Evaluation of a backward Lagrangian stochastic model for determining surface ammonia emissions

Agricultural and Forest Meteorology 234 (2017) 196–202

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet

Evaluation of a backward Lagrangian stochastic model for determining surface ammonia emissions Wenliang Yang, Anning Zhu ∗ , Jiabao Zhang, Xiuli Xin, Xianfeng Zhang Fengqiu Agro-ecological Experimental Station, State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, 71 East Beijing Road, Nanjing 210008, China

a r t i c l e

i n f o

Article history: Received 22 July 2016 Received in revised form 26 December 2016 Accepted 2 January 2017 Keywords: Ammonia emission Backward Lagrangian stochastic model Open-path tunable diode laser Integrated horizontal flux method Static chamber method

a b s t r a c t The rate of ammonia emissions from a small circular plot of maize was estimated by three procedures: (i) by inverse dispersion (based on upwind and downwind laser concentration-detectors interpreted using a backward Lagrangian stochastic model); (ii) by the height-integrated horizontal flux measured by point flux-detectors arrayed along the axis of the plot (i.e., the integrated horizontal flux method, IHF); and (iii) by extrapolation from static flux chambers (SC). The results indicated that the estimates made by the open-path tunable diode laser (OPTDL) system combined with the backward Lagrangian stochastic (BLS) model were statistically equivalent to those made by the IHF method. The ammonia fluxes estimated by the OPTDL-BLS technique were only 2.3% higher than those from the IHF method. Although the OPTDL technique failed to monitor concentration differences at low ammonia fluxes due to its detection limit, the OPTDL-BLS technique estimated the total ammonia loss to be only 10.9% less than the IHF results. The SC method was found to underestimate ammonia emissions and cumulative ammonia loss significantly compared with both the OPTDL-BLS and IHF methods Although the ammonia emissions estimated by the OPTDL-BLS technique showed a similar emission pattern to those estimated by the IHF method, the former provided an opportunity to estimate the diurnal pattern of ammonia emissions and to understanding the primary driving factors. A clear diurnal cycle and a dominant net solar radiation dependence in ammonia emissions were found. © 2017 Elsevier B.V. All rights reserved.

1. Introduction Ammonia (NH3 ) is the most abundant alkaline gas in the atmosphere. It has many negative effects on ecosystem function and health, and on air quality. Deposition of NH3 and NH4 + ions is an important contributor to soil acidification, eutrophication of natural ecosystems, and nitrate leaching (Binkley and Richter, 1987; Schulze et al., 1989). Further, NH3 is a chemically active gas and readily reacts with sulfates and nitrates to form particulates of mean aerodynamic diameter smaller than 2.5 ␮m (PM2.5) (Asman et al., 1998) which have been implicated in human respiratory problems. Agriculture is the main source of atmospheric NH3 , accounting for some 55–56% of global NH3 emissi ons (Bouwman et al., 1997; Schlesinger and Hartley, 1992). The volatilization of NH3 following fertilizer application is recognized as an important path of nitrogen movement to the atmosphere. This not only reduces the efficiency of applied nitrogen-based fertilizers, but it

∗ Corresponding author. E-mail address: [email protected] (A. Zhu). http://dx.doi.org/10.1016/j.agrformet.2017.01.001 0168-1923/© 2017 Elsevier B.V. All rights reserved.

also results in financial loss to farmers. Accurate measurement of NH3 emissions is therefore pertinent and necessary. There are many methods of measuring ammonia emissions: the dynamic chamber method, the static chamber method, the micrometeorological mass balance method, the eddy covariance method, the relaxed eddy accumulation method, the flux gradient method, the wind tunnel method, and so on. Depending on the purpose of gathering this information, these methods may be grouped into two categories: chamber methods and micrometeorological methods. Due to the modification of the environment by the chamber, chamber methods are mainly used to compare NH3 losses from different treatments; whereas micrometeorological methods provide reliable estimates of ammonia emissions and are often used for the quantitative determination of ammonia losses. Flesch et al. (1995) described a backward Lagrangian stochastic (BLS) model that calculates emission rates from measurements of wind speed and upwind and downwind concentrations at a single height and is independent of the size and shape of the source. With gas release experiments, Flesch et al. (2004), Gao et al. (2009), McBain and Desjardins (2005) and Ro et al. (2013) evaluated the accuracy of the BLS model and offered some guidelines for its use.

W. Yang et al. / Agricultural and Forest Meteorology 234 (2017) 196–202

197

This technique has been successfully used to estimate ammonia (Flesch et al., 2007; McGinn et al., 2007) and methane emissions (Loh et al., 2008). However, few reports are available on use of the BLS model for estimating NH3 emissions from fertilized cropland. This present study compares estimated NH3 emissions from a circular plot treated with urea using the BLS model against estimates using the integrated horizontal flux (IHF) method and the static chamber (SC) method. The objective was to assess the accuracy and capability of the BLS technique for estimating NH3 emissions from farmland.

