You are on page 1of 12

Agricultural and Forest Meteorology 228–229 (2016) 73–84

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology


journal homepage: www.elsevier.com/locate/agrformet

Ammonia emissions from anaerobic waste lagoons at pork production


operations: Influence of climate
Richard H. Grant a,∗ , Matthew T. Boehm a , Alfred J. Heber b
a
Purdue University departments of Agronomy, USA
b
Agricultural and Biological Engineering, USA

a r t i c l e i n f o a b s t r a c t

Article history: Swine waste storage is a major source of agricultural ammonia (NH3 ) emissions. Understanding the
Received 10 March 2016 factors that influence the magnitude of emissions is important to our understanding of NH3 loading of the
Received in revised form 22 June 2016 atmosphere and it’s deposition on the landscape. Since swine operations occur across the United States, a
Accepted 27 June 2016
better understanding of how differing climates influence NH3 emissions and how future climate change
Available online 13 July 2016
in might impact emissions is needed. Ammonia was measured from 2007 through 2009 at five swine
farms using anaerobic lagoons for manure storage and processing. Three farms, located in North Carolina
Keywords:
(NC), Indiana (IN), and Oklahoma (OK), handled the breeding to weaning of the pigs (sow farm). These
Ammonia emissions
Livestock emissions
farms had an annual mean sow live mass-specific emissions ranging from 62 g d−1 AU−1 (AU = 500 kg live
Climate change mass) to 120 g d−1 AU−1 . Two farms, located in NC and OK, handled the growing of the pigs to market
size (finisher farm). These farms had annual mean live mass-specific emissions of36 g d−1 AU−1 (OK) and
104 g d−1 AU−1 (NC). An emissions model incorporating the influence of winds and temperature accounted
for 74% of the measured emissions variability. A second emissions model including the influence of pH,
suspended organic matter, and farm type-specific nitrogen excretion rates accounted for 66% of the
measured emissions variability. An evaluation of the influence of warmer or cooler winters and summers
using the second model shows that a 1 ◦ C increase in maximum daily mean temperature results in a 6.9%
increase in annual emissions and a 9.5% increase in maximum emissions. Additional detailed studies
focused on better estimating the loading of the lagoons and chemistry of the surface of the lagoons are
needed to improve the estimation of NH3 emissions.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction facility emissions across the USA under different farm management
plans.
Measurements of ammonia (NH3 ) emissions from anaerobic Since NH3 emissions from swine lagoons are influenced by
treatment lagoons for swine manure (waste) are of interest because atmospheric turbulence, the emissions need to be measured using
farms are required to report emissions in excess of 220 kg/day micrometeorological approaches. However, there have been only
(100 lb/day) for the “Right-to-Know” regulations (Emergency Plan- a few studies of NH3 emissions using such methods from lagoons
ning and Community Right-to-Know Act) of the United States (USA) at sow farms. Harper and Sharpe (1998), using a micrometeoro-
(Centner and Patel, 2010). From a community and producer’s point logical method over an unspecified number of days, determined a
of view, it is important to know the corresponding size of farm that mean emission of 4.2 g hd−1 d−1 (hd, head representing1 finishing
would be expected to reach this threshold of emissions. However, hog, gestating sow, or sow and litter). Harper et al. (2004), from
it is difficult to accurately determine the threshold size since emis- twenty daily measurements over two years, determined an annual
sions are related to climate so that the threshold size in one part emission of 6.0 g hd−1 d−1 (hd representing unstated combination
of the country will not be the same as that in another warmer or of sow, litter and hogs) at a breed-to-wean sow farm in North Car-
colder part of the country. The National Air Emissions Monitoring olina. Grant et al. (2013a) observed an average daily mean emission
Study (NAEMS) was conducted to assist in characterizing livestock from a sow lagoon in Oklahoma of 69 g hd−1 d−1 over 181 d of
measurements across two years.
There have been more NH3 emission studies using micromete-
orological methods at finisher farms than at sow farms. Szögi et al.
∗ Corresponding author.
(2005) measured emissions of 9.75 g NH3 hd−1 d−1 during nine
E-mail address: rgrant@purdue.edu (R.H. Grant).

http://dx.doi.org/10.1016/j.agrformet.2016.06.018
0168-1923/© 2016 Elsevier B.V. All rights reserved.
74 R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84

Table 1
Summary of farm characteristics. Loading rates based on nominal farm capacity, farm-reported pig weights, and manure production estimates (ASAE, 2005).

IN4 NC3 NC4 OK3 OK4

Farm Capacity, number of sows 1400 2000 2784


Average Sow Weight, kg 216 197 223
Piglet Capacity 1960 2800 3898
Average Piglet Weight, kg 5 5 5
Farm Capacity, number of pigs 8000 3000
Average Finisher Weight, kg 135 170
Calculated Animal Units, AU 624 982 815 464 1279
Source Surface Area, m2 12,009 18,878 22,641 11,139 21,381
Source Capacity, m3 40,208 46,003 60,779 28,663 84,453
Mean Source Depth, m 3.3 2.4 2.7 5.9 3.9
Calculated N1 Excretion Rate3 , kg/d 63.9 202 83.9 95.3 131
Calculated N Loading Rate, g/d-AU3 102 205 102 205 102
Calculated N Loading Rate, g/m2 /d 5.32 10.7 3.71 8.55 6.11
Calculated N Loading Rate, g/m3 /d 1.59 4.39 1.38 3.32 1.55

1: Nitrogen.
2: Rotz (2004).
3: ASAE (2005).

