boundary-layer isotope dilution/mass balance methods for measurement of nocturnal methane emissions...
TRANSCRIPT
Atmospheric Environment 36 (2002) 4663–4678
Boundary-layer isotope dilution/mass balance methods formeasurement of nocturnal methane emissions from grazing
sheep
M.J. Harveya,*, G.W. Brailsforda, A.M. Bromleya, K.R. Lasseya, Z. Meia,I.S. Kristamenta, A.R. Reisingerb, C.F. Walkerc, F.M. Kelliherd
aNational Institute of Water and Atmospheric Research, P.O. Box 14-901, Kilbirnie, Wellington, New ZealandbMinistry for the Environment, P.O. Box 10-362, Wellington, New Zealand
cSOC, University of Southampton, European Way, Southampton SO14 3ZH, UKdManaaki Whenua—Landcare Research, P.O. Box 69, Lincoln, Canterbury, New Zealand
Received 20 July 2001; received in revised form 21 May 2002; accepted 12 June 2002
Abstract
Following advances with methods for 13C/12C isotopic analysis of methane in small (p4L) air samples, new isotope
dilution techniques are proposed for measurement of methane emissions at the paddock scale from grazing ruminant
animals. These techniques combine measurement of the isotopic d13CH4 composition of air samples with a non-
intrusive mass balance method applied in the nocturnal boundary layer. Flux estimates from trials using the isotope
dilution techniques are compared with estimates based on scaling up individual animal emission measurements using a
rumen gas tracer technique. The methane flux assessed by the latter technique ranged from 35 to 70mg (CH4) m�2 d�1
with a stocking density between 10 and 20 sheep ha�1. The isotope dilution based nocturnal boundary-layer estimates
generally agreed to better than a factor of 2 and usually to within 20% of the average of individual animal emission rate
per unit area of paddock. Both static and advecting mass balance methods are developed. In the advecting case, the
upwind/downwind contrast in d13C was typically 0.2–0.5%. Care was necessary with air sampling to avoid error in this
small contrast contributing to error in the flux. Agreement between concentration- and isotope-based nocturnal
boundary layer methods and the sheep breath measurements indicated that sample representativeness was generally
good. Factors which affect the accuracy of the method are examined and include variability in nocturnal mixing height,
the assumed d13CH4 composition of the source sheep breath and diurnal patterns in sheep emission. This paper
establishes new techniques useful in the paddock to landscape scale although widespread application awaits further
development of technology for rapid and repeatable field analysis of d13CH4 in small samples.
r 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Ruminant methane; Micrometeorology; Agricultural greenhouse gas; Methane isotopes; New Zealand
1. Introduction
Globally, ruminant methane accounts for about 15%
of the total methane flux to the atmosphere; more than
50% of this total flux is anthropogenic (Houghton et al.,
2001). However, in New Zealand, ruminant livestock is
thought to account for nearly 90% of anthropogenic
methane emission based on national inventory data for
2000 supplied under the United Nations Framework
Convention on Climate Change (New Zealand Climate
Change Project, 2002). On the basis of 100 year global
warming potentials (Houghton et al., 2001) methane is
New Zealand’s single most important greenhouse gas*Corresponding author. Fax: +64-4-386-2153.
E-mail address: [email protected] (M.J. Harvey).
1352-2310/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 2 - 2 3 1 0 ( 0 2 ) 0 0 4 1 0 - 7
emission accounting for about half of the CO2-
equivalent anthropogenic greenhouse gas emissions
(Ministry for the Environment, 1997). However, there
is significant uncertainty in these estimates. As part of
the process of improving the accuracy of the national
greenhouse gas emissions inventory, ruminant emissions
and their variability are being quantified under a range
of typical farming situations. The current study is one of
a number in New Zealand assessing methane emissions
over a range of measurement scales from individual
animal to region.
These studies include measurements at the individual
animal scale using a sulfur hexafluoride (SF6) tracer
technique developed for cattle by Johnson et al. (1994).
Based on a calibrated permeation source of SF6 inserted
in the animal rumen this technique measures, over a
period of hours to days, the average concentrations of
both SF6 and methane in exhaled air that is slowly
drawn into an evacuated container mounted on the
animal. Methane emission rates are estimated from
the ratio of methane to SF6 in the sampled breath and
the calibrated release rate of SF6. This technique has
been applied in New Zealand and Australia to both
cattle and sheep (Lassey et al., 1997; Leuning et al.,
1999). Emission variations between animals can limit the
certainty in scaled-up breath measurements. This
limitation has, in part, led to measurements that exploit
the atmosphere itself as a natural spatial integrator. At
the paddock scale (0.5–1 km), the micrometeorological
flux-gradient technique has been applied to sheep flocks
(Judd et al., 1999) using point profile gas sampling. For
improved spatial integration, a development of the flux-
gradient technique using long-path measurements with
Fourier transform infrared (FTIR) spectroscopy has
been suggested by Andreas et al. (1992) and applied in
New Zealand by Kelliher et al. (2002). At the regional
scale (20–40 km) convective boundary layer (CBL)
budget measurements have been made by measuring
vertical profile differences up- and downwind of a source
region using a light aircraft (Wratt et al., 2001). In this
paper, we focus on the paddock scale, where the flux-
gradient technique is attractive because it is relatively
easy to apply and non-intrusive. However, a number of
requirements need to be met. Firstly, the method
requires a homogeneous source region (fetch) upwind
and this is not possible to achieve without artificial
constraints on sheep that naturally wander and tend to
flock. Ideally, the landscape should be flat without
aerodynamic barriers (shelter-belts) upwind and suitable
sites are difficult to find in the New Zealand landscape.
Secondly, the technique is often applied close to the
lower limit of flux quantification as determined by the
signal-to-noise ratio, i.e. the magnitude of the concen-
tration gradient can be similar to the precision with
which methane can be measured. Thirdly, the technique
cannot be applied under conditions of light wind and
strong stability at night, when turbulent transfer
relationships break down.
As an alternative at night, sheep emission fluxes can
be quantified from the methane gas budget within the
nocturnal boundary layer (NBL), in either a static or an
advecting situation. The low-level inversion and stable
surface layer that forms, most strongly on still clear
nights, through loss of long-wave radiation to space
significantly inhibits the vertical mixing of trace gases
emitted at the surface. Nocturnal mixing heights are
typically less than 100m compared to daytime con-
vective boundary layers of 1000m, giving more than an
order of magnitude advantage in rate of concentration
increase for emissions within that layer. However,
because of poor mixing within the stable surface layer,
and the likelihood of intermittent entrainment of air
flowing above the inversion (Mahrt, 1999), inhomogene-
ities in methane concentration are highly likely. Never-
theless, nocturnal box and profile measurements have
been used to estimate methane emissions from wetlands
by Gallagher et al. (1994), Choularton et al. (1995) and
Beswick et al. (1998) and from extensive sheep grazing
by Denmead et al. (2000).
