Biogeosciences Exchange of reactive nitrogen compounds : concentrations and fluxes of total ammonium and total nitrate above a spruce canopy

Total ammonium (tot-NH+4 ) and total nitrate (tot-NO−3 ) provide chemically conservative quantities in the measurement of surface exchange of reactive nitrogen compounds ammonia (NH 3), particulate ammonium (NH + 4 ), nitric acid (HNO3), and particulate nitrate (NO − 3 ), using the aerodynamic gradient method. Total fluxes were derived from concentration differences of total ammonium (NH 3 and NH+4 ) and total nitrate (HNO3 and NO − 3 ) measured at two levels. Gaseous species and related particulate compounds were measured selectively, simultaneously and continuously above a spruce forest canopy in south-eastern Germany in summer 2007. Measurements were performed using a wetchemical two-point gradient instrument, the GRAEGOR. Median concentrations of NH 3, HNO3, NH + 4 , and NO − 3 were 0.57, 0.12, 0.76, and 0.48 μg m −3, respectively. Total ammonium and total nitrate fluxes showed large variations depending on meteorological conditions, with concentrations close to zero under humid and cool conditions and higher concentrations under dry conditions. Mean fluxes of total ammonium and total nitrate in September 2007 were directed towards the forest canopy and were −65.77 ng m−2 s−1 and−41.02 ng m−2 s−1 (in terms of nitrogen), respectively. Their deposition was controlled by aerodynamic resistances only, with very little influence of surface resistances. Including measurements of wet deposition and findings of former studies on occult deposition (fog water interception) at the study site, the total N deposition in September 2007 was estimated to 5.86 kg ha −1. Correspondence to: V. Wolff (veronika.wolff@art.admin.ch)

importance of the different deposition processes for a given chemical compound depends on whether the substance is present in gaseous or particulate form, its solubility in water, the amount of precipitation in the region and the terrain and land surface cover type (Seinfeld and Pandis, 1998;Erisman et al., 2005b;Foken, 2008).
Several methods exist to estimate bulk and dry N r deposition loads. These are 25 throughfall methods and micrometeorological methods to determine exchange fluxes of single N r compounds or sums of those. Throughfall, often used in forestry studies (Lovett and Lindberg, 1984;Berger et al., 2008;Hovmand and Andersen, 1995), may be regarded as the sum of wet and dry deposition and canopy exchange, as through-are mostly derived for flat homogeneous terrain and low vegetation and may not be adequate for complex sites, like highly structured, hilly, or forested sites (Hertel et al., 2006;Wesely and Hicks, 2000;Andersen and Hovmand, 1999). Additionally, parameterizations of surface related parameters are only valid for certain temperature and humidity ranges, fertilisation state and vegetation types (cf. Farquhar et al., 1980) and 5 may not always be applied for other ecosystems. The dry deposition of particles, such as particulate NH + 4 and NO − 3 is typically very different to the dry deposition of gaseous species (Nemitz et al., 2004a). Generally, deposition velocities depend on aerosol particle size and roughness of the underlying surface and they are about one order of magnitude smaller than those of gaseous compounds (Gallagher et al., , 2002. 10 From all N r compounds, NH 3 and HNO 3 and their particulate counterparts NH + 4 and NO − 3 are regarded as in the main contributors to atmospheric N r deposition (Andersen and Hovmand, 1999). NH 3 is emitted primarily by agricultural activities, such as volatilization from animal waste and synthetic fertilizers, but also from biomass burning, losses from soils, and fossil fuel combustion (Krupa, 2003). It is an important base 15 constituent in the atmosphere, and neutralises acids, such as sulphuric and nitric acid (HNO 3 ) and hydrochloric acid, forming ammonium (NH + 4 ) salts, whose major portion is present in the fine particle fraction (Finlayson-Pitts and Pitts, 1999). High concentrations of NH 3 are usually found close to sources since it is either effectively dry deposited close to its source and/or rapidly converted to NH + 4 (Ferm, 1998). Particulate NH + + 4 and NO − 3 form the majority of the long-range transported N r in the atmosphere (Hertel et al., 2006). NH 3 , HNO 3 and NH 4 NO 3 form a thermodynamic equilibrium between the gaseous species (NH 3 , HNO 3 ) and particulate NH 4 NO 3 (in solid or in aqueous form), which is 5 a function of temperature (T ) and relative humidity (RH) (Mozurkewich, 1993;Stelson and Seinfeld, 1982): NH 3(gaseous) + HNO 3(gaseous) T,RH ←→ NH 4 NO 3(solid/aqueous) Additionally, the equilibrium of React. (R1) depends on the chemical composition of aerosol particles, especially on concentrations of those ions that are competing with 10 NO − 3 for the NH + 4 , such as SO 2− 4 and Cl − (Wexler and Seinfeld, 1990). The application of the flux-gradient-similarity theory (see below) presumes that vertical fluxes of the measured compounds are constant with height within the atmospheric surface layer (Dyer and Hicks, 1970), which implies that they are considered chemically non-reactive tracers (e.g., Trebs et al., 2006). Phase changes due to shifts of 15 the thermodynamic equilibrium between gaseous and particulate species (as a result of fluctuations in T and RH) may, however, induce a chemical flux divergence of NH 3 , HNO 3 , NH + 4 and/or NO − 3 . Without correcting for these effects, dry deposition estimates derived from gradient measurements may substantially under-or overestimate the actual turbulent flux Nemitz et al., 2004a). The sum of both 20 phases, total ammonium (tot-NH + 4 ) and total nitrate (tot-NO − 3 ), are, however, conservative quantities in this respect (Kramm and Dlugi, 1994;Brost et al., 1988). While, for the quantification of the N r input in an ecosystem like in this study, the partitioning among gas and particulate phase is of minor importance, measurements of exchange fluxes of individual compounds (NH 3 , HNO 3 , NH + 4 and NO − 3 ) are a prerequisite for investigating 25 near-surface mechanistic processes required for atmospheric chemistry and transport models.
In this paper we will present measurements of concentrations of ammonia (NH 3 ), deposition and estimations on occult deposition, we estimate a total N r deposition rate for September 2007. In a subsequent paper, we will investigate chemical gas-particle conversion processes and the corresponding potential influence on fluxes of individual N r compounds.

