Field intercomparison of diVusive samplers for measuring ammonia M. Kirchner,a S. Braeutigam,a M. Ferm,b M. Haas,c M. Hangartner,d P. Hofschreuder,e A. Kasper-Giebl,f H. Ro�mmelt,g J. Striedner,c W. Terzer,f L. Tho� ni,h H.Werneri and R. Zimmerlingj aGSF-Forschungszentrum fu�r Umwelt und Gesundheit, Institut fu�r O� kologische Chemie, Ingolsta�dter Landstr. 1, D-85764 Neuherberg, Germany bSwedish Environmental Research Institute, P.O.Box 47086, S-40258 Gothenburg, Sweden cUmweltbundesamt, Zweigstelle Su�d, Siriusstr. 3, A-9020 Klagenfurt, Austria dEidgeno�ssische Technische Hochschule Zu�rich, Institut fu�r Hygiene und Arbeitsphysiologie, Clausiusstr. 21, CH-8092 Zu�rich, Switzerland eWageningen Agricultural University, Meteorology and Air Quality Group, Duivendaal 2, NL-6701 AP Wageningen, The Netherlands fInstitute for Analytical Chemistry, Vienna University of Technology, Getreidemarkt 9/151, A-1060 Vienna, Austria gUniversita�t Mu� nchen, Institut fu�r Balneologie und Klimatologie, Marchioninistr. 17, D-81377 Mu�nchen, Germany hForschungsstelle fu�r Umweltbeobachtung, Untere Bahnhofstr. 30, CH-8640 Rapperswil, Switzerland iUniversita�t Mu� nchen, Lehrstuhl fu�r Bioklimatologie und Immissionsforschung, Am Hochanger 11, D-85354 Freising, Germany jBundesforschungsanstalt fu�r Landwirtschaft, Institut fu�r Agraro�kologie, Bundesallee 50, D-38116 Braunschweig, Germany Received 24th March, Accepted 12th April 1999 Agricultural production systems are recognised as a major source of atmospheric ammonia.Deposition of ammonia and ammonium may contribute to undesired changes in oligotrophic ecosystems.The continuous measurement of atmospheric ammonia requires expensive and sophisticated techniques and is performed only in a very restrict number of ambient air stations in Europe. Therefore, the application of passive samplers, which have the advantage of being easy to handle and cost-eYcient, is useful.In the past the comparability of diVerent passive samplers must be considered as rather scarce. In a joint European project under the leadership of the GSF-Forschungszentrum fu� r Umwelt und Gesundheit, Neuherberg, in 1997 a comparison of diVerent passive ammonia monitoring methods was carried out in a prealpine rural site near Garmisch-Partenkirchen. It was considered valuable to include not only well established systems but also methods still being developed.For the comparative test ten working groups with diVerent methods took part. A wet annular denuder system, which has been developed by the Netherlands Energy Research Foundation for on-line measurement of atmospheric ammonia, served as reference of passive methods. The experiment, which started in June and finished in December, showed that most of the passive samplers fulfil the requirements and can be recommended for further measurements.Additional measurements of meteorological parameters were performed to check the influences of diVerent weather conditions on passive sampling. Ammonia deposition on ecosystems will result in Introduction acidification because of nitrification of the ammonium in the Ammonia emissions are high in Europe.The largest area soil7–9 and eutrofication.10 To calculate deposition and comaveraged emissions of over 6 tonne NH3 km-2 per year are pare results with critical loads,11 a typical resolution of encountered in the Netherlands and Belgium. Large areas of 1×1 km or less is needed.12 Absence of a detailed emission western France, the United Kingdom, Ireland, Denmark, the inventory and/or models to calculate concentrations and Po-valley in Italy and the eastern and southern part of deposition on this scale is a common problem.6 Measuring Germany also exhibit large emissions of NH3.1 Emissions concentrations in air is another option to determine air quality.from industry and use of fertilizer are small compared to The large number of measuring sites needed to obtain a emissions