In this paper, we describe the development of
techniques for the measurement of d13CH4 in small air
samples. We then develop and assess an extension to
NBL techniques using the isotopic methane measure-
ments. Sheep methane emission is quantified through
dilution of its distinctive isotope signal in the nocturnal
boundary layer and the method is applied to estimate
emission fluxes on five nights at two sites. The results are
compared with other techniques applied concurrently
and to predictions of a plume dispersion model.
Limitations and potential difficulties of the NBL
technique are discussed.
2. Methods
2.1. Methane analysis methods
Methane mixing ratio was measured by gas chroma-
tograph (GC-Hewlett Packard 5890 Series II) with flame
ionisation detection. Measurement precision was typi-
cally better than 1 ppbv. During 1998, high-precision
analysis techniques were developed for methane carbon
isotope ratio determination in small (4 L) air samples
through miniaturization of traps in the vacuum extrac-
tion line. Offline isotope ratio mass spectrometry
(IRMS) was done using the dual inlet technique on a
Finnigan-MAT 252. The GC and IRMS methods
including calibration are outlined by Lowe et al. (1999).
In 1999, the dual inlet technique was superseded by a
continuous-flow (GC-IRMS) technique which intro-
duced a nanomolar sample directly into the mass
spectrometer. The method described by Ferretti et al.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784664
(2000) for CO2 was modified for CH4. Interfering gases
were removed: CO through oxidation with iodine
pentoxide followed by removal of CO2, N2O, non-
methane hydrocarbons and water by cryogenic trapping.
The purified sample was then combusted to CO2 in a
furnace with puratronic grade platinum wire as the
catalyst. At 150ml of air per determination, three
determinations were made on small air samples to
0.06% precision. The total processing time by GC-
IRMS was about 30min compared with 2 h for
extraction and analysis by dual inlet. Accuracy of all
the mass spectrometer measurements was maintained
with regular working gas standard measurements
calibrated against IAEA-NBS19 standard.
For the atmospheric measurement, there is potentially
good signal to noise for both concentration and isotope
measurements. For example with an emission flux F of
2.5mmol(CH4) m�2 d�1 from a sheep flock and a mixing
height z of 100m, the predicted rate of concentration
change determined using Eq. (1) is 25 ppb h�1 which is
16 h�1 of the measurement precision of 1.5 ppb. From
NBL-IDT method 1 the predicted rate of d13C change is
�0.26% h�1 which is 4 h�1 of the measurement precision
of 0.06%.
2.2. NBL and mass balance techniques
If the NBL is considered as a well-mixed box, capped
by a mixing height z; in which the methane concentra-
tion, C; grows with time due to entrainment, the
methane surface source flux, F ; may be calculated as
F ¼ zðdC=dtÞ: ð1Þ
This simple zeroth-order relationship as applied by
Gallagher et al. (1994) to methane flux estimation asserts
mass balance in such a box, neglecting both horizontal
advection and diffusion through the box ceiling. More-
over, during light winds, a weakness in this technique is
the assumption that air within the box is well mixed. To
account for horizontal inhomogeneity, sufficient mea-
surements need to be made within the box to quantify a
‘‘representative’’ concentration. When vertical profiles
of methane concentration are available to characterise
the incomplete vertical mixing in the box, these lead to a
more accurate flux assessment given by
F ¼Z z
0
ðdC=dtÞ dz: ð2Þ
This method Eq. (2), denoted here NBL method 1, is
also limited to still conditions because it ignores
horizontal advection. In practice, this means nights with
windspeed at 2m of less than 2m s�1.
If there is significant advecting wind, (windspeed at
2m greater than 2m s�1) horizontal concentration
gradients will occur across the source region, and mass
balance then requires NBL method 2 Eq. (3) (e.g.
Denmead, 1995; Denmead et al., 1998; Leuning et al.,
1999):
F ¼1
X
Z z
0
uðCd � CuÞ dz: ð3Þ
Application requires that vertical profiles of gas
concentration be measured upwind, CuðzÞ; and down-
wind, CdðzÞ; of the source region. The overbar in Eq. (3)
denotes the time-averaged product of this profile
difference with the advecting windspeed, uðzÞ; across
the experimental plot with distance X between the
upwind and downwind measurement points. This mean
product will differ from the product of the mean profile
difference and mean wind speed because the latter
ignores correlations in the form of turbulent diffusive
fluxes. Denmead (1995) has examined these effects, and
showed that, as has been done in this study, ignoring
such correlations (i.e., replacing the mean of the product
by the product of the mean in Eq. (3)) incurs errors of
around 10–20%.
2.3. Isotope dilution mass balance techniques
As an extension to the NBL methods above, sheep
methane emissions are estimated by following the time
dependent increase in both mixing ratio and d13Cisotopic ratio of the ambient methane. We use the
standard delta notation, d13C, which relates the stable
isotope ratio of sample Rsa to that in the Vienna Peedee
belemnite (VPDB) isotope standard:
d13C ¼ ðRsa=RVPDB � 1Þ � 1000ð%Þ; ð4Þ
where the ratio RVPDB ¼13 CVPDB=12
CVPDB of around
0.0112 (Craig, 1957). The isotope dilution technique
(IDT) exploits the contrast between the d13C in methane
of background or upwind air, du; which is close to
�47%, and that of the agricultural source, ds: For sheepconsuming rye grass/clover pasture, the ‘‘enterically
fermented’’ methane has a d13C value which we have
measured at around �62%, consistent with values
reported by Rust (1981). Source isotopic composition
is discussed further in Section 4.2.
From mass balance considerations, the entrainment of
a surface flux of methane from grazing sheep (d13C value
ds) into upwind air (concentration Cu; d13C value du) to
produce downwind air (well-mixed concentration Cd;d13C value dd) is described by
Cddd ¼ Cudu þ Csds; ð5Þ
where Cs ¼ Cd � Cu is the concentration of sheep
methane in the downwind air. Rearrangement of
Eq. (5) produces an ‘‘isotope mixing line’’:
dd ¼ ds þ ðdu � dsÞCu=Cd ð6Þ
A plot of dd against 1=Cd is a straight line with
intercept at 1=Cd-0 that estimates ds; i.e. d13C of the
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–4678 4665
source (e.g. Stevens and Engelkemeir, 1988; Berga-
maschi et al., 1998).