measurement days distributed across a year. Shores et al. (2005) 2-m wind measurement. As a result of the proximity of the barns,
measured a NH3 emission of 44.1 g hd−1 d−1 during one day of mea- all measurements taken when the wind direction between 135◦
surements during July at a finishing farm in North Carolina. Zahn and 225◦ were excluded from analysis. The only significant pro-
et al. (2001) measured an average emissions of 22.7 g NH3 hd−1 d−1 ducer events at the farms during the measurement periods were a
during a 14-day measurement campaign in late summer and early pump-out of the lagoon and a period of empty barns (Table 3).
fall at a Missouri finishing operation. Grant et al. (2013a) found an The western swine finishing facility (OK3; Table 1) consisted
average daily mean emission from a finishing farm in Oklahoma of of three barns (Fig. 1d). Manure from the barns along with lagoon
18 g hd−1 d−1 over 155 d of measurements across two years. water recharge was transferred three times a week to the lagoon by
The NH3 emissions from lagoons have been modeled using pulling drain plugs in the barns’ underfloor pits. The waste lagoon
statistical regressions (Harper and Sharpe, 1998), semi-empirical was rectangular and located to the west of the barns (separated by
functions (Grant et al., 2013a), and detailed theoretical schemes a drainage swale). The clay-lined lagoon was oriented north-south.
(DeVisscher et al., 2002; Bajwa et al., 2006; and Sommer et al., Liquid was typically removed from the lagoon approximately every
2006). Statistical emissions models have little transportability since six months. Sludge from the bottom of the lagoon had not been
the statistical regressions depend on the location- and time-specific removed since construction in 1998. The barns resulted in a fetch
variations of the variables used in the regressions. Detailed theo- ratio of 10:6 for the east side of the lagoon. Fetches in all other direc-
retical models provide the greatest transportability across climates, tions were better than 100:1 (Fig. 1d). Due to the proximity of the
but typically require extensive information about the lagoon. Fur- barns, emissions from wind directions of 90◦ to 135◦ were excluded
thermore, ambient atmospheric turbulence introduces a significant from analysis. All fans exhausted barn ventilation air from the east
amount of chaos into the mixing and exchange processes within side of the barns and therefore did not influence the lagoon PIC
the liquid and between the liquid and air. Consequently modeling measurements except under easterly winds (which were not the
of the emissions into the turbulent atmosphere requires adequate prevailing winds at any time of the year). The nominal 120 day cycle
understanding so that the dominant processes can be theoretically of hogs through the farm was evident in the producer-reported
described and a wide range of conditions to evaluate the model. activity during the study (Table 3).
The objective of this study was the development of a theoretically- The southeastern sow farm (NC4; Table 1) consisted of three
based NH3 emissions model for swine lagoons across the USA. Five barns (Fig. 1a). Manure was transferred once a week from the barns
hog operations, including sow farms and finisher farms located to the clay-lined lagoon by pull plug with lagoon water recharge of
in the southeastern, mid-western, and southwestern USA were the pits. Wastewater pipes from all three buildings combined into
included in this analysis. one inlet pipe that discharged into the SW corner of the lagoon
(Fig. 1a. Sludge from the lagoon had not been removed since con-
struction in 1994. The sludge depth was approximately 0.7 m at the
2. Methods beginning of the study. Liquid was removed as weather permitted.
The fetch ratio of 10:3 to the south of the lagoon resulted in the
2.1. Farms exclusion of emissions when wind directions were between 225◦
and 315◦ . Fetches for wind measurements from the east (cropland)
The southeastern swine finishing lagoon facility (NC3; Table 1) and north were greater than 100:1 for all measurement heights.
consisted of five barns (Fig. 1c), an office, and an anaerobic treat- Fetch for wind measurements to the west was 100:1, 100:2.4, and
ment lagoon. Manure from the barns was drained by gravity daily 100:2 for wind measurements at the 2-m, 4-m, and 16-m heights.
to the lagoon by a pull plug and lagoon water recharge system in Consequently, all wind measurements were relatively unaffected
the barns. The clay-lined lagoon was oriented east-west and sepa- by upwind conditions. The fan exhaust blew from the northern-
rated by a drainage swale from the barns (Fig. 1c). Lagoon effluent most barn to the east and west. Ventilation fans exhausted from the
was removed for irrigation as weather permitted. Sludge from the west wall of the middle barn. The lagoon was pumped out several
lagoon had not been removed since construction in 1996. The barns times during the study (Table 2).
to the south of the lagoon resulted in a fetch ratio (distance: rise) The Midwestern sow farm (IN4; Table 1) consisted of nine barns
of 10:3 to the lagoon. Fetch ratio to the east for the wind mea- and a lagoon (Fig. 1e). In 1998 the facility was converted from a
surements was more than 4:1 for all measurement heights. While finisher farm to a sow farm. Liquid waste from the deep pits of
the 16-m measurement height was above the tops of the barns, the the barns was gravity drained once every two weeks to the lagoon
fetch ratio was 33:1 for the 4-m measurement height and 6:1 for the
R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84 75

Fig. 1. (a) Configuration of the NC4A farm. Image from GoogleEarthTM . Proximity of the buildings to the S restricted valid wind directions for calculating emissions estimates.
(b) Configuration of the OK4A farm. Image from GoogleEarthTM . Proximity of the building to the W restricted valid wind directions for calculating emissions estimates.
(c) Configuration of the NC3A farm. Image from GoogleEarthTM . Proximity of the buildings to the S restricted valid wind directions for calculating emissions estimates. (d)
Configuration of the OK3A farm. Image from GoogleEarthTM . Proximity of the buildings to the E restricted valid wind directions for calculating emissions estimates. (e)
Configuration of the IN4A farm. Image from GoogleEarthTM .

through a single inlet on the east side of the lagoon (Fig. 1e). The only producer-related change during the study was a shift from
surface of the clay-lined lagoon was 2–3 m above the surround- grain sorghum to corn in the feed ration (Table 2).
ing cropland. Sludge had not been removed from the lagoon. No Since the manure at these farms was stored in the lagoon for
reported lagoon pump out occurred during the study (Table 2). Corn multiple months as it decomposed, it was assumed that the number
completely surrounded the lagoon during its growing season. of animals contributing to the atmospheric emissions of the lagoon
The western sow farm (OK4; Table 1) consisted of three build- was the maximum capacity of the farm. The farm inventories were
ings (Fig. 1b): two gestation barns, and a 16-room farrowing at or near capacity during the study.
building. Manure from the barns was transferred weekly from the
two gestation units and every 21-d from the farrowing unit to
the lagoon by pull plug and lagoon water recharge. The clay-lined 2.2. Measurements
lagoon was separated from the barns by a drainage swale (Fig. 1b).
Sludge from the lagoon had not been removed since construction Continuous measurements of pH, oxidation-reduction poten-
in 1994. The nearest building obstruction resulted in a fetch ratio to tial, and temperature of the liquid were conducted from a float
the west of 100:8. Fetches in all other directions were better than located in the NE corner of the NC3, NC4, and OK4 lagoons, the
100:1 (Fig. 1b). Due to the proximity of the barns, emissions from SE corner of the IN4 lagoon, and the NW corner of the OK3 lagoon.
wind direction of 225◦ to 315◦ were excluded from analysis. The Analyses of the nitrogen (N) components of the lagoon liquid at the
OK4 and OK3 lagoons were provided by the producers every year.
76 R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84

Table 2
Measurement periods at each breed to wean farm.
The starting dates and number of days of emissions measurements after that date and the producer-supplied information on farm management events are indicated.

NC4 IN4 OK4

Year Start date # days Producer events Start date # days Producer Start date # days Producer events
events

2007 1 July 61 27 Jun 63 No events


4 Oct 18 no events 1 Sep 90 No events 7 Nov 20 11/6 Gestation ration changed
from milo to corn
1 Dec 89 No events 28 Nov 20 12/17 Gestation ration changed
from corn to milo
2008 29 Jan 13 no events
31 Mar 16 04/16 pump out 1 Mar 91 No events 23 Apr 13 No events
13 Aug 20 08/23 pump out 09/22 1 Jun 43 No events
pump out 09/23 pump out
09/24 pump out
4 Sep 19 No events No events 1 Oct 14 No events
2009 14 Jan 19 02/07 pump out 8 Jan 19 No events
28 Apr 13 no events 1 Apr 20 No events
1 Jul 20 no events 25 Jun 19 No events

Table 3
Measurement periods at each swine finishing farm. The starting dates and number of days of emissions measurements after that date and the producer-supplied information
on farm management events are indicated.

OK3 NC3

year Start date # days Producer events Start date # days Producer
events

2007 30 Aug 19 08/30 feed: Paylean 4.5 g 09/13 feed: 24 Oct 14 No events
Paylean 6.5 g 09/25 Barns emptied
09/27 and 10/04 Restocked

2008 24 Jan 26 1/29/2008 North barn emptied 13 Feb 21 No events


1/30/2008 Middle barn emptied
1/31/2008 S barn emptied 2/11/2008
North and middle barn restocked
2/14/2008 South barn restocked
7 May 22 05/07 feed: Paylean 4.5 g all 3 barns. 06 Mar 20 No events
05/20 feed: Paylean 6.5 g N barn 05/22
feed: Paylean 6.5 g middle and S barns.
5/28-6/11 2079 Barns emptied
29 May 12 No events
05 Nov 27 No events 25 Sep 19 9/29–9/30
pump-out
2 Dec 14 No events 04 Feb 19 02/04–02/23
Barns empty

2009 12 May 21 No events


23 Apr 21 No events 02 Jun 20 No events
15 July 20 No events 24 Sep 68 No events
24 Oct 14 No events

Table 4
Climate conditions.

