Bi-directional air-surface exchange of atmospheric ammonia: A review of measurements and a development of a big-leaf model for applications in regional-scale air-quality models



[1] The air-surface exchange of atmospheric ammonia (NH3) and measurements of the canopy and stomatal compensation points (χcp and χst, respectively) and the stomatal and soil emission potentials (Γst and Γg, respectively) are reviewed. A database of these values has been developed, to be used for the development of input parameters and for model evaluation. The compensation points are dependent on canopy type, nitrogen (N) status, temperature, growth stage, and meteorological conditions. Canopies that receive high atmospheric nitrogen input generally have high χcp values. χcp values also tend to be higher over intensively managed vegetated surfaces than semi-natural vegetation, due to the higher nitrogen content in these surfaces. Increased nitrogen concentrations from fertilization and cutting practices have been observed to increase the compensation points and therefore the emission from these canopies. The decomposition of litter leaves has been found to play a dominant role and significantly increase the values of χcp over agricultural vegetation and fertilized grasslands. By modifying an existing big-leaf dry deposition model to allow NH3 emission from leaf stomata and soil surfaces, a bi-directional air-surface exchange model has been developed for applications in regional-scale air-quality models. The model predicts χcp values that vary with canopy type, nitrogen content, and meteorological conditions. χcp values predicted by the model are around 1–2 μg m−3 over forest canopies and 4–10 μg m−3 over grasslands and agricultural canopies during a typical summer daytime; χcp values are ∼10 and ∼3 times lower over forests and agricultural lands, respectively, in nighttime and/or winter conditions. These results are similar to the range of measured values. The new bi-directional air-surface exchange model will reduce the dry deposition fluxes by 20–100 ng m−2 s−1 compared to the original dry deposition model over low-N forests and agricultural lands during typical summer daytime conditions, which can be the difference between whether deposition or emission occurs. For example, this new model produces emissions fluxes of 0–90 ng m−2 s−1 over croplands when the atmospheric NH3 concentrations are below 10 μg m−3.

1. Introduction

[2] Atmospheric ammonia (NH3) is one of the several key chemical species that needs to be included in regional-scale air-quality models. It is the most abundant alkaline gas in the atmosphere and contributes significantly to the formation of PM2.5 (particulate matter smaller than 2.5 μm in diameter) through chemical reactions with sulfuric, nitric, and hydrochloric acids [e.g., Ansari and Pandis, 1998; Blanchard and Hidy, 2003; Vayenas et al., 2005]. One process that needs improvement in many air-quality models is the treatment of the NH3 air-surface exchange. Although bi-directional air-surface exchange (dry deposition and emission) of NH3 has been frequently observed over a variety of land surfaces, the majority of the air-quality models treat the air-surface exchange of NH3 as dry deposition only. A Unified Regional Air-quality Modeling System (AURAMS) [Gong et al., 2006], developed by Environment Canada, does not address bi-directional exchange and has been found to underpredict the daytime ambient NH3 concentration but to predict a reasonable nighttime concentration. The Community Multiscale Air-Quality Modeling System (CMAQ), developed by the U.S. Environmental Protection Agency), has the same issue on the NH3 air-surface exchange, and a bi-directional air-surface exchange approach described in Cooter et al. [2010] is planned to be implemented into CMAQ. The underestimation of the daytime ambient NH3 concentration is likely caused by the overestimated dry deposition, considering that the observed bi-directional exchange of NH3 mainly occurs during the daytime. It is thus expected that using the bi-directional air-surface exchange approach for NH3 could improve these models' performances.

[3] NH3 and associated ammonium aerosols also contribute significantly to the total atmospheric nitrogen deposition to the ecosystem; thus, excessive amounts of NH3 can cause harmful effects to the ecosystem through wet and dry nitrogen deposition due to eutrophication and acidification [Fangmeier et al., 1994; Krupa, 2003; Sutton et al., 2009a]. Having a more accurate air-surface exchange model for NH3 is also important in assessing the atmospheric nitrogen input to the various ecosystems.

[4] A number of bi-directional air-surface exchange models, in a variety of complexities, have been developed and applied to field measurements [e.g., Sutton et al., 1998; Flechard et al., 1999; Nemitz et al., 2000, 2001b; Walker et al., 2006; Wu et al., 2009]. A comparison of several early models can be found in a paper by Nemitz et al. [2001b], the processes required for modeling χst have been identified in a review by Massad et al. [2010a], and a brief discussion of existing models is also presented by Wu et al. [2009]. Existing models are not ready for general applications in regional-scale air-quality models due to the lack of required input parameters over a large number of different land-use categories (LUCs). The purpose of this study is to conduct a detailed literature review on the NH3 bi-directional air-surface exchange processes and, based on the knowledge obtained from the review, to build a model that can be easily implemented in any regional-scale air-quality model. A database of stomatal and canopy compensation points from available measurements, and stomatal and soil emission potentials from both field and theoretical studies, is built through the literature review. The existing big-leaf dry deposition model of Zhang et al. [2003] is modified to simulate the bi-directional exchange through stomatal and soil surfaces, with input parameters based on the reviewed data. This updated model is evaluated using the reviewed measurements and compared to the original dry deposition model.

2. Measurements

[5] In this section, available measurements of the stomatal and canopy compensation points and stomatal and soil emission potentials are discussed, beginning with a brief description of current measurement methods. Definitions of these values are provided in section 3 with the description of the model.

2.1. Measurement Methods

[6] Various methods are used in the determination of the bi-directional fluxes and compensation points of NH3. The fluxes have been estimated through micrometeorological measurements (MM) such as the aerodynamic method (AER), the modified Bowen-ratio method (MBR), the relaxed eddy accumulation (REA) method, and enclosure method measurements such as the wind tunnel experiments (WT) and the dynamic chambers (DC). Canopy (or sometimes stomatal) compensation points can be generated from the flux measurements collected using the above methods. The apoplastic extraction approach (AE), also known as the bioassay approach, has also been used in extracting the compensation points. A brief description of the basic concept, strengths, and weaknesses of each measurement method that has been used for NH3 measurements are provided in Table 1. A detailed review of all measurement methods for various trace gas species can be found in a paper by Freibauer and Kaltschmitt [2000].

Table 1. Measurement Methods for NH3 Fluxes and Compensation Points, Along With Basic Strengths and Weaknesses of Each Methoda
  • a

    AER = aerodynamic, MBR = modified Bowen-ratio, REA = relaxed eddy accumulation, AE = apoplastic extraction, WT = wind tunnels, DC = dynamic chambers.

AERFlux is determined by concentration gradients and meteorological variables. χcp is determined by the change in the direction of the fluxes.Does not disturb the ecosystem, suitable for long-term measurements.Fetch requirement: a homogeneous area surrounding the measurement site with a size typically 100 times of the measurement height. Vertical concentration gradient can be small over rough surfaces. Lower χst and χcp measurements from conditions of high humidity and low NH3 concentrations.Duyzer et al., 1992; Sutton et al., 1995a; Flechard and Fowler, 1998; Schjoerring et al., 1998; Loubet et al., 2002; Wichink Kruit et al., 2007; Hensen et al., 2009; Milford et al., 2009
MBRSimilar to AER except that the sensible heat (or some gas species) is used as a tracer for calculating eddy diffusivity.Does not disturb the ecosystem, suitable for long-term measurements.Fetch requirement. Vertical concentration gradient can be small over rough surfaces. Under near neutral atmospheric conditions over forests, the temperature gradient is so small that it cannot be determined with sufficient accuracy.Duyzer et al., 1992; Walker et al., 2006; Bash et al., 2010
REAFlux is a function of the difference in the NH3 air concentrations in upward and downward moving eddies.Does not disturb the ecosystem, suitable for long-term measurements. Only needs concentration measurements at one height. Could be used to measure vertical flux divergence when used at two heights simultaneously.Concentration differences between the two reservoirs are small and may cause substantial error. Significant attention is needed in running the apparatus. Higher uncertainty for small fluxes.Nemitz et al., 2001a; Hensen et al., 2009; Milford et al., 2009
AEThe concentrations of NH4+ and H+ in the apoplast are measured and then used to calculate χst.Not affected by cuticular interactions. χst can vary locally, depending on type of plant species. Can be used in both greenhouses and in the field.Sensitive to cuticular interactions. χst can be considerably under- or overestimated compared to other methods, due to possible spatial variability of the pH and [NH4+] in the apoplast. No flux is measured.Husted and Schjoerring, 1995a; Sutton et al., 2001; Hill et al., 2002; Loubet et al., 2002; Massad et al., 2009; Mattsson et al., 2009a
WTCanopies are erected over field sites and used to measure the changes in the mass of NH3 traveling between the entrances and exits.Allows for varying conditions, such as canopy height and N status and is therefore well suited for comparative measurements.Turbulence in the wind tunnel is not representative of outdoor conditions. Light, temperature, and gas concentration in the tunnel may not be representative of outdoor conditions. Concentration differences between in- and outflow are small, and for that reason measurement methods with high accuracy are needed. Extrapolation of the wind tunnel-produced fluxes to real fluxes in the field is complicated.Ross and Jarvis, 2001a, 2001b
DCχcp is determined by the change in the concentration of NH3 between the entrance and exit of the chamber and the rate of the continuous flow of air through the enclosure.Allows for the study of growth stage on χst. Can distinguish between NH3 exchange from plant, soil, and litter. Is best suited for observations at the plant or canopy scale. Observation of the processes that control the NH3 fluxes is relatively straightforward. Suited for comparative measurements.Turbulence in the chamber is not necessarily representative of outdoor conditions. Condensation on the walls, changes in light levels, and increases in temperature can produce unrealistic results. Extrapolation of the chamber-produced fluxes to the field is complicated. Turbulence in the chamber is not representative of outdoor conditions. Seedlings used in enclosure methods do not exactly replicate the bi-directional exchange of young leaves from mature plants.Morgan and Parton, 1989; Husted and Schjoerring, 1995b; Sutton et al., 1995b; Bussink et al., 1996 and references therein; Hill et al., 2002; Köstner et al., 2008; Asman, 2009; David et al., 2009; Massad et al., 2009

2.2. Compensation Point Measurements

[7] A review of past measurements of the compensation point of NH3 for the stomata (unless otherwise indicated) using various experimental methods over different canopy types is provided in Table 2. Reviews of measured and modeled stomatal and ground emission potentials are provided in Tables 3 and 4, respectively. A brief discussion of these measurements is provided below. In Table 2, different units are used by different references to quantify the compensation points. It is worth noting that at 20°C and 1 atm pressure, 1 nmol mol−1 ≈ 0.7 μg m−3.

