432 E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445
coincidence factor C
i
between NH
4 +
and an an- ionic species i as a measure of the degree to which
the two species are found in the same size range
C
i
= 1 − 0.5
Z
∞
m
NH
4 +
R
p
m
tot NH
4 +
− m
i
R
p
m
tot i
dR
p
5 mR
p
denotes the mass distribution function and m
tot
the total mass of the species NH
4 +
or i. Values of C
i
range from 1 total coincidence down to 0 no co- incidence. Since equilibrium theoretically only holds
if the NH
4 +
is present as nitrate and sulphate, the au- thors suggest the restriction of the integration to size
classes in which the chloride equivalent concentration is small 0.1 [NH
4 +
] compared with the ammo- nium equivalent concentration.
3. Results
3.1. Air concentrations of acids and aerosols The analysis period of the batch denuder sam-
ples was selected due to the occurrence of southerly to westerly flow, which could be expected to result
in good fetch conditions and higher concentrations than onshore winds. Nevertheless, the concentrations
Fig. 2. Concentration χ of NH
3
, HNO
3
, HNO
2
and HCl interpolated for 1 m above then zero plane displacement as measured with the CEH AMANDA and batch denuders for 16–21 June 1995.
were small with median concentrations of 10 neq m
− 3
0.63 mg m
− 3
for HNO
3
, 4 neq m
− 3
0.2 mg m
− 3
for HNO
2
and 8.8 neq m
− 3
0.32 mg m
− 3
for HCl. Although concentrations in mg m
− 3
are most com- monly used, Fig. 2 shows the time course of these
concentrations in meq m
− 3
for intercomparison be- tween several species. Gaps in the data are partly
due to malfunctioning of the samplers, but mainly due to concentrations below the detection limit of ca.
0.2 mg m
− 3
HNO
3
, HNO
2
or HCl. There was no clear relation between concentrations and wind direc-
tions or time of day, with the exception of HCl, which showed largest air concentrations during the day.
Table 1 shows the total concentrations of partic- ulate NH
4 +
, NO
3 −
, Cl
−
and SO
4 2−
as measured with the cascade impactor, together with the mass
median diameter MMD and the coincidence fac- tor for NH
4 +
–NO
3 −
and NH
4 +
–SO
4 2−
as calculated according to Eq. 5. The values of the MMD of
NH
4 +
and SO
4 2−
are similar, and the MMD of NO
3 −
slightly larger. As a consequence, the size spectrum of NH
4 +
shows higher coincidence with SO
4 2−
average C
SO
4 2−
= 0.92 than with NO
3 −
average C
NO
3 −
= 0.69. Run 14 represents an exception with a poor ion
balance, very low NO
3 −
concentration and a higher coincidence of NH
4 +
with NO
3 −
than with SO
4 2−
. The MMD of Cl
−
is consistently larger than the MMD
E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445 433
Table 1 Aerosol concentration, mass median diameter MMD and coincidence factors as measured with the cascade impactor
a
Run number
Starting Duration
h Mean wind
direction
◦
Concentration neq m
− 3
Mass median diameter mm
Coincidence factor NH
4 +
Date Time
GMT NH
4 +
NO
3 −
Cl
−
SO
4 2−
NH
4 +
NO
3 −
Cl
−
SO
4 2−
NO
3 −
SO
4 2−
2 120695
21:47 8:02
1 ±39 52
0.45 3
130695 09:47
11:13 50 ±19
60 0.39
4 130695
22:29 8:25
347 ±20 46
0.34 5
140695 09:46
10:06 36 ±31
35 0.43
6 140695
22:02 7:05
184 ±84 59
0.50 7
150695 08:40
8:50 51 ±48
36 0.42
8 150695
21:00 8:00
229 ±63 53
0.50 9
160695 09:50
19:10 225 ±50
13 0.36
10 170695
09:42 8:26
263 ±8 29
10 23
17 0.28
0.72 2.33
0.28 0.68
0.95 11
170695 21:00
8:00 244 ±16
32 0.88
12 180695
11:00 22:45
230 ±34 38
0.45 13
190695 10:06
9:54 229 ±23
55 9.5
9.2 32
0.33 0.62
0.87 0.33
0.68 0.94
14 190695
20:18 12:08
228 ±28 19
5.8 7
7.2 0.30
0.19 0.19
0.34 0.79
0.82 15
200695 08:44
12:12 275 ±7
33 17
42 17
0.47 0.90
2.31 0.40
0.70 0.94
16 200695
21:13 22:47
273 ±21 31
0.39 17
210695 20:25
13:28 257 ±43
48 18
38 29
0.33 0.97
2.33 0.31
0.62 0.96
18 220695
10:25 23:10
112 ±87 58
0.33 Average
41 12.1
23.8 20.4
0.42 0.68
1.61 0.34
0.69 0.92
a
Fewer runs were analysed for anions than for NH
4 +
. The wind direction is a vector average stated in degrees from North, with the standard deviation of 10 min values in parentheses. The MMD was calculated by linear interpolation from the plot of the accumulated
mass on the impactor stages vs. 50 cut-off diameter e.g. Hinds, 1982.
