Evaluating Dispersion Modeling Options to Estimate Methane ...

Published January 8, 2015

Journal of Environmental Quality

TECHNICAL REPORTS Atmospheric Pollutants and Trace Gases

Evaluating Dispersion Modeling Options to Estimate Methane Emissions from Grazing Beef Cattle Sean M. McGinn,* Thomas K. Flesch, Trevor W. Coates, Ed Charmley, Deli Chen, Mei Bai, and Greg Bishop-Hurley

E

stimates of enteric methane (CH4) losses from graz-

Abstract

ing cattle on pasture are needed in support of accurate emission predictions for national greenhouse gas (GHG) inventories and to evaluate potential CH4 reduction strategies that influence the sustainability of the cattle industry. This information is especially required in regions where enteric CH4 is a major source of GHGs. For example, in Australia it is estimated that enteric CH4 emission accounts for 54.7 Mt CO2–e (CO2 equivalent) out of a total emission of 552.7 Mt CO2–e, which is about 10% of all the GHG sources (Australian National Greenhouse Accounts, 2013). Uncertainty in these emissions exists because enteric CH4 emissions from grazing cattle are difficult to measure directly. In vivo techniques, including the use of chambers and enteric tracer gas (Makkar and Vercoe, 2007), can provide insight into the magnitude and variability of CH4 emissions from individual animals. However, only micrometeorological approaches (Harper et al., 2011) can provide a nonintrusive measure of enteric CH4 losses. The application of micrometeorological techniques in a grazing environment is complicated by the size of cattle paddocks, the movement of animals, and the uneven distribution of animals across the paddock. The inverse dispersion technique is a micrometeorological technique that has shown potential for studying grazing animals (Flesch et al., 2005; Harper et al., 2011; McGinn et al., 2013). Gas concentration sensors are placed downwind of the animals, and their emission rate is calculated from the increase in gas concentrations over the background concentration as estimated by a dispersion model. The key to the emission calculation is an atmospheric dispersion model that relates the concentration rise (C) to the animal emission rate (Q). Although the inverse dispersion technique is straightforward in principle, there are practical questions in its application to grazing environments. An important decision when using the inverse dispersion technique with grazing cattle is how to treat the animal: (i) cattle can be treated as a moving point source, for which the C-Q calculation is made assuming the cattle are point sources with locations determined by global positioning system (GPS) collars

Enteric methane (CH4) emission from cattle is a source of greenhouse gas and is an energy loss that contributes to production inefficiency for cattle. Direct measurements of enteric CH4 emissions are useful to quantify the magnitude and variation and to evaluate mitigation of this important greenhouse gas source. The objectives of this study were to evaluate the impact of stocking density of cattle and source configuration (i.e., point source vs. area source and elevation of area source) on CH4 emissions from grazing beef cattle in Queensland, Australia. This was accomplished using nonintrusive atmospheric measurements and a gas dispersion model. The average measured CH4 emission for the point and area source was between 240 and 250 g animal-1 d-1 over the entire study. There was no difference (P > 0.05) in emission when using an elevated area source (0.5 m) or a ground area source (0 m). For the point-source configuration, there was a difference in CH4 emission due to stocking density; likewise, some differences existed for the area-source emissions. This study demonstrates the flexibility of the area-source configuration of the dispersion model to estimate CH4 emissions even at a low stocking density.

© Her Majesty the Queen in Right of Canada, as represented by the Minister of Agriculture and Agri-Food Canada. ASA, CSSA, and SSSA, 5585 Guilford Rd., Madison, WI 53711 USA. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.

S.M. McGinn and T.W. Coates, Agriculture and Agri-Food Canada, Lethbridge, AB, Canada T1J 4B1; T.K. Flesch, Univ. of Alberta, Dep. of Earth and Atmospheric Sciences, Edmonton, AB, Canada T6G 2H4; E. Charmley, CSIRO Livestock Industries, Townsville, Queensland, Australia; D. Chen and M. Bai, Univ. of Melbourne, Melbourne, Victoria, Australia; G. Biship-Hurley, CSIRO Livestock Industries, Brisbane, Queensland, Australia. Assigned to Associate Editor April Leytem.

J. Environ. Qual. 44:97–102 (2015) doi:10.2134/jeq2014.06.0275 Received 24 June 2014. *Corresponding author ([email protected]).