2. Materials and methods 2.1. Experimental site and set-up The experiment was conducted in a maize field adjacent to the Fengqiu Agro-Ecological Experimental Station of the Chinese Academy of Sciences in Fengqiu County, Henan Province, China (114◦ 24 E, 35◦ 00 N). The experimental site was flat, with no significant disturbing elements for several hundred meters in all directions. Commercial granular urea (total N ≥ 46.4%) was homogeneously applied by hand to the soil surface of a circular plot (radius 20 m) at a rate of 174 kg N ha−1 between 5:15 and 5:45 p.m. on June 18, 2012, during the seedling stage of maize. The experimental areas were irrigated with approximately 60 mm of water immediately after fertilizer application. The fields surrounding the experimental areas did not receive any N fertilizer during the experimental period to avoid the influence of NH3 emitted from the surrounding area.

2.2. Measurement of ammonia emissions Ammonia emissions were estimated using three methods: the BLS model, the IHF method and the SC method.

2.2.1. Backward Lagrangian stochastic model The free software WindTrax2.0 (Thunder Beach Scientific, Halifax, Canada) based on the BLS model was used to calculate ammonia emission rates. The emission rate was estimated by simulating the upwind transport trajectories of the gas particles from the concentration measurement location in the downwind emission plume back to the source area (Flesch et al., 1995, 2004). The BLS model is based on Monin–Obukhov similarity theory (MOST), which states that over short time intervals (e.g., 15–60 min) the wind properties in a horizontally homogenous surface layer (height z ≤ 50 m, but above a plant canopy) can be specified by four variables (Garratt, 1992): the atmospheric friction velocity u∗ , the Obukhov stability length L, the surface roughness length z0 , and the wind direction ˇ. Following Flesch et al. (2004), the line-averaged concentration (CL ) is assumed to be the average of P point concentrations spaced evenly along the measurement line, with the link to emission rate modeled by computing an ensemble of particle trajectories backward from each point. The BLS model then simulates the ratio of the concentration rise over the background to the emission rate, (C/Q)sim : 1 P P

(C/Q )sim =

i=1

1  2  N

|

wo

| ,

(1)

where N is the number of particles released at each point, P is the number of specific release points, and w0 is the vertical ‘touchdown’ velocity (i.e., the velocity of the gas particles as they reach the ground). The inner summation refers only to touchdowns within

Fig. 1. Illustration of the source-laser configuration. The ammonia emission source is shown by the shaded area.

the emission source. The emission rate QBLS (kg N ha−1 d−1 ) can therefore be calculated as: QBLS =

(CL − Cb ) (C/Q )sim

(2)

where Cb is the background concentration. In this study, the number of particles released N = 50, 000 and the number of release points P = 50 were used in the software. Each particle was followed horizontally for 500 m upwind. 2.2.2. Integrated horizontal flux method In this study, the IHF method, which is also known as the micrometeorological mass balance method (Denmead, 1983), was used as the standard reference method to estimate NH3 emissions. It equates the vertical flux of NH3 (i.e., the ammonia emissions) with the integrated horizontal flux (minus background flux) at a known downwind distance. As described by Denmead (1983) and Wilson et al. (1982), the net vertical flux of NH3 , QIHF (mg N m−2 s−1 ), was calculated using the trapezoidal rule: QIHF

1 = X



 0 zn

+· · ·

z

1 (F − Fb ) dz = X

 (F − Fb ) dz ,





z1

z2

(F − Fb ) dz + 0

(F − Fb ) dz z1

(3)

zn−1

where X (m) is the distance traveled by the wind over the fertilized area (i.e. the radius of the treated circle, 20 m in this case), z (m) is the height of the uppermost sampler, F (mg N m−2 s−1 ) is the mean horizontal flux at each sampling height and Fb is the background horizontal flux. Ammonia flux rates QIHF were reported in units of kg N ha−1 d−1 to be comparable with QBLS . The average horizontal flux of NH3 was measured by passive flux samplers coated internally with oxalic acid, which were described in detail by Leuning et al. (1985). In this study, six passive flux samplers at z = 0.4, 0.8, 1.2, 1.6, 2.0, and 2.5 m were mounted on a mast placed at the center of the circular plot. A sampling mast was placed 200 m west of the plot to make background measurements at the same height (Fig. 1). After exposure, ammonium oxalate in the sampler was eluted with 100 mL deionized water, and the solution was analyzed for NH4 + –N content by spectrophotometer (UV1601, Shimadzu, Japan) using the indophenol blue colorimetric method (Lu, 2000). The mean horizontal flux F was calculated as: F=

M , At

(4)

198

W. Yang et al. / Agricultural and Forest Meteorology 234 (2017) 196–202

2.3. Measurement of ammonia concentrations for the BLS model

Fig. 2. Diagram of static chamber instrument for capturing ammonia emitted from soil.

where M (mg N) is the mass (mg) of NH3 –N collected in the sampler during sampling period t (s) and A is the effective cross-sectional area of the sampler (m2 ) as determined in wind tunnel calibrations. Passive flux samplers were removed and replaced on the masts every 12 h from 6:00 p.m., June 18–6:00 p.m., June 27, 2012.