Weidenbrunnen research site
The experiment was conducted in summer/autumn 2007 (25 August-03 October) within the framework of the project EGER (ExchanGE processes in mountainous Regions) at the research site "Weidenbrunnen" (50 • 08 N, 11 • 52 E; 774 m a.s.l.), a Norway spruce forest site located in a mountainous region in south east Germany (Fichtel-15 gebirge). The surrounding mountainous area extends approx. 1000 km 2 and is covered mainly with forest, but also some agricultural land, meadows and lakes. Continuous air quality measurements indicate that the site is characterized as a rural site of Central Europe (Klemm and Lange, 1999). It is located in the transition zone from maritime to continental climates with annual average temperatures of 5.0 • C (1971Foken, 20 2003) and average annual precipitation sum of 1162. 5 mm (1971Foken, 2003). The study site is maintained for more than 10 years by the University of Bayreuth and a variety of studies have been conducted there (Falge et al., 2005;Held and Klemm, 2006;Klemm et al., 2006;Rebmann et al., 2005;Thomas and Foken, 2007;Wichura et al., 2004). The stand age of the Norway spruce (Picea abies) was approx. 54 23 m (Staudt, 2007), and the single sided leaf area index during the measurement campaign was approximately 5.3. Measurements were performed on a 31 m walk-up tower. For the Weidenbrunnen site, Thomas and Foken (2007) determined the roughness parameters displacement height (d ) and roughness length (z 0 ) as 14 m and 2 m, respectively.

Aerodynamic gradient method
The aerodynamic gradient method (AGM) is based on the gradient-flux similarity and derives fluxes from measured vertical concentration differences and micrometeorological exchange parameters (Ammann, 1998;Foken, 2008). The flux, F , is calculated as the product of a turbulent diffusion transfer coefficient, expressing mechanically and 10 thermally induced turbulence, and the vertical concentration difference, ∆C.
where u * is the friction velocity (m s −1 ), κ the von Kármán constant (0.4), Ψ H the integrated stability correction function for sensible heat (considered equal to that of trace compounds), and z/L is height z over the Obukhov length L, a measure of atmospheric 15 stability. The first term in the product on the right hand site of the equation is often referred to as the transfer velocity, v tr (m s −1 ). It represents the inverse resistance of the turbulent transport between the two heights z 1 and z 2 (Ammann, 1998). Note here, that we use all measurement heights z 1 , z 2 , and z, as aerodynamic heights above the zero plane displacement height, d . Equation (1) is strictly valid only for "smooth" surfaces 20 (e.g. pastures, meadows). However, when using this relationship close to a canopy to infer fluxes from measured vertical concentration gradients, corresponding fluxes have been found to be underestimated (Thom et al., 1975;Garratt, 1978;Hogstrom et al., 1989). Nevertheless, the flux-gradient relationship was found to hold above forest canopies, introducing a so-called enhancement factor into the left hand term of the right hand side of Eq. (1) (e.g., Simpson et al., 1998). However, as we have made use of directly measured u * , measured by an eddy covariance system (at 31 m height), it is not necessary to consider roughness sublayer enhancement factors in our evaluations 5 (Garratt, 1992). An underlying assumption of the AGM is that the flux to or away from the reference surface (e.g., canopy top) is identical to the vertical flux measured at a reference level in some distance above the surface (see above). This assumption may not hold due to chemical reactions occurring within the air layer between the surface and the reference 10 height (Fowler and Duyzer, 1989;Meixner, 1993). For the determination of total ammonium and total nitrate this is not of relevance, but does affect flux determination of individual gaseous and/or particulate compounds (NH 3 , HNO 3 , NH + 4 , NO − 3 ) as phase changes may lead to flux divergence (e.g., Nemitz et al., 2004a;Brost et al., 1988;Huebert et al., 1988). However, as long as the characteristic time scale of chemical 15 transformation is large in comparison to the turbulent timescale, fluxes of compounds that underlie rapid chemical transformation may be determined with sufficient accuracy (De Arellano and Duynkerke, 1992;Nemitz et al., 2004b). Photochemical reactions involving NH 3 and HNO 3 are slow compared to turbulence; however, timescales of phase changes within the NH 3 -HNO 3 -NH 4 NO 3 triad (see React. R1), may be comparable to 20 characteristic times of turbulent transport (Trebs et al., 2006). The turbulent timescales are estimated according to Mayer (2008). The timescales to achieve thermodynamic equilibrium between gas and aerosol phase of the NH 3 -HNO 3 -NH 4 NO 3 triad can be approximated as a function of the aerosol particle surface available for the equilibrium reaction Seinfeld, 1990, 1992).