Table 2. Existing Measurements of the Stomatal Compensation Point of NH3 (μg m−3) Over Different Canopy Types and Under Varying Conditions. Also Included (Where Noted) Are Available Measurements of the Canopy Compensation Point of NH3a
VegetationLocation / MethodConditionsχst (μg m−3)T (°C)CommentsReference
SavannaVenezuela 127–143 nmol mol−1  San Jose et al., 1991
Fagus sylvaticaGermany / DC90 yr old trees, summer3.2–3.4 nmol mol−1 Field experiments – High NGeβler et al., 2000
Fagus sylvaticaGlasshouse / DC12 wk old seedlings3.0 nmol mol−123Controlled conditionsGeβler et al., 2000
Avicennia alba, marina, etc.India / MMPremonsoon327 nmol m−326.85Full year measurementsBiswas et al., 2005
Avicennia alba, marina, etc.India / MMMonsoon178 (51–421) nmol m−329 Biswas et al., 2005
Avicennia alba, marina, etc.India / MMPostmonsoon145 nmol m−322.8 Biswas et al., 2005
Coniferous forestDC 0.3–2.210–27 Gravenhorst and Breiding, 1990b
Coniferous forestScotland / AERNovember0.086χcpSutton et al., 1993a
Pinus sylvestrisUSA / MMMontane-subalpine forest0.5620Low-N supplyLangford and Fehsenfeld, 1992
Pinus sylvestrisBelgium / MMSuburban, spring-autumn1.5–25.18.3–31.2High N, LW = 0, RH < 55%, select dataNeirynck and Ceulemans, 2008
Pseudotsuga menziesiiThe Netherlands / AER35 yr old stand48 nmol mol−1 N-polluted forest (2 yr study)Wyers and Erisman, 1998
Picea abies L. KarstGermany / DCEarly summer0.3  Kesselmeier et al., 1993c
Picea abies L. KarstDenmark/ AERMixed heath land and conifer forest0.2–1.92 – 25χcp, assumed [NH4+] and pHAndersen et al., 1999
Picea abies L. KarstGermany / DC90 yr old trees2.5 nmol mol−1 Exposed to high NH4+ and nitrateGeβler et al., 2002
Picea abiesGermany / AE1 yr old sun-exposed, healthy, summer0.11–0.27 nmol mol−125pH = 3.8–4.2Köstner et al., 2008
Semi-Natural Vegetation
Eriophorum vaginatumEngland / AERNonfrozen/frozen, spring−0.01 / 0.088 / 0χcp, far from sourcesSutton et al., 1992
Eriophorum vaginatumEngland / AERSpring0.017χcp, far from sourcesSutton et al., 1993a
Eriophorum vaginatumScotland / AERNonfrozen, November−0.065χcp, far from sourcesSutton et al., 1992
Eriophorum vaginatum, etc.Scotland / AERWinter/summer0 / ∼0.2 χcp, February to AugustFowler et al., 1998
Calluna/EriophorumScotland / AERNon-frozen, May0.0713χcp, far from sourcesSutton et al., 1992
Calluna vulgarisScotland / AERNon-frozen/frozen, February0.04 / 0.052 / –3χcp, far from sourcesSutton et al., 1992
Mixed moorlandScotland / AERTypical range0–0.3<0–31χcp, upwind sourcesFlechard and Fowler, 1998
Mixed moorlandScotland / AERAll conditions, all seasons0.44 (0–1.8) χcp, upwind N, range from graphFlechard and Fowler, 1998
Mixed moorlandScotland / AERWet surface: cool, wet, windy, September0.380–10χcp, upwind sourcesFlechard and Fowler, 1998
Mixed moorlandScotland / AERDry surface: warm, mod. dry, June–July0.576–22χcp, upwind sourcesFlechard and Fowler, 1998
Mixed moorlandScotland / AERSnow surface: frozen, snow-covered, March0.47<0χcp, upwind sourcesFlechard and Fowler, 1998
Mixed moorlandScotland / AERSummer green season0.11–1.295–25RH < 70%Flechard et al., 1999
Calluna vulgarisThe Netherlands / AERDay: dry, May–June4.9 (<018.7) χcp, local NH3 sourcesNemitz et al., 2004
Calluna vulgarisThe Netherlands / AERDay: wet, May–June0.59 (<0–10.3) χcp, local NH3 sourcesNemitz et al., 2004
Calluna vulgarisThe Netherlands / AERNight: dry, May–June1.3 (<0–15.2) χcp, local NH3 sourcesNemitz et al., 2004
Calluna vulgarisThe Netherlands/AERNight: Wet, May-June1.1 (<0–10.8) χcp, local NH3 sourcesNemitz et al., 2004
Calluna vulgaris L. HullLaboratory/DC2–4 days after plant harvest<0.1 nmol mol−18χg, pH = 6.9Schjoerring et al., 1998
Calluna vulgaris L. HullLaboratory/DC24–26 days after plant harvest8.0 nmol mol−122χgSchjoerring et al., 1998
Calluna vulgaris L. HullLaboratory/DC2 yr old, shoots elongated, flowers5.3 – 6.2 nmol mol−122χcpSchjoerring et al., 1998
Deschampsia flexuosa L. TrinLaboratory/DC5 month old plants3.0–7.5 nmol mol−124–31χcp, pH = 6.8Schjoerring et al., 1998
Calcareous grasslandEngland/AERSpecies rich neutral, August−0.1418χcp, no sourcesSutton et al., 1993a
Calcareous grasslandEngland/AERRecently harrowed chalk soil, March1.018χcp, pH = 8.4, high RcSutton et al., 1993a
Calcareous grasslandGermany/AERSemi-natural, late summer0.2–1.0 (max: 3.0)19χcpSpindler et al., 2001
Litter meadowSwitzerland/AERTall/short vegetation, spring–summer2 / 3.9 χcp, near intensive agricultureHesterberg et al., 1996
Arrhenatherum elatiusGlasshouse/AEVegetative stage, high N/low N0.04 / <0.16 nmol mol−120 Hanstein et al., 1999
Bromus erectus L. Huds.Glasshouse/AEVegetative stage, high N/low N0.32 / <0.16 nmol mol−120 Hanstein et al., 1999
Bromus erectus L. Huds.Laboratory/DC 0.9420Midrange agricultural fertilizationSutton et al., 2001
Luzula sylvatica L. Huds.Laboratory/AEVarying N treatment0.129 (0.017–0.54)25Infected with rustHill et al., 2001
Luzula sylvatica L. Huds.Laboratory/DCVarying N treatment0.70 (0.51–1.1)25Infected with rustHill et al., 2001
Luzula sylvatica L. Huds.Scotland/AEYoung/old leaves>0–0.4 / >0–0.2520From graphHill et al., 2002
Weed speciesGreenhouse/AEUntreated controls0.9–1.620All four speciesManderscheid et al., 2005
Chenopodium albumGreenhouse/AEPre – 6 days post PPT treatment4.1–7020Referred to 20°CManderscheid et al., 2005
Solanum nigrumGreenhouse/AEPre – 6 days post PPT treatment68–23620Referred to 20°CManderscheid et al., 2005
Tripleurospermum inordorumGreenhouse/AEPre – 6 days post PPT treatment70–10420Referred to 20°CManderscheid et al., 2005
Echinocloa crus-galliGreenhouse/AEPre – 6 days post PPT treatment22–10320Referred to 20°CManderscheid et al., 2005
Alpine tundraUSA/AERAugust–September< 0.02∼12 Rattray and Sievering, 2001
Fertilized Vegetation
Intensively Managed Grassland
Mixed grasslandScotland/MMSummer: grass nearing anthesis3.1 (−0.9–7.3)14 (10–21)Fertilized ungrazed grassSutton et al., 1993b
Mixed grasslandScotland/MMSummer: dry surface runs only4.8 (2–7)15Fertilized ungrazed grassSutton et al., 1993b
GrasslandDevon/MM and DCIntensively managed grassland>50 Downwind of slurry spreadingSutton et al., 1998
Lolium perenneThe Netherlands/AERNighttime/daytime, Ng = 40 g N kg−139.1/33.0 χcp, used for silage productionBussink et al., 1996
Lolium perenneThe Netherlands/AERNighttime/daytime, Ng = 20 g N kg−16.5/8.3 χcp, used for silage productionBussink et al., 1996
Lolium perenneThe Netherlands/AERSpring/summer (0000 to 2400 h)14.41 (13.39–18.34) χcp, intensively managed pastureHarper et al., 1996
Lolium perenneSwitzerland/AEFrequently cut grass/clover swards0.5–2.5 Summer higher than spring and fallHerrmann et al., 2001
Lolium perenneUK/WTUncut sward2.312.3χcp, swards with artificial urineRoss and Jarvis, 2001a
Lolium perenneUK/WTCut sward1.011.1χcp, large variation in dataRoss and Jarvis, 2001a
Lolium perenneUK/WT, AE 0.5–1.9 χcpRoss and Jarvis, 2001b
Lolium perenneLaboratory/DCIntensively managed grassland1.8620midrange agricultural fertilizationSutton et al., 2001
Lolium perenneScotland/AEIntensively managed grassland0.02–10 (max:14)20The max was after 2nd fertilizationLoubet et al., 2002
Lolium perenneThe Netherlands/AEIntensively managed grassland1.5–2.0 (0.5–4.0)0–25Full year (Jan–Nov)Van Hove et al., 2002
Lolium perenneLaboratory/AENewly excised detached leaves0.3–0.7 nmol mol−120(300 and 600 μmol m−2 s−1)Mattsson and Schjoerring, 2003
Lolium perenneLaboratory/AEAfter 4 days of senescence in darkness6–8 nmol mol−1 High N (300 μmol m−2 s−1)Mattsson and Schjoerring, 2003
Lolium perenneThe Netherlands/AERNonstable dry daytime conditions7.0 (0.5–29.7)7–29χcp, Nonfertilized (and χst)Wichink Kruit et al., 2007
Lolium perenneGermany/AEMay–June1–25 nmol mol−1 Pre- to postfertilizationMattsson et al., 2009b
GrasslandGermany/AEVariety of grass species0.20–6.57 nmol mol−1  Mattsson et al., 2009a
Lolium perenneLaboratory/DCGreen leaves0.1–0.4  David et al., 2009
Lolium perenneLaboratory/DCLitter/dried litter1000/160∼20χgDavid et al., 2009
Lolium perenneLaboratory/DCSoil just after cutting320 χgDavid et al., 2009
GrasslandSwitzerland/AERDry daytime conditions1.31516 month studyFlechard et al., 2009
Hordeum vulgare
Hordeum vulgareScotland/AERImm. before ear emergence, summer2.3 (0.6–7.4)13 (4–21)Fertilized cereal cropsSutton et al., 1993b
Hordeum vulgare LaevigatumLaboratory/DCAnthesis2.420 Husted and Schjoerring, 1995b
Hordeum vulgare GolfLaboratory/DC 2.820More active N metabolismHusted and Schjoerring, 1995b
Hordeum vulgare GolfLaboratory/DCEarly – generative growth3.0–6.4 nmol mol−120Related to N-conc. in shootsHusted et al., 1996
Hordeum vulgare LaevigatumLaboratory/DCEarly – generative growth< 0.9–5.3 nmol mol−120 Husted et al., 1996
Hordeum vulgare cv Wild-type and GSLaboratory/DCOld, senescent, and young leaves Intact shoots5.02–11.78 nmol mol−125(Referenced to 25°C from 10°C) Highest in 66% GS mutantMattsson et al., 1997
Hordeum vulgare cv Wild-type and GSLaboratory/AESingle, young leaves only Nonsenescent leaves0.75–7.72 nmol mol−125Highest in 66% GS mutantMattsson et al., 1997
Brassica napus
Brassica napusGreenhouse/AELate vegetative growth stages3.7–5.9 nmo1 mol−1 Ionic strength = 0–25 mM nmol mol−1Husted and Schjoerring, 1995a
Brassica napusGreenhouse/AENH3-free air0.91, 2.70, 6.31 nmol mol−1250.15, 0.3, 0.50 mol NHusted and Schjoerring, 1996
Brassica napusGreenhouse/AE15 nmol NH3 mol−1 Air0.46–14.20 nmol mol−125Varying vegetative stages, 2N–7NHusted and Schjoerring, 1996
Brassica napusGreenhouse/Measured 0.44–7.7 nmol mol−1250.15–0.5 mol NHusted and Schjoerring, 1996
Brassica napusGreenhouse/AESoil/sand mixture0.46–5.58 nmol mol−1200.15–0.5 mol NMattsson et al., 1998
Brassica napusGreenhouse/Measured 0.44–4.10 nmol mol−1 0.15–0.5 mol NMattsson et al., 1998
Brassica napusGreenhouse/AELow, med, high N0.46, 1.89, 5.59 nmol mol−1 Calculated from ΓstSchjoerring et al., 1998
Brassica napusGreenhouse/MeasuredLow, med, high N0.44, 1.72, 4.10 nmol mol−125 Schjoerring et al., 1998
Brassica napusScotland/AEMaturation stage<0.25–1.35 nmol mol−125Corrected for temperature effectsHusted et al., 2000
Brassica napusScotland/AELive leaves0.41–4.3010 – 30 Sutton et al., 2000a
Brassica napusScotland/AERPre-cutting–post-cutting1.3–6.8 χcpSutton et al., 2000b
Brassica napusGreenhouse/DCVarious N treatments, vegetative stage0.8–12.220Normalized to 20°CMassad et al., 2009
Additional Crops
Zea maysLaboratory/DC 1.5–4.225–26 Farquhar et al., 1980
Zea mays L. PioneerUSA/MMIrrigated, all day (0000–2400 h)6.9 (3.0–8.5) Tasseling to silking stagesHarper and Sharpe, 1995
Zea maysUSA/MMFully developed canopy6 χcpMeyers et al., 2006
Zea mays L. PioneerUSA/MBRSummer, developed canopy2.31 (0.95–7.43)30Little leaf litterBash et al., 2010
Phaseolus vulgarisLaboratory/DC 1.7–3.726–33Fertilized with nutrient solutionFarquhar et al., 1980
Amaranth edulisLaboratory/DC 2.132 Farquhar et al., 1980