2.2.3. Static chamber technique Static chambers, of the type designed by Wang et al. (2004) and widely used to estimate ammonia emission from farmland in China, were used to collect ammonia volatilized from the soil surface. The chambers, constructed from opaque PVC tube (15 cm internal diameter × 10 cm high), were pushed approximately 5 cm into the soil. Two circular pieces of sponge (16 cm diameter × 2 cm thick) were impregnated with 15 mL 5% phosphoric acid in 4% glycerol solution and then placed in the chamber (Fig. 2). The lower sponge was intended to absorb ammonia emitted from the soil in the chamber, and the upper sponge was intended to absorb atmospheric ammonia and thus prevent contamination. At the end of the exposure period, the lower sponge was removed from the chamber and immediately placed into a zip lock bag. In the laboratory, the ammonia absorbed by the lower sponge was extracted with 500 mL of 1 M KCl solution after 60 min of oscillation and then analyzed for NH4 + –N using the indophenol blue colorimetric method (Lu, 2000) and a spectrophotometer (UV1800, Shimadzu, Japan). The ammonia emission rate QSC (kg N ha−1 d−1 ), was estimated as: QSC =

M2 , A2 D

(5)

where M2 (kg) is the mass of NH3 –N captured by the static chamber during each sampling; A2 is the cross-sectional area (ha) of the static chamber; and D is the duration (d) of each sampling. Before fertilizer application, four static chambers were placed randomly in the circular plot to monitor ammonia emissions from the source. Urea was applied separately to the soil surface in the chambers to reduce the effect of spatial variability of fertilization. The fertilization and irrigation rates in the chambers were equal to those in the whole plot. Three chambers were placed in the background area to measure background ammonia emissions. In this study, the sampling interval of the static chambers was always 12 h. The lower sponge was replaced with a new one at the same time as IHF samples were gathered, and the upper sponge was changed every two days. Static chamber sampling continued until 6:00 p.m., June 27, 2012.

2.3.1. Ammonia laser sensor NH3 concentrations were continuously monitored by an openpath tunable diode laser system (OPTDL, resolution 5 ppm-m) provided by the Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, China. The system consisted of three transmitter heads, three retroreflectors and a central unit containing the laser source, a reference cell and the control and processing electronics. A laser beam focused at the absorption wavelength of ammonia (1544 nm) was produced by a tunable infrared diode laser and split into two beams. The weaker beam passed through a reference cell to provide a continuous calibration update. The other beam was transferred by optical fiber to a transmitter head; the head propagated the beam through the atmosphere to a retroreflector, which then returned the beam to the transmitter head. The returning beam and the signal from the reference cell were then analyzed to determine the ammonia concentration along the measurement path. The system controlled three laser transmitters capable of 1000 m maximum path length (separation between the transmitter and retroreflector). In this study, the lengths of all the laser paths were 100 m, corresponding to a detection limit of approximately 0.05 ppm. Ammonia concentrations were recorded every 2 s and averaged into 30 min means, commencing at 6:00 p.m., June 18, 2012, and continuing for 216 h.

2.3.2. Sensor positions Two open-path lasers were used simultaneously to monitor ammonia concentrations and were designated as ‘laser path 1 and ‘laser path 2 (Fig. 1). All the laser heads and retroreflectors were set 1.2 m above the ground surface. In anticipation of the dominant southerly wind, laser path 2 was placed 25 m north of the center of the experimental plot to measure the ammonia concentration in the emission plume. Laser path 1 was placed 35 m south of the plot to detect the background ammonia concentration. Occasionally, there were northerly winds, which then reversed. The ammonia concentration measurement height is necessary to estimate ammonia flux by the BLS model. Following Laubach (2010), the measurement heights were reduced by the zero-plane displacement (d), which was assumed to be two-thirds of the average height of plants covering the ground. A height of d = 0.13 m was assumed for maize and wheat stubble, which were approximately 20 cm tall during the course of the experiment.

2.3.3. Meteorological observations A three-dimensional sonic anemometer (CSAT3, Campbell Scientific, Inc., Logan, Utah), positioned at 3.0 m above the ground, was used to measure the wind velocity components (u, v and w), turbulence velocity fluctuation statistics ( u / u* ,  v / u* and  w / u* ) and acoustic temperature. These data were sampled at a frequency of 10 Hz and recorded as 30-min averages. These statistics enabled the WindTrax software to calculate the friction velocity (u∗ ), the Obukhov stability length (L), the surface roughness length (z0 ) and the wind direction (ˇ). Solar radiation (CNR4, Kipp & Zonen, Delft, Netherlands) was obtained from the meteorological station located approximately 300 m from the experimental site. Solar radiation was recorded every 2 min and averaged into 30-min means. Soil temperature at 5 cm depth was continuously measured with soil temperature thermocouples (RC-30B, Jingchuang Instruments Ltd., Shanghai, China) and recorded every 2 s. Soil moisture content was determined by the gravimetric method, with samples dried in an oven at 105 ◦ C for 8 h.