The GRadient of AErosol and Gases Online Registrator (GRAEGOR)
The GRAEGOR is a wet chemical instrument for semi-continuous two-point gradient measurements of water-soluble reactive trace gas species (NH 3 , HNO 3 (Thomas et al., 2009). GRAEGOR collects the gas and particulate samples simultaneously at two heights (for EGER: 24.4 m and 30.9 m) using horizontally aligned wet-annular rotating denuders and steam-jet aerosol collectors (SJAC), respectively. Air is simultaneously drawn through GRAEGOR's sample boxes, passing first the wet-annular rotating de-5 nuders, where water-soluble gases diffuse from a laminar air stream into the sample liquid. In both SJACs, the sample air (now containing only the aerosol particles) is then mixed with water vapour from double-deionized water and the supersaturation causes particles to grow rapidly (within 0.1 s) into droplets of at least 2 µm diameter. These droplets, containing the dissolved particulate species are then collected in a cyclone 10 (cf. Slanina et al., 2001). The airflow through the two sample boxes is ∼14Lmin −1 (at STP=0 • C and 1013.25 hPa) per box and is kept constant through a critical orifice downstream of the SJAC. The inlets of the sample boxes, directly connected to the wet-annular rotating denuders, consisted of PFA (perflouroalkoxy) Teflon tubing (I.D.=0.8cm, length=20cm), ended upstream in a home-made PFA Teflon rain protec- 15 tion, and were covered by a PFA gauze. Liquid samples (from both denuders and SJACs) are analyzed online using ion chromatography for anions and by flow injection analysis for NH + 4 . GRAEGOR provides one half-hourly averaged gas and particulate concentrations for each height for each species within each hour (cf. Thomas et al., 2009). The analytical performance of the instrument is continuously checked using 20 an internal bromide standard that is added to each sample. Calibrations of the ion chromatograph using Merck certiPUR ® standard solutions were performed twice, 6-7 September and 26 September. The FIA detector was calibrated weekly. The limit of detection (LOD) for the individual species was determined from in-field blanks once a week (for details see Wolff et al., 2009). The errors of the air concentrations of NH 3 , 25 HNO 3 , particulate NH + 4 , and particulate NO − 3 were calculated according to Trebs et al. (2004) and Thomas et al. (2009) using Gaussian error propagation. The precision of the measured concentration differences (σ ∆C /C) was investigated by extended sideby-side measurements in the beginning and at the end of the experiment (Wolff et

Supporting measurements
Vertical profiles of meteorological parameters, such as ambient temperature (T ), relative humidity (RH), and wind speed were measured at the tower (31 m high) at three heights above the canopy (at z=24.4, 26.6, and 30.9 m (T , RH, using psychrometers 5 (fine-wire thermocouples, custom build)) and z=24.3, 26.2 and, 31.2 m (wind speed, 3 cup anemometer (A100ML, Vector Instruments, UK))). On top of the tower (z=32m), an eddy covariance system (Gill R2 sonic anemometer in combination with a LI-COR-7000) measured three-dimensional wind speed, wind direction, friction velocity, stability, latent, sensible heat and CO 2 fluxes. Atmospheric visibility using a present weather  2007) was operated. During the field study, samples were collected event-based and stored cool until analysis (Bayerische Landesamt für Umwelt, Augsburg, ion chromatography). In September a total of six rain samples were analysed for chloride, nitrite, nitrate, phosphate, sulphate, sodium, ammonium, potassium, magnesium, and calcium.