of NH
4 +
and shows maximum values for northerly wind directions Runs 10, 15 and 17. On these oc-
casions NH
4 +
is more than balanced by the sum of the measured anions, and chloride must be expected
to represent mainly marine NaCl, which is typically found in larger size ranges than non-marine NH
4 +
salts. 3.2. Surface exchange flux of acids and aerosols
Since the concentrations of HNO
3
and HNO
2
were close to the detection limit, their measurements showed too much scatter for fluxes to be calculated
reliably. Although still rather uncertain, the HCl data were more consistent and fluxes could be calculated
using the AGM Section 2.2. Concentrations at 1 m x as well as the fluxes F are shown in Fig. 3. Both
x and F show a diurnal pattern and interestingly HCl
appears to be emitted for most of the time, with larger emission during daytime than during night-time. As
this daily variation could be an apparent effect of changes in the friction velocity u
∗
, the scaling pa- rameter χ
∗
= − F u
∗
, which essentially represents the slope of the log-linear concentration profile of
HCl, is shown in Fig. 3 for comparison. The fact that χ
∗
also showed larger values during daytime indi- cates that the detected gradient did change over the
day, supporting the estimates of the calculated F
HCl
. In addition, large values of F
HCl
are positively cor- related with elevated air concentrations χ
HCl
1 m. For a tracer that originates from distant sources and
is mainly deposited the magnitude of the deposition flux is governed by the air concentration: an increase
in χ
HCl
1 m should therefore have promoted depo- sition and reduced emission. The fact that at North
Berwick emission was linked to high HCl concentra- tions suggests that at this site the elevated concen-
trations are a result of the HCl emission from the oilseed rape canopy itself. An analogous conclusion
was drawn for NH
3
Sutton et al., 2000b. This cor- relation, therefore, strongly supports the occurrence
of significant HCl emission. Aerosol fluxes were calculated from the gradient
measurements above the canopy with both SJAC and
434 E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445
Fig. 3. HCl results measured with the automated batch denuder systems at three heights during the period 16–21 June. The concentration at 1 m χ
HCl
, the flux F
HCl
and the scaling parameter χ
∗ HCl
are shown with a 4-h running mean of hourly values, and the error bars represent the standard error of this running mean.