W. Yang et al. / Agricultural and Forest Meteorology 234 (2017) 196–202

40

(a)

3

Temperature (°C)

-1

Wind speed (m s )

4

2

1

Air temperature Soil temperature

(b)

35 30 25 20 15

0 0

40

24

48

72 96 120 144 168 Hours after fertilization

192

0

216

24

0.30

(c) Soil water content (kg·kg -1 )

Net solar radiation (W m-2 )

199

30

20

10

48

72 96 120 144 168 192 216 Hours after fertilization

(d)

0.25

0.20

0.15

0.10

0 0

24

48

72 96 120 144 168 Hours after fertili zation

192

216

0

24

48

72 96 120 144 168 192 216 Hours after fertilization

Fig. 3. Weather conditions after fertilizer application: (a) wind speed, (b) air and soil temperatures, (c) net solar radiation, (d) soil water content.

2.4. Statistical analysis

3.2. Accuracy of concentration measurements

A statistical analysis of the experiment results was performed using SPSS 17.0 software and Microsoft Excel 2003. The extent of quantitative agreement between the values estimated by the three methods was evaluated by the regression slope and the coefficient of determination. The paired Student’s t-test was also performed to evaluate the difference between ammonia emission rates estimated by the three methods. The Pearson correlation coefficient was used to determine the linear influence of weather conditions upon ammonia emissions.

The main limitation of the OPTDL technique is the precision achievable by the instruments used for concentration measurement. To date, the OPTDL technique has generally been used to monitor ammonia or methane concentrations at feedlots, where the concentration of emitted gas is substantially higher than the background level and the sensitivity of the laser system is of less concern. For estimating low ammonia emissions from fertilizer applied to cropland, low NH3 concentrations relative to background levels definitely challenge the detection limit of the open-path tunable diode laser. As Fig. 4 illustrates, the concentration differences (CL –Cb ) were always above the 0.1 ppm detection limit of the openpath laser system in the initial 72 h (3 days) after fertilization, which ensured that the BLS method could be compared with the IHF method for that period. After 72 h, the concentration differences fell below the nominal sensitivity of the open-path laser system, except for 84–93 h after fertilization. Data points <0.1 ppm were therefore omitted from further assessment.

3. Results and discussion 3.1. Environmental data During the nine days of the experiment, wind speed (halfhourly average) ranged between 0.2 and 3.7 m s−1 , with a mean value of 1.4 m s−1 (Fig. 3a). The average wind speed during the day was 1.7–2.4 m s−1 , and the average at night was between 0.4 and 1.4 m s−1 . The wind directions measured over the experimental period were southeasterly (64%), northeasterly (27%) and southwesterly (9%). Air temperatures (half-hourly average) ranged between 22.9 ◦ C and 38.1 ◦ C, with a mean value of 30.0 ◦ C; soil temperatures at 5 cm (half-hourly average) ranged between 17.0 ◦ C and 30.9 ◦ C, with a mean value of 23.2 ◦ C (Fig. 3b). Net solar radiation (half-hourly average) ranged from 0 to 38.0 W m−2 ; the variation of daily mean net solar radiation was relatively small, except for 107–119 h after fertilization (Fig. 3c). Soil moisture content declined gradually after urea application because no rain fell during the experimental period (Fig. 3d).

3.3. Impact of environmental factors on ammonia emission The accuracy of emission estimates using the BLS model depends on the accuracy of atmospheric description by MOST (Flesch et al., 2004). However, MOST-based atmospheric descriptions are unreliable under light wind or under extremely stable conditions, casting doubt on the estimation of QBLS (Flesch et al., 2004). Following Ro et al. (2013), subsequent analysis was carried out after eliminating data with u* ≤ 0.15 m s−1 or |L| < 5 m. One of the advantages of the OPTDL-BLS technique is its ability to investigate environmental influences on NH3 emissions from cropland. Weather conditions governing NH3 emissions include air

200

W. Yang et al. / Agricultural and Forest Meteorology 234 (2017) 196–202

Table 1 Effect of air temperature, soil temperature, wind speed and net solar radiation on ammonia emission: n is the number of observations.

*

Environmental factor n

0–12 h 15

12–24 h 21

24–36 h 17

36–48 h 24

48–60 h 14

60–72 h 24

84–96 h 18

Air temperature Soil temperature Wind speed Net solar radiation

−0.568* 0.578* 0.265

0.254 0.522* 0.283 0.851**

0.770** 0.642** 0.612**

0.463* 0.591** 0.783** 0.858**

0.971** 0.859** 0.934**

0.436* 0.639** 0.535** 0.815**

0.477 0.528* 0.776** 0.836**

, ** Significant at P < 0.05 and P < 0.01, respectively.

