Hatfield J.L., Gandini A. (eds.) Nitrogen in the Environment

Подождите немного. Документ загружается.

354 Nitrogen in the Environment

conditions. When combined with micrometeorological measurements, such data can

also be used to study and to model plant–atmosphere exchange processes with a detailed

time resolution (e.g., Fowler et al., 2001 ; Erisman et al., 2005 ; Sutton et al., 2007 ).

Daily measurements of NH

and NH



can be made with either filter packs

( EMEP, 1996 ) or annular denuder systems (ADS) ( Allegrini and De Santis, 1989 ;

EMEP, 1996 ). Due to phase uncertainties with filter pack sampling, NH

and NH



data have historically been reported to EMEP as Total Inorganic Ammonium (TIA).

ADS are designed for short-term sampling of 1–24 h, and this is a technique recom-

mended by the US Environmental Protection Agency ( USEPA, 1997 ) and also by

the European Monitoring and Evaluation Programme ( EMEP, 1996 ).

Automatic continuous monitors may also be appropriate for providing high-

resolution data where these can be employed at a reasonable cost. One of the earliest

continuous monitors for NH

is based on a chemiluminescence detector for nitric

oxide (NO), where the sampled air stream is passed through a catalytic converter to

convert NH

to NO ( Demmers et al., 1999 ). However, the method has a high detec-

tion limit and is sensitive to NH



aerosol.

Automatic batch denuders (e.g., Keuken et al., 1988 ) and continuous wet annu-

lar denuder methods with on-line analysis (AMOR: Buijsman 1998 ; AMANDA:

Wyers et al., 1993 ) are also available for monitoring NH

. The AMOR system oper-

ates with an hourly time resolution and is used in The Netherlands to provide NH

concentrations at eight stations ( Figure 9 , Van Pul et al., 2004 ). The AMANDA

gradient system ( Wyers et al., 1993 ; Erisman et al., 1998 ) has been used to provide

long-term average fluxes of NH

over grassland, moorland, and coniferous forest

( Sutton et al., 1995b, 1995c, 1997, 1998 ; Erisman et al., 1998 ).

Diffusion scrubbers use parallel plates to capture NH

by diffusion to a trapping

solution, followed by on-line chemical analysis. The term “ diffusion scrubbers ” has

320 240 160 80 0 80 160 240 320

SW Distance from source (m) NE

(µg/m

)

Period 1

Period 2

Period 4

Mean

Upwind site

Figure 8 . Monitored NH

concentrations along the SW-NE transect at a poultry

farm with natural ventilation for three 6–8-week bird cycles. Data from Tang et al.

(2005) .

CH11-P374347.indd 354CH11-P374347.indd 354 5/31/2008 6:11:12 PM5/31/2008 6:11:12 PM

Sources, Dispersion and Fate of Atmospheric Ammonia 355

often been used to refer to systems where the trapping solution is passed through

a semi-permeable tube located longitudinally in the center of an air sampling tube.

Such systems ( Dasgupta, 1993 ; Neftel et al., 1998 ) typically collect 10–30% of the

passing through, and are calibrated according to their collection efficiency.

These systems can also trap and analyze NH



particulate matter (e.g., Al-Horr et al.,

2003 ), but can be difficult to operate and are not generally suitable for monitor-

ing at multiple locations. An instrument based on the AMANDA system, but with

a steam-jet aerosol trapping device, measures hourly concentrations of NH

and



(GRAEGOR: Oms et al., 1996 ; Thomas et al., 2008 ). Wet denuder systems

such as the AMANDA collect 100% of the ammonia and therefore do not require

such calibration.

At a small number of “ super sites ” across Europe, highly complex time-resolved

measurements are carried out to allow interpretation and parameterization of proc-

esses for model improvement and the analysis of pollution events. State-of-the-art

instrumentation for high resolution, precise, and selective continuous monitoring

includes advanced optical methods, for example, Differential Optical Absorption

Spectroscopy (DOAS), Tuneable Diode Laser Absorption Spectroscopy (TDLAS),

and photo-acoustic monitors ( Figure 10 ). These high-frequency sampling devices

may in some cases be suited to flux measurements (e.g., by eddy covariance), vali-

dation of high resolution atmospheric chemistry/transport models, source–receptor

relationships, and rapid temporal changes.