from animal husbandry.2 At a sub-national scale representative picture of ammonia concentrations in a certain area averaged emissions vary considerably.3,4 The deposition area makes passive samplers the ideal instrument to do the velocity of ammonia is large.5,6 The combination of a large job.A number of both tube-type and badge-type passive variation in emission density, emissions close to the surface samplers for measuring ammonia were developed and are and a large deposition velocity will give rise to very local summarised by CEN.concentration patterns for ammonia in ambient air. These In the present paper we describe the results of an intercompaconcentration patterns can be obtained by modelling, using rision of 10 diVusive samplers and a continuously working an emission inventory coupled to a dispersion and deposition denuder system.The measurements were carried out in southern Bavaria in 1997. model, or by measuring concentrations. J. Environ. Monit., 1999, 1, 259–265 259to distinguish: (a) Tube type: the diVusion is determined Description of the experimental procedures through a static air column.DiVusion processes can be calcu- General sampler performance criteria lated according to Fick’s first law of diVusion provided that the diVusion coeYcient is known, as in the case of ammonia. The use of passive samplers is strongly supported by the (b) Badge type: permeation processes can also be described European Union. In the Council Directive 96/62/EC of by Fick’s first law of diVusion.In this case the diVusion is 27 September 1996 on Air Quality Assessment and determined by a static layer of air and a membrane of defined Management a frame work is set for preliminary assessments pore size. The collection rate is calculated analogous to that of air quality, station siting optimisation, supporting generalisof the tube type sampler.The resistance of the membrane is ation measurements, and evaluation of existing measurements. most of the time small compared to the resistance of the In the Guidance Report on Supplementary Assessment under stagnant air layer.19 As badge type samplers are more suscep- EC Air Quality Directives (1997)13 it is concluded that the tible to factors such as formation of stagnant layers in front low cost and easy operation of the diVusive sampling techof the turbulence damping membrane and the eYciency of the niques makes it an ideal tool for large scale air pollution membrane in damping turbulence (because of the low resist- surveys with a high spatial resolution. The diVusive sampler ance against uptake of the sampler), it is advised to test the is also of particular interest as an indicative technique.sampling rate experimentally. In the past, several passive To enhance the quality and comparability of measurements sampler intercomparison campaigns have been performed, but CEN decided to standardise passive sampling. This standardisnone for ammonia.20 ation is on performance not on instruments.The future standard on passive sampling will contain four parts. Part 1 Selected site, material and methods (general requirements)14 and Part 2 (specific requirements and test methods)15 will soon be sent to the member states for In 1997, the GSF-Forschungszentrum fu� r Umwelt und inquiry. Part 3 (guide for selection, use and maintenance)16 is Gesundheit invited European researchers working on passive in preparation. The performance criteria in Part 1 require monitoring methods for air pollution to participate in an unambiguity, selectivity, an overall uncertainty less than 30% international intercomparison of ammonia samplers.21 10 for specified measuring ranges and averaging times.Overall groups decided to take part by including not only well uncertainty (OU) is a combination of precision and bias.The established systems but also methods still being developed; as value of 30% is for components not specified in EC Daughter a compromise to