate from the Teflon filter in the filter pack method and the prefilters in the chemiluminescense continuous method. In this work, however, only filter packs were used for collection of particulate nitrate and nitric acid vapor and measurement of volatilization losses of ammonium nitrate and its conversion to nitric acid was not feasible. Based on the work of other researchers, it appears, however, that the volatilization losses of NH4NO3 would not significantly increase the HNO3 loading on the nylon filter. 4.3. Wet deposition flux Wet deposition fluxes of total nitrogen, nitrate (NO32), ammonium (NH4z) and organic nitrogen were calculated for the sampling period March 19, 1999–December 31, 2000, using weekly measured wet concentrations and the volume of water collected in precipitation collectors. From a total of 515 wet deposition samples, the data for 486 valid samples were analyzed and reported, resulting in a 94% capture rate. The maximum and minimum mean weekly wet deposition fluxes of total nitrogen were measured in Bridgeport and Hammonasset at 0.174 ¡ 0.032 and 0.135 ¡ 0.035 kg ha21 week21, respectively. There was no significant difference between the coastal and inland sampling locations for mean weekly wet precipitation fluxes of different nitrogen species (Table 4). This is similar to the findings of Carley et al.6 There was also no significant difference between the urban and rural sampling locations for mean weekly wet deposition fluxes of different nitrogen species (Table 5). Carley et al. reported significantly higher wet deposition flux of ammonium in urban sampling locations. There was a moderate positive correlation for total wet deposition flux measurements among all sampling sites. The highest correlation in rural sampling locations was found
Fig. 2 Montlhy comparison of wet, dry and bulk deposition fluxes of total nitrogen in Connecticut, March 19, 1999–December 31, 2000.
676
J. Environ. Monit., 2001, 3, 671–680
Table 5 Mean wet deposition flux of nitrogen at urban and rural sites in Connecticut, March 19, 1999–December 31, 2000 Urban Parameter NH4z NO32 Organic nitrogen Total nitrogen
Mean 0.052 0.083 0.028 0.162
Rural SD 0.053 0.061 0.062 0.144
Mean 0.045 0.085 0.022 0.150
SD 0.058 0.072 0.043 0.132
Tukey’s pairwise comparison test resultsa No No No No
difference difference difference difference
Table 6 Bulk deposition flux of nitrogen at coastal and inland sites in Connecticut, March 19, 1999–December 31, 2000 Coastal
Inland
Parameter
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
z
0.065 0.091 0.027 0.181
0.112 0.057 0.046 0.173
0.0977 0.083 0.036 0.207
0.339 0.062 0.056 0.288
No No No No
NH4 NO32 Organic nitrogen Total nitrogen
difference difference difference difference
a
a
between Hammonasset and Avery Point (r ~ 0.749) followed by Voluntown and Avery Point (r ~ 0.645). The highest correlation in urban sampling locations was found between Waterbury and Bridgeport (r ~ 0.616) followed by East Hartford and Bridgeport (r ~ 0.474). There was also a moderate positive correlation between urban and rural sites with the highest correlation found between Waterbury and Hammonasset (r ~ 0.681). The highest mean monthly wet deposition flux of total nitrogen occurred in the months of May and March of 2000. Similar to the dry deposition flux, the wet deposition flux of total nitrogen was lower during winter and fall than the summer and spring sampling periods (Fig. 2). The mean weekly concentration of total nitrogen in the seven sampling locations was plotted against the total weekly precipitation depth on a log-scale graph (Fig. 3). The scattered data points indicate that higher precipitation concentrations of total nitrogen occurred at total weekly precipitation depths of v1 cm. The sample correlation between the log-transformed precipitation data and the mean weekly concentration of total nitrogen was r ~ –0.61, implying that the total weekly precipitation depths were inversely correlated to total nitrogen concentrations. The analysis of variance rejected the null hypothesis H0: b1 ~ 0 (no significant linear association between the mean weekly precipitation concentration of total nitrogen and precipitation depth data sets), because using the logtransformed precipitation data the F-value was 203, which was much higher than the critical F(1, 509, 0.95), P v 0.001 ~ 3.86.