Measured Modeled Measured

Farm Days Mean Daily (◦ C) Maximum Minimum 1/2 Range of Mean Daily (◦ C) Annual Days Mean Daily Maximum
Mean Daily (◦ C) Mean Daily (◦ C) Mean Daily (◦ C) Amplitude (◦ C) Wind Speed Daily Wind
(m/s) Speed (m/s)

IN4 385 11.2 26.7 −15.2 21.0 9.9 13.4 275 3.76 13.7
NC3 198 15.2 28.5 −2.9 15.7 16.4 9.0 135 1.44 7.5
NC4 135 18.2 32.4 −5.2 18.8 17.0 9.5 125 2.16 11.2
OK3 150 13.0 34.3 −13.4 23.9 13.1 12.9 128 4.44 16.1
OK4 183 15.4 37.2 −11.6 24.4 13.2 14.1 152 4.78 17.5

Additional analyses of the lagoon liquid at the time of field applica- Diode Laser Absorption Spectrometer (TDLAS) open-path instru-
tions at OK3, NC3, and NC4 were provided by the producers; liquid ments and 3-dimensional (3D) sonic anemometers in conjunction
was applied to cropland by a center-pivot irrigation system when with meteorological measurements and vertical radial plume map-
levels reached the minimum freeboard of the lagoon (maximum ping (VRPM) and backward Lagrangian Stochastic (bLS) emission
capacity). The mean concentration of N in the liquid of each lagoon models. The path-integrated concentrations (PICs) of NH3 were
is indicated in Table 5. measured by TDLAS along optical paths defined by scanning TDLAS
The NH3 emissions from each lagoon were monitored during instruments (TDLAS/scanner) and retro-reflectors. The general con-
every quarter for two years (Tables 2 and 3) using scanning Tunable figuration of the TDLAS/scanners and retro-reflectors is illustrated
R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84 77

Table 5
Lagoon conditions during measurements.

Samples In situ lagoon measurements at 0.3 m depth

Farm na Mean N Days Mean daily (◦ C) Maximum Minimum daily Days Mean daily pH Maximum Minimum daily
concentration daily (◦ C) (◦ C) daily pH pH
(% wet weight)

IN4 0 0.29b 279 16.5 35.1 −0.6 218 7.9 8.3 7.5
NC3 19 0.43 131 18.2 31.6 5.6 161 7.7 8.4 6.9
NC4 19 0.29 130 20.7 35.7 2.3 112 7.8 8.2 7.5
OK3 3 0.57 116 16.0 31.9 0.6 104 7.9 8.2 7.3
OK4 1 0.44 128 19.7 36.2 2.6 108 7.8 8.4 6.5
a
n = number of samples.
b
set equal to NC4.

Fig. 2. Typical measurement configuration. Retro-reflectors indicated by squares with TDL OP indicated by solid lines. Sonic anemometers are indicated by bow-ties.

in Fig. 2. The TDLAS/scanners were mounted at 1-m heights above all instruments were also conducted. Inter-comparisons of sonic
the lagoon berm at the NE and SW corners of the NC3, NC4 and anemometers showed that all were within 0.1 ms−1 for any mean
OK4 lagoons (Figs. 1a–c) and at the NW and SE corners of the IN4 wind velocity component. No corrections or adjustments were
and OK3 lagoons (Figs. 1d, e). Towers for mounting retro-reflectors required for the sonic anemometer measurements. Calibration
were located on the NW and SE corners of the NC3, NC4, and OK4 checks assured that the TDLAS NH3 PIC measurements were main-
lagoons (Figs. 1a–c) and NE and SW corners of the IN4 and OK3 tained to a precision of ±10% (relative standard deviation) and an
lagoons (Figs. 1d, e). accuracy of 10% at 50 ppm-m. Atmospheric moisture interfered
Meteorological measurements (barometric pressure, air tem- with the NH3 measurements from 23 April, 2008 to 6 May, 2008 at
perature and relative humidity, solar radiation, and surface the OK4 sow farm, from 24 January, 2008 to 19 February, 2008 and 7
wetness) and 3D wind velocities (2.5 m height) were made near May, 2008 to 10 June, 2008 at the OK3 finisher farm, and throughout
the lagoon NE corner at NC3, NC4 and OK4, near the lagoon SW the study period at the IN4 sow farm. Inter-comparisons between
corner at IN4, and near the NW corner at OK3 (Fig. 1). The 3D sonic various TDLAS units experiencing atmospheric moisture interfer-
anemometers were located at nominal heights of 4 m and 16 m on ence and units without apparent interference revealed reduced
each tower on the SE towers at NC3, NC4, and OK4 and the SW responses by the moisture-affected units of 28%, 68%, 36% and 31%
towers at IN4 and OK3. for atmospheric moisture represented by dew point temperatures
Calibration verification checks of the TDLAS, sonic anemome- between −2 ◦ C and 20 ◦ C. The interference negatively affecting both
ters, and lagoon pH instruments were made at the beginning the magnitude of TDLAS response as well as the probability of the
and end of each measurement period. Semi-annual calibrations of measurement being NH3 (valid measurements had a coefficient of
78 R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84