Abbreviations: DMI, dry matter intake; DSM, digital stepping motor; GHG, greenhouse gases; GPS, global positioning system; LW, live weight; OPL, openpath laser.

97

(e.g., McGinn et al., 2011); (ii) cattle can be treated as a uniform area source, for which it is assumed that the paddock area is a uniform ground-level emitting source (i.e., cattle positions are not used); or (iii) the paddock can be treated as an elevated area source (i.e., cattle positions are not used). Treating cattle as point sources reflects the reality of cattle as enteric emission sources, but this treatment requires more effort than assuming the problem can be represented as a paddock with a uniform area source. Cattle must be handled to place and retrieve the GPS units. An additional consideration when applying the inverse dispersion technique to grazing cattle is the need to know the detectability of the target gas (above background concentration) relative to cattle stocking density (animals ha-1). The objectives of our study were to determine CH4 emissions from grazing cattle and to evaluate the impact of different source configurations when using a gas dispersion model. More specifically, this study reports on (i) the impact of cattle stocking densities on CH4 emissions, (ii) the influence of using either an area source (paddock area) or point source (animal position), (iii) the significance of elevated area source on CH4 emissions, and (iv) the magnitude of CH4 emissions from grazing cattle.

Materials and Methods Study Site This study was conducted at the Commonwealth Scientific and Industrial Research Organization Lansdown Research Station in Queensland, Australia (19° 39’ 40.93” S, 146° 50’ 46.42” E; elevation 66 m) in October 2012. A large uniform pasture (720 × 450 m with no obstruction to wind flow) contained six fenced paddocks (Fig. 1), each 1 ha in size. The pasture consisted of predominantly Rhodes grass (Chloris gayana Kunth), Sabi grass [Urochloa mosambiensis (Hack.) Dandy], and Verano Stylo [Stylosanthes hamata (L.) Taub], with some Buffel grass (Cenchrus ciliaris L.) and Seca Stylo (Stylosanthes scabra Vogel). Three of the paddocks (C1–C3; 100 × 100 m each) had a constant cattle stocking density, and three paddocks (T1–T3; 100 × 100 m each) had varying stocking densities. Brahman cattle (Bos indicus) with an initial average live weight (LW) of 425 ± 29 kg grazed in each paddock for approximately 96 h (Friday morning to Tuesday morning) before being moved to the next paddock in sequence. For the first period of the study (Period 1: 12–16 Oct.), paddocks C1 and T1 each held 30 animals; there were no animals in the remaining paddocks. In the second period (Period 2: 19–23 Oct.), there were 30 animals in paddock C2 and 20 animals in paddock T2. During the last period of the study (Period 3: 26–30 Oct.), there were 30 animals in paddock C3 and 10 animals in paddock T3.

Field Instrumentation Methane (CH4) concentrations along paths at each paddock were measured using two open-path lasers (OPLs) (GasFinder, Boreal Lasers Inc.) that recorded the average CH4 mixing ratio (ppm) between the OPLs and a retroreflector (Fig. 2). Before the start of the study, all four OPLs were set up side by side over a 350-m path to evaluate between-sensor differences. These data were used to correct CH4 measurements for each OPL (Table 98

Fig. 1. Schematic of study pasture showing paddocks C1, C2, and C3 and T1, T2, and T3 (each 100 × 100 m) and position of the weather station including the anemometer (triangle); north is toward the top of the image.

Fig. 2. Open-path laser positions (arrows) for (a) paddock C1, (b) paddock T1, and (c) paddocks C2, C3, T2, and T3. The westward looking path (dashed line in a and b) was used for the first 24 h and then shifted to the northward view (solid). North is toward the top of the image.

1). Temperature and pressure (weather station data) corrections provided by the manufacture were applied to the OPL CH4 data (Table 1). One OPL in each paddock was mounted on a digital stepping motor (DSM) (PTU D300, Directed Perceptions Inc.) that permitted multiple measurements of CH4 concentration sequentially along different paths radiating from the DSM. The DSM was located in the northwest corner of each paddock during measurement (Fig. 2). Paddock C1 (Fig. 2a) was set up differently than paddock T1 (Fig. 2b) regarding position of the OPL paths. A fourth path, facing north, was added to the DSM-OPL setup for paddocks C2-C3 and T2-T3 (Fig. 2c) to assist in determining background CH4 concentrations with non-southerly winds (the dominant wind direction was from the east). The longer DSM-OPL path looking north in Fig. 2c provided the opportunity to improve the resolution of the background concentration measurement when the wind direction was from the east, north, or west. All OPL paths were set at a height of 1.8 m above the soil surface. Table 1. Corrections applied to open-path CH4 concentration measurements. Laser unit no.