Meteorological conditions
Meteorological quantities, such as RH, T , wind speed (m s −1 ), wind direction ( • ), rain (mm), visibility (m) and global radiation (W m −2 ) for the month of September in 2007 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version Interactive Discussion are summarized in Fig. 1. Two episodes with several consecutive sunny days are in contrast to prevailing humid, foggy conditions with frequent rain and reduced visibility. During fog and rain, the temperature amplitude was reduced and temperatures ranged between below 5 and 15 • C, while RH almost always remained above 70-80%. During the sunny episodes (12-17 September and 19-24 September) temperatures show 5 a diel variation with increasing temperatures from day to day (up to more than 20 • C). During these days, RH dropped to below 70% during daytime. The wind speed was generally quite high with a median of 2.8 m s −1 , ranging between 2 and 4 m s −1 (inter quartile range). During the first half of September (foggy conditions), the wind blew frequently from northerly directions, turning towards south westerly directions afterwards.

Detection limits, precision and mean concentrations
To facilitate comparison between the concentrations and fluxes of the different nitrogen containing compounds all numbers are given in terms of nitrogen 1 . In this paper we focus on the measurements made in September 2007. Until the 26 September we measured concentrations at two levels, and after that concentration measurements at 15 one level are available (side-by-side measurements, see Wolff et al., 2009). Determined LOD values (3σ-definition) were 0.017 µgm 3 for NH 3 and particulate NH + 4 , and 0.029 µ g m −3 for HNO 3 and particulate NO − 3 , respectively (see Wolff et al., 2009). Less than 1% of the total NH 3 concentrations were below the detection limit, less than 2% of the total particulate NH + 4 concentrations, but 30% of the HNO 3 concentrations and 20 8% of the particulate NO − 3 concentrations were below corresponding detection limits. Below LOD concentrations were predominantly measured during rainy periods. For tot-NH + 4 and for tot-NO − 3 the precision was found to be 5.3% and 4.8%, respectively. The median error of the concentration difference (σ ∆C /∆C) was found to be 52.1% and 1 All concentrations and fluxes are given in µ g m −3 in terms of N. To convert these to µ g m −3 they must be multiplied with the ratio of the molar masses, i.e. for NH 3  37.9%. Measured concentration differences larger than the precision were considered to be significant and subsequently used for flux calculations, while those below the precision were determined to be insignificantly different from zero. For tot-NH + 4 , 55% of the 443 total measured concentration differences were found to be significant, for tot-NO − 3 , 77% of the 373 total concentration differences were significant. From the error of 5 the concentration difference, the flux error is derived in combination with an estimated error of v tr of around 10% (see Wolff et al., 2009 (inter quartile range between 0.32 and 1.39 µgm −3 ). HNO 3 varied between 0.05 and 15 0.28 µ g m −3 (median: 0.12 µ g m −3 ), while the particulate counterpart, NO − 3 was three to six times larger, varying between 0.18 and 0.80 µ g m −3 (inter quartile range; median: 0.48 µ g m −3 ). Concentrations varied with meteorological conditions and were generally higher during periods with higher temperatures and lower RH and reached their minimum in rainy periods. During the sunny and drier days ( Fig. 1, white bars) 20 the overall data coverage and the percentage of significant concentration differences were higher. High RHs, rain and fog droplets may cause problems with losses in the inlet lines (Wolff et al., 2009). Thus, our investigation of dry deposition and exchange processes of tot-NH + 4 and tot-NO − 3 will focus on the two sunny and dry episodes. 25 The diel variations of gaseous NH 3  Interactive Discussion position and phase partitioning), as well as horizontal and vertical transport within the planetary boundary layer. NH 3 and particulate NH + 4 share the same concentration range during the study and reveal a regular pattern of higher particle concentrations at night and higher gas phase concentrations during daytime, especially during the drier and sunnier episodes (cf. Fig. 2a, especially 22-25 September). HNO 3 and particulate 5 NO − 3 concentrations are very different, with particle concentrations of up to four times higher than gas phase concentrations (note different scales of the y-axes in Fig. 2b). Generally, HNO 3 and particulate NO − 3 concentrations also follow the pattern of high nighttime particle concentrations and high daytime gas phase concentrations. In the time series of tot-NH + 4 and tot-NO − 3 (Fig. 2c) the effect of gas-particle interactions is 10 removed. Their concentrations, however, still vary with time, reflecting meteorological conditions, with low values during rainy and foggy conditions and higher concentrations during the sunnier episodes (cf. Fig. 1). This difference between fair weather conditions and rainy/foggy episodes is more pronounced for tot-NO − 3 , reflecting to some extent the fact that production is linked with photochemical processes (oxidation of NO 2 with OH 15 to HNO 3 ). Tot-NH + 4 shows a regular pattern of higher concentrations towards the afternoons, while in the time series of tot-NO − 3 such a pattern can not be identified.