the CEN requirements at least 3 parallel Directives. Ammonia is a non-specified component. tube or badge type samplers were exposed in the campaign. The easiest way to test samplers is to perform tests in a Table 1 gives a brief summary of the diVerent diVusion sis should be plers. Four systems have already been described in the done to prove that the sampler meets the requirements which literature.22–25 are given in CEN (1998 b). Field tests are designed to evaluate The comparative measurements were performed at Aidling/ additional errors arising from the use of the sampler in a wider Riegsee in the district of Garmisch-Partenkirchen (Bavaria) range of environmental conditions existing in the field that are between June and December 1997.Aidling is a typical prealnot adequately covered by the laboratory tests. CEN (1998 b) pine region with agricultural activities such as pasture and requires the use of 6 parallel samplers and a single static application of liquid manure on meadows.In comparison to instrument. Consideration should be given to representa- other intensively used areas in Bavaria and Austria, the number tiveness of location (urban, rural, background), topography of cattle ha-1 near Aidling (ca. 1–1.5) is relatively small. (soil covering, degree of aVorestation), environmental conditions (temperature, relative humidity and wind speed) and Reference method likely interferences.The wet annular denuder system AMOR,26 which has been developed by the Netherlands Energy Research Foundation Theory on passive sampling (ECN) for continuous measurement of ammonia in ambient air, served as reference of the passive method. Ambient air is Integrating air monitoring methods are based on passive transport of gases from the atmosphere onto an absorbing pumped through a rotating annular denuder. Ammonia is absorbed by a weak acid solution passing the denuder in medium.17,18 There are basically two types of passive samplers Table 1 Description of diVusive methods used in the intercomparison Collection rate/ Integration Detection limit/ No.Type Inlet Collection medium ml min-1 Analytical method time/weeks mg m-3 week I Badge Membrane Impregnated filter 45 FIA 2, 4 0.8 (citric acid) II Badge Membrane Impregnated filter 45 Indophenol segment 1, 2, 4 0.3 (citric acid) flow analyser III Badge 5 mm teflon filter Stainless steel grid 77.3 Ion chromatography 1, 2, 4 0.2 (phosphoric acid) IV Ventilated sampler 5 wholes Impregnated filter 2000 Berthelot reaction 1, 4 0.05 Ø 5 mm (sulphuric acid) V Multi tube type 300 open Prepared membrane 53.8 Autoanalyser Not exposed 0.5 glass tubes (citrid acid) VI Palmes type Membrane Solution 13.4 Ion chromatography 1, 2, 4 0.6 (hydrochloric acid) VII Badge Membrane Impregnated filter 31.5 Indophenol 2, 4 0.8 (phosphoric acid) VIII Badge Teflon filter Impregn. glass surface 56 Ion chromatography 2, 4 1.0 (phosphoric acid) IX Palmes type Membrane Stainless steel grid 2.73 Conductance 2, 4 0.2 (sulfuric acid) (ECN-method) 260 J.Environ. Monit., 1999, 1, 259–265Fig. 1 Wet annular denuder system AMOR as a reference device in passive sampler intercomparison. counterflow direction. Downstream from the denuder the and yi=reference concentration of ammonia determined with the denuder.absorption solution is combined with an alkaline solution, which results in the formation of gaseous ammonia. Via a (ii) Correlation coeYcient r2 and regression coeYcients (k slope, d intercept) semi-permeable membrane, ammonia is transferred into a stream of demineralized water. The ammonium concentration For the determination of r2, k and d, the mean values obtained from 3 to 6 samplers exposed in parallel, were related in the water is determined conductometrically (Fig. 1). The calibration is made with aqueous solutions containing diVerent to the concentrations determined with the denuder. (iii) Precision. The precision of the diVerent passive sampler concentrations of ammonium. The method is very selective and independent from ambient humidity.Interferences were methods is calculated by the relative standard deviations (RSD): RSD(x)=[s(x)/x: ]×100 (%), where s(x)=standard only determined for volatile amines, which generally show concentrations much lower than ammonia. With a time reso- deviation of the results of 3–6 passive samplers exposed in parallel. lution of two minutes the measuring range is between 0.05 and 1000 mg m-3.