The IDT utilises the rate of change of d13C due to
entrainment of sheep methane within the boundary
layer, by analogy with the rate of change of concentra-
tion in NBL method 1. Taking the analogue of Eq. (1)
Fds ¼ zðdðCdÞ=dtÞ; ð7Þ
where dðtÞ is the changing value for d13C in the
accumulating atmospheric methane, and combining
with Eq. (1) itself gives
F ¼zCðdd=dtÞds � d
: ð8Þ
For a steady flux, C and d are evaluated at t ¼ 0: Theanalogue of NBL method 1, denoted NBL-IDT method
1, can be similarly derived as
F ¼Z z
0
Cð0Þdd=dt
ds � dð0Þdz: ð9Þ
The NBL-IDT method 1 requires measurement of an
initial concentration and time series of d13C in a well-
mixed box. A similar derivation based on NBL method
2 produces NBL-IDT method 2
F ¼1
X ðdu � dsÞ
Z z
0
uCdðdu � ddÞ dz; ð10Þ
where du is the mean upwind or background methane
d13C. If the background boundary-layer air is well mixed
and the isotope ratios of source and background are
quantified, then NBL-IDT method 2 enables a flux
estimation based on a single vertical profile of wind and
of methane isotope and mixing ratio downwind of the
source.
2.4. New Zealand field measurements
Isotope measurements were made on five nights
subsequently referred to as N1–N5, at two sites
(Table 1 and Fig. 1). The Aorangi Farm site (401210S,
1751290E) is in the fertile Manawatu lowlands of the
North Island, 9 km WNW of Palmerston North (and
21 km E of the Tasman sea). The site comprised three
contiguous flat and level paddocks of total length
B700m and width 150m. Westerly wind predominated
through all three measurement periods over a 1 km
upwind fetch of sheep with a stocking density of
approximately 20 ha�1. Comparisons are made with
the collocated flux-gradient measurements reported by
Judd et al. (1999) and CBL budget measurements by
Wratt et al. (2001) during Aorangi 1. The measurements
Table 1
Measurement periods at Aorangi and Springston Farms
Campaign Date and season Summary of measurementsa Nocturnal isotope dilution
experiment (Night number)
Aorangi 1 3–11 Apr 1997 (Autumn) SBb, NBL, FGb, CBLc 9–10 Apr (N1)
Aorangi 2 20–26 Sep 1997 (Early
Spring)
SB, NBL 20–21 Sep (N2)
25–26 Sep (N3)
Aorangi 3 21–27 Nov 1997 (Late
Spring)
FG
Springston 19 Mar–1 Apr 1999 (Early
Autumn)
SB, NBL, GR-FTIR 25–26 Mar (N4)
27–28 Mar (N5)
aSee Section 2.5, SB: sheep breath, NBL: nocturnal boundary layer, FG: gradient micrometeorological measurement, CBL: daytime
convective boundary layer budget, GR-FTIR: Gas ratio technique using open path Fourier transform infrared spectroscopy.b Judd et al. (1999).cWratt et al. (2001).
Fig. 1. New Zealand measurement sites.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784666
at Springston (43138.80S, 172124.70E) were on a com-
mercial farm on the Canterbury plains of the South
Island. An area of approximately 10 ha (380m� 250m)
was populated by 92 pregnant ewes maintained in sub-
paddocks to distribute the sheep at a more even paddock
density of around 10 sheep per hectare. In the summer
months before the experiment, the paddock was in
drought and stock were supported with supplemental
feed. Substantial grass regrowth occurred with rains
during the month before the measurements reported
here. Comparisons were made with other techniques as
summarised in Table 1.
The presence of a nocturnal surface inversion was
established from wind, temperature and humidity
profiles to 200m measured from kytoon (AIR Inc.
model TS-3A-SP tethered balloon) soundings through
each night. During periods when the winds were too
strong (>10m s�1) for using the kytoon, low-lift
airsondes (AIR Inc. model AS-1C-PTH) were used. At
Springston, a doppler acoustic sounder (Remtech PA2)
was also operated at 20m vertical resolution to give
20min average wind to 400m as well as estimated
mixing height. Gas samples were collected in 2L (or
1.5 L) glass flasks (Glass Expansion Pty, Australia)
which were flushed several times then filled with ambient
air over about 10min to approximately 200 kPa using
12V battery powered diaphragm pumps (Thomas
Industries Inc.). Glass flasks were used in preference to
Tedlar bags and allowed collection of pressurised
samples. Additionally, the samples had to be stored
for several months before isotope analysis and there is
evidence (Okada and Tezuka, 1989) of significant
isotope shifts over 1 month with samples in Tedlar bags
whereas glass has shown no such shift.
On N1, vertical gradients of methane mixing ratio
were measured at the downwind site and used in the
vertical flux integrations. On subsequent nights, the
vertical structure in the lowest 20m was examined by
kytoon temperature and humidity sounding. In the
absence of discontinuities, vertical profiles of methane
were estimated from measurement at a single 2m height
and the flux/profile relationship for the expected
magnitude of flux F :
F ¼ �k2 qu
qðz � dÞqC
qðz � dÞðfHfM Þ�1;
where k is the von Karman constant, d is zero plane
displacement and ðfHfM Þ�1 is a generalized stability
factor. Under stable conditions (�0.1oRio1) this
factor is equivalent to (1–5Ri)2 (Thom, 1975). In this
study, stability was determined by the bulk Richardson
number Ri ¼ ðg=%yÞð%y2 � %y1Þðz2 � z1Þ=ðu2 � u1Þ2 in the
lowest 20m of the atmosphere; %y2 and %y1 denote the
mean potential temperature measured along with mean
windspeeds u from mast or kytoon at heights z2 and z1; g
is acceleration due to gravity. In addition, 2 or 3
upwind/background gas samples were collected during
the night.
2.5. Comparison with other methods
In comparing different methods (Table 1), measure-
ments of the breath emissions of individual animals
using the SF6 tracer technique (Lassey et al., 1997), in
conjunction with stocking density, were considered as
the standard reference ‘‘bottom-up’’ flux. These sheep
breath measurements were made during all NBL
measurement periods. In order to compare nocturnal
measurements with daytime or 24 h measurements, any
systematic diurnality in flux needs to be accounted for.