Table 7 Bulk deposition flux of nitrogen at urban and rural sampling sites in Connecticut, March 19, 1999–December 31, 2000
a ~ 0.05; valid sampling events ~ 264.
4.4. Bulk deposition flux Bulk deposition fluxes of total nitrogen, nitrate (NO32), ammonium (NH4z) and organic nitrogen were calculated for the sampling period March 19, 1999–December 31, 2000, using the Swedish IVL Sampler discussed in Section 2.2.3. Overall, out of a potential 658 (7 sites 6 94 weeks) bulk deposition samples, the data for 459 valid samples were analyzed and
a ~ 0.05; valid sampling events ~ 260.
Urban Parameter
Mean
Rural SD
NH4z 0.069 0.071 0.081 0.048 NO32 Organic nitrogen 0.026 0.041 Total nitrogen 0.177 0.111 a a ~ 0.05; valid sampling events ~
Mean
SD
Tukey’s pairwise comparison test resultsa
0.095 0.090 0.037 0.211 263.
0.352 0.067 0.059 0.312
No difference No difference Rural w Urban No difference
reported, leading to a 70% capture rate. The maximum and minimum mean weekly wet deposition fluxes of total nitrogen were measured in Voluntown and Hammonasset at 0.299 ¡ 0.125 and 0.159 ¡ 0.025 kg ha21 week21, respectively. There was no significant difference between the coastal and inland sampling locations for mean weekly bulk fluxes of different nitrogen species (Table 6). The bulk deposition of organic nitrogen in rural locations was significantly higher than the urban sites (Table 7). The sample correlation coefficient for bulk deposition of total nitrogen between sites varied from site to site. The highest positive correlation was found between Mohawk Mountain and East Hartford (r ~ 0.720) followed by Waterbury and East Hartford (r ~ 0.482). The lowest negative correlation was found between Hammonasset and Bridgeport (r ~ 20.100) followed by Voluntown and Mohawk Mountain (r ~ 20.093). When urban and rural sites were separately grouped together, a moderate positive correlation was found between the urban and rural sampling locations (r ~ 0.343). The mean monthly bulk deposition of total nitrogen followed the same pattern as the wet and dry deposition fluxes. The bulk deposition fluxes peaked in late spring and summer and reached their lowest levels in late fall and winter. The highest
Fig. 3 Total weekly precipitation depth versus mean weekly concentration of total nitrogen in precipitation in Connecticut, March 19, 1999– December 31, 2000.
J. Environ. Monit., 2001, 3, 671–680
677
mean monthly bulk deposition flux of total nitrogen occurred in May 1999, and July 2000 (Fig. 2). 4.5. Comparison of bulk deposition fluxes with the sum of wet and dry deposition fluxes of nitrogen In order to assess the accuracy of the IVL bulk deposition sampler, a comparison between the mean weekly deposition fluxes of bulk total nitrogen with the sum of wet and dry deposition fluxes of total nitrogen at each sampling location was made. There was no significant difference between the bulk deposition flux of total nitrogen and the sum of wet and dry fluxes of total nitrogen in the four rural sampling locations. Significant difference was observed between the two fluxes for the urban sampling locations (Table 8) (Fig. 4). The comparison between the monthly averages indicated that the bulk deposition and the sum of wet and dry deposition fluxes of total nitrogen corresponded well during the months of June and November of 1999 (Fig. 2). 4.6. NADP and CASTNet atmospheric monitoring sites in Connecticut The National Atmospheric Deposition Program (NADP) has set up a sampling station in Abington, Connecticut to measure the weekly precipitation concentration of various chemicals including inorganic nitrogen species. The USEPA Clean Air Status and Trends Network (CASTNet) has a dry deposition sampling station in the same location (Fig. 1). Abington is a rural area located in northeastern Connecticut in Windham