(iv) Overall uncertainty (OU): OU={[|x: i-yi|+2s(x)]/ yi}×100, where x: i=mean concentration of ammonia obtained In order to guarantee the performance of the diVuse sampler intercomparison, a second less sophisticated denuder devel- from 3 to 6 diVusive samplers exposed in parallel and yi= reference concentration of ammonia determined with the oped by the Forschungsstelle fu� r Umweltbeobachtung (FUB), Rapperswil (Switzerland), delivered comparative data.The denuder. inside of this tubular denuder is coated with citric acid in methanol. After passing the denuder, aerosol compounds are Results and discussion collected on a filter. Extraction of the denuder and the filter, Conditions during the sampling period and the subsequent analyses by ion chromatography, are performed in the lab.The agreement between this denuder, Daily average concentrations of ammonia, determined with which has been exposed for one week periods, and the continu- the wet denuder system described above, varied between the ous flow denuder used as reference was rather good.21 limit of detection (0.05 mg m-3) and 20 mg m-3.Shorter time Regression analyses of 26 data points (weekly averages) gave averages reached up to 120 mg m-3 (30 min average). The site a linear relationship statistically significant at a 99% confi- was in close vicinity to several sources of ammonia. No marked dence. The correlation coeYcient (r2) is 0.83, the slope is 0.69 seasonal variations were visible (Fig. 2). Concentrations of and the intercept is 1.02. The annular denuder system AMOR particulate ammonium were determined with the denuder represents the x-axis. system operated by the FUB.21 Regarding weekly averages, In order to establish the influence of weather conditions on the ammonium concentrations ranged from 0.5 to 4 mg m-3; the ammonia measurements, the air temperature, relative there was no correlation between concentrations of particulate humidity, wind velocity and wind direction were registered by ammonium and ammonia.an automatic weather station. As the intercomparison started in April 1997 and lasted until December 1997 it covers late spring to early winter and Quality control thus diVerent meteorological conditions. Weekly averages of air temperature, humidity and wind velocity ranged from-0.4 Quality control included the handling of passive samplers and to 18 °C, 0.4 to 2 m s-1 and 64 to 99%, respectively.As could the statistical treatment of the data. Handling and storage of be expected, the comparison between ammonia concentrations passive samplers before and after exposure were controlled by measured by the denuder system and the meteorological GSF according to the individual requirements.21 parameters showed that low temperatures and high humidity The statistical analyses of the results obtained from the both favour low concentrations of ammonia in air.Lowest intercomparison comprises the following parameters: concentrations were determined when the ground was covered (i) Mean square error.The mean square error (MSE) with snow. describes the deviation between two measurements: Comparison between diVusion samplers and the active sampling MSE= 1 n]n i=1 bi2, where n=number of measurements (periods), system The concentrations of ammonia measured by the participating bi=x: i-yi, bi=bias, x: i=mean concentration of ammonia obtained from 3 to 6 diVusive samplers exposed in parallel groups are shown in Figs. 3–5. The