Eucalyptus paucifloraLaboratory/DC 3.232 Farquhar et al., 1980
Triticum aestivumLaboratory/DCPresenescent period> 15  Parton et al., 1988
Spring wheatLaboratory/DCEarly grain filling—end season13–2525 Morgan and Parton, 1989
Wheat canopyScotland/MM 215 Sutton et al., 1995a
Wheat cropEngland/MM and DCMineral-fertilized (Data not for a specific method)3–4 (max: 9) χcp, peak a few days after fertilizationYamulki and Harrison, 1996
Trifolum repensGreenhouse/AETreated with nitrates∼0.17–0.92 nmol mol−125(from graph)Herrmann et al., 2002
Glycine max L. Merr.USA/MMDaytime< 2  Harper et al., 1989
Glycine max. L. Merr.USA/MBRExtremely dry conditions7.4 χcpWalker et al., 2006
Glycine max. L. Merr.USA/MBRElevated atm. NH311.5  Walker et al., 2006
Glycine max. L. Merr.AEDry daytime periods8.3–13.826.5–33.3 Walker et al., 2006
Winter wheat-riceChina/DCJointing stage31–3810–17 Fang and Zhang, 2006
Winter wheat-riceChina/DCTassel stage19–2615–32 Fang and Zhang, 2006
Medicago sativaUSA/AERSummer1.4  Dabney and Bouldin, 1990
Table 3. Available Measurements and Modeled Values of Stomatal Emission Potentials for NH3 Over Different Types Of Vegetation
Coniferous 310 Gravenhorst and Breiding, 1990a
Pinus sylvestrisLow N155 Langford and Fehsenfeld, 1992
Pinus sylvestrisSuburban, high N, spring3,300Calculated from measured χstNeirynck and Ceulemans, 2008
Pinus sylvestrisSummer/fall1,375 Neirynck and Ceulemans, 2008
Pseudotsuga menziesiiN-polluted forest3,470pH = 6.8, [NH4+] = 550 μMSutton et al., 1995b
Pseudotsuga menziesiiN-polluted forest8,500pH = 7, [NH4+] = 850 μM (fitted data)Wyers and Erisman, 1998
Picea abies Karst 290Assumed pH = 6.8, [NH4+] = 46 μMAndersen et al., 1999
Picea abiesSummer0.8 (0.505–1.268)AE: pH = 4, [NH4+] = 80 μMKöstner et al., 2008
Semi-Natural Vegetation
MoorlandAll conditions132.1Best fit of all dataFlechard and Fowler, 1998
MoorlandSummer, dry, green180Best-fit, (RH<70%)Flechard et al., 1999
Calluna vulgarisHigh N1,200Estimated from measurements (RH < 50%)Nemitz et al., 2004
Calluna vulgaris ∼500Laboratory measurementsSchjoerring et al., 1998
Calluna vulgaris 50–130 Milford et al., 2001
Calcareous grasslandLate summer150–1,000MM measurementsSpindler et al., 2001
Semi-natural grasslandVarying N138–310AE: pH = 5.7–6.3, [NH4+] = 136–297 μMHanstein et al., 1999
Semi-natural grasslandHorses in fall, geese in winter6,000AE: pH = 6, [NH4+] = 6,000 μMMosquera et al., 2001
Semi-natural grasslandAlpine lake290 Tarnay et al., 2001
Semi-natural grasslandSpring7,240 (100–18,000)ModelingHorvath et al., 2005
Semi-natural grasslandSummer200 (100–400)ModelingHorvath et al., 2005
Semi-natural grasslandFall5,186 (100–12,000)ModelingHorvath et al., 2005
Semi-natural grasslandWinter1,733 (100–5,000)ModelingHorvath et al., 2005
Bromus erectusGreen leaves257Steady-state, pH = 5.2, [NH4+] = 1,620 μMHanstein and Felle, 1999
Bromus erectusFlushed with 281 nmol NH3 mol−1 air237Laboratory measurements, linear regressionSutton et al., 2001
Luzula sylvatica 3,150ModelingAsman et al., 1998
Luzula sylvaticaAE method18.7 (2.5–78.4)N treatmentsHill et al., 2001
Luzula sylvaticaDC method101.1 (73.7–159)N treatmentsHill et al., 2001
Luzula sylvaticaYoung leaves136.5AE: pH = 6.63, [NH4+] = 32 μMHill et al., 2002
Luzula sylvaticaOld leaves6.7AE: pH = 5.41, [NH4+] = 26 μMHill et al., 2002
Weed speciesControl weeds230–510AE, 1 day before–6 days afterManderscheid et al., 2005
Weed speciesPPT-treated0–100,000AE, 1 day before–6 days afterManderscheid et al., 2005
Tropical pasture 100–200Modeling, from measurements over grassTrebs et al., 2006
Alpine tundra 16DerivedRattray and Sievering, 2001
Fertilized Vegetation
Intensively Managed Grassland
Trifolium repens 20–120 Herrmann et al., 2002
GrasslandLong grass3,150 Sutton et al., 1997
GrasslandShort grass—fertilized12,620 Sutton et al., 1997
GrasslandOutside fertilization events, dairy cattle4,900pH = 6.8, [NH4+] = 775 μMPlantaz, 1998
Lolium perenneWhole year3,800Assumed value for current national deposition modelsSmith et al., 2000b
GrasslandCut grassland50–100AEHerrmann et al., 2001
Lolium perenne 477Laboratory measurements, linear regressionSutton et al., 2001
Lolium perennePre-cut630Modeling, approximate value, good agreementMilford et al., 2001
Lolium perennePost-cut6,000Modeling, approximate value, good agreementMilford et al., 2001
Lolium perennePostfertilization40,000Modeling, approximate value, good agreementMilford et al., 2001
GrasslandBefore–after fertilization>0–∼40,000Modeled values, from graphRiedo et al., 2002
Lolium perenneBefore 2nd fertilization6–70AELoubet et al., 2002
Lolium perenneDay after 2nd fertilization2,000+AELoubet et al., 2002
Lolium perenneBelow 12°C1,156AE, average value, close to best-fit valuevan Hove et al., 2002
Lolium perenneAbove 12°C588AE, average value, close to best-fit valuevan Hove et al., 2002
Lolium perenneAttached leaves20–300AE, 4 growth stages, highest Γ in green leavesMattsson and Schjoerring, 2003
Lolium perenneExcised leaves, high N<1,000AE, after senescence in darknessMattsson and Schjoerring, 2003
GrasslandPre-cut, winter630 Milford, 2004
GrasslandPost-cut, postfertilization28,000 Milford, 2004
GrasslandGrazing4,000 Milford, 2004
Lolium perenneHigh atm. N2,200Derived from MMWichink Kruit et al., 2007
Lolium perenneGreen leaves, pre-cut period305Modeling, from Mattsson et al., 2009bBurkhardt et al., 2009
GrasslandExcised green leaves of cut plants>23,000 David et al., 2009
GrasslandGreen leaves1,300–2,600 (∼50–2,600)No immediate effect from cutting, pH = 6.0David et al., 2009
GrasslandInt. managed, dry daytime620 (150–4,000)MMFlechard et al., 2009
GrasslandUnfertilized, dry daytime585 (220–1,400)MMFlechard et al., 2009
Mixed grasslandPrefertilization< 400Young leaves had highest ΓHerrmann et al., 2009
Mixed grasslandPostfertilization<400–∼1,600Data from graphHerrmann et al., 2009
GrasslandIncrease after fertilization100–600Modeling, based on measurementsPersonne et al., 2009
Grassland8 grass species10–750AEMattsson et al., 2009a
Lolium perenne 10–150AEMattsson et al., 2009b
Lolium perenneAfter fertilization1,000AEMattsson et al., 2009b
Lolium perenneGreen leaves3,827AE: pH = 6.33, [NH4+] = 1,790 μMMattsson et al., 2009b
Lolium perenneStems2,696AE: pH = 6.37, [NH4+] = 1150 μMMattsson et al., 2009b
Agricultural Crops
Brassica napusHigh N, late vegetative growth stage, young leaves820AE: pH = 5.8, [NH4+] = 1,300 μMHusted and Schjoerring, 1995a
Brassica napusHigh N, late vegetative growth stage, old leaves669AE: pH = 5.8, [NH4+] = μMHusted and Schjoerring, 1995a
Brassica napusVegetative growth stage, 3 N levels55–672AEHusted and Schjoerring, 1996
Brassica napusAnthesis growth stage, 3 N levels325–502AEHusted and Schjoerring, 1996
Brassica napusSenescense growth stage, 3 N levels381–1,712AEHusted and Schjoerring, 1996
Brassica napusLow, med, high N55, 227, 672AESchjoerring et al., 1998
Brassica napus 100, 30,000[NH4+]steady-state = 1,000 μM, 300 μM, pH = 5.0, 8.0Nielsen and Schjoerring, 1998
Brassica napusMaturation stage, lower attached leaves136–190AEHusted et al., 2000
Brassica napusMaturation stage, Middle attached leaves400–503AEHusted et al., 2000
Brassica napus 350Modeling based on measurements by Husted et al., 2000Sutton et al., 2000a
Brassica napus 1,200Modeling—assumed, better fitNemitz et al., 2000
Brassica napus 1,200Modelling – estimated from extractionsNemitz et al., 2001b
Brassica napusDark and light conditions, various N treatments0–8,000AE and DCMassad et al., 2009
Glycine maxT < 28.51,013Calculated from χst and T (range = 798–1,054)Walker et al., 2006
Glycine max28.5 < T < 32.31,054Calculated from χst and TWalker et al., 2006
Glycine maxT > 32.3798Calculated from χst and TWalker et al., 2006
Glycine maxFertilized927From Walker et al., 2006Wu et al., 2009
Glycine maxFertilized549–1,568Modeled valuesWu et al., 2009
Wheat 631Assumed pH = 6.8 [NH4+] = 100 μMSutton et al., 1995a
WheatDew wetted294ModelingSutton et al., 1995a
Wheat 1,200Modeling (from measurements, pH = 6.3, [NH4+] = 600 μM)Sutton et al., 1995a
WheatDownwind of slurry spreading631ModelingSutton et al., 1998
Zea mays 3,000Modeling, based on maize field managementLoubet et al., 2006
Zea maysSummer221 (40–429)Mean of AE measurementsBash et al., 2010
Phaseolus vulgaris 290pH = 6.8, [NH4+] = 46 μMFarquhar et al., 1980
Hordeum vulgare, wild and GS mutantsNonsenescent leaves75–782pH = 5.32–5.86, [NH4+] = 360–1,080 μMMattsson et al., 1997
Hordeum vulgareVarying N levels199–5,233Apoplastic extractionMattsson et al., 1998
Table 4. Available Measurements and Modeled Values of the Emission Potentials for Bulk Litter and Soil for NH3 Over Different Types of Vegetation
Forest SoilSoil20Measured, next to swine production facilityWalker et al., 2008
Semi-Natural Vegetation
Weed speciesPPT-treated bulk tissue∼0–1,000,000 Manderscheid et al., 2005
Fertilized Vegetation
Lolium perenneBulk tissue attached500–9004 growth stages, highest Γ in partly yellow leavesMattsson and Schjoerring, 2003
Lolium perenneBulk tissue excised, high-N60,0004 days after senescence in darknessMattsson and Schjoerring, 2003
Lolium perenneSenescent leaves in the post-cut period5,193Modeling, from Mattsson et al., 2008bBurkhardt et al., 2009
Lolium perenneAfter cutting200,000Modeling, from Mattsson et al., 2008bBurkhardt et al., 2009
GrasslandLong grass litter∼140,000 (6,000–410,000)Γlitter = 5–8× Γsoil and 100× Γplant, pH = 6.4David et al., 2009
GrasslandShort grass litter∼260,000pH = 7.0David et al., 2009
GrasslandHay—wet litter400,000pH = 6.4David et al., 2009
GrasslandBare soil with excised shoots100,000 David et al., 2009
GrasslandSoil—cut grass85,000 David et al., 2009
GrasslandSoil—uncut grass60,000 David et al., 2009
GrasslandBare soil with litter50,000, 75,000 David et al., 2009
Mixed grasslandLitter5,000Not a direct measurementHerrmann et al., 2009
GrasslandSenescent leaves, litter173,000pH = 7.03, [NH4+] = 16,200 μMMattsson et al., 2009b
GrasslandPostfertilization—10 days later40,000–300,000Modeling, based on measurementsPersonne et al., 2009
Brassica napusDecaying leaves630–14,200Daytime measurementsHusted et al., 2000
Brassica napusDecomposing litter13,000Modeling, from Nemitz et al., 2000Sutton et al., 2000a
Brassica napusLitter leaves3,000, 6,000, 13,000ModelingNemitz et al., 2000
Brassica napusSenescing leaves on soil13,000Modeling, estimated from extractionsNemitz et al., 2001b
Soybean, winter wheat, cornRegular crop soil1,514Next to swine production facility, crop rotationWalker et al., 2008
Soybean, winter wheat, cornCrop soil sprayed with swine waste8,935Next to swine production facility, crop rotationWalker et al., 2008
WheatSoil900–2,000 Neftel et al., 1998
Wheat stubbleSoil630FitNemitz et al., 2001b
Winter wheatSoil24,651 Fang and Zhang, 2006
Winter wheat 4,200MeasuredFang and Zhang, 2006