Between-laser bias†

Pressure correction‡

Temperature (T) correction

CH4–1012 CH4–1013 CH4–1034 CH4–1042

1.073 1.048 1.000 0.895

0.9946 0.9917 0.9873 0.9873

0.0047 × T + 0.9064 0.0043 × T + 0.9140 0.0037 × T + 0.9257 0.0037 × T + 0.9257

† Using CH4–1042 as the standard. ‡ Using 1000 hPa as the base air pressure. Journal of Environmental Quality

The OPL sampled CH4 concentration approximately once every second and recorded the 1-min average sequentially along each path. The OPL and DSM were connected to a computer that enabled the interface software (Boreal Laser Inc.) to control the DSM. An option in the software was used to optimize the OPL return light level at each retroreflector using small movements in the DSM that maintained an ideal range between 2500 and 10,000 (no units). The OPL and DSM system was powered by 12 V batteries, which were recharged using solar panels during the daylight, and a generator overnight. The second OPL in each paddock was dedicated to a single fixed path and was powered with a 12 V battery. A tower-mounted, three-dimensional anemometer (CSAT3, Campbell Scientific Inc.) was co-located at the site at a height of 2.8 m (Fig. 1). The average values of the wind speed and temperature, and their variance and cross products, were recorded (sampling at 10 Hz) every 10 min by a datalogger (CR10X, Campbell Scientific Inc.). These data were used to estimate the surface roughness (zo), the friction velocity (u*), and the Monin-Obukhov length (L). Wind direction was also monitored. A vertical air temperature gradient was measured using two shielded type T thermocouples at heights of 1 and 3 m. Before the cattle entered each paddock, they were fitted with a collar containing a GPS device that recorded the time, latitude, and longitude once every second. The GPS data were downloaded from each GPS device, the easting and northing (Universal Transverse Mercator) values were calculated, and average animal position over 10-min intervals was determined. The 10-min interval was a compromise between the need to accurately measure animal position (shorter averaging time) and the ability to measure meaningful values of zo, u*, and L (longer averaging time). The 10-min averaged OPL concentrations and the anemometer and GPS datasets were merged to predict 10-min average CH4 emission (g d-1) using a dispersion model (WindTrax version 2.0.8.8, Thunder Beach Scientific). The WindTrax simulation is based on a Lagrangian stochastic procedure (Flesch et al., 2004) where, in our study, 10,000 trajectories were used to predict the growth of the CH4 plume from cattle in a paddock. This cattle source was treated as a uniform area source (at 0 or 0.5 m heights) (an elevated area source is included in the latest version of WindTrax) or as point sources (at 0.5 m height) based on the GPS animal locations. In WindTrax, the atmospheric pressure was calculated from the elevation, and the background (upwind of the source) CH4 concentration was predicted by the WindTrax model. The WindTrax-simulated plume is used to predict an emission-to-concentration relationship (Q/C)sim. This simulated relationship is used with measured 10-min CH4 concentrations (OPL multiple paths) to infer the daily herd CH4 emission. For each paddock, the mean animal CH4 emission (g animal-1 d-1) was calculated by dividing by the number of animals in the paddock.