Our results contrast those of Judd et al. (1999) and show
that diurnality in the methane emission of pasture-
grazing sheep is complex and not predictable (Section
4.3). We therefore have not made corrections to our
NBL data on the basis of diurnality. At Aorangi,
comparisons with average flux-gradient and aircraft
CBL profile measurements were made. At Springston,
NBL measurements of gas concentration over a 100m
open-path, were made by FTIR spectrometer (Kelliher
et al., 2002). From these measurements, fluxes were
estimated through both mixing ratio increase (NBL
method 1) and an atmospheric gas ratio technique (GR-
FTIR) (Kuhlmann et al., 1998). The latter technique
uses the multi-gas capability of the FTIR spectrometer,
to calculate an indirect estimate of the methane flux
from the ratio of CO2 and CH4 mixing ratios and an
independent measurement of CO2 respiration flux at
night. An assumption is made that the sheep methane
emission and CO2 respiration flux are both homoge-
neously distributed surface sources. The nocturnal
surface emission of CO2 from soil and plant respiration
was measured at sample points around the paddock
using a portable chamber, covering an area of 0.008m2,
and infrared gas analyser system (coupled SRC-1 and
EGM-3, PP Systems, Ltd., Hitchin, Hertfordshire,
England).
2.6. Modelling methane sheep breath dispersion
For each night, the Ausplume Gaussian plume
dispersion model (VEPA, 1986) was used to simulate
sheep breath dispersion and predict the methane vertical
profile immediately downwind of the paddock. Sheep in
the immediate paddocks were modelled as a uniform
grid of point sources. At Aorangi, sheep in distant
paddocks upwind were modelled as area sources.
Meteorological input data were generated from the
measured hourly sonde and kytoon vertical profiles and
output concentration fields were produced at hourly
resolution.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–4678 4667
3. Analyses and results
3.1. Methane isotope mixing lines
Fig. 2 shows the isotopic mixing lines for each night.
The correlation between d13C and inverse concentration
is weakest for the Springston measurements where the
local sheep methane was a less dominant source than at
Aorangi for two reasons. Firstly, both the paddock
sheep density and the regional sheep density was lower
than Aorangi. Secondly, nearby local agricultural burn-
offs were observed during the time of the experiment
with the possible introduction of an additional methane
source. The lighter intercept value for N4 (Table 2) is
evidence of the latter. However, the absence of elevated
CO in samples indicated that major contamination from
agricultural burning was unlikely. The average intercept
value of d13C (excluding N4 having low r2) was about
�64%, and close to our expected value for sheep
methane of �62% based on some individual measure-
ments of eructated gases (see Section 4.2). The
uncertainty in this estimate as calculated from the
standard error of the regression intercept was around
1% for nights with the greatest methane build up.
Table 2 summarises the mixing lines and makes a
comparison with regional baseline measurements made
at Baring Head (BHD 41124.60S, 174152.20E) baseline
monitoring station (Lowe et al., 1994). BHD provides
methane data sampled in an unpolluted marine bound-
ary layer and can be regarded as characterising
the background methane arriving at New Zealand.
The comparison (in the last column) is made between
the d13C predicted by the mixing line at the baseline
(BHD) mixing ratio at the time of these measurements
(Column 6: BHD mixing ratio) with the actual baseline
d13C measured at Baring Head (Column 7: d13C(BHD))
at the same time. Assuming a homogeneous back-
ground, a significant difference between d13C(fitted) andd13C(BHD) indicates that there has been some degree of
contamination of the methane upwind of the sample
4.00e-4 4.50e-4 5.00e-4 5.50e-4 6.00e-4
1/[CH4] (ppbv-1)
-53
-51
-49
-47
δ13
C(‰
)
N1 (Aorangi 9-10 Apr 1997 downwind)N1 (Aorangi 9-10 Apr 1997 upwind)N2 (Aorangi 20-21 Sep 1998)N3 (Aorangi 25-26 Sep 1998)N4 (Springston 25-26 Mar 1999)N5 (Springston 27-28 Mar 1999)
(2500 ppbv) (1670 ppbv)(1818 ppbv)
Fig. 2. Mixing lines of d13C methane. Each line represents the relationship between isotope ratio and mixing ratio for one night.
Table 2
Isotope mixing line summary and concurrent baseline methane at Baring Head (BHD)
Night
no.
d13C (source)
(%)
Std. Error
d13C (source)
(%)
Gradient
(% ppbv)
Correlation
(r2)
BHD CH4
mixing ratio
(ppbv)
d13C (BHD)
(%)
d13C (fitted)–
d13C (BHD)
(%)
N1 �63.4 2.2 26440 0.72 1705 �47.1 �0.8
N2 �66.1 0.9 33290 0.97 1740 �47.1 0.1
N3 �63.9 1.2 29000 0.95 1740 �47.1 �0.2
N4 �52.6 1.3 9220 0.38 1720 �47.03 �0.2
N5 �61.3 4.1 24710 0.59 1720 �47.03 0.1
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784668
site, or that upwind trajectories deliver different air
masses to the two sites. On most days d13C(fitted) is
within 0.2% of d13C(BHD). However, for N1 the
predicted value is 0.8% lighter (less enriched in 13C)
than d13C(BHD). There were several pieces of evidence
that on N1 at Aorangi there was a complex situation
with an airmass containing a significant mixing ratio of
isotopically lighter methane aloft. Firstly, the time series
of concentration and d13C showed a significant spike of
high concentration light methane around the time of the
break-up of the nocturnal inversion. Secondly, around
noon on 10 April 1997 an aircraft overflight at Aorangi
reported by Wratt et al. (2001) recorded methane mixing
ratios in the onshore winds at the site that were slightly
greater than those simultaneously recorded at heights
between 4 and 1m from ground-based measurement.
Finally, Wratt et al. (2001) also found significantly
greater methane concentrations at 1500m height off-
shore to the west than were present 20 km inland at
Aorangi at 1500m in westerly winds. In this case the
source of the methane aloft is uncertain although Wratt
et al. (2001) did find other instances of methane
draining off the west coast at night to be advected back
in again in a developing daytime westerly breeze. This
example and others in the general discussion below
indicate that spatial variability and complexities of
additional sources can be an important factor limiting
the ease of application of all boundary-layer mass
balance methods.
3.2. Nocturnal trials and flux estimation
The largest methane build-up was observed on N2
with light winds and strong stability. Table 3 sum-
marizes the meteorological conditions during periods of
gas build-up suitable for nocturnal flux determination.
Conditions on the very stable N2 are illustrated in
Fig. 3 which shows the potential temperature y differ-
ence between 100, 20 and 4m and the surface and the
bulk Richardson number calculated between 20 and 4m
and the surface. Fig. 4 shows the time series of methane
mixing ratio and d13C for this night. Periods with the
greatest atmospheric stability showed the greatest
increase in concentration and reduction in d13C. The
most dramatic instance was around 22:00 when con-
centration was approximately 1.5 times background.