2.2.1. Semi-Natural Vegetation

[8] Semi-natural vegetation such as forests are generally not exposed to large amounts of N and therefore tend to be a sink for NH3 rather than a source, with χst values close to zero [Sutton et al., 1992, 1993a, 1995b; Flechard and Fowler, 1998; Fowler et al., 1998; Husted et al., 2000]. Forests and other semi-natural vegetation that are in close proximity to agricultural fields or industrial sites will have higher N status due to the emissions and subsequent deposition of this additional N. Semi-natural vegetation with higher N supplies have been found to have higher compensation points. Examples of this include observations over forest canopies [Sutton et al., 1995b; Geβler et al., 2000, 2002], moorlands upwind from agricultural sources [Flechard and Fowler, 1998], and the results from the application of an herbicide to four weed species [Manderscheid et al., 2005]. In addition, stomatal emission potentials (Γst) are larger over high-N canopies, where the emission potential is defined as the ratio of the concentration of NH4+ to H+ in the apoplast (in mol L−1/mol L−1) [Nemitz et al., 2000].

[9] The dependence of the compensation point on the temperature means that diurnal and seasonal variations are expected, with the compensation point increasing with increasing temperature. Observations have shown higher χcp values over moorland in the summer than in the winter [Flechard and Fowler, 1998; Fowler et al., 1998], as well as a smaller than expected χst, related to the cooler temperatures (∼12°C), over alpine tundra [Rattray and Sievering, 2001]. Diurnal variations have been observed over semi-natural vegetation, with the peak in the compensation point concentration occurring in the middle to late afternoon [Nemitz et al., 2004]. Limited data over forest canopies prevents any discussion on the diurnal variation over this type of canopy. Despite this temperature-dependence, the compensation point also changes over the seasons due to the seasonal variation in Γst. While independent of temperature, Γst is dependent on [NH4+] and the apoplastic pH, both of which change with season [Loubet et al., 2002]. An additional dependence is on the atmospheric turbulence, where the compensation point has been observed to decrease with decreasing wind velocity (u) [Biswas et al., 2005].

[10] Canopy compensation points of moorland sites in Scotland have been measured using MM methods with lower values in cooler, wet conditions and higher values in warmer, drier conditions, due to the decrease in the cuticular resistance with increasing canopy wetness [Flechard and Fowler, 1998; Fowler et al., 1998]. In the Netherlands, χcp values over heathland, using the AER method, were measured during the day and night for both wet and dry conditions [Nemitz et al., 2004]. Similar to the studies over moorland, the dry conditions produced the highest average value.