results are given for 8 J. Environ. Monit., 1999, 1, 259–265 261Fig. 2 30 min average ammonia concentration measured by AMOR 6/97–12/98. Fig. 3 1 week exposure of selected passive samplers compared with an active monitor at the Aidling station. sampler types, grouped for the diVerent times of exposure (1, Mean square error 2 and 4 weeks).No results are given for method V. This The MSE calculated for the single diVusion samplers versus method was introduced too late into the investigation. the denuder method ranges from 0.1to 2.59. Minimum values Similar trends were determined by all diVusion samplers were obtained for PS I and PS III (4 week exposure) and PS and the active denuder method, but some diVerences become IX (6 samplers in parallel, 2 week exposure).These parameters visible. As expected, the time pattern of the integrated quantify the deviation of passive samplers measurements from ammonia concentrations for 2 week and monthly those obtained with the denuder as demonstrated in Figs. 3–5. measurements are not as pronounced as those of 1 week exposures. Regression analyses Tables 2–4 summarise the statistical analyses of the comparison of the single diVusion samplers with the denuder A linear relationship between the diVusion samplers and the reference denuder method, being statistically significant at the system AMOR and, in brackets, the second denuder delivered by FUB.The results are grouped for the samplers exposed for 95% level, was obtained mainly for shorter periods of exposure. Just one diVusion sampler (PS III ) meets these requirements 1, 2 and 4 weeks, respectively. 262 J. Environ. Monit., 1999, 1, 259–265Fig. 4 2 week exposure of selected passive samplers compared with an active monitor at the Aidling station. Fig. 5 4 week exposure of selected passive samplers compared with an active monitor at the Aidling station.at exposure periods of 4 weeks. This seems to be due mainly and elevated concentrations is at about 4 mg m-3. Although this trend can be seen for most of the samplers, we want to to the small number of data points available for the longer exposure periods. In most of the cases slope and intercept point out that the extent of the deviation is very diVerent for each type of sampler.were significantly diVerent from 1 or 0, respectively, at the 95% confidence level. This indicates both a proportional and a constant bias between the diVusion samplers and the denuder Relative standard deviation method. Most of the cases in which no diVerence could be obtained were cases with an extremely small number of data The RSD calculated from multiple sampling was in the range 2–49%.Regarding the single diVusion samplers, longer expo- points and a rather high overall uncertainty. Several of the diVusion samplers tend to overestimate low and to underesti- sure periods generally lead to smaller values of the RSD. RSDs below 5% were obtained for PS I, PS III, PS IV and mate high ammonia concentrations compared to the results of the active sampling method.The discrimination between low PS VI (4 weeks), PS I (2 weeks) and PS IV (1 week). Table 2 Comparison of the diVusion samplers and the reference denuder method (in brackets: FUB denuder); 1 week exposurea n bm s(b) MSEr 2 k d RSD (%) OU (%) PS II 26 (28) 0.49 (0.67) 1.31 (1.00) 1.89 (1.44) 0.45b (0.56) 0.57c (0.84) 2.11c (1.24) 15.2 75 (64) PS III 25 (27) 0.37 (0.53) 1.28 (1.08) 1.71 (1.42) 0.45b (0.52) 0.62c (0.91) 1.75c (0.84) 7.8 54 (46) PS IV 26 (28) 0.34 (0.53) 0.92 (0.62) 0.92 (0.68) 0.75b (0.78) 0.58c (0.92) 1.91c (1.20) 4.7 41 (34) PS VI 23 (25) 0.03 (0.14) 0.79 (0.71) 0.60 (0.50) 0.77b (0.78) 0.80c (1.02) 0.76c (0.05) 10.9 45 (38) an: number of measurements, bm: bias, s(b): standard deviation of bias, MSE: mean square error, r2: coeYcient of correlation, k: slope, d: intercept (mg m-3), RSD: relative standard deviation, OU: overall uncertainty.bLinear relationship statistically significant