Data Management All data were managed using SAS software (SAS Inst.). The collar GPS data included 10-min easting and northing values that were converted to positional data relative to the southwest corner of each paddock, designated as the origin (0, 0). These

GPS data were merged with the OPL concentrations and the CSAT3 anemometer data, creating 10-min averaged datasets for each of the six paddocks (C1–C3 and T1–T3). These six datasets were sequentially used by WindTrax to estimate mean animal CH4 emission for each paddock. The CH4 emissions were first filtered (Filter A) for intervals when the OPL return light level was outside the recommended levels of 10,000 (Table 2). A second filter (Filter B) corresponded to atmospheric boundary conditions that are known to violate the underlying assumptions in the dispersion model. Previous studies (Flesch et al., 2009; Harper et al., 2011) have used the criteria u* < 0.15 m s-1 and |L| < 10 m to identify these invalid conditions but resulted in a severe loss of data. An alternate protocol developed for invalid atmospheric conditions by Flesch et al. (2013) that retained more data was used in our study. Filter B eliminated data based on u* < 0.05 m s-1 and the difference (>1°) (this difference is slightly more conservative than the value of 0.5 used by Flesch [personal communication]) between the measured air temperature gradient (∆T) and the Monin-Obukhov gradient (∆TMO) as calculated in WindTrax:

DTMO =

ù u*2 Tavg é æç z2 ö÷ ê lnç ÷-j ( z ) +j ( z )ú H 2 H 1 ú ÷ ê 2 ç kv g L ëê çè z1 ÷ø ûú

where Tavg is the average air temperature (at z = 2 m; °K), k is von Karman’s constant (0.4), g is the gravitational acceleration (9.81 m s-2), and jH is the stability correction function given as:

(

)

j H = 2 ln éê 1 + 1-16 z / L /2ùú (for L < 0) ë û =-5 z / L (for L > 0) The final filter (Filter C) eliminated outliers identified using the Univariate procedure (SAS Inst., Inc.) and by deviations from the mean greater than three times the standard deviation of the mean.

Statistical Analysis The statistical test for significant differences (P ≤ 0.05) and trends (P ≤ 0.10) between CH4 emissions between paddocks Table 2. Methane emission data (point-source configuration) available after application of filtering criteria A, criteria B, and criteria C.† Paddock‡ C1 T1 C2 T2 C3 T3 Mean

Maximum (counts)

Filter A (counts)

Filter B (counts)

Filter C (counts)

558 558 576 576 552 552 562

400 405 111 237 294 350 299

331 373 96 223 282 317 270

265 332 76 174 251 251 224

† Criteria A: return laser light level 10,000 for all laser paths. Criteria B: criteria A plus u* < 0.05, L < 1, and DDT > 1, where u* is friction velocity (m s-1), L is the Lagrangian length scale (m), and DDT is the difference between calculated and measured vertical air temperature gradient (°C). Criteria C: criteria B plus check for outliers >3 × SD. ‡ 1, 2, 3, treatment period; C, control; T, treatment.

www.agronomy.org • www.crops.org • www.soils.org 99

was evaluated using the Mixed procedure (SAS Inst.). The significant interaction of technique (1 = point source, 2 = area source at z = 0 m, 3 = area source at z = 0.5 m), period (Period 1–3), and paddock (1 or 2) was found, and contrasts were used to make the comparisons of interest (e.g., C1 vs. T1, C2 vs. T2, and C3 vs. T3).

Results and Discussion Data Filtering The maximum number of 10-min intervals where cattle were in the paddocks was 558, 576, and 552 for Periods 1, 2, and 3, respectively (Table 2). As data filtering criteria were applied, the percent of useable data decreased accordingly. For example, for the point-source configuration, the amount of retained data was reduced to 53% (average) once the CH4 emission data were checked for OPL light level and other instrument signal errors. The further loss of data (48% retained on average) occurred when criteria in filter B was applied, which has been recognized (Flesch et al., 2013) to coarsely remove invalid data due to underlying assumptions in the dispersion model. Additional outliers, where the emissions were greater than three times the standard deviation of the mean, were removed with filter criteria C (Table 2). Filter C reduced the retained data amount to 40% of the maximum.

Impact of Stocking Density The CH4 emissions for the different stocking densities (T1, T2, and T3) were not as consistent for the point-source configuration (201–299 g animal-1 d-1) as compared with the area-source configuration (263–281 g animal-1 d-1) (Table 3). The relatively greater CH4 emission of 269 to 299 g animal-1 d-1 at the lowest stocking density (10 animals ha-1) may relate to a greater dry matter intake (DMI) of the cattle resulting from more available forage per animal. These greater emissions may also relate to increased error in calculating CH4 emissions at low stocking densities. For example, the error in estimating the animal positions becomes more critical at low stocking densities when determining the dispersion plume. In addition, McGinn et al. (2013) reported, for the lowest stocking density, that the fraction of the time that the horizontal CH4 gradient concentration was detectable (downwind minus upwind OPL reading >0.02