[11] The growth stage of the plant is an additional factor affecting the air-surface exchange of NH3. χst values of young Calluna vulgaris plants have been measured to be smaller for shoots that had recently begun growing than for older shoots with developed flowers [Schjoerring et al., 1998]. Interestingly, the opposite has also been observed. Luzula sylvatica (Greater woodrush) has been measured by Hill et al. [2002] for young and old leaves using apoplast extraction. χst values were larger for the younger leaves due to the higher pH in the apoplast of the younger leaves. They observed no age-related changes in the apoplastic [NH4+].

2.2.2. Fertilized Vegetation Intensively Managed Grassland

[12] Measured compensation points of fertilized vegetation are higher than those for nonfertilized canopies. It was proposed by Hill et al. [2001] that this could be due to the ability of agricultural crops to adapt to increased supplies of N. The values of Γst for intensively managed grasslands have been observed to vary significantly between species [Mattsson et al., 2009a] and degree of management [Loubet et al., 2002; van Hove et al., 2002; Herrmann et al., 2009; Mattsson et al., 2009a]. [NH4+] levels have been observed to change in relation to the temperature, the growth of the vegetation, and fertilization practices [Loubet et al., 2002; Riedo et al., 2002; van Hove et al., 2002]. In the field, compensation points over this type of agricultural crop using the AER method [Bussink et al., 1996; Harper et al., 1996; Wichink Kruit et al., 2007] have been higher than those determined using either AE [Herrmann et al., 2001; Loubet et al., 2002; van Hove et al., 2002; Mattsson and Schjoerring, 2003] or wind tunnels [Ross and Jarvis, 2001a, 2001b]. High χcp values have been observed over ryegrass fields high in N or downwind from N sources [Bussink et al., 1996; Harper et al., 1996; Sutton et al., 1998; Wichink Kruit et al., 2007]. The effects of management practices have recently been investigated. The calculated χst values of Mattsson et al. [2009b] did not increase with cuttings but did increase postfertilization. They observed significant increases in the apoplastic [NH4+] levels, and therefore increases in Γst, immediately after fertilization, with a return to prefertilization values within a number of days, demonstrating the short-term effect of fertilization practices. A trend of higher compensation points in the spring and summer, consistent with observations over semi-natural vegetation, have been observed [Flechard and Fowler, 1998; Herrmann et al., 2001; Loubet et al., 2002; van Hove et al., 2002]. Diurnal variations have also been observed over this type of vegetation by some [Harper et al., 1996] but not others [Bussink et al., 1996a]. Agricultural Crops

[13] Stomatal compensation points for Hordeum vulgare (barley) have been found to differ with developmental stage, being larger toward the end of the growing season compared with during the early vegetative growth [Sutton et al., 1993b; Husted et al., 1996], similar to the trend observed over Calluna vulgaris [Schjoerring et al., 1998] and by Husted and Schjoerring [1996] over Brassica napus. In addition, χst has also been found to vary depending on the type of barley. For example, the cultivar Laevigatum has a smaller stomatal compensation point at anthesis than the high-yielding cultivar Golf, due to higher N concentrations in its shoots [Husted and Schjoerring, 1995b; Husted et al., 1996]. In a study of a wild-type of Maris Mink and glutamine synthetase (GS)-reduced mutants of this species, higher compensation points were observed for the mutant varieties than the wild-type variety, and for the shoots containing both young and old leaves than those of young leaves only [Mattsson et al., 1997]. These results of higher NH3 emission correlated to decreased GS are similar to those of Manderscheid et al. [2005] over the PPT-treated weed species.

[14] Detailed studies on Brassica napus (oilseed rape) have shown that rapeseed exposed to atmospheric concentrations of NH3 had higher stomatal compensation points than those exposed to NH3-free air [Husted and Schjoerring, 1996]. As mentioned above, plants in the later stages of growth (e.g., senescence) were observed to have higher compensation points [Husted and Schjoerring, 1996]. Also, plants supplied with higher amounts of N; at higher atmospheric temperatures; and after cuttings, had higher compensation points [Husted and Schjoerring, 1996; Sutton et al., 2000a, 2000b]. Diurnal variations have been measured over crops, with peaks in the middle to late afternoon [Husted et al., 2000; Walker et al., 2006].

[15] Measured χst values over winter-wheat rice and other wheat crops are significantly larger than measured values over semi-natural vegetation [Farquhar et al., 1980; Parton et al., 1988; Morgan and Parton, 1989; Harper and Sharpe, 1995; Yamulki and Harrison, 1996; Fang and Zhang, 2006; Walker et al., 2006]. Francis et al. [1997] did not observe any significant diurnal trend in the NH3 exchange of irrigated corn plants. However since no compensation point was measured in this study, it is not known whether a diurnal dependence in the stomatal compensation point existed or not. The measurements over the corn, soybean, and wheat crops all produced diurnal variations in the compensation point, with higher values in the day when the temperature was the highest [Harper and Sharpe, 1995; Yamulki and Harrison, 1996; Walker et al., 2006].

2.2.3. Ground Surfaces

[16] Recent measurements of the ground surface have shown that the ground surface plays a significant role in the emission/deposition of atmospheric NH3, with measured ground emission potentials (Γg) from the decomposing leaf litter being considerably higher than the stomata [Neftel et al., 1998; Nemitz et al., 2001b; Walker et al., 2008; David et al., 2009; Herrmann et al., 2009; Mattsson et al., 2009b]. Factors that affect the χg include the type of ground surface, where Γg has been observed to be higher over soil and litter than grass (∼100 times) [David et al., 2009]; the N levels, where Γg increases with increasing N; the degree of decomposition of the litter, with Γg increasing as the litter decomposes; as well as the length of the litter, where Γg has been observed to be larger for shorter canopies (i.e., higher pH and the [NH4+] in short vs. long grass cuttings) [Walker et al., 2008; David et al., 2009; Herrmann et al., 2009; Mattsson et al., 2009b; Personne et al., 2009; Sutton et al., 2009b]. In addition to cutting practices, fertilization practices have also been found to increase the emission potential of the ground [Walker et al., 2008]. Although, to our knowledge, there are no measurements of deposition of NH3 over desert soils, an increase in the emission flux over the Mojave Desert has been observed to occur with rainfall [McCalley and Sparks, 2008]. Perhaps measurements of NH3 dry deposition over this land type in the future might provide an additional data set.

2.2.4. Summary of Measurements

[17] To make use of the above discussion for modeling development and evaluation purposes (sections 3 and 4), major findings of the four key parameters (χst, χcp, Γst, and Γg) are summarized here. The major factors affecting χst and χcp are the canopy type, the level of N, the growth stage of the plant, and the meteorological conditions. χst and χcp are generally larger over agricultural crops and fertilized vegetation with bulk tissue litter than semi-natural vegetation, with a maximum measured χcp value of 29.7 μg m−3 over cut perennial ryegrass close to fertilized fields [Winchink Kruit et al., 2007], a maximum litter compensation point of 1,000 μg m−3 for perennial ryegrass [David et al., 2009], and a minimum χcp value of ∼0.01 μg m−3 over moorlands [Sutton et al., 1992]. Typical compensation points are generally less than 2 μg m−3 over semi-natural vegetation and in the double-digits over fertilized vegetation. The compensation point increases with increased N supply from fertilization and proximity to industrial sources. The highest χst value measured was 236 μg m−3 for Solanum nigrum, a weed species that had an observed peak in the stomatal compensation point six days after PPT treatment. The stages of vegetative growth have also been observed to affect the stomatal compensation point, with χst values having been observed to decrease from tillering to anthesis and then increase postanthesis to senescence [Husted et al., 1996]. Meteorological conditions also affect the compensation point, with values higher in warm conditions than cold conditions and in dry climates than wet climates. For example, as discussed in section 2.2.1, the observed compensation point changed dramatically from pre- to postmonsoon [Biswas et al., 2005]. Nemitz et al. [2004] observed an increased χcp from 0.59 to 4.9 μg m−3 in the canopy compensation point of Calluna vulgaris between wet and dry conditions.

[18] Similar to the compensation point, the Γst and Γg have also been observed to vary depending on the factors discussed above. The soil and bulk litter have significantly larger emission potentials over fertilized vegetation and agricultural crops than the stomatal emission potentials. For example, the median ιg value over croplands is 5,100, whereas the median Γst value over the same vegetation is only ∼630. This demonstrates the importance of the ground surface in the air-surface exchange of NH3 at the canopy scale over fertilized vegetation. On the other hand, the stomata appear to have a dominant role in the canopy exchange over forests, where the median Γst value is 300 while the median (and only) Γg value is 20. However, due to the lack of data reported for forest canopy floors, there is potential that forest litter leaves play a more significant role at the canopy scale. As discussed in section 3.2, the measured and modeled values of the emission potentials reviewed in this paper (see Tables 3 and 4) have been used to derive the input parameters for our modeling study, shown in Table 5. In addition to the range of Γst and Γg values selected as our input parameters, we have also given the minimum, maximum, mean, and median range of published measured and modeled Γst and Γg values over the various land-use categories, which provided the basis for the development of our parameters. More detailed discussion on the selection of the input parameters for the different LUCs is included in section 3.2.

Table 5. Empirical Input Stomatal and Ground Emission Potentials for NH3 for Different LUCs to be Used in the Bi-directional NH3 Air-Surface Exchange Modela
LUC ΓstbΓgc
InputMean, Median, Min, MaxInputMean, Median, Min, Max
  • a

    These values are based on the measured and modeled values from Tables 3 and 4, along with the mean, median, and minimum and maximum values from these tables.

  • b

    If LAI < 0.5, Γst = 0.

  • c

    If there is snow cover, Γg = 0.

  • d

    The min is for low-N canopies, and the max value is for high-N canopies.