at 95% confidence level. cSlope or intercept significantly diVerent from 1 or 0 respectively at 95% confidence level. J. Environ. Monit., 1999, 1, 259–265 263Table 3 Comparison of the diVusion samplers and the reference denuder method (in brackets: FUB denuder); 2 week exposurea n bm s(b) MSEr 2 k d RSD (%) OU (%) PS I 13 0.12 (0.32) 0.72 (0.55) 0.49 (0.38) 0.74b (0.77) 0.69c (0.91) 1.32c (0.65) 4.5 25 (21) PS II 12 0.40 (0.57) 0.65 (0.60) 0.54 (0.67) 0.81b (0.77) 0.77c (0.98) 1.29c (0.64) 10.3 41 (42) PS III 12 0.13 (0.29) 0.64 (0.58) 0.39 (0.40) 0.69b (0.74) 0.87c (1.16) 0.61c (-0.25) 6.7 33 (31) PS VI 12 -0.25 (-0.03) 1.01 (0.71) 0.99 (0.47) 0.48 (0.58) 0.53c (0.76) 1.62c (0.90) 11.3 40 (36) PS VII 11 0.07 (0.30) 1.00 (0.77) 0.92 (0.64) 0.57 (0.62) 0.61c (0.86) 1.55c (0.80) 10.5 45 (43) PS VIII 12 1.36 (1.45) 0.72 (0.73) 2.33 (2.78) 0.79b (0.80) 0.92c (1.23) 1.68c (0.61) 13.1 79 (77) PS IX (3) 7 0.30 (0.73) 1.59 (1.64) 2.26 (2.85) 0.66 (0.76) 1.27 (1.81) -0.77 (-2.17) 48.9 147 (155) PS IX (6) 7 -0.16 (0.26) 0.40 (0.45) 0.16 (0.24) 0.95b (0.91) 0.93c (1.08) 0.49c (-0.05) 13.1 35 (41) an: number of measurements, bm: bias, s(b): standard deviation of bias, MSE: mean square error, r2: coeYcient of correlation, k: slope, d: intercept (mg m-3), RSD: relative standard deviation, OU: overall uncertainty.bLinear relationship statistically significant at 95% confidence level.cSlope or intercept significantly diVerent from 1 or 0 respectively at 95% confidence level. Table 4 Comparison of the diVusion samplers and the reference denuder method (in brackets: FUB denuder); 4 week exposurea n bm s(b) MSEr 2 k d RSD (%) OU (%) PS I 6 0.07 (0.18) 0.56 (0.28) 0.26 (0.10) 0.79 (0.95) 0.81c (1.13) 0.81 (-0.30) 3.1 16 (12) PS II 6 0.16 (0.44) 0.65 (0.32) 0.38 (0.28) 0.72 (0.86) 0.60c (0.94) 1.73c (0.66) 11.5 43 (39) PS III 6 0.05 (0.15) 0.36 (0.56) 0.11 (0.28) 0.87b (0.65) 0.87c (0.89) 0.54 (0.56) 4.2 18 (22) PS IV 6 0.25 (0.48) 0.80 (0.57) 0.60 (0.50) 0.45 (0.61) 0.56c (0.81) 2.09 (1.21) 1.9 22 (22) PS VI 7 -0.11 (0.08) 0.67 (0.35) 0.39 (0.11) 0.63 (0.84) 0.65c (0.94) 1.31 (0.31) 4.8 23 (18) PS VII 3 -0.71 (-0.22) 0.77 (0.81) 0.90 (0.48) 0.70 (0.41) 0.58 (0.64) 1.07 (1.13) 9.0 31 (34) PS VIII 6 0.61 (0.89) 0.94 (0.66) 1.11 (1.15) 0.53 (0.39) 0.22 (0.27) 3.67 (3.54) 17.4 77 (76) PS IX (3) 3 -1.44 (-0.95) 0.88 (0.76) 2.59 (1.30) 0.64 (0.36) 0.42 (0.45) 1.04 (1.14) 31.1 74 (67) PS IX (6) 3 -1.18 (-0.69) 0.44 (0.34) 1.52 (0.56) 0.99 (0.85) 0.68c (0.91) 0.18 (-0.34) 18.4 54 (46) an: number of measurements, bm: bias, s(b): standard deviation of bias, MSE: mean square error, r2: coeYcient of correlation, k: slope, d: intercept (mg m-3), RSD: relative standard deviation, OU: overall uncertainty. bLinear relationship statistically significant at 95% confidence level. cSlope or intercept significantly diVerent from 1 or 0 respectively at 95% confidence level.Overall uncertainty The OU ranged from 16 to 147%.Disregarding the maximum value above 100% because of the low number of comparisons, the upper limit of the overall uncertainty is 79%. An overall uncertainty below 30% (as demanded by the CEN) is obtained Table 5 Dependence of the diVerence between active and passive by PS I (2 and 4 week exposure), PS III (2 and 4 week measurements (2 week sampling) on averaged air temperature, wind velocity and relative humidity exposure), PS IV (4 week exposure) and PS VI (4 week exposure). Temperature Performance of the diVusion samplers compared to <5 °C 5–15° C >15 °C environmental conditions Bias/mg m-3 Bias/mg m-3 Bias/mg m-3 Furthermore the results from 1 and the 2 week sampling were PS I 0.32 -0.03 0.10 grouped according to the average temperature, relative OS II -0.04 0.80 0.42 humidity and wind speed determined at the site.To examine PS III -0.41 0.16 0.48 PS VI -1.50 0.11 0.23 any variations of the performance of the sampler, the average PS VII -0.78 0.62 0.50 bias was related to the average environmental conditions. PS VIII 0.87 1.57 1.66 These comparisons, which could not be performed for PS IX Wind velocity due to the delayed beginning, are given for 2 week sampling in Table 5.