filter-packs. The fluxes were filtered for low wind speeds u 1 m 1 m s
− 1
, insufficient fetch con- ditions by footprint analysis as described by Sutton
et al., 2000b and highly stable situations L
− 1
0.5 m
− 1
. From the remaining values the deposition velocities V
d
were calculated as V
d
1 m = − F
χ
χ 1 m 6
The medians of the results are presented in Table 2, where the results of the SJAC measurements are di-
vided into three classes of u
∗
. The median values of V
d
for NH
4 +
increased with u
∗
, whereas the arith- metic mean of all SJAC data points and in particu-
Table 2 Median deposition velocities V
d
for aerosols as measured at North Berwick with SJAC and filter-packs Aerosl
species Measuring
technique Sampling
time min u
∗
class m s
− 1
Median V
d
mm s
− 1
Standard deviation V
d
mm s
− 1
No. of observations
Fraction of emission
NH
4 +
SJAC 30
All 1.7
23.0 164
47 SJAC
30 0.3
0.2 13.0
35 49
SJAC 30
0.3–0.6 1.9
22.9 93
47 SJAC
30 0.6
10.4 30.3
12 44
NH
4 +
Filter-packs 120
All 0.8
19.8 12
50 NO
3 −
Filter-packs 120
All 10.1
26.7 7
14 Cl
−
Filter-packs 120
All 6.5
38.7 7
14 SO
4 2−
Filter-packs 120
All 4.9
18.9 7
29
lar for 0.3 u
∗
0.06 m s
− 1
was negative. Up- ward gradients of NH
4 +
were measured for half of the time with both SJAC and filter-packs, independent
of u
∗
. However, owing to the low concentrations, all filter-pack NH
4 +
emission gradients turned out not to be significant at P = 0.05, although many only just
failed this criterion. The mean aerosol deposition ve- locities measured by filter-packs increased in the or-
der NH
4 +
, SO
4 2−
, NO
3 −
, Cl
−
, which is consis- tent with the increase in the mass median diameter
Table 1. The profiles of an example filter-pack run Run
18: 17 June 16:00–18:30 GMT are presented in Fig. 4. The concentration profiles should form straight
E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445 435
Fig. 4. Example above canopy profiles of NH
4 +
, NO
3 −
, SO
4 2−
and Cl
−
aerosol as measured with filter-packs on 17 June 1995, 16:00–18:30 GMT. The error bars are the sum of the error
due to analytical procedures ca. 6 and the standard error in the blanks N = 4. The corresponding fluxes ±standard er-
ror of the regression were 0.35 ± 0.29 neq m
− 2
s
− 1
for NH
4 +
, −
0.23 ± 0.07 neq m
− 2
s
− 1
for NO
3 −
, −1.9 ± 0.78 neq m
− 2
s
− 1
for SO
4 2−
and −5.8 ± 1.4 neq m
− 2
s
− 1
for Cl
−
, where negative values represent deposition.
lines when plotted versus lnz–d–ψ
H
, where z is the height, d the zero plane displacement and Ψ
H
the inte- grated stability correction function for heat and other
scalars e.g. Sutton et al., 2000b. During this pe- riod NO
3 −
and SO
4 2−
showed slow deposition, Cl
−
was deposited rapidly whereas NH
4 +
was probably emitted. If the NH
4 +
emission is real, it probably rep- resented NH
4
Cl. From the equivalent concentrations it is clear that Cl
−
was dominated by species other than NH
4
Cl. Consistent with the MMD values from Table 1, a fraction of these were contained in larger
and therefore more rapidly depositing particles. A small emission gradient of NH
4
Cl would easily have been masked by the deposition of non-ammonium
Cl
−
. 3.3. Within-canopy profiles and sourcesink analysis
Nemitz et al. 2000a reported measurements of in-canopy profiles of NH
3
for the oilseed rape and cal- culated the associated vertical distribution of sources
and sinks together with the height-dependent flux. In this section, the same procedure is applied to gradients
of aerosols as well as HCl. Fig. 5 shows in-canopy profiles of NH
4 +
, NO
3 −
, Cl
−
and SO
4 2−
aerosols as well as gaseous NH
3
and HCl in meq m
− 3
for four runs covering the period from 21 June 22:00 GMT to
22 June 13:30 GMT. The sourcesink profiles for Cl
−
and NH
4 +
as calculated with the ILT for the same runs are presented in Fig. 6.