3Inland lake0000
4Evergreen needleleaf trees300–3,000d1,740, 300, 0.51, 8,50020–1,000d20
5Evergreen broadleaf trees300–3,0001,740, 300, 0.51, 8,50020–1,00020
6Deciduous needleleaf trees300–1,740, 300, 0.51, 8,500200–2,00020
7Deciduous broadleaf trees300–3,0001,740, 300, 0.51, 8,500200–2,00020
8Tropical broadleaf trees300–3,0001,740, 300, 0.51, 8,50020–1,00020
9Drought deciduous trees300–3,0001,740, 300, 0.51, 8,500500–2,00020
10Evergreen broadleaf shrubs300–3,0001,740, 300, 0.51, 8,50020–1,00020
11Deciduous shrubs300–3,0001,740, 300, 0.51, 8,500200–1,00020
12Thorn shrubs300–3,0001,740, 300, 0.51, 8,50020–1,00020
13Short grass and forbs300–3,0001,740, 300, 0.51, 8,5002,000–200,000162,035, 750,000, 0, 1,000,000
14Long grass300–3,0001,740, 300, 0.51, 8,5002,000–100,000154,000, 100,000, 6,000, 410,000
15Crops8001,714, 631, 0, 30,0005,0007,547, 5,100, 630, 24,651
16Rice8001,714, 631, 0, 30,0005,0007,547, 5,100, 630, 24,651
17Sugar8001,714, 631, 0, 30,0005,0007,547, 5,100, 630, 24,651
18Maize8001,156, 429, 40, 3,0005,0007,547, 5,100, 630, 24,651
19Cotton8001,714, 631, 0, 30,0005,0007,547, 5,100, 630, 24,651
20Irrigated crops8001,714, 631, 0, 30,0003,0007,547, 5,100, 630, 24,651
23Swamp100164, 180, 132, 18020
25Mixed wood forest300–3,0001,740, 300, 0.51, 8,50020–3,00020
26Transitional forest300–3,0001,740, 300, 0.51, 8,50020–3,00020

3. Model Description

3.1. Theory

[19] The model proposed here is similar in concept to existing bi-directional air-surface exchange models developed for NH3 [Sutton et al., 1998; Flechard et al., 1999; Nemitz et al., 2000, 2001b; Spindler et al., 2001; Riedo et al., 2002; Walker et al., 2008; Wu et al., 2009]; also see reviews in Massad et al. [2010a]. In particular, the model is very similar to the so-called two-layer model of Nemitz et al. [2001b]; that is, the flux exchange over leaf stomata and over soil is bi-directional. The schematic view of the model's pathways is shown in Figure 1, which is a modification from the big-leaf model of Zhang et al. [2003].

Figure 1.

Scheme of the bi-directional exchange model. Ft is overall flux at a reference height above the canopy, and Fst and Fg are bi-directional fluxes through stomata and above soil surface, respectively. χa is the ambient concentration at the reference height; χc is the concentration at the top of canopy (the canopy compensation point χcp is the specific value of χc at which the fluxes of Ft change directions); χst and χg are stomatal and soil compensation points, respectively. Resistance terms include Ra, aerodynamic; Rb, quasi-laminar; Rac, in-canopy aerodynamic; Rg, soil; Rcut, cuticle; and Rst, stomatal resistance.

[20] The overall flux (μg m−2 s−1) at a reference height above the canopy can be calculated as

equation image

where χa and χc are the ambient concentrations (μg m−3) at the reference height and at the canopy top, respectively. χc can be calculated according to

equation image

Apparently, when χa is higher than χc, the flux will be downward (negative value of Ft), and when χa is lower than χc, the flux will be upward (positive value of Ft). The specific value of χc at which the fluxes change directions is the so-called canopy compensation point (χcp) commonly observed in the field studies (as discussed in section 2). χcp can be estimated using equation (2) by assuming χa = χc = χcp:

equation image

Note that equation (2a) for the calculation of χcp will not be needed in the model. It is only for the purpose of the model evaluation discussed in section 4.2. It is noted that the formula of χc here is in a much more simplified form than the one in Nemitz et al. [2010b] due to the different position of the resistance term Rb.

[21] Resistance terms include Ra, aerodynamic; Rb, quasi-laminar; Rac, in-canopy aerodynamic; Rg, soil; Rcut, cuticle; and Rst, stomatal resistance (all in s m−1). Formulas for all of the resistance terms can be found in Zhang et al. [2003]. Note that the cuticle resistance term is important in modeling bi-directional NH3 exchange. Various formulas for Rcut can be found in a recent review of Massad et al. [2010a]. In the present study, we choose to use the formula from Zhang et al. [2003] for Rcut (and all other resistance terms) considering that (1) using all formulas from the same model can ensure their internal consistence, and (2) the formula considers the factors of canopy wetness, leaf area, and meteorological conditions (relative humidity, friction velocity), which are all important to cuticle uptake. It is worth noting that Ra, Rb, and Rac decrease with the increase in the friction velocity; Rg decreases with soil wetness; Rcut decreases with the increase in the friction velocity, relative humidity, and canopy wetness; and Rst increases with canopy wetness. This information is needed for the explanation of the model results presented in section 4.

[22] In this modified model, two new parameters are required: the compensation point over the stomata (χst) and the compensation point over the soil (χg). χst is defined chemically as the concentration at which there is both thermodynamic equilibrium between NH3 in the liquid and gas phases (NH3(aq)equation imageNH3(g)) and acid-base equilibrium between NH4+ and NH3 in the liquid phase (NH4+(aq)equation imageNH3(aq) + H+). These equilibriums are given by the Henry's law coefficient (KH = 10−1.76 L mol−1) and the dissociation constant for NH4+ (Kd = 10−9.25 mol L−1), respectively, at 25°C [Asman, 1992; Husted and Schjoerring, 1996; Herrmann et al., 2009; Personne et al., 2009].

[23] χst (mol NH3 mol−1 air) can either be measured directly or be calculated according to

equation image

where Tref is the reference temperature (K), ΔH°dis is the enthalpy of NH4+ dissociation (52.21 kJ mol−1), ΔH°vap is the enthalpy of vaporization (34.18 kJ mol−1), and R is the gas constant (0.00831 kJ K−1 mol−1). With these parameters inserted, χst (converted to micrograms per cubic meter) is given by [Nemitz et al., 2004]:

equation image

where Tst is the temperature (K) of the leaf stomata, and Γst is the stomatal emission potential (also known as the apoplastic ratio) at 1 atmosphere. This ratio is given by [Nemitz et al., 2000]

equation image

where [NH4+]st is the concentration of NH4+ (mol L−1) in the apoplastic fluid, and [H+]st is the stomatal concentration of H+ (mol L−1), defined as [H+]st = 10−pH, where the pH of the intercellular fluid is given at a pressure of 1 atmosphere.

[24] The calculations for χg (micrograms per cubic meter) and Γg are similar to equations (4) and (5) for the stomata [Nemitz et al., 2004]:

equation image
equation image

where Tg is the temperature of the ground surface (K), and [NH4+]g and [H+]g are the concentrations of NH4+ and H+ (moles per liter), respectively, in the ground cover. The ground cover includes bare soil, leaf litter, and snow cover, depending on the canopy and season.

3.2. Input Parameters: Γst and Γg

[25] As mentioned above, the modified model (from dry deposition only to bi-directional exchange) needs two new parameters: χst and χg. However, calculating χst and χg using equations (4)(7) requires information on [NH4+] and the pH within the leaf stomata and in the soil. Although such information can be measured at selected locations, they are not available at regional scales, nor are they calculated in regional-scale air-quality models. Here we have chosen to use empirically derived values for Γst and Γg to generate χst and χg (only equations (4) and (6) are needed), considering that the uncertainties in the empirically derived Γst and Γg should not be larger than the calculated ones using equations (5) and (7) with assumed values of [NH4+] and the pH (as demonstrated by the sensitivity tests in Wu et al. [2009]). As the first attempt of incorporating the bi-directional exchange model into regional-scale air-quality models, a constant Γst and Γg for each land-use category is chosen. Since the main goal of the present bi-directional exchange model is to better simulate the NH3 dry deposition over long periods (e.g., on an annual basis) and over regional scales instead of capturing local/short episodes/periods, the Γst and Γg values chosen here are relatively conservative or from long-time averages (knowing that most of the extremely high values were from human practices such as fertilization and grass cutting). The empirical values of Γst and Γg, as listed in Table 5 for the different land-use categories simulated in our model, have been determined based on the findings from section 2.2.

[26] It is known from section 2.2 that, for forests and grasslands, the same land-use category can have quite different nitrogen contents (due to the receipt of different amounts of dry and wet nitrogen deposition) and, thus, different emission potentials. To reflect this difference, two sets of Γst and Γg values, one for high-N canopies and one for low-N canopies, are chosen for these land-use categories (Table 5). The high- and low-N canopies for each model grid within a model domain can be decided from total N deposition maps produced either from previous model runs or from monitoring networks. As a first estimation, the critical loads (commonly on the order of several tens kg N ha−1 yr−1) might be used to distinguish high- and low-N canopies.

3.2.1. Semi-Natural Vegetation

[27] Forest canopies are semi-natural vegetation and therefore generally have a low NH3 stomatal emission potential [Gravenhorst and Breiding, 1990, cited in Sutton et al., 1993a; Langford and Fehsenfeld, 1992; Köstner et al., 2008]. However, for any forest canopy that receives high atmospheric N input (through dry and wet deposition) and/or is managed intensively, higher Γst values have been measured [Sutton et al., 1995b; Wyers and Erisman, 1998; Neirynck and Ceulemans, 2008]. A range of values is provided in Table 5, with the lowest value (Γst = 300) for forest canopies low in N and the highest value (Γst = 3000) for any N-polluted forest. For these canopies, when the leaf is gone or is covered in snow, Γst is assumed to be zero (i.e., LAI < 0.5). The values of Γg used for low-N forests are 20 for evergreens, 200 for deciduous forests, and 500 for drought deciduous forests. The high-N values are 1,000 for evergreens and shrubs and 2000 for deciduous forests. The conservative value (Γg = 20) is based on the measurement by Walker et al. [2008] of the soil of a forest. The higher Γg values are to account for the N in the leaf litter of high-N forests and the possibility of substantial leaf litter on the ground of the deciduous canopies. Due to a lack of measurements over forest canopies, no distinctions are made in the values of the stomatal and ground emission potentials used for the different seasons, with the exception of the lack of emission from snow-covered surfaces. It should be noted that a significant part of the seasonal variation in the stomatal compensation point is due to the change in the temperature throughout the year. This seasonal temperature-dependence is calculated in the model and will still be reflected in the results, despite the constant emission potential.