400

350

300

250

200

150

100

AMOR NH

(µg/m

)

01/08/2003

01/02/2004

01/08/2004

01/02/2005

01/08/2005

01/02/2006

Figure 9 . Continuous hourly monitoring data (NH

) from the Dutch air monitoring

network (data courtesy of RIVM, The Netherlands).

CH11-P374347.indd 355CH11-P374347.indd 355 5/31/2008 6:11:12 PM5/31/2008 6:11:12 PM

356 Nitrogen in the Environment

3.3.3 . An integrated monitoring approach

In developing strategies for monitoring atmospheric ammonia (the same applies

for other atmospheric pollutants), the use of an integrated approach including sev-

eral “ levels ” of sophistication can be an effective use of available resources ( Sutton

et al., 2004b ; Torseth and Hov, 2003 ). In this strategy, measurements at each level

are matched to the purpose (i.e., most appropriate methodology for the resolution

and sensitivity required), such as providing information on temporal and spatial

changes in air concentrations and deposition or the development and validation of

models and source–receptor relationships. A three-level approach appropriate for

the monitoring of NH

and NH



is outlined in Table 1 (cf. Sutton et al., 2004b ).

3.4 . Modeling Concentrations and Deposition

3.4.1 . Introduction

In addition to the measurement and monitoring of NH

and NH



concentra-

tions and deposition, numerical modeling of the behavior of these compounds fur-

ther extends our knowledge. Whilst monitoring of chemical concentrations can be

carried out at only a restricted number of sites, computer models can be applied to

estimate concentrations and deposition at a large number of modeled locations (e.g.,

grid cells), providing a useful tool for the generation of continuous maps. Transport

models allow the prediction of the fate of atmospheric pollutants in the environ-

ment and to correlate regional concentrations with emissions sources in different

geographical locations and from different source types. Furthermore, models permit

800

600

400

(µg/m

)

200

06:00

27/09/2005

12:00

Date/time (GMT)

18:00 00:00

28/09/2005

AMANDA

Nitrolux

Figure 10 . Comparison of field measurements of NH

close to an agricultural

source, using a continuous wet ADS (AMANDA, ECN, NL) and a photo-acoustic

monitor (NitroLux-100, Pranalytica Inc., CA, USA), both as 5-min averages

(M. Twigg and E. Nemitz, unpublished data, CEH Edinburgh).

CH11-P374347.indd 356CH11-P374347.indd 356 5/31/2008 6:11:13 PM5/31/2008 6:11:13 PM

Sources, Dispersion and Fate of Atmospheric Ammonia 357

Table 1.

Outline of a three-level monitoring strategy for reactive nitrogen compounds.

Level Objective Measurement requirements

1 – To form a baseline for

monitoring temporal and spatial

changes in air concentrations

and deposition at a national level

– Assess effectiveness of

abatement strategies

– Development and validation

of models

Low-cost manual time-integrating

methods that can be implemented

at a large number of sites, with low

frequency monitoring:

e.g., DELTA system ( Sutton et al.,

2001b ), passive diffusion samplers

( Tang et al., 2001 ), filter packs

(but cannot distinguish between

gaseous NH

and particle NH



)

2 – Evaluation of source–receptor

relationships

– Dry deposition monitoring

– Quantify long-range transport

of pollution and transboundary

fluxes

Daily or continuous monitoring

at a reduced number of sites:

e.g., Annular denuder systems

( EMEP, 1996 ; USEPA, 1997 ),

AMANDA/AMOR ( Erisman et al.,

2001 ), photo-acoustic instruments

Low-cost long-term dry deposition

monitoring (e.g., New COTAG

system) ( Fowler et al., 2001 )

3 – Detailed understanding of

atmospheric and chemical

processes

– Interpretation and

parametrization of processes

for model

development

– Detailed analysis of

pollution events

State of the art instrumentation

for high resolution, precise, and

continuous monitoring at a small

number of “ super sites ” :

Coupled multi-phase measurements

at half hourly to hourly resolution,

e.g., DOAS, TDLAS ( Clemitshaw,

2004 ) or continuous flux

measurements:

e.g., AMANDA gradient system

( Wyers et al., 1993 ), GRAEGOR

( Thomas et al., 2008 ), MARGA

( Trebs et al., 2004 ; ten Brink et al.,

2007 )

assessment of future environmental change through scenario simulations consider-

ing the application of policies to reduce NH

emissions.