ppm) declined significantly. This would introduce an error in CH4 emissions for lower cattle stocking density. Because even the lowest stocking density of 10 animals ha-1 is high for extensive grazing conditions, this highlights a limitation of the method, which can be overcome in practice by corralling cattle at high densities around a water point for several hours a day (Phillips et al., 2013). A comparison of CH4 emissions between the control and treatment for each period (C1-T1, C2-T2, and C3-T3) indicated a significant difference due to stocking density for the point-source configuration (Table 3). The same comparison for the area-source configuration was not as conclusive, yielding a difference at the medium stocking rate (20 animals) but not at the low stocking rate (10 animals).

Point-Source vs. Area-Source Configuration Over all paddocks, the mean CH4 emission was 240 g animal-1 d-1 for the point-source configuration (SD = 116) and 247 g animal-1 d-1 for the area-source (SD = 127) configuration (Table 3), which was significantly different (P < 0.05). In this latter configuration of the source, the height of the release was assumed to be the surface (z = 0). The average CH4 emissions rates for C1, T1, C2, and C3, which were associated with the high stocking density, were 230 g animal-1 d-1 (average of two area-source configurations) and 217 g animal-1 d-1 (point-source configuration). For the medium stocking density (20 animals ha-1), the emission rates for the area- and point-source emissions were 264 and 201 g animal-1 d-1, respectively. Much higher emission rates of 299 g animal-1 d-1 (point source) and 269 to 277 g animal-1 d-1 (area source) were calculated for the lowest stocking density (10 animals ha-1).

Elevated Area-Source vs. Surface Area-Source Configuration The elevation of the area source to a position of 0.5 m above the surface (0 m) was done to emulate an average height of the head of the animal above the surface (when lying down or standing and during eating, ruminating, or idle activities). The exhaled CH4 from the nostrils and mouth of ruminants is the primary source of enteric CH4 loss to the atmosphere (Murray et al., 1976). The overall difference in the estimated CH4 emissions

Table 3. Methane emissions (±SD) at different cattle stocking densities using point-source and area-source configurations. Paddock† C1 T1 C1-T1 C2 T2 C2-T2 C3 T3 C3-T3 Mean

Stocking density animals ha-1 30 30 30 20 30 10

Point source (z = 0.5 m)

Area source (z = 0 m)

Area source (z = 0.5 m)

———————————————— g animal-1 d-1 ———————————————— 225 ± 112 214 ± 107 214 ± 109 240 ± 101 276 ± 128 281 ± 126 NS S S 156 ± 112 167 ± 100 152 ± 78 201 ± 120 264 ± 148 268 ± 146 S S S 248 ± 120 274 ± 115 272 ± 114 299 ± 120 277 ± 132 269 ± 132 S NS NS 240a‡ ± 116 247b ± 127 250b ± 126

† 1, 2, 3, treatment period; C, control; T, treatment; NS, not significant from zero (P > 0.05); S, significant from zero (P < 0.05). ‡ Values within row with different superscripts are significantly different (P < 0.05). 100


attributed to source height (247 vs. 250 g animal-1 d-1) (Table 3) was not significant (P > 0.05). There was no systematic difference in CH4 emissions between the area source elevation (z = 0 or z = 0.5 m) simulations due to time of day (e.g., daytime vs. nighttime; data not shown).