temperature, soil temperature, wind speed, solar radiation, and rainfall (Meade et al., 2011; Sommer and Hutchings, 2001). Because no precipitation occurred during the experimental period, only the influences of air temperature, soil temperature, wind speed, and net solar radiation on the variability of ammonia emissions were investigated. As shown in Table 1, the effects of weather conditions on ammonia emissions were found to be different on each of the selected days, but a dominant net solar radiation relationship was always indicated in the daytime. There was no predominant factor influencing ammonia emissions at night. Clearly, the diel pattern of ammonia emissions was closely tied to atmospheric conditions. 3.4. Methodology assessment The IHF method is often used as the standard method when validating new methods for measuring ammonia emissions from farmland or feedlots (Sherlock et al., 1989; Sommer et al., 2005; Pacholski et al., 2006; Sanz et al., 2010). Sommer et al. (2005) and Sanz et al. (2010) used the IHF method as the reference method to assess the BLS model for estimating ammonia emissions from field-applied fertilizers or animal manure. Pacholski et al. (2006) used the IHF method as the micrometeorological reference method to calibrate a dynamic chamber method. In this study, twice-daily IHF ammonia flux measurements QIHF were made throughout the experimental period, and detailed OPTDL-BLS ammonia fluxes QBLS were calculated at 30-min intervals. To compare the two sets of measurements, QBLS was averaged at the same time scale as QIHF (i.e., 12-hourly means). The 12-h periods extended from 6:00 a.m. to 6:00 p.m. (daytime) or 6:00 p.m. to 6:00 a.m. (nighttime). Due to the limitations imposed by the detection limit of the laser system, only those ammonia emissions occurring in the initial three days were selected for validation of the OPTDL-BLS technique. In the initial 72 h and 84–96 h after fertilization (i.e., 168 half-hour periods), 27 observations were associated with u* ≤ 0.15 m s−1 or |L| < 5 m and 12 observations were below the

0.1 ppm detection limit, of which only 4 observations were taken during the day. More emissions data were filtered out during the night due to low wind speeds and more stable conditions. This may cause a bias in averaging the half-hourly QBLS over the 12-h period, especially for 12-hourly means at night. To create a proper 12hourly average emission rate, the observations were extrapolated to estimate half-hourly QBLS for periods with rejected data using the linear regression model. As shown in Table 1, the relationship of QBLS with air temperature, soil temperature, wind speed, and solar radiation was first determined by correlation analysis. Then the regression equation between QBLS and the predominant environmental factor affecting ammonia emissions was built for each 12-h period. Finally, the missing QBLS were estimated by the regression equations and measured meteorological data. The precision of the OPTDL-BLS technique relative to the IHF method was expressed by the coefficient of determination (R2 ); the relative accuracy was shown by the slope of the regression line, where 0 = no accuracy and 1 = maximum accuracy. Fig. 5 shows that the fitted regression line based on eight data points has a slope of 1.023 with an R2 of 0.993, which indicates a strong linear relationship between the ammonia flux densities determined by the two methods. The OPTDL-BLS technique tended to estimate larger emissions than the IHF method, but the slope of its fitted regression line did not differ significantly from unity (P < 0.01). A paired-comparison t-test also indicated no significant difference between the methods (P = 0.59). The results suggest that the OPTDLBLS technique can be used to estimate ammonia emission rates from cropland-applied fertilizers with acceptable accuracy. Until now, the OPTDL-BLS technique has mainly been used to estimate ammonia emissions from feedlots. Sommer et al. (2004) found that ammonia emission rates from a composting stockpile of cattle manure measured by the OPTDL-BLS technique agreed well with those from the IHF method. Laubach and Kelliher (2005) also found a strong correlation (R2 = 0.80) between the two methods, with a slope of 1.09 for measuring methane emissions from a fenced

0.6

0.4

0.2 0.1 0.0 0

24

48

72 96 120 144 Hours after fertilization

168

192

216

Fig. 4. Time series of 30-min average concentration differences (differences between measured concentrations and background) during the entire experimental period. The detection limit of the tunable diode laser system for concentration differences in this experiment (0.1 ppm) is shown by the dashed line.

BLS-technique (kg N ha-1 d-1)

NH3 concentration (ppm)

30 y = 1.023x

1:1

R2 = 0.993 20

10

0 0

10 20 -1 -1 IHF-technique (kg N ha d )

30

Fig. 5. Comparison of ammonia emissions using the BLS model and the IHF method. Solid line: linear regression constrained to pass through the origin. Dotted line: 1:1 line with slope of 1, indicating perfect agreement of techniques.

W. Yang et al. / Agricultural and Forest Meteorology 234 (2017) 196–202

40

30

20

10

0

40 -1

BLS-30min BLS-12h IHF SC

Cumulative NH3 loss (kg N ha )

NH3 emission (kg N ha -1 d-1)

50

201

30

20 BLS-30min IHF SC

10

0 0

24

48

72 96 120 144 168 192 216 Hours after fertilization

Fig. 6. Daily trend of ammonia emissions estimated with the BLS model, the IHF method and the SC method: solid line, 30-min average QBLS ; solid circles, QBLS at time intervals of 12 h; open squares, QIHF ; solid triangles, QSC .