Periods of rapid mixing ratio growth in the early part of
the evening in the most stable periods of the NBL were
used to give flux estimates by NBL method 1 and NBL-
IDT method 1. Table 4 gives the flux estimates
compared with those of other techniques. The sheep
breath measurements are based on a sub-flock sample of
12 sheep with results given on a per unit area and per
head basis. In the light winds of N2, NBL estimates were
in good agreement with the scaled sheep-breath esti-
mates. In contrast, significant advective loss on N3 has
reduced the NBL estimate to about half that from the
sheep breath.
On more windy nights (N1, N4 and N5), flux
estimates by NBL method 2 and NBL-IDT method 2
were made. Figs. 5 and 6 show times lines of mixing
ratio and d13C for N4 and N5, respectively. With
simultaneous measurement at four sites, large nocturnal
variability is apparent. On N4, the general trend in
mixing ratio increase is similar at three of the four sites
at around 2.5–3 ppbv h�1 over the whole period. The
most dramatic build up was at the downwind (S) site
through to the break up of the nocturnal inversion. On
N5, a similar mixing ratio increase was observed at all
four sites in the first part of the night from 17:30 to
01:00. In the second half of the night, mixing ratio
dropped; this drop is likely to have been caused by
the wind backing slightly in the middle of the night
from around 501 to 301 so that the downwind sample
site was displaced from the centre of the sheep methane
plumes.
On these two nights at Springston (Figs. 5 and 6),
downstream concentrations were not always largest and
upstream concentrations were not always smallest. d13Cshowed consistent anomalies with upstream data some-
times being more depleted in 13C compared to values
downstream. Although the winds were fairly steady on
these nights with a persistent north-easterly flow at
the surface, there was some sheep movement though the
night which could have accounted for much of the
variability. In general though, Table 4 shows the NBL
Table 3
Summary of nocturnal meteorology and maximum methane concentration
Night Average wind (10m)
(m s�1)
Potential
temperature gradient
(20–0m) (Km�1)
Bulk Richardson
number (4m)
Maximum nocturnal
[CH4] at 2m (ppbv)
N1 5 +0.04 0.03 (4–2m) 1809
N2 1.3 +0.09 1.02 2527
N3 2.0 +0.05 0.01 1975
N4 5.0 No data 0.02 1831
N5 4.3 +0.02 0.04 1875
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–4678 4669
methods gave good agreement to within 720% of the
‘‘bottom up’’ sheep breath assessments excluding N3
which underestimated flux due to unexpected advective
loss and N4 where estimates were likely to have been
affected by a biomass burning contamination.
4. Discussion
The NBL static techniques (NBL method 1 and NBL-
IDT method) 1 worked well with a well defined
nocturnal boundary layer, strong stability and light
18:00 21:00 00:00 03:00 06:00
Time of day
-2
-1
0
1
2
3
4
5P
oten
tial t
empe
ratu
re d
iffer
ence
(K
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Bul
k R
icha
rdso
n nu
mbe
r
dθ (4m) dθ (20m) dθ (100m) Ri (4m) Ri (20m)
Fig. 3. Time course of atmospheric stability for N2 (20–21 September 1997).
18:00 21:00 00:00 03:00 06:00
Time of day
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
CH
4(p
pbv)
-54.0
-53.5
-53.0
-52.5
-52.0
-51.5
-51.0
-50.5
-50.0
-49.5
-49.0
-48.5
-48.0
-47.5
-47.0
δ13C
(%)
Measured CH4 mixing ratio downwind (ppbv) δ13C downwindInversion height (m)Ausplume CH
4 mixing ratio (ppbv, 2m height)
0102030405060708090
100
Inve
rsio
nhe
ight
(m)
Fig. 4. Time course of methane mixing ratio and d13C for N2 (20–21 September 1997).
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784670
winds below 2m s�1. The mass balance technique NBL-
IDT method 2 was applied to give acceptable estimates
under windier conditions. Table 4 also shows that
agreement between all the techniques is generally good
to within a factor of 2, and usually within B20%. We
have not expressed the magnitude of the error in
individual NBL trial assessments although we estimate
uncertainty in the flux estimate to be of the order of
720% from analysis of the propagation of error. The
contribution from the analytical error in the methane
mixing ratio and d13C measurements is insignificant
(typically 0.1% and 0.06%, respectively). The overall
error is dominated by the variability that is present due
to meteorological factors, such as poor mixing, which
causes horizontal variability in the concentration fields
from an inhomogeneous source and through difficulty in
measuring mixing height z: Some of the source
inhomogeneity is due to differences in emission rate of
individual animals. The coefficient of variation (1s s.d.)
of individual animal measurements is quite high at up to
25%. However, a far greater source of spatial inhomo-
geneity is caused by the tendency of the sheep to flock. A
careful sampling strategy is required to ensure that this
natural variability is captured and spatially averaged by
the NBL techniques. In order to collect upwind and
downwind air samples that are representative, we
analysed several samples of each collected along
horizontal crosswind transects to be confident that air
sampling uncertainties were minimal.
The IDT techniques are susceptible to error that could
arise through unrepresentative sampling. With upwind–
downwind contrasts in d13C as small as 0.2%, it is
important that errors in this contrast be minimised. For
example, an error of around 0.2% in the downwind d13Ccould give rise to a 50% or more error in the inferred
flux. The source-air contrast in d13C is much larger at
B15%, so that an inaccuracy of 2–3% in the value
adopted for the sheep source translates to around –20%
to +30% error in the inferred flux. Table 4 records good
agreement between the conventional NBL techniques
and the IDT methods, and their concurrence with
individual sheep measurements, on nights with favour-
able meteorology. This conformity suggests that the air
sampling did not unduly limit confidence in the inferred
fluxes. Nevertheless, for the benefit of future applica-
tions, we discuss specific sources of uncertainty in more
detail.