determination of 0.9 between the measured absorption feature and


internal cell absorption feature.) All measurements at the IN sow
farm were potentially affected by a 31% bias in one sensor resulting
in a potential error in emissions of +/− 31%. Thirty six percent of the
measurements at the NC sow farm and 31% of the measurements
at the NC finisher farm were potentially affected by a 68% bias in
one sensor and a 36% bias in the second sensor for a net potential
emissions error of +/− 23%. No adjustments to these measurements
were possible since it was not possible to separate the occurrence
of a biased low response but adequate probability of NH3 versus
one with inadequate probability of NH3 . Calibration adjustments
to the raw measurements were made based on the multipoint cal-
ibrations. Multipoint calibrations of the affected sensors did not
exhibit the moisture interference during calibration. All concen-
tration measurements were normalized to 101.325 kPa and 20 ◦ C
(STP). Concentration measurements for a given path were consid-
ered representative of the 30-min averaging period if at least one
dwell on the retro-reflector defining the path had valid measure-
ments during the period. The TDLAS (Model GasFinder2TM NH3-OP,
Boreal Laser Inc., Spruce Grove, Alberta, Canada) had a dwell time on
each retro-reflector of 15 s. PIC measurements were made approx-
imately every 1.2 s.
Half-hourly emissions were estimated using two models:
an inverse Lagrangian dispersion model (Windtrax, Thunder
Fig. 3. Comparison of daily mean air temperature and daily mean lagoon tempera-
Beach Scientific, http://www.thunderbeachscientific.com, Edmon-
ture. Solid line represents a power function with R2 of 0.87.
ton, Alberta, Canada) and an integrated horizontal plume
mapping model (Vertical Radial Plume Model, VRPM). The inverse
Lagrangian dispersion model used the twelve near-surface NH3 emissions with the VRPM emissions adjusted for equivalence to
PIC measurements at each lagoon (Fig. 2). In this model, the NH3 the bLS emissions according to the relationship between the two
emission rate (Q) was calculated by single value decomposition emissions models described in Grant et al. (2013b).
of the matrix of flux versus PIC relationships obtained from each Emissions from the lagoons were considered in the context of
valid near-surface PIC using a backward Lagrangian Stochastic (bLS) a conceptual equilibrium two-film theory. Disregarding chemical
model (Crenna et al., 2008). The turbulence statistics used to esti- reactions in the lagoon or air, the ammonia flux from the lagoon
mate the trajectory of each parcel backward in time and space wastewater to the air can be expressed as:
was derived from Monin Obukov theory using turbulence statis-  
Q = −KL Cbkg − H (T ) Cliq (1)
tics based on data from the on-site 3D sonic anemometers. Validity
of ½ h bLS emissions estimates required absolute values of Monin where, at sufficient height, Cbkg is the background NH3 concen-
Obukov lengths (L) of greater than 2 m, friction velocities (u∗ ) tration (upwind concentration for the RPM emissions model and
greater than 0.15 m/s, standard deviation of wind direction less derived from the simultaneous equation of the bLS emissions
than 30◦ , and touchdown fractions greater than 0.1 (Grant et al., model), Cliq is the lagoon NH3 concentration at the interface (not
2013a). Assuming a theoretical random error of 22% (Laubach and measured), H(T) is Henry’s law constant (H) for NH3 solubility in the
Kelliher, 2005) and a bias of −40% for TDLAS units with moisture lagoon water as influenced by the liquid temperature at the inter-
interference, the expected error in the bLS-measurement of NH3 face (T), and KL is the overall mass transfer coefficient including
emissions was ±24%, with a bias of −40% for TDLAS NH3 mea- diffusion through the liquid and the air over the liquid.
surements subject to moisture interference and no bias for TDLAS Differences between the liquid and atmospheric environments
NH3 measurements free of moisture interference. This was consis- need to be considered to understand the factors controlling the
tent with tracer-estimated errors of the bLS emission calculation lagoon emissions. The solubility of NH3 and dissociation of NH4 +
method of between 5% and 36% (Gao et al., 2008). to NH3 in the lagoon liquid is temperature dependent. The daily
The VRPM method assumed a bivariate Gaussian function to mean air temperature at 2-m height above the lagoon berm was
describe the distribution of mass across vertical measurement correlated with the daily mean lagoon liquid temperature at 0.3-m
planes along each of the four sides of the lagoons (Hashmonay et al., depth using a power function (Fig. 3; coefficient of determination
2008). At least five NH3 PIC measurements defining one measure- R2 = 0.87). Since the lagoon temperature was not recorded during
ment plane and usually 10 NH3 PICs defining two measurement periods of potential freezing of the lagoon surface, and the liquid
planes at each lagoon were used to determine the lagoon emis- temperature at the air: liquid interface was probably closer to that
sions (Fig. 2). A valid VRPM estimate for a given measurement plane of the air at 2 m height than that at 0.3 m liquid depth.
required the mean wind direction to be less than 60◦ off the per- Daily mean emissions were estimated from the 1/2 h NH3 emis-
pendicular of the measurement plane, an upwind source fraction sions measurements at each farm. The fraction of 1/2 h emissions
of 0.9, a minimum wind speed of 1 ms−1 , and various other criteria needed to estimate a daily mean emission was estimated by first
to assure detection and capture of the emission plume within the simulating emissions at each farm based on the fit of the mea-
VRPM measurement plane. Complete criteria for the validation of surements to the Van’t Hoff’s equation describing the temperature
VRPM-derived NH3 emissions are described by Grant et al. (2013b). influence on NH3 solubility and NH4+ disassociation (Ni, 1999), and
A comparison between the bLS and VRPM emissions method- then comparing the simulated daily mean emissions for each day
ologies for farm lagoons on level land with minimal local wind with air temperature measurements to that resulting from an aver-
influences indicated that bLS emissions were on average 15% age of a subset of measurements during the day. The error in the
greater than VRPM emissions (Grant et al., 2013b). Emissions for estimated emissions at each farm for each subset of measurements
a given ½ h period were based on the average of bLS and VRPM (48 measurements each day) was less than 25% provided at least 25
R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84 79

½ h measurements were used to estimate the daily mean emissions


(Grant et al., 2013a).
Ammonia emissions in part depend on the concentration of NH3
dissolved in the lagoon liquid, which is dependent on a pH con-
trolled equilibrium with ammonium in the liquid. The ammoniacal
N represents between 80% and 92% of the total N in the lagoons
(DeRouchey et al., 2002; Szögi et al., 2005). Assuming that the
ammoniacal N fraction of the measured total N in the lagoon liq-
uid was 90%, the median upwind Cbkg (0.033–0.045 ␮mol mol−1 :
Table 6) was approximately 0.001% of the infrequently measured
values of Cliq (3.7–7.3 mmol mol−1 : Table 6). If there are no differ-
ences in the characteristics of lagoon: air interface so that the bulk
transport coefficient K function and NH3 solubility H function is the
same at each farm (the difference in emissions between farms must Fig. 4. Variation in pH with lagoon temperature.
be largely due to differences in meteorological conditions and Cliq ).
However since the lagoon N concentrations (Cliq ) were only mea- pended organic material in the lagoon. The Kads was set to 2 for
sured when lagoon liquid was being removed for land application sow farms and 4 for swine finishing lagoons based on the propor-
and no information was available for IN4, another measure of the tion of suspended organic material between farm types (Chastain
source strength at each farm was needed. Ammonia in the bulk liq- et al., 1999).
uid is a result of the breakdown of liquid and solid manure from The lagoon pH for each farm type was estimated from the lagoon
the pigs. The production of manure is commonly based on the diet, temperature as sigmoidal functions using TableCurve 2D® (SYSTAT
age, sex, gestational state, and mass of the animal (ASAE, 2005; Software Inc., Richmond, CA):
Koelsch, 2006). Estimates of the manure N excretion were made  
for each farm based on the age, sex, gestational state, and mass of pH = 7.8 + 0.32/ (1 + e(T −12)/1.07 ) (R2 = 0.56) (5)
the animal (Table 1). The LM-specific emissions can be described
for the sow farms and
as (Grant et al. 2013a):
1   
  1 pH = 7.6 + 0.52/ (1 + e(T −12)/1.21 ) (R2 = 0.46) (6)
QLM = Si /Mi ˛1 KL He−ˇ T
− 298.15K
, (2)
for the swine finishing farms.
where Si is area of the lagoon (m2 ) and Mi is the total live mass The annual emissions for each farm was estimated from daily
(AU) on farm i, KL is the bulk exchange coefficient, T is the air tem- emissions modeled using Eq. (3). The annual variation in daily
perature, ␣1 represents the nominal source strength of the lagoon mean temperature at the five farms was modeled as a first-order
liquid to emit NH3 at 25 ◦ C and nominal transport efficiency from Fourier series function of the measured daily mean temperatures
the bulk liquid to the turbulent boundary layer over the lagoon using TableCurve 2D® . The modeled annual temperature regimes
(4.39 g NH3 (liq) s m−3 ), and ␤ describes the influence of tempera- were typically similar in mean temperature but much different in
ture on the availability of NH3 gas at the liquid interface to overlying temperature range (Table 4). Mean wind speeds were assumed
air. The values for ␣1 and ␤ were determined by non-linear regres- as constant throughout the year (Table 4). Lagoon temperature
sion using TableCurve® 3D (SYSTAT Software Inc., Richmond, CA). at 0.3 m depth was modeled as a power function of air tempera-
The bulk transfer coefficient KL was assumed to be dominated by ture (Fig. 3). The NH3 -NH4 + dissociation constant Ka was modeled
transfer through the air above the liquid surface and equal for all according to DeVisscher et al. (2002) utilizing the modeled lagoon
lagoons given identical wind speed in accordance with Grant et al. temperature and the modeled pH according to Eqs. (5) and (6)
(2013a). The proposed KL of Ro and Hunt (2006) based on 13 studies below.
was also evaluated.
In an effort to better link processes with emissions, ␤ was set
3. Results and discussion
to the best estimate according to NIST (2011) and ␣1 , the source
strength of the lagoon liquid to emit NH3 , of Eq. (2) was expanded
Ammonia concentration, lagoon, and atmospheric measure-
to consider the anticipated NH3 available for emission based on the
ments were made at four of the five farm lagoons for approximately
calculated N loading rate and the farm type:
  1 1
 two weeks each quarter of the year for two years (2007–2009) and
QLM = Si /Mi ˛2 Lf FNH3 KL He−ˇ T
− 298.15K
, (3) over the course of an entire year (2007–2008) at the other farm
lagoon (Table 1). In general, operations at each farm were consis-
where ␣2 is a regression-fitted coefficient (3.45 s m−3 ), and Lf is tent throughout the study (Tables 2 and 3). Since emissions depend
N loading rate per LM for a farm f that is either a sow or finisher on total manure accumulation in the lagoons, there was no exclu-
farm (Table 1). The N loading rate was assumed equivalent to the sion of measurement periods during which there were no hogs due
N excretion rate and varies from a generic 205 g AU−1 d−1 (Rotz, to empty finisher barns (Table 3). Daily mean air temperatures and
2004) to a more animal-specific 0.280 g AU−1 d−1 for a finisher winds varied across the five farms with maximum temperatures
farm and 102 g AU−1 d−1 for a sow farm (ASAE, 2005). Changes ranging from 33 ◦ C to 42 ◦ C, minimum temperatures from −21 ◦ C to
in crude protein and phosphorus in the feed as well as feed effi- −7 ◦ C, and mean wind speeds from 1.4 ms−1 to 4.8 ms−1 (Table 4).
ciency can change the excretion rate by 50% (Koelsch, 2006). The The Koeppen-Geiger climate classification for the NC, IN and OK
fraction of N loading (as NH3 ) in the lagoon that was not adsorbed farm locations are warm temperate humid with warm summer
onto organic matter (FNH3 ) was estimated according to DeVisscher (Cfb), continental (Snow) humid with warm summer (Dfb), and
et al. (2002) and DeRouchey et al. (2002) based on the measured cold arid steppe (BSk), respectively. Lagoon conditions at 0.3-m