The NH
3
flux above the canopy changed from strong deposition on the evening of 21 June to small
deposition during the night and then to emission in the morning of 22 June. During Runs 25 and 27
the NH
4 +
concentration appeared to be balanced by NO
3 −
whereas later on, the NH
4 +
exceeded the NO
3 −
concentration, indicating that at least a part of the NH
4 +
was present as SO
4 2−
or Cl
−
Fig. 5. During the late evening Run 25 all species were
deposited to the canopy Fig. 5a and the sink distri- butions of both Cl
−
and NH
4 +
corresponded to the leaf area distribution see Nemitz et al., 2000a in
the canopy Fig. 6a. During the night, all concen- trations increased within the canopy, which can be
seen in the early morning run Fig. 5b. The large concentration of Cl
−
aerosol in the middle of the canopy, which was found to be caused by a source in
the range 0.33–1.1 m Fig. 6b, was particularly dis- tinct. There was indication that NH
4 +
, SO
4 2−
and NO
3 −
were also emitted from the top of the canopy, but unlike the Cl
−
peak within the canopy this was the consequence of the concentration measured with
only one filter-pack. Later in the morning the aerosol concentrations were again smaller Fig. 5c. The Cl
−
peak, though less pronounced, was still clearly visi- ble, which was reflected in a smaller but still positive
source strength between 0.33 and 0.75 m Fig. 6c. At this time all other aerosol gradients at the top of the
canopy indicated deposition to the canopy and NH
4 +
was absorbed by the whole canopy. By noon the Cl
−
aerosol concentration had decreased even further, but the measurements of gaseous Cl compounds, most
probably representing HCl, showed a clear maximum within the canopy Fig. 5d. Fig. 6d shows that at this
436 E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445
Fig. 5. Example within-canopy profiles of aerosols, NH
3
as well as HCl for run 29 only measured with filter-packs during the period 21–22 June. The error bars are sum of the error due to analytical procedures ca. 6 and the standard error in the blanks N = 4. The
values of u
∗
are presented to indicate the magnitude of turbulence within the canopy.
Fig. 6. Example distributions of the sourcesink density S in the oilseed rape canopy for Cl
−
and NH
4 +
aerosol as well as HCl Run 29 only calculated for the same period as Fig. 5 21–22 June. Negative values of S denote sinks whereas positive values represent sources.
E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445 437
Fig. 7. Time course of measured K
m
and equilibrium K
e
concentration products between HNO
3
and NH
3
for 16–21 June. The concentrations of NH
4 +
1 m as measured with SJAC is shown for comparison. Some gaps in the SJAC time-series have been filled with data from the filter-packs and cascade impactor.
small concentration the sourcesink analysis for Cl
−
aerosol did not reveal a clear structure, while NH
4 +
was clearly deposited to the top canopy and HCl ap- peared to be released at mid-canopy. A general feature
of all within-canopy profiles was the concentration increase near the ground.
3.4. Concentration products The time courses of the NH
4 +
concentration and the measured concentration product [NH
3
] × [HNO
3
] K
m
are shown in Fig. 7, together with the theoretical equilibrium product for pure NH
4 +
salts K
e
. Since K
e
is a function of T and h, it follows a clear diurnal pattern. Although K
m
was often larger during the day- time than at night, it only exceeded K
e
on the morning of 20 June. Consequently, during this period forma-
tion of NH
4
NO
3
aerosol could be expected, whereas at other times, NH
4
NO
3
ought to have evaporated. Dur- ing some mornings 17 and 19 June the increase in
K
m
followed K
e
closely. This could either indicate that during this time the concentrations adjusted to equilib-
rium or that K
m
increased due to daytime emission of NH
3
. Bearing in mind that NH
4 +
aerosol represents not only NH
4
NO
3
and NH
4
Cl but also NH
4 2
SO
4
, it seems nevertheless likely that K
m
did not gener- ally adjust to K
e
despite the presence of considerable amounts of volatile NH
4 +
. The h and T dependence of the measured concentration product with HNO
3
at 1 m Fig. 8a shows that K
m
was smaller than K
e
ex- cept for h 90. In contradiction to theory, con-
centration products were smaller for h 70 than 70. In the case of HCl Fig. 8b, measured concen-
tration products appeared to be independent of h, but purely governed by T, attaining values which would
be expected for h = 95. Using the measured fluxes of HCl as well as sensible H and latent heat λE
presented by Sutton et al. 2000b, K
m
can be calcu- lated for the notional height of the mean canopy ex-
change z
′
according to Eq. 2. These values Fig. 8c show a different picture altogether: K
m
tended to ex- ceed K
e
at canopy height for h 95, whereas for low h 60 K
m
≪ K
e
was found. Hence, at high h there was potential for aerosol production or growth
within the canopy at the same time as NH
4
Cl above the canopy should have evaporated.