<1 m s-1 1–15 m s-1 >1.5 m s-1 Calculating the average bias for three temperature classes Bias/mg m-3 Bias/mg m-3 Bias/mg m-3 (<5, 5–15 and >15 °C) a temperature dependence could be observed for several diVusion samplers. At average tempera- PS I 0.23 -0.02 0.37 tures below 5 °C these samplers show a negative or a relatively PS II 0.24 0.38 0.71 PS III -0.34 0.37 0.34 small positive bias, while a larger positive bias is calculated PS VI -0.78 -0.18 0.60 for higher temperatures.Two possible explanations are given PS VII -0.25 0.00 0.89 for this phenomenon. Thus PS IV was biased by the freezing PS VIII 1.39 1.17 1.99 of the absorbent (sulfuric acid). An improvement of the anti- Relative humidity freeze reduced these problems.For PS VI the underestimation of ammonia at low temperatures can be explained by the <75% 75–85% >85% condensation and the subsequent freezing of water on the Bias/mg m-3 Bias/mg m-3 Bias/mg m-3 entrance membrane of the sampler. This leads to a blocking of the entrance opening. No explanation was given for the PS I -0.39 0.13 0.61 PS II -0.07 0.63 0.31 other diVusion samplers, which showed similar behaviour.No PS III 0.06 0.23 -0.41 uniform trend could be observed for the comparison between PS VI -0.69 0.25 -1.16 the average bias and the relative humidity as well as the wind PS VII -0.26 0.74 -0.58 velocity classes. A negative bias at wind velocities below PS VIII 1.31 1.87 0.92 1 m s-1 might point to a starvation eVect.Since slightly 264 J. Environ. Monit., 1999, 1, 259–265Symp. Den Bosch, Acid Rain Research; Do we have enough contradicting results are obtained for the 1 week and the 2 answers?, (ed. Erisman Heij) Elsevier, Amsterdam, 1995, p. 81–90. week averages, it is diYcult to derive any reliable conclusions. 4 M. A. Sutton, C. J. Place, M.Eager, D. Fowler and R. I. Smith, Due to the relative low number of observations (<25 periods) Atmos. Environ., 1995, 29(12), 1393. the statistical treatment of concentration and weather data did 5 M. A. Sutton, PhD Thesis, University of Edinburgh, 1990. not result in clear dependencies. 6 W. A. J. van Pul, C. J. M. Potma, E. P. van Leeuwen, G. P. J. Draaijers and J. W. Erisman, EDACS: European deposition maps of acidifying components on small scale: model descrip- Sampler handling tion and preliminary results, Report RIVM no 722401005, 1995.To apply a diVusion sampler in a monitoring network, 7 N. van Breemen and H. F. G. van Dijk, Environ. Pollut., 1988, installation and handling in the field as well as the mailing of 54, 249. 8 H. Ellenberg, O� kologische Vera� nderungen in Biozo�nosen durch the samplers should be as simple as possible.To fulfill all these StickstoVeintrag. Symposium ‘Ammoniak in der Umwelt’, 10. bis 12. requirements some improvements had to be made during the Oktober 1990, Bundesforschungsanszalt fu� r Landwirtschaft intercomparison. These improvements included modifications (FAL) Braunschweig-Volkenrode, 1991.of the holding device of several samplers to allow a more 9 La� nderausschuss fu� r Immissionsschutz (LAI), Bewertung von rapid and easier change or the use of air tight contain for the Ammoniak- und Ammoniumimmissionen, Bericht des Untermailing of the samplers. In the final versions of exposure and ausschusses ‘Wirkungsfragen’ Erich Schmidt Verlag, 1995. 10 J. W. Erisman, R.Bobbink and L. van der Eerden, Nitrogen handling all samplers met the above mentioned requirements. pollution on the local and regional scale: the present state of knowledge and research needs, Report RIVM no 722108010, 1996. Conclusions 11 A. Fangmeier, A. Hadwiger-Fangmeier, L. van der Eerden and H. J. Jaeger, Environ. Pollut., 1994, 86, 43. For ambient air monitoring, simple, but reliable instruments 12 J.P. Hettelingh, R. H. Gardner and L. Hordijk, Environ. Pollut., are needed. The field measurements of ammonia conducted in 1992, 77, 177. 