Modeling the emissions, transport, transformation, and deposition of NH

is a

challenging task, due to the complexity of the underlying chemical, physical, and

CH11-P374347.indd 357CH11-P374347.indd 357 5/31/2008 6:11:13 PM5/31/2008 6:11:13 PM

358 Nitrogen in the Environment

meteorological processes, which need to be accurately parameterized to simulate

the environmental behavior of NH

. The degree of complexity depends on the pur-

pose of the model, the available computational power, and the state of knowledge

of the relevant process and input parameters. Models representing the transport and

transformation of NH

need to consider seasonal and diurnal cycles in emissions

due to livestock management and changes in meteorological conditions. Modeling the

dry deposition of NH

requires calculation of the deposition velocity, which is depend-

ent on vegetation type, meteorology, and the equilibrium between the gas phase con-

centration in air and the NH



concentration in the stomatal cavity (Section 3.2).

Inclusion of land cover maps is important in numerical models, as calculations show

that deposition rates to forest and moorland vegetation types are significantly higher

than those to grassland and arable land types, which often act as net sources. In many

ATMs, the wet removal rate is represented relatively simply by the product of the

scavenging coefficient (i.e., the ratio between the concentration of a chemical com-

pound in air and in water droplets) and the rate of precipitation.

Due to the very different distances associated with the transport of gas phase

and particulate phase NH



, numerical models to estimate the concentration

and deposition of reduced N have been developed at a range of spatial resolutions

to satisfy different objectives. These models include local, national, continental, and

global scales. Gridded data generated with local- and national-scale models can be

used to assess the exceedance of thresholds for environmental effects, both accord-

ing to air concentrations, as “ critical levels ” for NH

concentrations, and as total

nitrogen deposition as “ critical loads. ” The concept of critical loads and critical

levels is described in Section 4.3.

3.4.2 . National-, regional-, and global-scale modeling

ATMs can be broadly grouped into two types: Lagrangian and Eulerian. In an

Eulerian framework, the calculation of physical and chemical variables is under-

taken simultaneously for all the grid points in the model domain. With a Lagrangian

approach, calculations are made along a pre-defined trajectory which a parcel of air

is assumed to follow. Large numbers of trajectories (typically tens of thousands) are

required to generate statistically significant results. A major difference between the

Eulerian and the Lagrangian approach is that, whilst calculations in Lagrangian tra-

jectories are independent, the calculations at the grid locations of an Eulerian model

are interdependent.

An example of a national-scale Lagrangian model used to simulate atmos-

pheric ammonia is the Fine Resolution Atmospheric Multi-pollutant Exchange

model (FRAME), which was applied to the British Isles with a 5-km grid resolution

( Singles et al., 1998 ; Fournier et al., 2005 ). The atmosphere is represented in the

model by a column of air with 33 layers of vertical resolution varying from 1 m

at the surface up to 200 m. Such a detailed vertical treatment is very helpful for

ammonia, which is mainly emitted from sources at or near ground level, giving rise

to strong vertical concentration profiles. A statistical representation of meteorology

CH11-P374347.indd 358CH11-P374347.indd 358 5/31/2008 6:11:13 PM5/31/2008 6:11:13 PM

Sources, Dispersion and Fate of Atmospheric Ammonia 359

in FRAME is made by using directional wind frequency and wind speed roses. The

model includes 12 prognostic chemical variables with dry and aqueous phase sul-

fur and nitrogen chemistry. Dry deposition of ammonia is simulated with a land-use

dependent canopy resistance formulation, while wet deposition of ammonia gas and

ammonium aerosol is calculated assuming constant drizzle, driven by a map of annual

precipitation. The annual UK budget estimated by FRAME (for 1996) includes 103 k t

-N dry deposition and 105 k t N H

-N wet deposition. A map of annual mean NH

concentrations estimated with FRAME for the UK is illustrated in Figure 11 .