pH and temperature  depth according to:
of the lagoon at 0.3 m depth varied across the five farms with maximum temperatures
FNH3 = 0.9/ (1 + 10−pH )(1 + Kads )/Ka , (4)where 0.9 is the ranging from 32 ◦ C to 36 ◦ C, minimum temperatures from −1 ◦ C to
ammonical fraction of total N, Ka is the dissociation constant for 6 ◦ C, maximum pH from 8.2 to 8.4, and minimum pH from 6.5 to 7.5
the equilibrium of NH4 + and NH3 in a dilute solution, and Kabs (Table 5). Lagoon pH decreased with increasing lagoon temperature
is the equilibrium constant for the adsorption of NH3 onto sus- (Fig. 4).
80 R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84

Table 6
Comparison of emission methods.

Farm Median NH3 Median total N bLS emissions VRPM emissions bLS emissions VRPM emissions VRPM adjustmenta Combined
Cbg Cliq estimates estimates estimates estimates emissions
estimates

␮mol−1 mmol mol−1 # ½ h values # ½ h values # days # days (%) # days

OK4 0.035 5.7 142 114 50 40 +8 67


OK3 0.034 7.3 136 108 43 23 −5 59
IN4 0.044 3.7 71 31 18 2 +12 18
NC4 0.033 3.7 128 93 28 13 +57 30
NC3 0.045 5.5 114 132 10 8 +2 16
a
Grant et al. (2013a).

Fig. 5. Influence of air temperature on emissions. Circles, triangles, and diamond


indicate emissions with mean wind speeds less than 2 m/s at IN4, NC3, and NC4
respectively. Solid line represents Eq. (2) with ␣ = 66.08 g NH3 AU−1 d−1 and ␤ = 6408.

Due to the exclusion of measurement for some wind directions


and instrument failures, the number of emissions estimates was
significantly limited relative to the total time that measurements
were taken at a given farm. Approximately an equal number of Fig. 6. Modeled NH3 emissions. Measured values (filled diamonds) and modeled
values based on Eq. (2) (filled circles) and Eq. (3) (open circles and open diamonds)
½ h NH3 emissions measurements using the VRPM and bLS emis-
are indicated. Model bias is indicated as measured-modeled.
sion models passed quality control and assurance protocols at any
given farm (Table 6). Daily mean emissions were derived from the
½ h emission measurements using both the VRPM and bLS mod- less than the one-day emission measurement from another study
els, resulting in 59 daily mean emission measurements at OK3, 16 of 44 gd−1 hd−1 in North Carolina (Shores et al., 2005). The summer
at NC3, 67 at OK4, 30 at NC4, and 18 at IN4 (Table 6). Daily mean NH3 emissions of 17.0 g d−1 hd−1 at the OK finisher farm was com-
lagoon pH and temperature were calculated from continuous mea- parable to that found at swine finishing farms in Missouri (Zahn
surements taken during 31 of the measurement days at OK3, 14 at et al., 2001) and in North Carolina (Szögi et al., 2005).
NC3, 32 at OK4, 24 at NC4, and 0 at IN4. The annual mean live mass- and sow-specific emissions from
the sow lagoons varied from 61.9 g d−1 AU−1 (27.6 gd−1 hd−1 )
3.1. Measured daily mean emissions to 120 g d−1 AU−1 (55.1 gd−1 hd−1 ). The mean emission corre-
sponded to 50% to 99% of an estimated nitrogen (N) excretion of
The mean annual live mass- and hog-specific NH3 emis- 102 g d−1 AU−1 for sow and litter farms (ASAE, 2005). Sow-specific
sions from the finisher farm lagoons were 36.1 g d−1 AU−1 emissions varied from a summer maximum of from 113 g d−1 AU−1
(4.44 g d−1 hd−1 ) at NC3 and 104 g d−1 AU−1 (16.1 g d−1 hd−1 ) at (50.4 gd−1 hd−1 ) to 198 g d−1 AU−1 (90.8 gd−1 hd−1 ). The annual
OK3. The mean emission corresponded to between 14% and 42% mean sow-specific emissions at NC4 of 86.1 gd−1 hd−1 were much
of an estimated excretion of 205 g d−1 AU−1 (ASAE, 2005). The non- higher than the 1.9–5.2 gd−1 hd−1 reported for a farm in North
linear relationship between emissions and air temperature (Fig. 5) Carolina (Harper and Sharpe, 1998). Discrepancies between the
suggests that part of this difference in farm emissions is due to emissions may be due in part to the measurement approach and
the higher summer air temperatures and much higher mean wind the meteorological conditions on specific days of measurement,
speeds at the Oklahoma farms compared with those in North Car- as considerable variability existed (Fig. 6), and similar emissions
olina. In addition, the North Carolina lagoons emitted all winter due were found for the NC swine finishing farm reported here and those
to the lack of freezing while the freezing of the lagoon surface in reported in the literature (see above).
Oklahoma limited significant emission from the lagoon liquid dur- Lagoon pump-out events at the NC farms (NC3 and NC4) due
ing the winter resulting in a larger springtime NH3 concentrations. to the low evaporation/precipitation ratio and feed changes at the
The one valid day of NH3 emissions during the summer at the NC OK farms (OK3 and OK4) (Table 3) did not appear to influence NH3
finisher farm was 122 g d−1 AU−1 (15.0 gd−1 hd−1 ) (Table 4), much emissions.
R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84 81

Fig. 7. Daily mean emissions distributions. The measured emissions (panel A) are compared against the estimated emissions based on Eq. (2) (panel B) and Eq. (3) (panelC).