paddock. Given the uncertainty in the accuracy of the IHF method, the absolute accuracy of QBLS remains unknown. 3.5. Efficiency of estimating diurnal patterns of ammonia emissions As Fig. 6 shows, the BLS and IHF methods resulted in similar flux rates and patterns. Due to the lack of air exchange, the microclimatic conditions generated in the static chamber minimized the influence of weather conditions on ammonia emissions. The SC method gave a lower emission rate and a completely different dynamic of ammonia emissions compared with the IHF method (Fig. 6). Based on 18 data points over the entire experimental period, the fitted regression line between QSC and QIHF had a slope of 0.532 with an R2 of 0.759. QIHF rose during the day and fell at night, whereas QSC had one emission peak only—that is, QSC increased quickly, reached a peak on the next day after fertilization, then gradually declined. The advantage of the OPTDL-BLS technique over the other two methods is its ability to evaluate the diurnal pattern of ammonia emissions. As Fig. 6 shows, the half-hourly ensemble of ammonia emissions in the initial 72 h and 84–96 h after fertilization revealed a dramatic diurnal emission cycle coincident with atmospheric conditions: a maximum at noonday and a minimum at midnight. Variability in emission rates was larger during the day than at night, which was attributed to the strong diurnal cycle in atmospheric conditions. The results of this research also display a similar diurnal pattern of NH3 emissions. Considering that the weather conditions were similar on each day of the experiment, the high degree of consistency in the diel emission cycles was not surprising. 3.6. Efficiency of estimating total ammonia loss Loubet et al. (2010) pointed out that cumulative ammonia losses following fertilization, evaluated as ammonia emissions integrated over time, is an important output of ammonia emission measurements and is also the basis for creating an ammonia emission inventory. Fig. 7 compares cumulative ammonia losses measured by the three methods. The total ammonia emissions measured by the IHF and SC methods were calculated every 12 h throughout the whole period of the experiment. Due to the detection limit of the open-path laser system, only ammonia emissions in the initial 72 h and 84–96 h after fertilization were counted in the BLS model.

0

24

48

72 96 120 144 168 192 216 Hours after fertilization

Fig. 7. Cumulative ammonia emission estimated with the BLS model, the IHF method and the SC method: solid line, cumulative ammonia loss calculated from 30min QBLS ; open squares, cumulative NH3 loss calculated from QIHF ; solid triangles, cumulative NH3 loss calculated from QSC .

Although the OPTDL technique failed to measure ammonia concentration differences at low flux, the total ammonia loss from the BLS model was comparable with that from the IHF method. The BLS cumulative ammonia loss calculated half-hourly amounted to 32.8 kg N ha−1 (18.9% of the total applied N), which was slightly less than the total 36.8 kg N ha−1 loss measured by the IHF method. The level of consistency between the total ammonia losses determined by the two methods provides a measure of confidence in the OPTDL-BLS technique for estimating total ammonia loss from farmland. As expected, the static chambers were found to underestimate cumulative ammonia loss compared with the IHF method. During the overall observation period, the total ammonia loss determined by the SC method was 24.3 kg N ha−1 , accounting for 14.0% of the total applied Ni et al. (2015) reported that the dynamic chamber method underestimated average total ammonia loss by 39.0% compared with estimates using the BLS model with open-path Fourier-transform infrared spectroscopy when ammonia emissions were measured after applying various fertilizers to farmland. Pacholski et al. (2006) also found cumulative NH3 loss after urea application to maize, amounting to 2.8 kg N ha−1 as measured by the Dräger-Tube Method, were much lower than those found by the IHF method, which reached 38.2 kg N ha−1 . In the North China Plain, ammonia emissions from fertilizer application have received significant attention from both agricultural and environmental scientists. However, only a few studies have used micrometeorological methods to measure ammonia emissions after fertilization in this region. Cai et al. (2002) used the IHF method to measure ammonia emissions following urea application (75 kg N ha−1 ) at the seedling stage of maize and found that 18% of the applied N was lost by NH3 volatilization when using surface broadcast with irrigation treatment. Using the Bowen ratio method, Zhang et al. (2005) found that the total ammonia loss was 26.6% when 157 kg N ha−1 of urea was surface-applied to maize at the jointing stage after rain. Su et al. (2007) used the wind tunnel method to estimate ammonia emissions after urea surface application followed by irrigation at the seedling and jointing stage of maize and found that cumulative ammonia loss was 22.1% and 23.7% at fertilization rates of 100 and 200 kg N ha−1 respectively. The relatively high ammonia loss in the study of Zhang et al. (2005) was primarily due to different fertilizer application methods. Many studies have demonstrated that rainfall or irrigation after