Timescale analysis
The diel variations of NH 3 , HNO 3 , NH + 4 and NO − 3 were most likely influenced by changes of T and RH (see Figs. 1 and 2) and subsequent changes of the thermo-20 dynamic equilibrium (React. R1) between gaseous and particulate phase. To estimate the effect of the system striving towards equilibrium on the determination of exchange fluxes, we performed a timescale analysis for the time of available aerosol particle size distribution measurements Seinfeld, 1990, 1992). The Damköhler ratio (Fig. 3), the ratio of characteristic turbulent timescales to equilibration timescales 25 (Da=τ turb /τ equi ), is a measure of the degree to which chemical conversion may affect the determination of exchange fluxes by micrometeorological methods (Foken et al., 1995). Assuming that all surfaces of the particles take part in the equilibrium reaction 10676 Introduction  Fig. 3), Da often approaches and exceeds unity (10 0 ), especially during nighttime. For a smaller fraction of the aerosol particle surface taking part (e.g., 10% red line in Fig. 3), turbulent transport would be fast enough to exclude the influence of chemical divergence affecting the concentration gradients, at least during daytime (Da<0.1). Consequently, for flux measurements of individual N r compounds of the 5 NH 3 -HNO 3 -NH 4 NO 3 triad, the gas-particle partitioning processes need to be considered. In this paper, focusing on the total dry deposition of ammonium and nitrate, we confine ourselves to the derivation of fluxes of the conservative sums of gaseous and particulate phase, tot-NH + 4 and tot-NO − 3 . As stated above, chemical divergences and single compound fluxes will be investigated in a subsequent publication.

Fluxes and deposition velocities
Due to the high roughness of the forest, the transfer velocity was quite large with 0.45±0.25ms −1 . Maximum transfer velocities of 0.70±0.18ms −1 were found at noon, while during nighttime minimum values were 0.25±0.22ms −1 . From the measured concentrations at two levels above the forest canopy, we calculated the fluxes for total 15 ammonium and total nitrate. In September, concentration measurements at two levels were available until the morning of the 26th. Fluxes varied significantly according to the prevailing meteorological conditions (see Fig. 4). Fluxes of tot-NH + 4 were generally smaller during fog and humid conditions and showed emission events from wet or drying surfaces and large deposition fluxes during the sunny days. Fluxes of tot-NO and tot-NO − 3 , remain below the aerodynamically maximum possible value, indicated as 1/R a . 15 Major constituents of dry N r deposition are particulate NH + 4 and NO − 3 and gaseous compounds NH 3 , HNO 3 and to a minor extent also nitrogen dioxide (NO 2 ), nitrous acid (HONO), peroxyacetyl nitrate (PAN), and nitric oxide (NO) (Andersen and Hovmand, 1999). Surface-atmosphere exchange fluxes of NO and NO 2 measured by eddy covariance at our site were found to be one order of magnitude lower than the fluxes of (inter quartile range). The directions of NO and NO 2 fluxes are in contrast to what would be expected for low vegetation (Delany and Lenschow, 1987), however they are mainly due to chemically induced flux divergence (Meixner, 1993). Concentrations of HONO were in the order of 0.03 to 0.09 µgm −3 (inter quartile range), and concentra-5 tions of PAN were not measured at the site. Hence, the high importance of tot-NH + 4 and tot-NO − 3 in N r deposition generally recognized (Erisman and Draaijers, 2003;Erisman et al., , 2005aErisman et al., , 2007Andersen and Hovmand, 1999;Asman et al., 1998;Galloway et al., 2008;Sutton et al., 2007) is confirmed by our study. Median diel tot-NH + 4 and tot-NO  (30). This method was used because we do not have uninterrupted flux measurements, especially under foggy and rainy conditions, when the instrument worked less reliable (see above and Wolff et al., 2009). Flux values derived from non-significant concentration differences were also included in the calculation, as these were usually close to zero and had thus a significant influence on the median.  Fig. 1). In most of the rain samples NH + 4 was dominating over NO − 3 (Fig. 7), with concentrations being about one third larger. In the rain storm after the six days of sun (19)(20)(21)(22)(23)(24)(25)