3.2. Modeled daily mean emissions

The combined LM-specific emissions for the lagoons


(n = 182 days) were strongly related to the approximated temper-
ature of the lagoon liquid:air interface (Fig. 3). The semi-empirical
model (Eq. (2)) accounted for 74% (model R2 = 0.74) of the daily
mean emissions variance (Fig. 6). This is comparable to the
explained variance of a similar semi-empirical model for the OK4
and OK3 farm alone (Grant et al., 2013a). This indicates that the
assumption of the LM as a proxy for lagoon loading and the dom-
inance of temperature and wind speed influences on emissions
from lagoons is likely valid across the country and not just in arid
Oklahoma. Measurements with the previously-discussed potential
moisture interference tended to be on the lower bound of the
generalized annual emissions pattern suggesting that the bias was
generally low rather than high (Fig. 6). The model had a mean
root mean squared error (RMSE) of ±29.98 g d−1 AU−1 and a mean
bias error (MBE) of −1.10 g d−1 AU−1 across all five farm lagoons
(Table 7). This MBE and RMSE is lower than the model reported
in Grant et al. (2013a) in part due to the greater number of daily
emission values included in the assessment. Overall, the model
described emission from the finisher lagoons (combined R2 = 0.79)
better than for the sow lagoons (combined R2 = 0.75). The greatest
bias in NH3 emission estimate was for the OK sow lagoon (Table 7).
Fig. 8. Influence of pH on modeled emissions bias (Panel A) and ammonia avail-
A comparison of the KL model (Eqs. (2), (3)) empirically derived ability (panel B). Values for swine finishing farms (filled circles) and breed-to-wean
by Grant et al., 2013a) and that proposed by Ro and Hunt (2006) farms (open circles) are indicated. Bias is indicated as measured-modeled.
showed a much lower coefficient of determination (R2 ) for the Ro
and Hunt model. This suggests that the KL describing the mixing OK3 around day of year 250 better than the semi-empirical model
within and over the relatively small lagoons was probably not as (Fig. 6). The mean RMSE across the four farms for the theo-
non-linearly related to wind speed as within and over open water retical model was ±46.31 g d−1 AU−1 , greater than that for the
without distinct berms. The regressed value for the temperature semi-empirical model of ± 29.98 g d−1 AU−1 . The theoretical model
coefficient (␤) in Eq. (2) was 6408 K. This coefficient was similar to accounted for 66% of the overall variance in the emissions mea-
that in Grant et al., 2013a; but greater than that expected to describe surements at four of the five farms (IN4 had no valid days with
NH3 solubility in the dilute water solution of the lagoon (nomi- emission estimates due to missing lagoon pH and temperature
nally 4200; NIST, 2011). The additional temperature influence may data). The RMSE was greater for the sow farms than the fin-
be associated with the dissociation constant for the equilibrium of isher farms: ±54.01 g d−1 AU−1 for the breed-to-wean farms and
NH4 + and NH3 . The dissociation constant (Ka ; Eq. (4)) for the equi- ±44.18 g d−1 AU−1 for the finishing farms. The inverse of the com-
librium of NH4 + and NH3 influences the fraction of N in the lagoon parison of modeled with measured emissions was used in the
that is present in solution as NH3 . In general, the pH influences fNH3 evaluation of the more complete mechanistic model of daily mean
(Eq. (4)), with greater pH associated with a higher fNH3 . However NH3 emissions for three farm lagoons in North Carolina and Georgia
since fNH3 is also influenced by temperature (Eq. (4)), low temper- (DeVisscher et al., 2002). Reversing the ordinates for the assessment
ature conditions reduce fNH3 under high pH conditions (Fig. 7a). of the more theoretical model resulted in an R2 of 0.67; comparable
DeVisscher et al. (2002) used a Kads (Eq. (4)) of 3 for their model- to the 0.69 determined by DeVisscher et al. (2002).
ing of the Harper and Sharpe (1998) measurements at a waste sow The MBE of the theoretical model for the five farms was
lagoon, while Zhang et al. (1994) used a value of 4 for finisher hog 0.30 g AU−1 d−1 , with no tendency for bias in a given direction by
waste. Since the total solids content of manure from feeder to fin- type of farm (Table 7). The model bias was inversely correlated with
ishing hogs is approximately twice that from breed to wean sows lagoon pH (Fig. 8): suggesting that the equilibrium of NH4 + and NH3
(Chastain et al., 1999), the Kads was optimized for the finisher and influenced the emissions. Lagoon pH (Eq. (4)) results primarily from
sow farms separately. The optimum Kads values were 4. the interaction of carbonate, ammonium, ammonia, carbonic acid,
The theoretical model (Eq. (3)), including the influence of NH3 carbon dioxide, acetate, and acetic acid (Sommer et al., 2006).
availability and solubility, simulated the positive skew of the emis- The higher RMSE for the swine finishing farms was a result of
sions distribution better than the semi-empirical model (Fig. 7). extreme model errors for three days- one at the NC3 farm and two at
Specifically, the theoretical model fit the extreme emissions of the OK3 farm. The high emissions overestimate at the NC3 lagoon
82 R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84

Table 7
Mean annual ammonia emissions.

Farm Measured emissions days Model Biasa (Eq. (2)) RMSE days Model bias (Eq. (3)) RMSE
−1 −1 −1 −1 −1 −2
g NH3 d AU g NH3 d hd g NH3 d m # g NH3 d−1 AU−1 g NH3 d−1 AU−1 # g NH3 d−1 AU−1 g NH3 d−1 AU−1

OK3 104.04 16.09 4.334 59 2.21 52.96 31 1.33 55.10


NC3 36.17 4.44 1.881 16 −9.50 15.62 14 −25.01 48.63
OK4 119.96 55.01 7.175 67 24.52 34.31 32 27.99 45.60
NC4 86.09 35.08 3.099 30 −16.52 33.95 24 −3.10 35.90
IN4 61.85 27.57 3.214 18 −6.23 13.04 0 – –
a
Modeled-measured.

Table 8
Predicted change in mean annual emissions for all sow farms with change in maximum and minimum annual daily mean temperature.

Maximum daily temperature

-2 o C -1 o C Nominal +1 o C +2 o C Mean

Minimum daily o
+2 C −10.6% −2.80% 4.40% 11.20% 17.60% 4.00%
temperature +1 o C −13.3% −5.30% 2.20% 9.20% 15.80% 1.70%
Nominal −16.00% −7.70% 0.00% 7.20% 14.00% −0.50%
-1 o C −18.5% −10.00% −2.10% 5.30% 12.30% −2.60%
-2 o C −21.1% −12.40% −4.30% 3.40% 10.50% −4.80%
mean −15.90% −7.70% 0.10% 7.30% 14.00%