202

W. Yang et al. / Agricultural and Forest Meteorology 234 (2017) 196–202

fertilization suppresses NH3 emissions by filling the pores in the soil matrix, which hinders diffusion of NH3 gas from the soil to the air and increases infiltration of fertilizer into the soil (Roelcke et al., 2002; Mkhabela et al., 2009; Sanz-Cobena et al., 2011). However, surface application of fertilizer after rain will leave fertilizer on the soil surface and promote ammonia emissions. This study, as well as those of Cai et al. (2002) and Su et al. (2007), indicate that total ammonia emission losses after fertilization during the maize growing season are approximately 20% of applied N for the locally common fertilizer application technique (i.e., followed by irrigation). 4. Conclusions The BLS model combined with the open-path tunable diode laser system for estimating ammonia emission was successfully field-tested in an experiment where ammonia emissions after urea application to maize were continuously measured over a period of nine days. The OPTDL-BLS technique and the IHF method were comparable when estimating ammonia emission rate and total ammonia loss. The OPTDL-BLS technique was unable to estimate low ammonia flux, but the total ammonia loss obtained by the OPTDL-BLS technique was only 10.9% less than that from the IHF method. The SC method significantly underestimated ammonia flux and cumulative ammonia loss compared with the IHF method. The SC method also measured a completely different dynamic of ammonia emissions. Measurements of ammonia concentration with high temporal resolution helped to quantify diurnal variations in ammonia emissions and also made it possible to identify the dominant factor influencing ammonia emissions. The results demonstrated a clear diurnal cycle in ammonia emissions and confirmed that the pattern of ammonia emissions was closely tied to atmospheric conditions. Acknowledgements This study was financially supported by the National Key Research and Development Program of China (No. 2016YFD0200304), the National Natural Science Foundation of China (No. 41501327, 41471239), the Natural Science Foundation of Jiangsu Province (No. BK20151055), and the Science and Technology Service Network Initiative (No. KFJ-SW-STS-142). References Asman, W.A.H., Sutton, M.A., Schjørring, J.K., 1998. Ammonia Emission, atmospheric transport and deposition. New Phytol. 139, 27–48. Binkley, D., Richter, D., 1987. Nutrient cycles and H+ budgets of forest ecosystems. Adv. Ecol. Res. 16, 1–51. Bouwman, A.F., Lee, D.S., Asman, W.A.H., Dentener, F.J., Van Der Hoek, K.W., Olivier, J.G.J., 1997. A global high-resolution emission inventory for ammonia. Global Biogeochem. Cycles 11, 561–587. Cai, G.X., Chen, D.L., White, R.E., Fan, X.H., Pacholski, A., Zhu, Z.L., Ding, H., 2002. Gaseous nitrogen losses from urea applied to maize on a calcareous fluvo-aquic soil in the North China Plain. Aust. J. Soil Res. 40, 737–748. Denmead, O.T., et al., 1983. Micrometerological method for measuring gaseous losses of nitrogen in the field. In: Freney, J.R. (Ed.), Gaseous Loss of Nitrogen from Plant-Soil Systems. Martinus Nijhoff/Dr W. Junk Publishers, Hague, pp. 133–158. Flesch, T.K., Wilson, J.D., Yee, E., 1995. Backward-time Lagrangian stochastic dispersion models and their application to estimate gaseous emissions. J. Appl. Meteorol. 34, 1320–1332. Flesch, T.K., Wilson, J.D., Harper, L.A., Crenna, B.P., Sharpe, R.R., 2004. Deducing ground-air emissions from observed trace gas concentrations: a field trial. J. Appl. Meteorol. 43, 487–502. Flesch, T.K., Wilson, J.D., Harper, L.A., Todd, R.W., Cole, N.A., 2007. Determining ammonia emissions from a cattle feedlot with an inverse dispersion technique. Agric. For. Meteorol. 144, 139–155. Gao, Z.L., Mauder, M., Desjardins, R.L., Flesch, T.K., Van Haarlem, R.P., 2009. Assessment of the backward Lagrangian Stochastic dispersion technique for