4.1. Uncertainties in mixing height and comparisons with
Ausplume model results
As with the measurements, modelled methane profiles
were highly sensitive to the stability class used in the
model. The effect is seen on N2 in the two peaks in the
Table 4
Comparison of flux estimates
Location Night
number
Flock
density
(ha�1)
Sheep breath
(mgm�2 d�1)
(g head�1 d�1
in brackets)
Mixing
height
(m)
NBL
method
NBL flux
(mgm�2 d�1)
Flux/gradient
(mgm�2 d�1)
Other
(mgm�2 d�1)
Aorangi N1 20 39.072.8 20 NBL 2 38 46a 73733b
(19.571.4) NBL-IDT 2 42
Aorangi N2 20 62.673.1c 20 NBL 1 56 NA NA
(31.371.6) NBL-IDT 1 59
Aorangi N3 20 62.673.1c 20 NBL 1 32 NA NA
(31.371.6) NBL-IDT 1 28
Aorangi
21–27 Nov
1997
— 20 68.473.1c — None NA 96766d NA
(34.271.6)
Springston N4 10 33.472.4c 2 NBL 2 79 NA NA
(35.272.5) NBL-IDT 2 29
Springston N5 10 33.472.4c 2 NBL 2 40 NA 20e
(35.272.5) NBL-IDT 2 38 39712f
Units (mg (CH4)m�2 d�1), NA: not available.
aFlux gradient (Judd et al., 1999).bDaytime aircraft CBL budget (Wratt et al., 2001).cAverage for week on both per area and per animal basis7standard error of mean (n=12).d22 half hour flux-gradient (between 4 and 2m) tedlar bag samples.eNBL method 1 by FTIR 29–30 March 1999 assuming 20m mixing height.fNBL gas ratio (GR-FTIR) CO2/CH4 29–30 March 1999.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–4678 4671
Ausplume prediction (Fig. 4) which correspond to
periods when the windspeed was less than 1m s�1 in a
highly stable boundary layer. The general trends of
methane mixing ratio are similar in model and
measurement although details in the timing are
different; e.g. the dip around 22:00 was not seen at
19:00 22:00 01:00 04:00 07:00 10:00 13:00
Time of day
1700
1750
1800
1850
1900
[CH
4](p
pbv)
NE site (upstream) Centre paddockS site (downstream) SW site (downstream)
19:00 22:00 01:00 04:00 07:00 10:00 13:00
Time of day
-48.0
-47.8
-47.6
-47.4
-47.2
-47.0
-46.8
-46.6
δ13
C(‰
)
NE site (upstream) Centre paddock
S site (downstream) SW site (downstream)
Fig. 5. Time course of methane mixing ratio and d13C for N4 (25–26 March 1999).
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784672
the measurement frequency. In general, for all nights
(not shown) the modelled methane concentrations
were significantly less than the measured values. One
possible explanation may be the unrealistic model
assumption of a uniform flux that does not recognise
sheep flocking, a second could be poor performance of
the plume model under highly stable light wind
conditions.
With the strong vertical gradients in methane
that exist within the stable NBL, a depth-integrated
16:00 19:00 22:00 01:00 04:00 07:00 10:00 13:00
Time of day
1700
1750
1800
1850
1900
[CH
4]
(ppb
v)
NE site (upstream)
Centre paddock
S site (downstream)
SW site (downstream)
Ausplume SW prediction
19:00 22:00 01:00 04:00 07:00 10:00 13:00
Time of day
-48.0
-47.8
-47.6
-47.4
-47.2
-47.0
-46.8
-46.6
δ 13C
(‰)
NE site (upstream)
Centre paddock
S site (downstream)
SW site (downstream)
Fig. 6. Time course of methane mixing ratio and d13C for N5 (27–28 March 1999).
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–4678 4673
approach to all of the NBL methods provided the best
accuracy. Without its direct measurement, mixing height
is hard to obtain and introduces a major source of error.
Other scalars can be used to quantify mixing. We
estimated mixing height from discontinuities in vertical
soundings of potential temperature and water vapour
mixing ratio. As previously discussed, the best agree-
ment in NBL method 1 and NBL-IDT method 1 with
the sheep breath technique was found on N2. When
compared with modelled vertical profiles of methane on
N2, the extent of vertical diffusion predicted by
Ausplume (Fig. 7a) throughout the surface 20m layer
was similar to the mixing depth determined from kytoon
measurements. The kytoon showed a very stable layer to
20m in the first half of the night, containing a uniform
water vapour mixing ratio as an indicator of the well
mixed nature of this surface layer. A well defined mixing
height and strong near surface stability on this night led
to the best agreement between measurement and model.
In contrast, on nights N4 and N5, the mixing height
was less clearly defined and other sources of methane
were present above the NBL from daytime stubble
burning. On these nights, the Doppler sounder did not
prove useful. The predicted mixing height from acoustic
echo was typically of the order of 100m. However, from
methane vertical profiles, we found that an accurate flux
estimate, as compared with the sheep breath technique,
was achieved with vertical integration to 2m with the
restricted upwind extent of sheep at Springston of
B300m compared to B1 km extent at Aorangi. Fig. 7b
shows the modelled vertical profiles with concentrations
falling to background at around z ¼ 10m. This model,
using standard plume dispersion formula, broadly
agrees with the limited vertical dispersion observed
and predicts sz; the vertical dispersion parameter is
about 0.3 at 20m downwind from the Springston
paddock under the most stable conditions (Hanna
et al., 1982). In general, accurate information on mixing
height is important and this requirement is best met
through vertical profile measurement.
4.2. d13C signature of sheep breath
The range of d13C in eructed methane will be
dependent on d13C of the feed and fractionation
occurring in methanogenesis. Our NBL flux estimates
are sensitive to the d13C contrast between the source and
atmosphere. Uncertainties should be able to be kept
small compared to this contrast and it is therefore
important to measure the source signature for the
environment under study. Comparing feeds, the carbon
in C3 photosynthesis plants is relatively depleted in 13C
compared with C4 photosynthesis plants. Rust (1981)
found correspondingly that d13C of CH4 from ruminants
with a C3 diet ranged from �58.1% to �76.0%compared with those on a C4 diet of �45.4% to
�52.8%. However, there can be significant differences,
even within the C3 group. Bilek et al. (2001) show how
differing d13C in C3 hay/concentrate feeds leads to
corresponding differences in eructed d13CH4. Thus a
range of values appear in the literature. With the typical
C3 diet of temperate grasslands, Wahlen (1994) shows a
range between �67% and �72%, Gupta et al. (1996)
cite –62%, Hein et al. (1997) cite –6273% and Quay
et al. (1999) summarise data at –6075%. We have
measured for a typical New Zealand pasture diet of rye
grass (Lolium perenne) and white clover (Trifolium
repens), during Aorangi 2 and Aorangi 3 campaigns a
range of d13CH4 in 24 breath samples between �58.2%and �66.8% with a mean of �62.870.6 (s.e.m.). A
value of –62% has been used in our calculations.
4.3. Diurnality in sheep eructation
In using a nocturnal method we were concerned for
any bias that might occur due to diurnal patterns in
sheep emission. Judd et al. (1999) report on two
experiments where the daytime or afternoon emission
flux is 40% or 50% larger than the 24 h average flux.