Fluxes and deposition velocities
Concentrations levels of gaseous NH 3 and HNO 3 and particulate NH + 4 and NO − 3 , ob-10 served in our study (Fig. 2) are comparable to previous observations at the Weidenbrunnen site (Held et al., 2002). Also the dominance of the particulate phase over the respective gas phase was observed by Held et al. (2002). Flux values and values of v d derived in this study are relatively large. There are indications that the exchange of reactive species at our site is not limited by any surface 15 resistance. In summer 2001 exchange fluxes of hydrogen peroxide (H 2 O 2 ) were determined at our site using a relaxed eddy accumulation technique . The characteristics of H 2 O 2 are comparable to HNO 3 , as it is produced above the canopy in the gas phase (mainly by recombination of two HO 2 radicals) with sinks in the particulate phase and efficient dry deposition to surfaces due to its high sol-20 ubility and reactivity (Hall et al., 1999;Walcek, 1987). A modelling analysis indicated that H 2 O 2 exchange was largely controlled by turbulent transport to and into the canopy, and also by the supply from above where chemical production occurs (Ganzeveld et al., 2006) Most previous studies derived fluxes of individual N r compounds only or derived total fluxes using the inferential method that is constrained by required parameterizations of surface related exchange parameters (see Sect. 1).We found only one study (Sievering et al., 1994) that reported directly measured tot-NO − 3 fluxes (using the AGM) and v d above the forest. They found evidence for large deposition rates of tot-NO

mgm
−2 s −1 , which is very similar to what we found. The observed geometric median mass diameters of NO − 3 and NH + 4 were 2.24±0.85µm and <0.9µm, respectively. For these particle diameters, Peters and Eiden (1992) Sievering et al. (1994) were in the range of 2-9 cms −1 , being approximately 15 equal for HNO 3 and NO − 3 . Due to large uncertainties in particulate NH + 4 measurement, they could not determine exchange fluxes and/or v d for tot-NH + 4 . Since the particulate phase dominated our measured concentrations (see Sect. 3.2) we presume that they also dominate the deposition fluxes, at least in tot-NO − 3 . Aerosol particle fluxes and v d depend on particle diameter, atmospheric conditions (friction velocity and stability) 20 and surface conditions, such as roughness and canopy morphology (Erisman et al., 1997;Gallagher et al., 1997;Peters and Eiden, 1992;Fowler et al., 2009). Reported v d range from some mms −1 for small particles (<1µm) and for low wind speeds to more than 10 cms −1 for larger particles (>10µm) and high wind speeds. The high roughness and the large surface of the needles of the spruce forest at our site com- 25 bined with the aerodynamic regime of high friction velocities (inter quartile range 0.32-0.63 m s −1 ) partly explain the finding of large aerosol particle v d (Fowler et al., 2009). Furthermore, the 50% theoretical particle cut-off diameters of the GRAEGOR due to inlet design and denuder airflow regime is 0.2 nm and 18 µm (Thomas et al., 2009). According to Peters and Eiden (1992) v d for aerosol particles captured by the GRAE-GOR at typical wind speeds at our site (2-4 m s −1 ) may have varied between 0.08 and more than 10 cms −1 . Measurements at the Weidenbrunnen site by Held et al. (2002) revealed consistent patterns in the size distributions of particulate NH + 4 and NO − 3 , with 5 the former one dominating the fine particle concentrations (mean particle diameter of 0.25 and 0.71 µm), and the latter dominating the coarse particles ranges (mean particle diameter of 0.71 and 2 µm). For particles with mean diameters of 0.25, 0.71 and 2 µm, v d at the encountered wind speeds would theoretically range between 0.01-0.6, 0.01-1.08, and 0.06-3.5 cms −1 , respectively (Peters and Eiden, 1992). According to 10 Gallagher et al. (1997) these values would be 0.05-1, 0.03-2, and 0.11-0.8 cms −1 .
Thus, another possible explanation for the large v d found in our study (Sect. 3.5 and Fig. 5), could be the presence of large (≥10µm) NH + 4 and NO − 3 containing particles. Deposition velocities of NH + 4 and NO − 3 (derived from eddy covariance measurements) larger than those of SO 2− 4 and those derived from particle number flux mea- 15 surements have been reported (Nemitz et al., 2004b;Thomas, 2007). These were explained by changes in the thermodynamic equilibrium towards the more rapidly depositing gaseous species between the measurement height and the vegetated surface (Fowler et al., 2009). The quick removal processes of NH 3 and HNO 3 just above and within the canopy together with warm surface temperatures would favour aerosol 20 evaporation, consequently enhancing total deposition of both, particulate and gaseous phase. We observed highest v d during daytime of the sunny episodes during our measurement period. During these times, temperature was highest at the canopy top. Although using the chemically conservative quantities of tot-NH + 4 and tot-NO − 3 , the derived total deposition fluxes were probably influenced by this additional sink, resembling 25 thus more v d of gaseous species than those of particles. Assuming such a mechanism we can follow up reports which claimed the importance of particulate N species in N deposition estimates and the use of effective deposition velocity parameterisations for highly volatile aerosol compounds rather than using parameterisations regardless of the particulates' chemical composition (Erisman et al., 1995(Erisman et al., , 1997Fowler et al., 2009).