occurred on 12 February, 2009 (day of year 43, Fig. 6); the first The Eq. (3) emissions model utilizing the pH models of Eq. (5)
valid daily mean emissions measurement after all animals were and (6) tended to underestimate the annual NH3 emissions. The
removed from the farm due to a disease outbreak (Table 3). The annual NH3 emissions at the IN, NC, and OK sow farms were esti-
lagoon pH during this period was relatively high (approx. 8.3), but mated by Eq. (3) at 9.6 kg NH3 AU−1 , 19.3 kg NH3 AU−1 , and 12.7 kg
not atypical for the time of year and the lagoon temperatures were NH3 AU−1 respectively. The corresponding mean measured emis-
not unusual. The theoretical model estimated a high fNH3 based on sions over the entire period of measurements was 22.6 kg NH3
the high pH and consequently estimated high emissions. Actual AU−1 , 31.4 kg NH3 AU−1 , and 43.8 kg NH3 AU−1 for the IN, NC, and
emissions were low and similar for the three other measurement OK sow farms respectively. The higher annualized mean measured
days with no hogs on site. Large underestimates of the emissions emissions compared to the modeled emissions was in part due to
at the OK3 lagoon occurred on 15 and 17 September 2007 (day of the variability in winds and the greater range in air temperatures
year 258 and 260; Fig. 6). Between 30 August and 18 September, during the measurements than modeled (Table 4). The emissions
2007 (Table 3), the lagoon temperature increased but the measured estimate for the OK sow farm is much lower than the estimate of
emissions were not correlated with lagoon temperature, pH, or fNH3 55.1 kg NH3 AU−1 for the same farms using different methodology
(Correlation coefficient, R < = 0.1). The theoretically-modeled emis- (Grant et al., 2013a). This may be in part due to narrower ranges
sions were of course correlated with all three variables through in modeled temperatures (maximum temperatures measured at
Eqs. (3) and (4). These measurements were made during a period 37.2 ◦ C and modeled at 26 ◦ C), modeled pH (modeled 7.9–8.1 versus
of feed change to Paylean® prior to the market weight finishing measured 6.5–8.4), and no range in modeled winds (Table 4), and/or
hogs being loaded out (Table 3). The addition of Paylean® enhances due to error in the assumed N excretion rate versus unknown actual
N retention in the feed and decreases swine manure N entering the rate associated with the specific diet used at the OK farm (Lynch
lagoon (DeCamp et al., 2001). Consequently, the effect of the feed et al., 2007).
change should be a decrease, not increase in emissions and does not The annual NH3 emissions at the NC and OK finisher farms
explain the modeled emissions discrepancy. Unfortunately there were estimated at 19.9 kg NH3 AU−1 and 32.3 kg NH3 AU−1 respec-
were no other measurement periods during which Paylean® was tively. The corresponding mean measured emissions over the entire
added to the feed. Unlike the theoretically-modeled emissions for period of measurements was 13.2 kg NH3 AU−1 , and 38.0 kg NH3
the swine finishing farms, there were no distinct outliers in the AU−1 for the NC, and OK sow farms respectively. The higher OK farm
errors in modeled emission for the sow farms. annualized mean measured emissions compared to the modeled
emissions was in part due to the variability in winds and the greater
range in air temperatures during the measurements (Table 4). The
estimated annual emission estimate for the NC finisher farm is
3.3. Climate impacts
much lower than the estimated 27.9 kg NH3 AU−1 for another NC
finishing farm (Szögi et al., 2005). This may be due to the limited
The impact of differing air temperatures and winds on maxi-
emission measurements by Szögi et al. (2005), a narrower range
mum and mean emissions of each farm was evaluated based on
in modeled pH (modeled 7.7–8.0 versus measured 6.9–8.4), and/or
the annual modeled daily emissions. The annual variation in daily
due to error in the assumed N excretion rate versus unknown actual
mean temperature at the five farms was modeled as a first-order
rate associated with the specific diet used at the NC farm (Lynch
Fourier series function of the measured daily mean temperatures
et al., 2007). The annual emission estimate for the OK finisher farm
using TableCurve 2D® . The modeled annual temperature regimes
is similar to the 38.4 kg NH3 AU−1 for the same farms using different
were typically similar in mean temperature but much different in
methodology (Grant et al., 2013a).
temperature range (Table 4). Mean wind speeds were assumed
Ammonia emissions is strongly influenced by the lagoon tem-
as constant throughout the year (Table 4). Lagoon temperature
perature, so year to year variations in air temperatures will strongly
at 0.3 m depth was modeled as a power function of air tempera-
influence predicted NH3 emissions. A warmer summer (increasing
ture (Fig. 3). The NH3 − NH4 + dissociation constant Ka was modeled
the annual maximum daily air temperature by 1 ◦ C) would increase
according to DeVisscher et al. (2002) utilizing the modeled lagoon
the maximum daily NH3 emission at both sow and finisher farms by
temperature.
R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84 83

Table 9
Predicted change in mean annual emissions for all finisher farms with change in maximum and minimum annual daily mean temperature.

Maximum daily temperature

-2 o C -1 oC Nominal +1 oC +2 oC Mean

Minimum daily +2 oC −10.5% −3.50% 3.20% 9.60% 15.60% 2.90%


temperature +1 oC −12.3% −5.20% 1.50% 8.00% 14.10% 1.20%
Nominal −14.00% −6.80% 0.00% 6.50% 12.70% −0.30%
-1 oC −16.0% −8.80% −1.80% 4.90% 11.20% −2.10%
-2 oC −18.1% −10.70% −3.60% 3.20% 9.60% −3.90%
mean −14.20% −7.00% −0.10% 6.40% 12.60%

istry was likely due to variations in turbulent mixing in and over


the lagoon. The wide range in possible N excretion rates strongly
influences the estimated NH3 emissions and needs to be measured
to validate the proposed emissions model.
The more theoretical model was used to simulate the effects
of changes in air temperature and wind speeds on hog emissions
for the five farms. If the maximum mean daily air temperature
increased 1 ◦ C (a warmer summer) the annual NH3 emissions was
estimated to increase approximately 7%, while if the entire year’s air
temperatures were warmer by 1 ◦ C, the annual NH3 emissions was
estimated to increase approximately 8.6%. As anticipated, increased
wind speeds result in increased emissions of both sow and finisher
lagoons- varying to some degree on the temperature climate.

Acknowledgements

Assistance with field measurements was made by J. Wolf, S. Cor-


tus, A. Lawrence, C. Fullerton, and D. Snyder. Without their help this
work could not have been possible. Thanks also go to the anony-
mous producers who were very helpful in providing information
and assistance at various points during the study. This research
was funded by livestock producers, the Agricultural Air Research
Fig. 9. Influence of wind speed on maximum daily emissions under the tempera-
ture conditions of the five measured farms based on theoretical model. Predicted Council, Inc. and the National Pork Board.
emissions for sow (open circle) and finisher (closed circle) farm lagoons are labeled
with Koeppen Climate classification of the location.
References