County. The NADP and CASTNet methods of wet and dry deposition collections in Abington were similar to the methods used in this study. For calculation of dry deposition velocities of gaseous and particulate nitrogenous compounds, however, CASTNet used a multilayer DDIM model,26 whereas in this study a singlelayer DDIM model was used. The mean annual concentrations of dry and wet deposition fluxes of nitrogen estimated for Abington in 1999 were compared with the mean values obtained for rural sites (January 1–December 31, 1999) in this study. The mean weekly wet deposition flux of nitrogen (0.101 ¡ 0.027 kg ha21 year21) measured by NADP in Abington in 1999 was comparable to the mean weekly wet deposition flux of nitrogen (0.125 ¡ 0.022 kg ha21 year21) measured in rural areas in this study. The mean weekly dry deposition flux of nitrogen (0.046 ¡ 0.033 kg ha21 year21) measured by CASTNet in Abington was comparable to the total dry deposition flux of nitrogen (0.071 ¡ 0.008 kg ha21 year21) measured in rural areas in this study. The total nitrogen fluxes reported by NADP and CASTNet were 35 and 20% lower, respectively, than the mean deposition fluxes reported in rural areas of this study. Various reasons may be responsible for this phenomenon. For wet deposition flux measurements of nitrogen, NADP reported NH4z and NO32. For dry deposition, CASTNet reported NH4z, NO32 and HNO3. Neither CASTNet nor NADP reported organic nitrogen that was reported in this study. Abington is located in a rural area northeast of Connecticut, where no major towns, industrial activities or heavy traffic that could be a major source of oxidized nitrogen exist within a 40 km radius of this site.
Table 8 Comparison between the bulk deposition flux of nitrogen with the sum of wet and dry fluxes, March 19, 1999–December 31, 2000 Bulk deposition flux
Sum of wet and dry deposition flux
Mean
SD
Mean
SD
Tukey’s pairwise comparison test resultsa
Bridgeport (urban coastal)— NH4z NO32 Organic nitrogen Total nitrogen
0.067 0.099 0.026 0.192
0.040 0.052 0.049 0.095
0.099 0.124 0.030 0.262
0.060 0.082 0.041 0.155
Valid sampling events ~ 87 Wet and dry w bulk Wet and dry w bulk No difference Wet and dry w bulk
Waterbury (urban inland)— NH4z NO32 Organic nitrogen Total nitrogen
0.077 0.073 0.030 0.180
0.106 0.044 0.042 0.143
0.103 0.105 0.024 0.232
0.609 0.076 0.027 0.140
Valid sampling events ~ 92 No difference Wet and dry w bulk No difference No difference
East Hartford (urban inland)— NH4z NO32 Organic nitrogen Total nitrogen
0.063 0.074 0.023 0.160
0.041 0.046 0.030 0.082
0.097 0.101 0.026 0.225
0.078 0.068 0.082 0.189
Valid sampling events ~ 91 Wet and dry w bulk Wet and dry w bulk No difference Wet and dry w bulk
Mohawk Mountain (rural inland)— 0.056 NH4z 0.116 NO32 Organic nitrogen 0.054 Total nitrogen 0.224
0.061 0.086 0.072 0.153
0.074 0.114 0.033 0.221
0.051 0.076 0.067 0.142
Valid sampling events ~ 90 No difference No difference No difference No difference
Hammonasset (rural coastal)— NH4z NO32 Organic nitrogen Total nitrogen
0.053 0.076 0.030 0.157
0.050 0.049 0.044 0.105
0.069 0.089 0.013 0.169
0.070 0.080 0.015 0.150
Valid sampling events ~ 88 No difference No difference Bulk w wet and dry No difference
Avery Point (rural coastal)— NH4z NO32 Organic nitrogen Total nitrogen
0.074 0.097 0.026 0.194
0.181 0.067 0.045 0.261
0.0584 0.091 0.016 0.164
0.059 0.086 0.016 0.150
Valid sampling events ~ 90 No difference No difference Bulk w wet and dry No difference
Voluntown (rural inland)— NH4z NO32 Organic nitrogen Total nitrogen
0.193 0.069 0.036 0.262
0.657 0.049 0.066 0.526
0.092 0.098 0.018 0.208
0.086 0.098 0.020 0.183
Valid sampling events ~ 91 No difference Wet and dry w bulk Bulk w wet and dry No difference
a
a ~ 0.05.