continuous measurements of CH4 emissions. Agric. For. Meteorol. 149, 1516–1523. Garratt, J.R., 1992. The Atmospheric Boundary Layer. Cambridge University Press, Cambridge, pp. 316. Laubach, J., Kelliher, F.M., 2005. Measuring methane emission rates of a dairy cow herd (II): results from a backward-Lagrangian stochastic model. Agr. For. Meteorol. 129, 137–150. Laubach, J., 2010. Testing of a Lagrangian model of dispersion in the surface layer with cattle methane emissions. Agr. For. Meteorol. 150, 1428–1442. Leuning, R., Freney, J.R., Denmead, O.T., Simpson, J.R., 1985. A sampler for measuring atmospheric ammonia flux. Atmos. Environ. 19, 1117–1124. Loh, Z., Chen, D.L., Bai, M., Naylor, T., Griffith, D., Hill, J., Denmead, T., McGinn, S., Edis, R., 2008. Measurement of greenhouse gas emissions from Australian feedlot beef production using open-path spectroscopy and atmospheric dispersion modeling. Aust. J. Exp. Agric. 48, 244–247. Loubet, B., Génermont, S., Ferrara, R., Bedos, C., Decuq, C., Personne, E., Fanucci, O., Durand, B., Rana, G., Cellier, P., 2010. An inverse model to estimate ammonia emissions from fields. Eur. J. Soil Sci. 61, 793–805. Lu, R.K., 2000. Analysis Methods in Soil Agrochemistry. Agricultural Science Press, Beijing, pp. 129–133 (in Chinese). McBain, M.C., Desjardins, R.L., 2005. The evaluation of a backward Lagrangian stochastic (bLS) model to estimate greenhouse gas emissions from agricultural sources using a synthetic tracer source. Agric. For. Meteorol. 135, 61–72. McGinn, S.M., Flesch, T.K., Crenna, B.P., Beauchemin, K.A., Coates, T., 2007. Quantifying ammonia emissions from a cattle feedlot using a dispersion model. J. Environ. Qual. 36, 1585–1590. Meade, G., Pierce, K., O’Doherty, J.V., Mueller, C., Lanigan, G., Mc Cabe, T., 2011. Ammonia and nitrous oxide emissions following land application of high and low nitrogen pig manures to winter wheat at three growth stages. Agric. Ecosyst. Environ. 140, 208–217. Mkhabela, M.S., Gordon, R., Burton, D., Smith, E., Madani, A., 2009. The impact of management practices and meteorological conditions on ammonia and nitrous oxide emissions following application of hog slurry to forage grass in Nova Scotia. Agric. Ecosyst. Environ. 130, 41–49. Ni, K., Köster, J.R., Seidel, A., Pacholski, A., 2015. Field measurement of ammonia emissions after nitrogen fertilization—a comparison between micrometeorological and chamber methods. Eur. J. Agron. 71, 115–122. Pacholski, A., Cai, G.X., Nieder, R., Richter, J., Fan, X.H., Zhu, Z.L., Roelcke, M., 2006. Calibration of a simple method for determining ammonia volatilization in the field − comparative measurements in Henan Province, China. Nutr. Cycl. Agroecosyst. 74, 259–273. Ro, K.S., Johnson, M.H., Stone, K.C., Hunt, P.G., Flesch, T.K., Todd, R.W., 2013. Measuring gas emissions from animal waste lagoons with an inverse-dispersion technique. Atmos. Environ. 66, 101–106. Roelcke, M., Li, S.X., Tian, X.H., Gao, Y.J., Richter, J., 2002. In situ comparisons of ammonia volatilization from N fertilizers in Chinese loess soils. Nutr. Cycl. Agroecosyst. 62, 73–88. Sanz, A., Misselbrook, T., Sanz, M.J., Vallejo, A., 2010. Use of an inverse dispersion technique for estimating ammonia emission from surface-applied slurry. Atmos. Environ. 44, 999–1002. Sanz-Cobena, A., Misselbrook, T., Camp, V., Vallejo, A., 2011. Effect of water addition and the urease inhibitor NBPT on the abatement of ammonia emission from surface applied urea. Atmos. Environ. 45, 1517–1524. Schlesinger, W.H., Hartley, A.E., 1992. A global budget for atmospheric NH3 . Biogeochemistry 15, 191–211. Schulze, E.D., De Vries, W., Hauhs, M., Rosén, K., Rasmussen, L., Tamm, C.-O., Nilsson, J., 1989. Critical loads for nitrogen deposition on forest ecosystems. Water Air Soil Pollut. 48, 451–456. Sherlock, R.R., Freney, J.R., Smith, N.P., Cameron, K.C., 1989. Evaluation of a sampler for assessing ammonia losses from fertilized fields. Fert. Res. 21, 61–66. Sommer, S.G., Hutchings, N.J., 2001. Ammonia emission from field applied manure and its reduction—invited paper. Eur. J. Agron. 15, 1–15. Sommer, S.G., McGinn, S.M., Hao, X., Larney, F.J., 2004. Techniques for measuring gas emissions from a composting stockpile of cattle manure. Atmos. Environ. 38, 4643–4652. Sommer, S.G., McGinn, S.M., Flesch, T.K., 2005. Simple use of the backward Lagrangian stochastic dispersion technique for measuring ammonia emission from small field-plots. Eur. J. Agron. 23, 1–7. Su, F., Ding, X.Q., Gao, Z.L., Huang, B.X., Chen, X.P., Zhang, F.S., Kogge, M., Römheld, V., 2007. Ammonia volatilization from nitrogen fertilization of winter wheat-summer maize rotation system in the North China Plain, China. Environ. Sci. 27 (3), 409–413 (in Chinese). Wang, Z.H., Liu, X.J., Ju, X.T., Zhang, F.S., Malhi, S.S., 2004. Ammonia volatilization loss from surface-broadcast urea: comparison of vented- and closed-chamber methods and loss in winter wheat-summer maize rotation in north China plaini. Commun. Soil Sci. Plant 35 (19/20), 2917–2939. Wilson, J.D., Thurtell, G.W., Kidd, G.E., Beauchamp, E.G., 1982. Estimation of the rate of gaseous mass transfer from a surface source plot to the atmosphere. Atmos. Environ. 16, 1861–1868. Zhang, Y.M., Hu, C.S., Dong, W.X., 2005. Ammonia volatilization from wheat-maize rotation field in the piedmont of Taihang. Plann. Nutr. Fert. Sci. 11 (3), 417–419 (in Chinese).