[28] For our modeling study, it is assumed that the swamp land-use category (LUC = 23) is similar in nature to the moorland vegetation used in past measurement studies. Measurements over this type of unfertilized vegetation have produced relatively low values of Γst [Flechard and Fowler, 1998; Flechard et al., 1999]. A sensitivity study by Jones et al. [2007] of Γst over moorland vegetation using a range of values (Γst = 0–360) found the dependence to be minor and determined that the value of 180 measured by Flechard et al. [1999] was sufficient for their modeling purposes. We have therefore chosen to use a similar Γst value of 100 for this modeling study. Due to a lack of measurements over the ground surface for this type of canopy, the conservative value used for the low-N forest ground (Γg = 20) is also used for this land-use category, except during snow cover when it is equal to zero.

[29] A measurement study by Rattray and Sievering [2001] over an alpine tundra canopy observed very small emissions similar to montane subalpine forests. Due to the low-growth nature of tundra canopies and the lack of existing measurements, the same small value (20) is assumed for both Γst and Γg.

3.2.2. Fertilized Vegetation

[30] The fertilized vegetation canopies include both short and long grasses and agricultural crops. Observations over mixed grassland canopies have shown that shorter grass has a higher stomatal emission potential than longer grass and that this value also increases with increasing N [Sutton et al., 1997; Loubet et al., 2002]. We have therefore chosen a range of Γst and Γg values for the grassland canopies depending on the N status. A Γst value of 3,000 has been chosen for high-N short and long grasses, which is within the range of values of available measurements over intensively managed grasslands [e.g., Sutton et al., 1997; van Hove et al., 2002; Mattsson and Schjoerring, 2003; David et al., 2009; Mattsson et al., 2009b]. Similar to the forest canopies, a lower Γst value of 300 has been chosen for the low-N grassland canopies, which is also within the range of measured Γst values of this type of vegetation [Flechard et al., 2009]. These values have been chosen empirically to be constant throughout the year, regardless of the season, management practices such as cutting and fertilization, and growth stage, due to the already large uncertainty in our estimate and the small number of long-term measurement studies. As with the other canopies, if there is snow cover, this value is assumed to be zero. Recent measurements of [NH4+] and pH by David et al. [2009] for short and long grass litter have provided a base for our high-N grass land-use categories (Γg = 200,000 and 100,000 for high-N short and long grass canopies, respectively). We have chosen Γg to be 2,000 for the low-N grasslands, similar to past lower-end measurements over this type of vegetation [e.g., Mattsson and Schjoerring, 2003].

[31] As discussed in section 2.2.3, there have been a significant number of field studies over oilseed rape, soybean, barley, and wheat canopies. These studies have produced Γst values anywhere between 55 and 30,000, depending on the growth stage and N level, with an average value of ∼630 [e.g., Sutton et al., 1995a; Husted and Schjoerring, 1995a, 1996; Mattsson et al., 1997, 1998; Nielsen and Schjoerring, 1998; Schjoerring et al., 1998; Nemitz et al., 2001b; Loubet et al., 2006; Walker et al., 2006]. Since the available measurements are relatively similar over the different canopy types, a value of Γst = 800 has been chosen for the crops canopy (LUC = 15) and extended to all of the other agricultural canopies (LUCs = 16–20) since there are no available measurements over these canopy types. The value of Γg has been chosen to be 5,000. This value is in the range of values measured over fertilized crop soil [Husted et al., 2000; Nemitz et al., 2001b; Fang and Zhang, 2006; Walker et al., 2008]. As with all of the other canopies, the value in the winter is zero if there is snow.

4. Model Evaluation and Example Output

4.1. Stomatal and Ground Compensation Points

[32] The dependence of the stomatal and ground compensation points on the leaf or soil temperature (degrees Celsius), respectively, is shown in Figure 2 for different stomatal and soil emission potentials selected in the model (see Table 5). These plots were produced using equations (4) and (6), as described in section 3.1. As can be seen in Figure 2, the compensation point increases exponentially with increasing temperature. The modeled χst values shown in Figure 2 agree well with the range of measurements shown in Table 2. For example, the low-N forest is assigned a Γst of 300 (Table 5), which produces a χst of 0.1–3.7 μg m−3, similar to the value obtained over a forest known to have low-N input [e.g., Langford and Fehsenfeld, 1992]. On the other hand, the high-N forest is assigned a Γst of 3,000, which produces a χst of 1–37 μg m−3, which is also within a similar order of magnitude as available measurements [e.g., Wyers and Erisman, 1998; Geβler et al., 2000; Neirynck and Ceulemans, 2008]. The agricultural crop is assigned a Γst of 800, which produces a χst of 0.2–10 μg m−3, very close to the majority of measurements over this LUC.

Figure 2.

Modeled χst and χg (micrograms per cubic meter) as a function of leaf or soil temperature (°C) for different values of the input stomatal or soil emission potentials (Γst/g).

[33] The very small value of 20 for Γst and Γg used as the low end for several of the semi-natural vegetative canopies produces an almost negligible compensation point at the lower temperatures, thus the past observations of low-N semi-natural vegetation acting as almost perfect sinks for NH3. On the other end of the emission potential range, a high emission potential, representative of the ground litter of an intensively managed grassland canopy, produces a high χg value of ∼100 μg m−3 at 0°C. This means that even during the winter, if there is no snow cover, emission could still be the dominant exchange of NH3 for this type of vegetation. At 20°C, a Γst value of 300 produces a stomatal compensation point close to 1 μg m−3, and a Γst value of 3,000 results in a χst value of almost 10 μg m−3. This difference in the stomatal compensation point could mean the difference between whether there is emission or deposition from the canopy during the summer or warm spring/fall season under typical ambient NH3 concentrations (more discussions below). The high values of the compensation points produced at all temperatures for the high emission potentials (Γ = 100,000–200,000) demonstrate the importance of including the ground surface in our bi-directional model.

4.2. Canopy Compensation Point

[34] The canopy compensation points over all land-use categories with nonzero emission potentials are shown in Figure 3 for six sets of conditions: four are overly dry canopies (solid columns in Figure 3), and two are overly wet canopies (dashed columns in Figure 3). Note that, for the dry canopies, the typical summer noon conditions should represent the upper-end values, whereas the nighttime values should represent the lower-end values.

Figure 3.

Modeled canopy compensation points micrograms per cubic meter) for various land-use categories for six sets of conditions. Summer noon dry and wet canopies: T = 25°C, (solar radiation) SR = 800 W m−2; Summer morning dry canopies: T = 10°C, SR = 200 W m−2; Summer night dry and wet canopies: T = 10°C; Winter day dry canopies: T = 2°C, SR = 200 W m−2. Wind speed at 20 m is chosen to be 6 m s−1 for all tests; relative humidity is chosen as 70% for dry canopies and 95% for wet canopies. Wet canopies are controlled by a nonzero precipitation amount.

[35] For the dry canopies, the summer midday χcp values are higher over all LUCs than the summer morning and night and winter day values due to the high temperatures and lower stomatal resistance caused by higher solar radiation. The winter day values are the smallest values of χcp over all of the canopies and are almost negligible over the forest canopies and tundra and swamp vegetation, due to the low temperatures and low soil emission potentials. Despite the lack of a stomatal compensation point in the summer night (the stomata are assumed to be closed at night), the canopy compensation point is still over 2 μg m−3 for the agricultural crops due to the contribution of the soil emissions.

[36] Over the forest canopies under dry conditions, the daytime χcp values range between <0.01 and 2 μg m−3, with the lower values in the winter at 2°C and the higher values at 25°C. These concentrations are similar to the χcp range measured over a coniferous forest in Denmark by Andersen et al. [1999] between 0.2 and 1.9 μg m−3 for temperatures between 2°C and 25°C. A low χcp value of 0.08 μg m−3 at 6°C over a Scottish coniferous forest, in November, is also close to our modeled values [Sutton et al., 1993a]. In a modeling and measurement study, Neirynck and Ceulemans [2008] calculated a χcp value of 5.2 μg m−3, for the month of July, over a high-N pine forest in suburban Belgium. While their value is higher than our summer noon value over low-N forests, it is definitely much lower than our canopy compensation points calculated assuming high-N forest canopies (Figure not presented).

[37] Over the semi-natural swampland vegetation under dry conditions, the χcp values are: 0.5 μg m−3 at summer noon, ∼0.1 μg m−3 in the morning, and ∼0.01 μg m−3 both during the summer nighttime and in the winter daytime. These values are in the range of values measured over summertime semi-natural moorland vegetation in Scotland (0.2–0.57 μg m−3) [Flechard and Fowler, 1998; Fowler et al., 1998] and over winter/frozen moorland canopies in both Scotland and England (0–0.08 μg m−3) [Sutton et al., 1992, 1993a; Fowler et al., 1998].

[38] The χcp values over the low-N grassland canopies under dry conditions ranged from ∼0.5 μg m−3 in the daytime winter, to ∼ 1 μg m−3 in the summer night and morning, to 3 and 5 μg m−3 in the summer daytime for long and short grasses, respectively. Our χcp values of ∼1 μg m−3 in the summer morning at 10°C are close to the canopy compensation point of 1.01 μg m−3 at a similar temperature of 8°C measured by Sutton et al. [1993a] in March over calcareous grassland at Harwell in England. Our daytime values are slightly larger than the spring/summer χcp measurements by Hesterberg et al. [1996] of 2 and 3.9 μg m−3 for long and short meadow grasses, respectively, in the lower Reuss valley in Switzerland. They are also larger than the χcp range of 0.2–1 μg m−3 at 19°C measured by Spindler et al. [2001] in the late summer in Germany over semi-natural calcareous grassland. On the other hand, our values are smaller than the χcp measurements by Bussink et al. [1996] at a perennial ryegrass field in The Netherlands that is used for silage production. They measured daytime χcp values between 8.3 and 33.0 μg m−3 and nighttime χcp values between 6.5 and 39.1 μg m−3 for grass N concentrations between 20 and 40 g kg−1. Our grassland χcp values are also smaller than the measurements of Wichink Kruit et al. [2007] in Wageningen, The Netherlands over nonfertilized perennial ryegrass. Despite the lack of fertilization in the past ten years, the canopy is in close proximity to fields receiving animal manure and mineral fertilizer. They measured dry daytime χcp values between 0.5 and 29.7 μg m−3 at temperatures between 7°C and 29°C.