These measurements were based on a small subset of the
flock at Aorangi assessed using the SF6 tracer method.
We repeated the small trial to test diurnal variability of
individual animals using the SF6 tracer technique at
Springston. Conversely, for six sheep over a period of 5
days, we found an average 1.2570.30 g head h�1 during
the working day (9:30–17:00) compared with
1.4470.35 g head h�1 (i.e. 15% more) for the period
(17:30–08:30). Fig. 8 shows that four of the six sheep
tested had significantly lower daytime emissions. During
the experiment, sheeps were observed eating and resting
during both day and night. Denmead et al. (2000)
summarised the results of other studies that show
methane emission is greatest during rumination and
rest and least when animals are actively grazing. In
general, the diurnal pattern of methane emissions is
linked closely to animal feeding patterns, which in turn
may be linked to times of availability of fresh feed. Our
trial results highlighted that there is the possibility of
significant differences between sites and flocks, such that
daytime emissions may not always be largest. For this
reason, we have not applied any correction in scaling to
a 24 h average from our nocturnally based estimates.
5. Conclusions
As with all the micrometeorological approaches, care
needs to be taken in sampling representative concentra-
tions and this will always be a difficulty when these
methods are applied to an inhomogeneous source
presented by free-roaming animals within a poorly
mixed nocturnal boundary layer. With the flocks we
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784674
12:00 18:00 00:00 06:00
Time of day
0
20
40
60
Hei
ght
(m)
174
2
1745
1750
1800 1800
1742 ppbv 1745 ppbv 1750 ppbv1800 ppbv 1850 ppbv 1900 ppbv
1950 ppbv 2000 ppbv 2050 ppbv
2100 ppbv 2150 ppbv 2200 ppbv
Sonde mixing height Bulk Ri
-0.2
0.3
0.8
1.3
1.8
2.3
Bul
k R
icha
rdso
n nu
mbe
r (4
m)
12:00 15:00 18:00 21:00 00:00 03:00 06:00 09:00 12:00
Time of day
0
20
40
60
Hei
ght (
m)
1721.0
1725.0
1730
.0
1730
.0
1721 ppbv1725 ppbv1730 ppbv1735 ppbv
Sonde mixing heightRi
-0.05
-0.01
0.03
0.07
Ric
hard
son
Num
ber
(3m
)
(a)
(b)
Fig. 7. (a) Modelled methane vertical profiles 10m downwind of the Aorangi paddock on N2. Background concentration 1740 ppbv.
(b) Modelled methane vertical profiles 10m downwind of the Springston paddock on N5. Background concentration 1720 ppbv.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–4678 4675
have measured, the observed ‘between animal’ varia-
bility in emission was quite large (coefficient of variation
up to 25%). If high- and low-emitters are evenly
distributed within the flock, an advantage of the
proposed method is that it performs some spatial
averaging. In this assessment of the IDT for ruminant
methane emission measurement, the advantages of
providing a second somewhat independent and specific
assessment of sheep methane emission at the paddock
scale need to be considered against the added complica-
tion and expense of d13C determination. However, as the
technology develops for precise methane isotopic
analysis of small air samples, more widespread use of
this isotope dilution/mass balance approach is likely.
The IDT technique requires a careful strategy to ensure
representative samples are collected, especially with the
advecting technique where the upwind/downwind con-
trasts in d13C are small. In this study of nocturnal
boundary layer budget techniques, good agreement in
flux estimates to within a factor of 2 was achieved
compared with the ‘‘bottom-up’’ individual animal
measurements using the SF6 tracer technique. This
factor was closer to unity under ideal sampling
conditions and is considered to be excellent given the
many sources of uncertainty discussed above.
Acknowledgements
This research was funded by the New Zealand
Foundation for Research, Science and Technology
(FRST) contract CO9636 from the Landcare Research
programme ‘‘Greenhouse gas emissions from the terres-
trial biosphere’’. The FTIR measurements and develop-
ment of GC-IRMS techniques were made by the
National Institute of Water and Atmospheric Research
through ‘non-specific output’ funding.
Thanks are due to the many who made these
measurements possible including Ross Martin, Robert
Knobben, Andrew McMillan and Errol Lewthwaite at
NIWA and to Bill Ussler (currently University of North
Carolina), for initiating the work. Support was provided
at Aorangi Farm by AgResearch and thanks are due to
Alan Garrett and Brian Alison for accommodating our
research at Hollymount Farm, Springston. Finally we
thank the two anonymous reviewers for their useful
comments.
References
Andreas, E.L., Gosz, J.R., Dahm, C.N., 1992. Can long-path
FTIR spectroscopy yield gas flux measurements through a
variance technique. Atmospheric Environment 26A,
225–233.
Bergamaschi, P., Lubina, C., K .onigstedt, R., Fisher, H.,
Veltkamp, A.C., Zwaagstra, O., 1998. Stable isotope
signatures of methane from European landfill sites. Journal
of Geophysical Research 103, 8251–8265.
Beswick, K.M., Simpson, T.W., Fowler, D., Choularton, T.W.,
Gallagher, M.W., Hargreaves, K.J., Sutton, M.A., Kaye,
A., 1998. Methane emissions on large scales. Atmospheric
Environment 32, 3283–3291.
1 2 3 4 5 6
Sheep identification number
0.5
0.9
1.3
1.7
Ani
mal
em
issi
on (
g he
ad-1
h-1)
Day: 09:30 to 17:00 Night 17:30 to 08:30
Fig. 8. Hourly individual sheep methane emission for daytime and nighttime periods.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784676
Bilek, R.S., Tyler, S.C., Kurihara, M., Tagi, K., 2001.
Investigation of cattle methane production and emission
over a 24-hour period using measurements of d13C and dDof emitted CH4 and rumen water. Journal of Geophysical
Research 106, 15405–15413.
Choularton, T.W., Gallagher, M.W., Bower, K.N., Fowler, D.,
Zahniser, M., Kaye, A., 1995. Trace gas flux measurements
at the landscape scale using boundary-layer budgets.
Philosophical transactions of the Royal Society of
London Series A: Physical Sciences and Engineering 351,
357–369.
Craig, H., 1957. Isotopic standards for carbon and oxygen
and correction factors for mass spectrometric analysis of
carbon dioxide. Geochimica et Cosmochimica Acta 12,
133–149.
Denmead, O.T., 1995. Novel meteorological methods for
measuring trace gas flux. Philosophical transactions of the
Royal Society of London Series A: Physical Sciences and
Engineering 351, 383–396.