9.5% with the mean annual daily emission at sow farms increasing ASAE, 2005. Manure Production and Characteristics Standard D384.2. American
Society of Agricultural and Biological Engineers, St. Joseph, MI 49085.
by 7.2% (Table 8) and at finisher farms increasing by 6.5% (Table 9). Bajwa, K.S., Aneja, V.P., Arya, S.P., 2006. Measurement and estimation of ammonia
A cooler summer (1 ◦ C decrease in maximum daily air temperature) emissions from lagoon-atmosphere interface using a coupled mass transfer
would correspond with a 10.7% decrease in the maximum daily NH3 and chemical reactions model, and an equilibrium model. Atmos. Environ. 40,
S275–S286.
emission. The corresponding change in mean annual emissions at Centner, T.J., Patel, P.G., 2010. Reporting air emissions from animal production
the sow farms was a decrease of 7.7% (Table 8) while that at the activities in the United States. Environ. Int. 36, 237–242.
finisher farms a decrease of 6.8% (Table 9). The typically lower NH3 Chastain, J.P., Camberato, J.J., Albrecht, J.E., Adam, J., 1999. Swine manure
production and nutrient content. In: South Carolina Confined Animal Manure
emissions in the winter due to the lower temperatures reduced
Managers Certification Program. Clemson University, SC, pp. 1–17 (Chapter 3).
the impact of a warmer winter (1 ◦ C increase in minimum daily air Crenna, B.P., Flesch, T.K., Wilson, J.D., 2008. Influence of spurce-sensor geometry
temperature) would increase the mean annual emissions by only on multi-source emissions. Atmos. Environ. 42, 7373–7383.
DeCamp, S.A., Hankins, S.L., Carroll, A.L., Ivers, D.J., Richert, B.T., Sutton, A.L.,
2.1% at the sow farms (Table 8) and by only 1.8% at the finisher
Anderson, D.B., 2001. Effect of ractopamine and dietary crude protein on
farms (Table 9). The mean annual NH3 emissions were predicted to nitrogen and phosphorus excretion from finishing pigs. J. Anim. Sci. 79 (Suppl.
increase 9.2% at the sow farms and 8.0% at the finisher farms if the 1), 61.
climate warmed 1 ◦ C (annual mean temperature) across the coun- DeRouchey, J.M., Goodband, R.D., Nelssen, J.L., Tokach, M.D., Dritz, S.S., Murphy, J.P.,
2002. Nutrient composition of Kansas swine lagoons and hoop barn manure. J.
try, only 2.0% (sow farms) and 1.5% (finisher farms) more than that Anim. Sci. 80, 2051–2061.
predicted for a 1 ◦ C warmer summer. DeVisscher, A., Harper, L.A., Westerman, P.W., Liang, Z., Arogo, J., Sharpe, R.R.,
Increasing mean annual wind speed by 10% increased the max- VanCleemput, O., 2002. Ammonia emissions from anaerobic swine lagoons:
model development. J. Appl. Meteorol. 41, 426–433.
imum daily NH3 emissions from lagoons by 2–3% (Fig. 9) Gao, Z., Desjardins, R.L., van Haarlem, R.P., Flesch, T.K., 2008. Estimating gas
emissions from multiple sources using a backward lagrangian stochastic
model. J. Air Waste Manage. Assoc. 58 (11), 1415–1421.
4. Summary and conclusions
Grant, R.H., Boehm, M.T., Lawrence, A.F., Heber, A.J., 2013a. Ammonia emissions
from anaerobic treatment lagoons at sow and finishing farms in Oklahoma.
The semi-empirical model accounted for 73% of the variability in Agric. For. Meteorol. 180, 203–210, http://dx.doi.org/10.1016/j.agrformet.
2013.06.006.
daily mean emissions from five hog farms. A more theoretical model
Grant, R.H., Boehm, M.T., Lawrence, A.F., 2013b. Comparison of a
including nominal N excretion rates, theoretical NH3 solubility, and backward-Lagrangian Stochastic and Vertical Radial Plume Mapping methods
NH4 + dissociation chemistry based on lagoon temperatures and for estimating animal waste lagoon emissions. Agric. For. Meteorol. 180,
pH did a poorer job of explaining daily mean emissions variability 236–248, http://dx.doi.org/10.1016/j.agrformet.2013.06.013.
Harper, L.A., Sharpe, R.R., 1998. Ammonia emissions from swine lagoons in the
(approximately 50%). This might be expected as the large amount southeastern U.S. coastal plains. Final Report. Southern Piedmont Conservation
of ‘noise’ in the day-to-day emissions with similar lagoon chem- Agricultural Research Center, Watkinsville, GA. Submitted to Division of Air
84 R.H. Grant et al. / Agricultural and Forest Meteorology 228–229 (2016) 73–84

Quality, North Carolina Department of Environment and Natural Resources, Ro, K.S., Hunt, P.G., 2006. A new unified equation for wind-driven surficial oxygen
Raleigh, NC, December. transfer into stationary water bodies. Trans. Am. Soc. Agric. Biol. Eng. 49,
Harper, L.A., Sharpe, R.R., Parkin, T.B., de Visscher, A., van Cleemput, O., Byers, M., 1615–1622.
2004. Nitrogen cycling through swine production systems: ammonia, Rotz, A.C., 2004. Management to reduce nitrogen losses in animal production. J.
dinitrogen, and nitrous oxide emissions. J. Environ. Qual. 33, 1189–1201. Anim. Sci. 82 (E. Suppl), E119–E137.
Hashmonay, R.A., Varma, R.M., Modrak, M.T., Kagman, R.H., Segall, R.R., Sullivan, Shores, R.C., Harris, D.B., Thompson, E.L., Vogel, C.A., Natschke, D., Hashmonay, R.A.,
P.D., 2008. Radial plume mapping: a US EPA test method for area and fugitive Wagoner, K.R., Modrak, M., 2005. Plane-integrated open path Fourier
source emission monitoring using optical remote sensing. In: Kim, Y.J., Platt, U. transform infrared spectrometry methodology for anaerobic swine lagoon
(Eds.), Advanced Environmental Monitoring. Springer-Verlag Press, pp. 21–36. emission measurements. Appl. Eng. Agric. 21, 487–492.
Koelsch, R.K., 2006. Updated ASABE standard manure excretion standard. Sommer, S.G., Zhang, G.Q., Bannink, A., Chadwick, D., Misselbrook, T., Harrison, R.,
Conference Presentations and White Papers: Biological Systems Engineering, Hutchings, N.J., Menzi, H., Monteny, G.J., Ni, J.-Q., Oenema, O., Webb, J., 2006.
Paper 6 http://digitalcommons.unl.edu/biosysengres/6. (accessed 20.01.16). Algorithms determining ammonia emission from buildings housing cattle and
Laubach, J., Kelliher, F.A., 2005. Measuring methane emission rates of a dairy cow pigs and from manure stores. Adv. Agron. 89, 261–335.
herd (II): results from a backward-Lagrangian stochastic model. Agric. For. Szögi, A.A., Vanotti, M.B., Stansbery, A.E., 2005. Reduction of ammonia emissions
Meteorol. 129, 137–150. from treated anaerobic swine lagoon. In: Havenstein, G. (Ed.), Proceedings of
Lynch, M.B., Sweeney, T., Callan, J.J., Flynn, B., O’Doherty, J.V., 2007. The effect of the 2005 Animal Waste Management Symposium. Research Triangle Park, NC,
high and low dietary crude protein and inulin supplementation on nutrient USA, 5–7 October 2005, North Carolina State University, NC, USA.
digestibility, nitrogen excretion, intestinal microflora and manure ammonia Zahn, J.A., Tung, A.E., Roberts, B.A., Hatfield, J.L., 2001. Abatement of ammonia and
emissions from finisher pigs. Animal 1, 1112–1121, http://dx.doi.org/10.1017/ hydrogen sulfide emissions from a swine lagoon using a polymer biocover. J.
S1751731107000407. Air Waste Manage. Assoc. 51, 562–573.
NIST, 2011. Chemistry Webbook: Ammonia Material Measurement Laboratory. Zhang, R.R., Day, D.L., Christenson, L.L., Jepson, W.P., 1994. A computer model for
National Institute of Standards and Technology, http://webbook.nist.gov/cgi/ predicting ammonia release rates from swine manure pits. J. Agric. Eng. Res.
cbook.cgi?ID=C766441. (accessed 11.02.11). 58, 223–229.
Ni, J., 1999. Mechanistic models of ammonia release from liquid manure: a review.
J. Agric. Eng. Res. 72, 1–17.

You might also like