CH4 Emissions

The overall mean CH4 emissions were 240 and 247 to 250 g animal-1 d-1 for the point- and area-source configurations in our study (Table 3). These emissions are greater than that found by McGinn et al. (2011) of 141 ± 147 for beef cattle grazing Rhodes grass and Leucaena (Leucaena leucocephala Lam) in Queensland, Australia (Table 4). Leucaena is reported to have high tannin content that is associated with reduced enteric CH4 emissions. Tomkins et al. (2013) reported CH4 emissions at the herd scale of 136 to 231 g animal-1 d-1 in northern Australia, whereas Phillips et al. (2013) reported emissions of 113 to 147 g animal-1 d-1 for beef cattle in northwestern Western Australia (Table 4). The lower CH4 emissions are generally associated with younger cattle and improved pastures (Tomkins et al., 2013). There may also be a bias due to daytime/nighttime differences when measuring CH4 emissions. The greater CH4 emissions in our study were not attributed to differences in sampling frequency between daytime and nighttime because there was little change in emissions between these periods (Fig. 3). However, the small peak in CH4 emission during the morning does coincide with a typical grazing bout for cattle (Gregorini et al., 2006) and cattle walking (grazing) activity (Tomkins et al., 2009). This grazing–emission association is not evident in the pre-dusk period, typical of intensive cattle grazing (Gregorini et al., 2006). Forage quality and quantity may also influence CH4 emission differences found in the literature. Using a chamber technique and feeding a high-grain diet, McCrabb and Hunter (1999) measured enteric CH4 emission of 160 g aimal-1 d-1 (Brahman cattle). In a comprehensive study by Kennedy and Charmley (2012), who examined CH4 emissions using chambers for a variety of pasture species typical of northern Australia, the CH4 emissions ranged from 42 to 159 g animal-1 d-1, whereas the CH4 emissions per kg DMI varied from 15.8 to 22.4. In a similar study to ours (same paddock but a year later), DMI was estimated to be 1.8% of LW using the fecal NIRS method (Luciano González, personal communication, 2013). Assuming a relationship of 19.7 g CH4 kg-1 DMI (Kennedy and Charmley, 2012), the mean emission for these cattle would have been 153 g animal-1 d-1. In a model study (Bentley et al., Table 4. Reported methane emissions from beef cattle fed forage in Australia. Study Present study McGinn et al. (2011) Tomkins et al. (2013) Phillips et al. (2013) McCrabb and Hunter (1999) Kennedy and Charmley (2012) Bentley et al. (2008) Dong et al. (2006)

Methane emission g animal-1 d-1 240–250 141 136–231 113–147 160 42–159 358–417 164

Fig. 3. Average enteric CH4 emission for each hour of the day as calculated using all CH4 emission data (dots) with error bars showing SD and a fifth-order polynomial fitted line.

2008) that was based on DMI, the CH4 emission of cattle in the pastoral system of Northern Australia ranged from 358 to 417 g animal-1 d-1. In comparison, the Intergovernmental Panel on Climate Change (Dong et al., 2006) emission rate (Tier 1 category) is 164 g animal-1 d-1 for nondairy cattle in the Oceania region. In addition to the difference in forage composition between these studies that affects CH4 emissions, there is an effect due to LW and DMI. The higher CH4 emission in our study coincides with a higher LW of 425 ± 29 kg, compared with that of 271 ± 22 kg reported in McGinn et al. (2011). The ratio of CH4 emission to kg LW in our study (0.56) was greater than that reported by McGinn et al. (2011) of 0.52. These ratios are higher than reported for beef cows (0.43–0.51) grazing different spring forages (Olson et al., 1997).

Conclusions This study shows that estimated CH4 emission using an inverse dispersion model is generally not sensitive to spatial differences in the source configuration. For the roaming cattle in the 1-ha paddock, there was little advantage in using GPS units mounted on the animal to locate the precise source of the CH4 plume compared with assuming a uniform source of CH4 across the paddock. The effect of source height (0 or 0.5 m above the surface) on the calculated CH4 emissions using a uniform area source was not significant. The highest CH4 emission coincided with the lowest stocking density for the pointsource configuration, which may reflect the underlying error in determining the CH4 emissions at low stocking densities as it relates to the detectability of the concentration sensors.

Acknowledgments This study was supported by the CSIRO Sustainable Agriculture Flagship Program, the Canadian AGGP Program, and Agriculture and Agri-Food Canada’s Growing Forward Program.