678
J. Environ. Monit., 2001, 3, 671–680
Fig. 4 Comparison of mean total deposition of nitrogen with wet and dry deposition fluxes in Connecticut, March 19, 1999–December 31, 2000.
5. Conclusions Wet, dry and bulk deposition fluxes of atmospheric nitrogen were measured in seven sampling locations in Connecticut for a period of 94 weeks. The overall mean weekly wet, dry and bulk deposition fluxes of nitrogen in Connecticut were 0.157, 0.103 and 0.202 kg ha21 week21, respectively. The data collected in this study indicated that there is no discernible spatial gradient for wet, dry and bulk deposition fluxes of total nitrogen in Connecticut. The dry deposition fluxes of nitrogen species were higher in urban areas than the rural sampling locations. No significant difference was observed between the wet deposition fluxes of total nitrogen between rural and urban sampling locations. When inland and coastal sites were compared, the dry deposition fluxes of NH4z and total nitrogen were significantly higher in inland locations and no significant difference was observed between the coastal and inland sampling locations for wet deposition fluxes of nitrogen species. The highest wet and dry deposition fluxes of total nitrogen were measured in May 2000 and July 1999, respectively. Wet and dry deposition fluxes of total nitrogen peaked in summer and reached their lowest values during late fall and winter sampling periods. Many industrial activities take place along the shores of Hudson River north of New York metropolitan area. In the past few years, the population of New York City and its suburbs has had an increasing trend resulting into more congested traffic and more vehicular emissions. Mohawk Mountain is a rural site located in northwestern part of Connecticut. This site is approximately 50 km east of Hudson River and 100 km northeast of the New York City area. Therefore, high dry, wet and bulk deposition fluxes of nitrogen observed at this sampling location maybe due to the proximity of the site to areas with high emission sources. Voluntown is another rural site located in eastern part of Connecticut about 35 km west of Providence (capital of Rhode Island) that is a center of many industrial activities. Rhode Island and southeastern Connecticut are major tourist attraction areas all year around. Heavy traffic and industrial emissions could have been the main causes of high deposition fluxes of nitrogen observed at Voluntown sampling location. The lifetime of NH3 ranges from y0.5 h to 5 d. This short lifetime is attributed to the rapid gas-to-particle conversion of
NH3 to NH4z and deposition of NH3 to natural surfaces. The lifetime of particulate NH4z is 5–10 days. If the conversion of NH3 to NH4z proceeds slowly, most of the NH3 emissions will be deposited locally and less NH4z will be made available for long-range transport.27 It maybe possible that a major portion of the ammonia gas transported to Connecticut as long-range transport will be converted to ammonium particles. In this study, NH3 and NO2 were not quantitatively captured, which may have led to low bias for nitrogen dry and total deposition calculations. Agricultural and livestock activities are not very common in Connecticut and forests cover more than 60% of the land surface. To identify the spatial distribution of ammonia and nitrogen dioxide in Connecticut, the authors suggest that measurements of atmospheric concentrations of NH3 and NO2 in at least one rural and one urban sampling location in Connecticut be added to the program schedule.
Acknowledgements We acknowledge the financial support of the Connecticut Department of Environmental Protection (CTDEP) and the Environmental Research Institute (ERI) for this project. We also acknowledge the collaborative efforts and feedback from Carmine DiBattista, Thomas Morrissey, Robert Smith, Teresa Gutowski and Paul Stacey of CTDEP and Hugo Thomas and David Miller of the University of Connecticut. This project could not have been completed without the sampling help from CTDEP Air Toxics Monitoring Group personnel. Finally, we acknowledge the efforts of the ERI personnel who conducted sampling and analytical duties for this study: Janet E. Heitert, Steve Chirdon, Mark Groszek, Stephan Nowakowski, Mike Morrison, Dan Tubbs, Gary Ulatowski and Todd Wheeler.