Deposition of reactive nitrogen
Wet deposition rates of NH + 4 and NO − 3 measured previously in summer 2001 at the Weidenbrunnen site were very similar to ours, with September sums being 0.906 kg ha −1 5 NH + 4 and 0.835 kg ha −1 NO − 3 (see Klemm and Wrzesinsky, 2007). In order to determine the total nitrogen deposition at our site we also have to take into account the occult deposition through the interception of fog water. The Weidenbrunnen research site shows a high frequency of fog events throughout the year and throughout our measurement campaign (see meteorological conditions, Fig. 1). Several studies on 10 fog meteorology and chemistry have been conducted at that site (Klemm and Wrzesinsky, 2007;Wrzesinsky and Klemm, 2000), in which the importance of the so called occult deposition was stressed. For a study between April 2001 and March 2002, most fog events were found to be associated with clouds being advected from westerly directions, intercepting with the forest vegetation of the site as it is located on a moun- 15 tain range of about 1000 ma.s.l. (Klemm and Wrzesinsky, 2007). The most important ionic constituents of fog water were found to be NH + 4 , NO − 3 and SO 2− 4 . These three ions were significantly enriched in the fog water compared to the rain water samples (ratios fog/rain concentrations 18.1, 12.7 and 11.8, respectively), which led to the conclusion that during the study of April 2001 and March 2002 the occult deposition of 20 NH + 4 , NO − 3 and SO 2− 4 in fog water was similar and often larger than the wet deposition, although more liquid water was deposited through rain than through fog. Accounting these findings, we estimate the occult deposition in September 2007 as equal to the wet deposition. Consequently, the total N r deposition in September 2007 sums up to 5.86 kg ha −1 .