References Australian National Greenhouse Accounts. 2013. National inventory report 2011. Vol. 1. http://www.climatechange.gov.au/sites/climatechange/files/ documents/05_2013/AUS_NIR_2011_Vol1.pdf (accessed Nov. 2013). Bentley, D., R.S. Hegarty, and A.R. Alford. 2008. Managing livestock enterprise in Australia’s extensive rangelands for greenhouse gas and environmental outcomes: A pastoral company perspective. Aust. J. Exp. Agric. 48:60–64. doi:10.1071/EA07210 Dong, H., J. Mangino, T.A. McAllister, J.L. Hatfield, D.E. Johnson, K.R. Lassey,

www.agronomy.org • www.crops.org • www.soils.org 101

M. Aparecida de Lima, and A. Romanovskaya. 2006. Emissions from livestock and livestock manure. In: 2006 IPCC Guidelines for national greenhouse gas inventories, volume 4: Agriculture, forestry and other land use. Chap. 10. http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4.html (accessed 4 May 2011). Flesch, T.K., L.A. Harper, J.M. Powell, and J.D. Wilson. 2009. Inverse-dispersion calculation of ammonia emissions from Wisconsin dairy farms. Trans. ASABE 52:253–265. doi:10.13031/2013.25946 Flesch, T.K., S.M. McGinn, D. Chen, J.D. Wilson, and R.L. Desjardins. 2013. Data filtering in bLS calculation of emissions rates: Improvements to increase data retension. Adv. Anim. Biosci. 4:472. Flesch, T.K., J.D. Wilson, L.A. Harper, and B.P. Crenna. 2005. Estimating gas emissions from a farm with an inverse-dispersion technique. Atmos. Environ. 39:4863–4874. doi:10.1016/j.atmosenv.2005.04.032 Flesch, T.K., J.D. Wilson, L.A. Harper, B.P. Crenna, and R.R. Sharpe. 2004. Deducing ground-to-air emissions from observed trace gas concentrations: A field trial. J. Appl. Meteorol. 43:487–502. doi:10.1175/1520-0450(2004)0432.0.CO;2 Gregorini, P., S. Tamminga, and S.A. Gunter. 2006. Review: Behavior and daily grazing patters of cattle. Prof. Anim. Sci. 22:201–209. Harper, L.A., O.T. Denmead, and T.K. Flesch. 2011. Micrometeorological techniques for measurement of enteric greenhouse gas emissions. Anim. Feed Sci. Technol. 166–167:227–239. doi:10.1016/j. anifeedsci.2011.04.013 Kennedy, P.M., and E. Charmley. 2012. Methane yields from Brahman cattle fed tropical grasses and legumes. Anim. Prod. Sci. 52:225–239. doi:10.1071/ AN11103 Makkar, H.P.S., and P.E. Vercoe. 2007. Measuring methane production from

102

ruminants. Springer, Dordrecht, The Netherlands. McCrabb, G.J., and R.A. Hunter. 1999. Prediction of methane emission from beef cattle in tropical production systems. Aust. J. Agric. Res. 50:1335– 1339. doi:10.1071/AR99009 McGinn, S.M., T.K. Flesch, T.W. Coates, M. Bai, G. Bishop-Hurley, E. Charmley, N. Tomkins, and D. Chen. 2013. Detectability of methane emissions from grazing cattle. Adv. Anim. Biosci. 4:308. McGinn, S.M., D. Turner, N. Tomkins, E. Charmley, G. Bishop-Hurley, and D. Chen. 2011. Methane emissions from grazing cattle using point-source dispersion. J. Environ. Qual. 40:22–27. doi:10.2134/jeq2010.0239 Murray, R.M., A.M. Bryant, and R.A. Leng. 1976. Rates of production of methane in the rumem and large intestine of sheep. Br. J. Nutr. 36:1–14. doi:10.1079/BJN19760053 Olson, K.C., R.D. Wiedmeier, and R. Banner. 1997. Reducing methane emissions from beef cow herds in range based management systems. In: D. Conneely and M. Gibbs, editors, Ruminant Livestock Efficiency Program Annual Conference Proceedings, USEPA USDA, Washington, DC. Phillips, F.A., N. Tomkins, M. Bai, T. Naylor, and P. Vercoe. 2013. Quantifying methane mitigation potential for beef cattle at the herd scale in NW Western Australia. Adv. Anim. Biosci. 4:432. Tomkins, N., P.J. O’Reagain, D. Swain, G. Bishop-Hurley, and E. Charmley. 2009. Determining the effect of stocking rate on the spatial distribution of cattle for the subtropical savannas. Rangeland J. 31:267–276. doi:10.1071/ RJ07070 Tomkins, N., M. Bai, and E. Charmley. 2013. Measuring methane emissions at the herd scale in northern Australia. Adv. Anim. Biosci. 4:386.