13 R. van Aalst, L. Edwards, T. Pulles, E. de Saeger, M. Tombrou Bavaria in 1997 by the GSF-Forschungszentrum fu� r Umwelt and D. Tonnesen, Guidance report on supplementary assessment und Gesundheit demonstrated that several of the diVusion under EC Air Quality Directives, second draft, April 1997, p. 51. samplers for ammonia presently in use fulfil the CEN quality 14 CEN (1998 a) Ambient Air Quality: DiVusive samplers for the requirement and are suitable for applications in rural areas. determination of concentrations of gases and vapours; Requirements Nevertheless, the intercomparision showed that no general and test methods.Part 1: General requirements, draft report, statement can be given about the performance of all passive January 1998. 15 CEN (1998 b) Ambient Air Quality: DiVusive samplers for the samplers, even if very similar designs are used by diVerent determination of concentrations of gases and vapours; Requirements groups. On the other hand it is impossible to favour one and test methods.Part 2: Requirements and test methods, draft special design or analytical method. Furthermore, quite varireport, August 1998. able results were obtained when one sampler type was exposed 16 CEN (1998 c) Ambient Air Quality: DiVusive samplers for the for rather short (1 week) or long (4 week) time periods. determination of concentrations of gases and vapours; Requirements The application of several passive samplers tested during and test methods.Part 3: Guide for selection, use and maintanance, draft report, May 1998. the field campaign give the possibility of extensive screening 17 E. D. Palmes A. F. Gunnison, J. Di Mattio and C. Tomczyk, Am. surveys in large areas. Under certain circumstances, i.e. in lack Ind. Hyg. Assoc.J., 1976, 37, 570. of calibration, passive monitoring can provide an indication 18 R. H. Brown, Pure Appl. Chem., 1993, 65(8), 1859. about reliability of continuous monitors. In any case detailed 19 P. Hofschreuder, W. van der Meulen, P. Heeres and J. Slanina, quality assurance and control is necessary not only in the lab, J. Environ. Monit., 1999, accepted. but also in field intercomparisons. 20 M. Kirchner, Durchfu�hrung von Vergleichsversuchen zur Austestung von Passivsammlern, 1997,Wetter und Leben 49, H.4. 21 M. Kirchner, S. Braeutigam and G. Welzl, Validierung von Acknowledgements Passivsammlern zur Messung von Ammoniak im Freiland, 1998, GSF-Bericht 18/98. This work was funded by the Bavarian Ministry for State 22 A. Blatter, M. Fahrni and A. Neftel, CEC Air Pollut. Res. Development and Environmental AVairs and the European Rep. 41, 1992. Fund for Regional Development. Infrastructional support was 23 M. Ferm and H. Rodhe, J. Atmos. Chem., 1997, 27, 17. given by the Regional Government of Salzburg. One of the 24 A. Kasper and H. Puxbaum, Fresenius’ J. Anal. Chem., 1994, 350, 448. participating groups (IAC) thanks the Austrian Ministry of 25 L. De Temmerman and P. Coosemans, Ammonia and ammonium Environmental AVairs and Regional Government of Salzburg deposition on pine stands in Belgium. Proc. 14th meeting for for financial support. Spezialists in Air Pollution eVects on forest ecosystems (IUFRO p2.05): ‘Air pollution and forest decline’, Interlaken. ed. J. B. Bucher and I. Bucher-Wallin, Interlaken, Austria, 1989, vol. 1, References p. 91–96. 1 W. A. H. Asman, Nova Acta Leopold., NF70, 1994, 288, 263. 26 G. P. Wyers, R. P. Otjes and J. Slanina, Atmos. Environ. Part A, 2 E. Mattews, Global biogeochem. cycles, 1994, 8(4), 411. 1993, 27(13), 2085. 3 J. M. M. Aben, P. S. C. Heuberger, R. C. Acharya and A. L. M. Dekkers, Preliminary validation of ammonia emission data using a combination of monitoring and modelling. Proceedings Acidification Paper 9/02378J J. Environ. Monit., 1999, 1, 2