25
To compare our measurement results with results from other deposition studies of reactive nitrogen (N r ), we derive estimates on potential annual deposition sums. Held et al. (2002) found substantial differences in concentrations of N r species in winter as compared to summer values at the Weidenbrunnen site, with winter values being less than half of the summer values. Concentrations of NH 3 were below the detection limit during the winter 2001 measurements. Obviously, the total N r deposition varies throughout the year. If we would consider that six month of a year the wet, dry and The comparison between our short-term measurements based on the AGM method with the long-term monitoring data is not straight forward. Firstly, the estimation of annual deposition from measurements during one month only is very rough, especially 20 in comparison with long-term observations. The throughfall approach is very different to a micrometeorological approach. It is an integrated, straight forward and relatively cheap approach for deposition monitoring, but drawbacks comprise spatial representativeness and the missing knowledge on canopy exchange, especially of N r species. The canopy exchange includes both leaching and consequently efflux from the canopy 25 as well as uptake or retention and therefore influx to the canopy. Throughfall methods may therefore easily either over-or underestimate total deposition fluxes. Although Klemm and Wrzesinsky (2007) found similar wet deposition values in September 2001 compared to our September measurements, the annual wet depo-sition for April 2001 to March 2002 was much higher. 2001 was a comparatively wet year with high fog occurrence (on 233 days compared to an annual mean (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007) of 200 days). The estimate we made using similar deposition rates through rain and fog/cloud water interception is, however, also valid for the annual budget. For the "Fichtelgebirge" mountains, Matzner et al. (2001) reported average through-5 fall fluxes of mineral N of 21 kgha −1 yr −1 , pointing out that this value underestimates the actual atmospheric deposition due to the canopy uptake, which they estimate as high as 28 kgha −1 yr −1 . Our two estimates of the annual N r deposition flux are both larger than their throughfall flux data but the lower estimate is within the range of estimated throughfall plus canopy uptake. Zimmermann et al. (2006) measured deposition 10 of NH + 4 and NO − 3 in the Erzgebirge in south east Germany for several years (2001)(2002)(2003)(2004). He used both micrometeorological and throughfall methods and measured dry, wet, and occult deposition. The wet deposition is comparable to our results, being more important than dry deposition. Differences between the dry deposition estimate of Zimmermann et al. (2006) and our study could be due to the use of the inferential 15 model by Zimmermann et al. (2006) and, of course, to our very rough estimation of the annual deposition on only one month measurements. The fog/cloud water deposition was not very significant in the study of Zimmermann et al. (2006), but it is pointed out that it is important for the forest at higher altitudes of the Erzgebirge (Zimmermann and Zimmermann, 2002). Throughfall fluxes were roughly two thirds of the sum of dry, wet, 20 and occult deposition, suggesting possible canopy uptake of NH + 4 and NO − 3 . Rothe et al. (2002) reported N r bulk deposition fluxes of 10.7 kgha −1 yr −1 and N r throughfall fluxes of 29.6 kgha −1 yr −1 to a spruce forest in south west Bavaria, Germany, about 190 km south southwest of our site. Their study includes N r deposition loads of 15 European spruce stands (all determined using throughfall methods) ranging from 11.2 to 25 56 kgha −1 yr −1 . Sievering et al. (1994) reported a range of diel deposition fluxes of tot-NO − 3 and annual wet deposition of total inorganic N r from which we derived an annual deposition range (cf. Fig. 8) Sievering et al. (1994) point out that their results indicate an exceedance of the critical N loads. Berger et al. (2009) measured throughfall fluxes in a nutrient poor spruce forest in northern Austria and found relatively small values, slightly less than deposition values estimated by Matzner et al. (2001) for our site. Berger et al. (2009) used a canopy exchange model to estimate total deposition fluxes 5 and modelled the canopy to be a source for NO − 3 and a sink for NH + 4 . Direct deposition estimates (AGM) of different N r species for the year 1995 at the Speuld forest in The Netherlands were reported by Erisman et al. (1996). Dry deposition rates of tot-NH + 4 and tot-NO − 3 were 22.60 kgha −1 yr −1 and 4.67 kgha −1 yr −1 , respectively. Reported wet deposition rates were 11.3 kgha −1 yr −1 for NH + 4 and 5.2 kgha −1 yr −1 for NO − 3 . Total N 10 deposition (the sum of dry and wet deposition) yielded 43.77 kgha −1 yr −1 , which is very similar to our estimates (when looking at dry and wet deposition). Andersen and Hovmand (1999) report dry deposition estimates derived using the AGM (NH 3 ) and inferential modelling (HNO 3 , NO mark. In Denmark the dry deposition of N r compounds to forest ecosystems contributes with 50-67% to the total composition (Andersen and Hovmand, 1999). Thus, the total deposition would yield an annual N deposition flux of 17.1 to 22.8 kgha −1 yr −1 . Although we estimated the annual dry deposition to our site from one month measurements only, the comparison to long term direct measurements in The Netherlands 20 (Erisman et al., 1996) and Denmark (Andersen and Hovmand, 1999) is remarkable good. Additionally the ratio of dry to wet deposition compares very well with these studies. Occult deposition was only measured in two studies and the importance of this deposition pathway at our site was stressed (Zimmermann et al., 2006;Klemm and Wrzesinsky, 2007). All of the studies listed in Fig. 8 were conducted above/in is also reflected in higher seepage rates of NO 3 . The high surface to volume ratio of the canopy may thus also enhance dry deposition. The deposition load also largely depends on the pollution levels of the surrounding air masses, i.e. the proximity of agriculture, industry and traffic. The effect of this becomes evident when comparing the studies by Andersen and Hovmand (1999) and Erisman et al. (1996) in Denmark (west 5 of Denmark, close to the sea, few sources nearby) and The Netherlands (surrounded by intensive agriculture), respectively.

Conclusions
An intensive field campaign with hourly resolved two-level measurements of ammonia (NH 3 ), nitric acid (HNO 3 ), particulate ammonium (NH + 4 ) and nitrate (NO − 3 ) were conducted above a spruce forest in southeast Germany. For the first time, the complete NH 3 -HNO 3 -NH 4 NO 3 triad was measured continuously and simultaneously at two levels above a forest canopy, allowing for the calculation of surface-atmosphere exchange fluxes using the aerodynamic gradient method. However, indications for rapid phase changes in the NH 3  (c) potential strong chemical sink below our measurement height due to phase 25 changes between NH 3 , HNO 3 and NH 4 NO 3 towards the more efficiently deposited gases Consequently, we like to state, that the measurement of concentrations of only one single compound at one single level above forest using the inferential model to calculate fluxes (deriving R c values) may lead to underestimation of the actual deposition load, 5 especially in the case of aerosol particles. In our study, tot-NH + 4 dry and wet deposition was larger than the respective tot-NO − 3 deposition. An estimated annual total nitrogen deposition load, approximated from dry, wet, and occult (fog interception) deposition is at the upper end of the range reported in literature. Thus, our site is a significant receptor region for reactive nitrogen.