Two other national-scale models applied to simulate ammonia are the OPS

model (for The Netherlands) and the DAMOS model (for Denmark). The OPS

model represents a combination of a Gaussian plume model for local-scale appli-

cation and a trajectory model for long-range transport operating on grid scales of

0.0–0.5

0.5–1.0

1.0–1.5

1.5–2.0

2.0–2.5

2.5–3.0

3.0–3.5

3.5–4.0

4.0–4.5

4.5

µg N/m

Figure 11 . Surface concentration of NH

for the UK in 2002 (FRAME model ver-

sion 5.9). Units are  g N/m

CH11-P374347.indd 359CH11-P374347.indd 359 5/31/2008 6:11:13 PM5/31/2008 6:11:13 PM

360 Nitrogen in the Environment

500 m and 5 km ( Van Pul et al., 2004 ). The model was used to simulate concentra-

tions, deposition, and budgets of NH

gas and NH



aerosol. The implementation

of national regulations to control NH

emissions in The Netherlands is estimated to

have led to a reduction of 37% in emissions between 1990 and 1996. The change in

emissions was, however, not matched by a decrease in measured NH

concentra-

tions. This mismatch between the expected and observed change in concentration

is referred to as the Dutch “ ammonia gap ” ( Erisman et al., 1998 ) and may in part

be due to the reduced gas to particle conversion rates for NH

driven by a parallel

reduction in emissions of SO

( Sutton et al., 2003 ).

The Danish Ammonia Modelling system (DAMOS) uses a combination of a

long-range transport model ( Christensen, 1997 ) and a Gaussian local-scale transport-

deposition model for dry deposition. The model operates on a variety of scales with

two-way nesting, from 150 km for the northern hemisphere, 50 km for Europe, and

16.7 km for Denmark. Ammonia emissions are computed with high spatial and tem-

poral resolution at a single farm and field level ( Gyldenkaerne et al., 2005 ). The

high resolution in the inventories was shown to be important for the model perform-

ance ( Hertel et al., 2006 ).

A European scale model was developed by EMEP under the Convention on

Long-Range Transboundary Air Pollution (CLTRAP) for international co-operation

to solve transboundary air pollution problems ( Simpson et al., 2003 ). The EMEP

model is a 3D Eulerian ATM with a domain which includes all of Europe at a

50  50-km grid resolution. The model uses 20 vertical layers to describe the tropo-

sphere, with the vertical domain extending up to 16 km altitude. By setting the emissions

of pollutant gases (NH

, NO

, and SO

) from individual countries to zero, the model

generates source–receptor matrices of the contribution to dry and wet deposition

in one country associated with emissions from another country ( Tarrasón, 2003 ).

The analysis shows that NH

emissions lead to significant trans-national transport

of air pollution (mainly in the form of NH4



) aerosol in Europe. Figure 12 illus-

trates the modeled NH

and NH



concentrations.

An example of a global simulation of NH

dry deposition is shown in Figure 13 ,

using the STOCHEM model, a global 3D Lagrangian particle chemistry transport

model ( Derwent et al., 2003 ). The spatial pattern of deposition is broadly simi-

lar to that shown in Figure 3 (Section 2.4.) for NH

emissions (although an older

inventory was used by Derwent et al. (2003) compared with the recent improved

inventory of Bouwman et al., 2005 ), which again illustrates the relatively rapid

deposition of NH

near sources.

3.4.3 . Local-scale modeling of ammonia

To assess accurately the effects of NH

deposition to individual sites (e.g.,

nature reserves) it is usually necessary to use local-scale atmospheric dispersion

modeling and to take into account the high spatial variability of NH

concentrations

and deposition fluxes. Regional- or national-scale modeling at a spatial resolution

of several kilometers cannot represent this high spatial variability and can significantly

under- or overestimate concentrations and deposition fluxes ( Dragosits et al., 2002 ).