Denmead, O.T., Harper, L.A., Freney, J.R., Griffith, D.W.T.,
Leuning, R., Sharpe, R.R., 1998. A mass balance method
for non-intrusive measurements of surface-air trace gas
exchange. Atmospheric Environment 32, 3679–3688.
Denmead, O.T., Leuning, R., Griffith, D.W.T., Jamie, I.M.,
Esler, M.B., Harper, L.A., Freney, J.R., 2000. Verifying
inventory predictions of animal methane emissions with
meteorological measurements. Boundary-Layer Meteorol-
ogy 96, 187–209.
Ferretti, D.F., Lowe, D.C., Martin, R.J., Brailsford, G.W.,
2000. A new gas chromatograph-isotope ratio mass spectro-
metry technique for high-precision, N2O-free analysis of
d13C and d18O in atmospheric CO2 from small air samples.
Journal of Geophysical Research 105, 6709–6718.
Gallagher, M.W., Choularton, T.W., Bower, K.N., Stromberg,
I.M., Beswick, K.M., Fowler, D., Hargreaves, K.J., 1994.
Measurements of methane fluxes on the landscape scale
from a wetland area in North Scotland. Atmospheric
Environment 28, 2421–2430.
Gupta, M., Tyler, S., Cicerone, R., 1996. Modeling atmospheric
d13CH4 and the causes of recent changes in atmospheric
CH4 amounts. Journal of Geophysical Research 101,
22923–22932.
Hanna, S.R., Briggs, G.A., Hosker, Jr., R.P., 1982. National
Technical Information Series: Handbook on Atmospheric
Diffusion, TIC-11223 ed. USDOE, Sprinfield, VA.
Hein, R., Crutzen, P.J., Heiman, M., 1997. An inverse modeling
approach to investigate the global atmospheric methane
cycle. Global Biogeochemical Cycles 11, 43–76.
Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der
Linden, P.J., Dai, X., Maskell, K., Johnson C.A. (Eds.),
2001. Climate Change 2001: The Scientific Basis.
Contribution of Working Group 1 to the Third Assessment
Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge, United
Kingdom.
Johnson, K., Huyler, M., Westberg, H., Lamb, B., Zimmer-
man, P., 1994. Measurement of methane emission from
ruminant livestock using a SF6 tracer technique. Environ-
mental Science and Technology 28, 359–362.
Judd, M.J., Kelliher, F.M., Ulyatt, M.J., Lassey, K.R., Tate,
K.R., Shelton, D., Harvey, M.J., Walker, C.F., 1999. Net
methane emissions from grazing sheep. Global Change
Biology 5, 647–657.
Kelliher, F.M., Reisinger, A.R., Martin, R.J., Harvey, M.J.,
Price, S.J., Sherlock, R.R., 2002. Measuring nitrous oxide
emission rate from grazed pasture using Fourier-transform
infrared spectroscopy in the nocturnal boundary layer.
Agricultural and Forest Meteorology 111, 29–38.
Kuhlmann, A.J., Worthy, D.E.J., Trivett, N.B.A., Levin, I.,
1998. Methane emissions from a wetland region within the
Hudson Bay Lowland: an atmospheric approach. Journal of
Geophysical Research 103, 16009–16016.
Lassey, K.R., Ulyatt, M.J., Martin, R.J., Walker, C.F.,
Shelton, I.D., 1997. Methane emissions measured directly
from grazing livestock in New Zealand. Atmospheric
Environment 31, 2905–2914.
Leuning, R., Baker, S.K., Jamie, I.M., Hsu, C.H., Klein, L.,
Denmead, O.T., Griffith, D.W.T., 1999. Methane emission
from free-ranging sheep: a comparison of two measurement
methods. Atmospheric Environment 33, 1357–1365.
Lowe, D.C., Brenninkmeijer, C.A.M., Brailsford, G.W.,
Lassey, K.R., Gomez, A.J., 1994. Concentration and 13C
records of atmospheric methane in New Zealand and
Antarctica: Evidence for changes in methane sources.
Journal of Geophysical Resarch 99, 16913–16925.
Lowe, D.C., Allan, W., Manning, M.R., Bromley, T.,
Brailsford, G., Ferretti, D., Gomez, A., Knobben, R.,
Mei, Z., Moss, R., Koshy, K., Maata, M., 1999. Shipboard
determinations of the distribution of 13C in atmospheric
methane in the Pacific. Journal of Geophysical Research
104, 26125–26135.
Mahrt, L., 1999. Stratified atmospheric boundary layers.
Boundary-layer Meteorology 90, 375–396.
Ministry for the Environment, 1997. Climate change—the New
Zealand response II. New Zealand’s Second National
Communication under the Framework Convention on
Climate Change. Ministry for the Environment, Wellington,
New Zealand, 191p.
New Zealand Climate Change Project, 2002. National inven-
tory report—greenhouse gas inventory 1990–2000. New
Zealand Climate Change Project, Wellington, 190p.
Okada, S., Tezuka, M., 1989. Some problems of estimation of
Carbon Stable Isotope Ratio of methane in natural gases
and in cuttings gases. Journal of the Japanese Association
for Petrolium Technology 54, 1.
Quay, P., Stutsman, J., Wilbur, D., Snover, A., Dlugokencky,
E., Brown, T., 1999. The isotopic composition of atmo-
spheric methane. Global Biogeochemical Cycles 13,
445–11461.
Rust, F., 1981. Ruminant methane delta (13C/12C)
values: Relation to atmospheric methane. Science 211,
1044–1046.
Stevens, C.M., Engelkemeir, A., 1988. Stable carbon isotopic
composition of methane from some natural and anthro-
pogenic sources. Journal of Geophysical Research 93,
725–733.
Thom, A.S., 1975. Momentum, mass and heat exchange of
plant communities. In: Monteith, J.L. (Ed.), Vegetation and
the Atmosphere. Academic Press, London.
VEPA, 1986. The Ausplume Gaussian Plume Dispersion
Model. Victorian Envionmental Protection Authority,
Melbourne, Victoria.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–4678 4677
Wahlen, M., 1994. Carbon dioxide, carbon monoxide and
methane in the atmosphere: abundance and isotopic
composition. In: Lajtha, K., Michener, R.H. (Eds.), Stable
Isotopes in Ecology and Environmental Science. Blackwell
Scientific Publications, Oxford.
Wratt, D.S., Gimson, N.R., Brailsford, G.W., Lassey, K.R.,
Bromley, A.M., Bell, M.J., 2001. Estimating regional
greenhouse gas emissions from agriculture using aircraft
measurements of concentration profiles. Atmospheric
Environment 35, 497–508.
M.J. Harvey et al. / Atmospheric Environment 36 (2002) 4663–46784678