3.5. Chemical time-scales Potential chemical time-scales are estimated here
under the assumption that all NH
4 +
represented NH
4
NO
3
and that condensation was limited to these aerosols. The results calculated according to Eq. 3
for two different values of α 0.1, 1 are presented in Table 3. Also shown are critical values of u
∗
u
∗ crit
, above which the influence of GPIC on the NH
3
flux
438 E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445
Fig. 8. Comparison of measured concentration products for different temperatures T and classes of relative humidity h with values predicted from thermodynamic theory for specified relative humidities lines. a According to the products of [HNO
3
] × [NH
3
] at 1 m, aerosol evaporation is expected for h 95. b The product [HCl] × [NH
3
] at 1 m continuously stays below the equilibrium value, whereas c the product [HCl] × [NH
3
] extrapolated to the surface z
′
exceeds K
e
for h 90. Data points represent hourly values.
Table 3 Chemical time-scales τ
∞
for two different values of the accommodation coefficient α as calculated from the NH
4 +
size spectra assuming NH
4
NO
3
only, as well as from the estimated size spectrum of the total mass of hydrophilic aerosol
a
Run number Starting
NH
4
NO
3
only α = 0.1 NH
4
NO
3
only α = 1 Estimated total hydrophilic values at α = 1
Date Time GMT
τ
∞
h u
∗ crit
m s
− 1
τ
∞
h u
∗ crit
m s
− 1
τ
∞
h u
∗ crit
m s
− 1
2 120695
21:47 0.30
0.005 0.05
0.029 3
130695 09:47
0.36 0.004
0.07 0.021
4 130695
22:29 0.40
0.004 0.07
0.020 5
140695 09:46
0.65 0.002
0.12 0.012
6 140695
22:02 0.39
0.004 0.07
0.019 7
150695 08:40
0.63 0.002
0.12 0.012
8 150695
21:00 0.46
0.003 0.09
0.016 9
160695 09:50
1.97 0.001
0.43 0.003
10 170695
09:42 0.58
0.002 0.11
0.013 0.09
0.016 11
170695 21:00
0.92 0.002
0.17 0.008
12 180695
11:00 0.63
0.002 0.12
0.012 13
190695 10:06
0.39 0.004
0.08 0.018
0.07 0.020
14 190695
20:18 0.90
0.002 0.16
0.009 0.10
0.015 15
200695 08:44
0.48 0.003
0.08 0.018
0.07 0.021
16 200695
21:13 0.64
0.002 0.12
0.012 17
210695 20:25
0.40 0.004
0.07 0.020
0.07 0.022
18 220695
10:25 0.32
0.004 0.06
0.024
a
Also shown are the critical values of u
∗ crit
, above which the profile measurements of NH
3
can be expected to be unaffected by chemical reactions.
E. Nemitz et al. Agricultural and Forest Meteorology 105 2000 427–445 439
can be ruled out, calculated according to Eq. 4. Since condensation is expected to take place to any
hydrophilic aerosol Wexler and Seinfeld, 1992, the limitation to NH
4 +
aerosol may overestimate τ
∞
. The aerosol was only analysed for NH
4 +
, Cl
−
, NO
3 −
and SO
4 2−
but not, for instance, for Na
+
, Ca
2+
or Mg
2+
. Thus, a best estimate of the mass of total hy- drophilic aerosol was obtained as follows: if NH
4 +
was more than balanced by the sum of the measured anions, any Cl
−
in excess was thought to represent NaCl. By contrast, excess NH
4 +
was interpreted as resulting from an uncertainty in the NO
3 −
measure- ment. Although, the concentration of additional Cl
−
interpreted as NaCl was high during some runs, it only contributed a minor fraction to the total aerosol
surface owing to its large MMD Table 1. Hence, the net effect was only marginally smaller values of
τ
∞
larger u
∗ crit
Table 3. For comparison, measured values of u
∗
ranged from 0.004 to 1.0 m s
− 1
. Values of u
∗
0.03 m s
− 1
, the largest values of u
∗ crit
derived here, were found for less than 2 of the time. How-
ever, the applicability of the aerodynamic gradient technique is restricted at such low turbulence and
fluxes are very small.
4. Discussion