CH11-P374347.indd 360CH11-P374347.indd 360 5/31/2008 6:11:14 PM5/31/2008 6:11:14 PM

Sources, Dispersion and Fate of Atmospheric Ammonia 361

(a)

Figure 12 . Predicted surface concentration for Europe in 2002: (a) Ammonia (NH

)

and (b) Ammonium (NH



), calculated with the EMEP Unified model (data cour-

tesy of David Simpson, Norwegian Meteorological Institute). Units are  g N/m

(b)

0.0–0.3

1.3–1.5

0.3–0.5

1.5–1.8

0.5–0.8

1.8–2.0

0.8–1.0

2.0–2.2

1.0–1.3

2.2

CH11-P374347.indd 361CH11-P374347.indd 361 5/31/2008 6:11:14 PM5/31/2008 6:11:14 PM

362 Nitrogen in the Environment

Local-scale modeling is typically carried out over a spatial range of up to sev-

eral kilometers, since this represents the scale where most impacts of dry deposition

of the NH

emitted by an individual source occur. A range of local-scale mode-

ling approaches have been applied over different spatial scales, from the interac-

tion between an individual source and receptor, over distances of less than 100 m

( Loubet et al., 2006 ), to the impacts of several sources on different receptors within

a landscape of, for example, 5 k m  5 km ( Dragosits et al., 2002 ). Such simula-

tions are often conducted to estimate NH

concentrations and deposition in order

to assess the impacts of existing or proposed emissions on nearby semi-natural eco-

systems. Local-scale modeling may also be used to provide concentration predic-

tions to assess health risks to people, as well as a research tool to better understand

the processes of dispersion and deposition.

The choice of which local-scale model to use depends on the objectives of the

exercise. The study of short-scale interactions between sources and sinks (such as

recapture of NH

by an adjacent belt of trees) requires a detailed model with a high

spatial resolution (e.g., Loubet et al., 2006 ). Such models require a detailed descrip-

tion of airflows and are therefore most suited to simple situations (e.g., single source

and receptor). For the study of interactions between a network of sources and recep-

tors within a landscape, a “ local-scale ” model is required. Modified Gaussian plume

(e.g., AERMOD and ADMS 3) and Lagrangian (e.g., LADD: Hill, 1998 ; described in

Figure 13 . Global NH

dry deposition map for 2000, produced from output of

the STOCHEM model (data courtesy of David Stevenson, Edinburgh University).

Units are g N / m

gN m

2

0.0 – 0.2 0.2 – 0.4 0.4 – 0.6 0.6 – 0.8 0.8 – 1.0  1.0

CH11-P374347.indd 362CH11-P374347.indd 362 5/31/2008 6:11:18 PM5/31/2008 6:11:18 PM

Sources, Dispersion and Fate of Atmospheric Ammonia 363

Dragosits et al., 2002 ) models are most often used for these assessments and domain

sizes up to several kilometers are common. Figure 14 shows the spatial distribution of

simulated NH

concentrations and deposition fluxes within a 5 k m  5 km domain sur-

rounding several farms as estimated by the LADD Lagrangian model ( Dragosits et al.,

2002, 2006 ). The sources of ammonia emissions from this landscape are described in

Figure 2 . Figure 14 highlights the large spatial variability at this spatial scale due to

the dispersion of NH

away from sources and the different land cover types (i.e.,

different deposition velocities). Chemical reactions of NH

can mostly be neglected

Figure 14 . Predictions of NH

(a) concentrations and (b) deposition within an agri-

cultural area using a local-scale atmospheric dispersion model. The critical level

shown (8 g/m

) is that set in 1992 ( Ashmore and Wilson, 1994 ), and has recently

been updated ( UNECE, 2007 ) .

 0–1

1 – 2.5

2.5–5

Critical level

0 1 2 km

Nature reserves

Farms

5–10

10–50

 50

Concentration (ug NH

-Nm

)

(a)

CH11-P374347.indd 363CH11-P374347.indd 363 5/31/2008 6:11:19 PM5/31/2008 6:11:19 PM