References 1 Connecticut Department of Environmental Protection, The Atmosphere and Hypoxia in Long Island Sound, Hartford, Connecticut, http://www.dep.state.ct.us/wtr/lis/atmoslis.htm, 2000. 2 H. L. Hu, H. M. Chen, N. P. Nikolasdis, D. R. Miller and X. Yang, Water Air Soil Pollut., 1998, 105, 521. 3 Connecticut Department of Environmental Protection, Water Quality of Long Island, Hartford, Connecticut, http://www.dep. state.ct.us/wtr/lis/wtrqual.htm, 2000. 4 L. J. Puckett, Nonpoint and Point Sources of Nitrogen in Major
J. Environ. Monit., 2001, 3, 671–680
679
5 6
7 8 9 10 11 12 13 14
680
Watersheds of the United States, Water Resources Investigations Report 94-001, US Geological Survey, Reston, VA, 1994. Connecticut Department of Environmental Protection, Long Island Sound: Fact Sheet, DEP-FS-08, Connecticut Department of Environmental Protection, Hartford, CT, 1999. R. J. Carley and C. Perkins, Nitrogen Deposition Monitoring in Connecticut: Final Report, submitted to Connecticut Department of Environmental Protection, Hartford, Connecticut, www.eri. uconn.edu, 2001. M. A. Geigert, N. P. Nikolaidis, D. R. Miller and J. Heitert, Atmos. Environ., 1994, 28, 1689. National Atmospheric Deposition Program (NADP), NADP/NTN Sites in Connecticut, CT15, Abington, http://nadp.sws.uiuc.edu/, 1999. USEPA Clean Air Status and Trends Network (CASTNet), Sites/ Abington, Connecticut, Monitoring Station ABT147, http://www. epa.gov/castnet/, 1999. A. Iverfeldt, Water Air Soil Pollut., 1991, 56, 553. P. M. Berthouex and L. C. Brown, Statistics for Environmental Engineers, Lewis Publishers, CRC Press, Boca Raton, FL, 1994. D. G. Kleinbaum, L. L. Kupper and K. E. Muller, Applied Regression Analysis and Other Multivariable Methods, PWS-Kent Publishing, Boston, 1988. W. J. Lyman, W. F. Reehl and D. H. Rosenblatt, Handbook of Chemical Property Estimation Methods, American Chemical Society, Washington, DC, McGraw Hill, New York, 1990. E. S. Edgerton, T. F. Lavery and H. S. Prentice, Project Summary:
J. Environ. Monit., 2001, 3, 671–680
15 16 17 18 19 20
21 22 23 24 25 26 27
Nation Dry Deposition Network Third Annual Progress Report, ESE, Gainesville, FL, 1990. T. P. Meyers, B. B. Hicks, J. P. Hosker Jr., J. D. Womack and L. C. Satterfield, Atmos. Environ., 1991, 25, 2361. G. Lear, U.S. Environmental Protection Agency, Clean Air Markets Division, Washington, DC, personal communications. R. I. Smith, D. Fowler, M. A. Sutton, C. Flechard and M. Coyle, Atmos. Environ., 2000, 34, 3757. H. V. Anderson and M. F. Hovmand, Atmos. Environ., 1994, 28, 3495. B. B. Hicks, D. D. Baldocchi, T. P. Meyers, R. P. Hosker and D. R. Matt, Water Air Soil Pollut., 1987, 36, 311. J. F. Clarke and E. S. Edgerton, Dry Deposition Flux Calculations for the National Dry Deposition Network, Atmospheric Research and Exposure Assessment Laboratory, USEPA, Research Triangle Park, NC, EPA/600/R-93/065, 1993. X. Yang, D. R. Miller, X. Xu, L. H. Yang, H. M. Chen and N. P. Nikolaidis, Atmos. Environ., 1996, 30, 3801. M. A. Geigert, Masters Thesis, University of Connecticut, Storrs, CT 1991. T. J. Butler and G. E. Likens, Atmos. Environ., 1998, 32, 3749. A. G. Allen and R. M. Harrison, Atmos. Environ., 1989, 23, 1591. R. M. Harrison and M. I. Msibi, Atmos. Environ., 1994, 28, 255. T. P. Meyers, P. Finkelstein, J. Clarke, T. G. Ellestad and P. F. Sims, J. Geophys. Res., 1998, 103, 22645. J. T. Walker, P. V. Aneja and D. A. Dickey, Atmos. Environ., 2000, 34, 3407.