[39] The modeled χcp values ranged from < 2 μg m−3 in the winter day and summer morning, to ∼2.5 μg m−3 in the summer night, to 8–10 μg m−3 in the summer daytime at 25°C over the agricultural crops under dry conditions. The summer values are in the same range as canopy compensation points observed over a rapeseed crop using the AER method [Sutton et al., 2000b]. During a summer study, average values were 1.3 μg m−3 for the pre–cutting campaign and 6.8 μg m−3 for the post–cutting campaign. Our summer daytime range is also on the same order as the median χcp value of 7.4 μg m−3 measured by Walker et al. [2006] over a soybean crop during a dry daytime period. The effect of the ground surface on the canopy compensation point can be seen by the large summer daytime values and by the relatively large canopy compensation point during the cold winter conditions.

[40] It is noted from Figure 3 that, for dry canopies, the summer night compensation points are slightly larger than the summer morning measurements over the majority of the agricultural lands, but much smaller over forest canopies. This can be explained by different soil emission potentials between forests and agricultural lands and the differences in Rst and Rcut between morning and nighttimes (see equation (2a)).

[41] χcp can be substantially reduced by canopy wetness due to the high solubility of NH3. For example, during a typical summer rain day, χcp is decreased by 50–70% over forests and 30–50% over agricultural lands (compare the red dashed column with the red solid column in Figure 3); and, during a typical summer rain night, χcp is decreased by 70–85% over forests and 50–70% over agricultural lands (compare the blue dashed column with the blue solid column in Figure 3). Canopy wetness increases Rst and decreases Rcut, and both of these will decrease χcp (see equation (2)). On the other hand, soil wetness decreases Rg and thus increases the soil emission, which will increase χcp. The latter also explains the very small changes in χcp between wet and dry canopies for some land-use categories (e.g., short grass, tundra). The results shown in Figure 3 seem to agree qualitatively when compared to the limited measurements for both dry and wet conditions shown in Table 2 [e.g., Nemitz et al., 2004].

4.3. Fluxes

[42] The new bi-directional model calculates net fluxes (Ft) according to equation (1); in contrast, the original model calculates dry deposition fluxes (Fdry) according to

equation image

Note that the treatments of leaf and soil wetness on different resistance components are the same in the new and the original model (not presented here for simplicity in the discussions below). The differences in the fluxes between the new and the original model (ΔF = FtFdry) can then be estimated from equations (1), (2), and (8):

equation image

By comparing equation (9) with equation (2a), it can be seen that ΔF = χcpVd, with Vd being the deposition velocity calculated from the original model.

[43] Figure 4 shows the modeled fluxes (nanograms per square meter per second) from the old dry deposition model (dashed lines) and the new bi-directional model (solid lines) as a function of ambient NH3 concentration (micrograms per cubic meter) under typical summer daytime conditions. Apparently, using the bi-directional exchange model instead of the original dry deposition model decreases the downward fluxes. For low-N canopies (Figure 4, left), the new bi-directional model predicts emissions during typical summer days at ambient NH3 concentrations below ∼1.5 μg m−3 for a forest canopy (LUC = 7), ∼5 μg m−3 for a grassland (LUC = 13), and ∼9 μg m−3 for a cropland (LUC = 15). If using the model parameters for high-N canopies (Figure 4, right), the model predicts emissions at ambient NH3 concentrations below ∼10 μg m−3 for a forest (LUC = 7) and emissions for grasses and croplands even at ambient NH3 concentrations as high as 100 μg m−3. The emissions fluxes at the lower NH3 concentrations are between 20 and 100 ng m−2 s−1 for the low-N canopies and between 100 and 1500 ng m−2 s−1 for the high-N canopies, with the higher fluxes for the grassland and agricultural crops. The emission fluxes for LUC = 15 (crops) produced by the model agrees very well with the daytime fluxes observed over a soybean field [Walker et al., 2006]. The results presented here show that intensively managed grasslands and fertilized agricultural crops are net emitters of NH3. This is an important atmospheric exchange that the old model is not able to reproduce.

Figure 4.

Fluxes (nanograms per square meter per second) as a function of the ambient NH3 concentration (micrograms per cubic meter), calculated using the bi-directional model (solid lines) and the original dry deposition model (dashed lines) under typical summer day dry conditions (the same summer conditions as in Figure 3), using the emission potentials from Table 5 for low-N canopies (left) and high-N canopies (right). These simulations are over four types of dry canopies: deciduous broadleaf trees (LUC = 7), short grass and forbs (LUC = 13), long grass (LUC = 14), and agricultural crops (LUC = 15).

[44] According to equation (9), the flux differences between the new and the original model depend on the stomatal and soil compensation points (which are strongly dependent on T) and the resistances (which are strongly dependent on the friction velocity and other meteorological parameters). Figure 5 shows an example of the dependence of ΔF on temperature (T) and friction velocity (u*) under full leaf conditions. An increase of T by 20°C can increase ΔF by one order of magnitude due to the exponential increases of χst and χg, as shown in Figure 2. Note that the increase of ΔF with T is approximately exponential for LUC = 13, but it is not the case for LUC = 4 and LUC = 9. Such a variation among the different LUCs is caused by their different Γstg ratios (a ratio higher than 1 represents that stomatal emission dominates over soil emission, and vice versa). Thus, the increase of ΔF with T deviates from the original exponential increase due to the change in Rst with T if stomatal emission dominates (other resistances do not change with T directly if u* is fixed).

Figure 5.

Modeled flux decreases (ΔF) (nanograms per square meter per second) using the bi-directional exchange model compared to the original dry deposition model under full leaf daytime (SR = 600 W m−2) conditions: (a) as a function of temperature (degrees Celsius) assuming a wind speed of 6 m s−1 at a 20 m height, and (b) as a function of friction velocity (meters per second) assuming a temperature of 20°C. These simulations are over four types of dry canopies: evergreen needleleaf trees (LUC = 4), drought deciduous trees (LUC = 9), short grass and forbs (LUC = 13), and agricultural crops (LUC = 15).

[45] ΔF also increases rapidly with u*, e.g., by a factor of 2–5 within a reasonable range of u* values (0.1–1.0 m s−1). For canopies with a small soil emission potential (e.g., LUC = 4), the increase of ΔF with u* is only important at small u* values, while for canopies with a large soil emission potential (e.g., LUC = 9 and LUC = 15), ΔF increases with u* at all values of u*. This is because Rac decreases with the increase in u* (equations (2) and (9)) thereby increasing the soil emissions in the modified model, which in turn increases ΔF. Other meteorological variables can also affect ΔF through changes in any of the resistance components. For example, an increase in solar radiation may decrease Rst and therefore increase ΔF. The above discussions suggest that the daytime NH3 dry deposition fluxes can be substantially reduced by using the new bi-directional exchange model in replacement of the original dry deposition model in regional-scale air-quality models, especially at locations where the ambient NH3 concentrations are close to the canopy compensation points.

5. Conclusions

[46] The bi-directional air-surface exchange of atmospheric NH3 has been reviewed. Deposition has been observed to occur more often at lower temperatures in wet conditions, while emission dominates the hotter, drier conditions. The canopy compensation point at which the exchange flux changes between deposition and emission is higher over agricultural crops and fertilized grassland than over semi-natural vegetation, such as forests and moorlands. Fertilization processes, past exposure to N in the soil surface, cutting practices, and proximities to local point sources of N all increase these compensation points.

[47] A bi-directional air-surface exchange model, developed for NH3, has been found to produce canopy compensation points, over different canopies and under varying meteorological conditions, comparable to existing measurements. Typical summer daytime χcp values, assuming a low-N status, were less than 2 μg m−3 over the forests and other semi-natural canopies, below 5 μg m−3 over the grasslands, and between 5 and 10 μg m−3 over the agricultural crops. In the winter, these values decreased to almost zero over the forests and to below 3 μg m−3 over the crops. The application of the new bi-directional air-surface exchange model in replacement of the original dry deposition model will reduce the dry deposition fluxes simulated in the regional scale air-quality models, especially during the daytime and for canopies with high-N status. The decreases in the simulated dry deposition fluxes will also be larger at higher temperatures, stronger wind speeds, and drier conditions.

[48] The model should be further evaluated using field measurements and can only be improved if extensive field measurements covering various canopies and under different chemical and meteorological conditions are available. The criterion for defining high- and low-N canopies suggested in this study is very crude at this stage and needs further investigation. Ideally, the emission potentials should be controlled by the accumulated atmospheric N input and thus varies on the time scale of season for semi-natural canopies or even within days for fertilized fields. Several more sophisticated theoretical approaches, although very useful in explaining field data at individual sites where more specific information is known, are not practical for use in regional-scale models [e.g., Wu et al., 2009; Massad et al., 2010b]. They can, however, be used as references to evaluate the current model in conjunction with field measurements.

[49] At the present stage, the model can first be evaluated through applications in regional-scale air-quality models. The spatial and temporal budget of ammonia and associated ammonium particles produced by the air-quality models will shed some light on the effectiveness of this bi-directional exchange approach. For those air-quality models that have explicitly included the biogenic ammonia emissions as inputs, the double counting of ammonia emissions at locations with high-N status should be avoided when implementing the bi-directional exchange model. In this case, the bi-directional exchange model should only provide deposition fluxes (at a reduced rate compared to the original deposition model), but not additional emission fluxes.


[50] L. Zhang greatly appreciates M. A. Sutton for helpful discussions and R. Vet for initiating the project and for providing continuous support. Constructive comments from anonymous reviewers are also greatly appreciated.