Published June 25, 2014

Journal of Environmental Quality

TECHNICAL REPORTS Atmospheric Pollutants and Trace Gases

Particulate Emissions from a Beef Cattle Feedlot Using the Flux-Gradient Technique Henry F. Bonifacio, Ronaldo G. Maghirang,* Steven L. Trabue, Laura L. McConnell, John H. Prueger, and Edna B. Razote

A

ir pollutant emissions from concentrated

Data on air emissions from open-lot beef cattle (Bos taurus) feedlots are limited. This research was conducted to determine fluxes of particulate matter with an aerodynamic diameter ≤10 mm (PM10) from a commercial beef cattle feedlot in Kansas using the fluxgradient technique, a widely used micrometeorological method for air emissions from open sources. Vertical PM10 concentration profiles and micrometeorological parameters were measured at the feedlot using tapered element oscillating microbalance PM10 samplers and eddy covariance instrumentations (i.e., sonic anemometer and infrared hygrometer), respectively, from May 2010 through September 2011, representing feedlot conditions with air temperatures ranging from −24 to 39°C. Calculated hourly PM10 fluxes varied diurnally and seasonally, ranging up to 272 mg m−2 h−1, with an overall median of 36 mg m−2 h−1. For warm conditions (air temperature of 21 ± 10°C), the highest hourly PM10 fluxes (range 116–146 mg m−2 h−1) were observed during the early evening period, from 2000 to 2100 h. For cold conditions (air temperature of −2 ± 8°C), the highest PM10 fluxes (range 14–27 mg m−2 h−1) were observed in the afternoon, from 1100 to 1500 h. Changes in the hourly trend of PM10 fluxes coincided with changes in friction velocity, air temperature, sensible heat flux, and surface roughness. The PM10 emission was also affected by the pen surface water content, where a water content of at least 20% (wet basis) would be sufficient to effectively reduce PM10 emissions from pens by as much as 60%.

animal feeding operations (CAFOs) such as openlot beef cattle feedlots can adversely affect air quality locally in downwind areas (National Research Council, 2003); these emissions also may affect air quality on a regional scale. Emissions from CAFOs generally include NH3, H2S, greenhouse gases (GHGs), volatile organic compounds (VOCs), and particulate matter (PM) (National Research Council, 2003). Based on model farms, NH3 makes up the majority (70–94%) of air emissions from CAFOs, followed by PM and N2O, each contributing up to 20% of the total emissions, while H2S and VOCs can have contributions of up to 15 and 5%, respectively (USEPA, 2001). The National Research Council (2003) stated the need for accurate pollutant emissions estimates for CAFOs that can be used to assess their impact on the environment and regulate them effectively. In 2005, the USEPA initiated the National Air Emissions Monitoring Study (NAEMS) that aimed to address uncertainties in emissions calculated for CAFOs (USEPA, 2005a). The NAEMS was a 2-yr study in which emissions of regulated pollutants were measured from different types of CAFOs for development and improvement of emission estimating methodologies for CAFOs (USEPA, 2005a; Purdue Applied Meteorology Laboratory, 2009). The NAEMS also intended to bring the participating CAFOs, which represented the layer, broiler, swine, and dairy industries, into compliance with applicable environmental regulations (USEPA, 2005a). Open-lot beef cattle feedlots were not included because NAEMS focused on nonfugitive emissions only; however, as stated by the USEPA (2005a), fugitive emissions, which would include feedlot emissions, would be studied next. To ensure that a wide range of scientific data would be used, in 2011, the USEPA solicited quality-assured CAFO emissions data on PM (≤2.5-mm particulate matter, PM10, and total suspended particulates), H2S, NH3, and VOCs from broiler, layer, turkey, swine, dairy, and this time, beef operations to supplement those collected through NAEMS (USEPA, H.F. Bonifacio and R.G. Maghirang, Dep. of Biological and Agricultural Engineering, Kansas State Univ., Manhattan, KS 66506; S.L. Trabue and J.H. Prueger, USDA–ARS, National Lab. for Agriculture and the Environment, Ames, IA 50011; L.L. McConnell, USDA–ARS, Environmental Management and Byproduct Utilization Lab., Beltsville, MD 20705; and E.B. Razote, Bureau of Air, Kansas Dep. of Health and Environment, Topeka, KS 66612. H.F. Bonifacio, present address: USDA–ARS, Pasture Systems and Watershed Management Research Unit, University Park, PA 16802. Assigned to Associate Editor Robert Dungan.

Copyright © American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. 5585 Guilford Rd., Madison, WI 53711 USA. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. J. Environ. Qual. 42:1341–1352 (2013) doi:10.2134/jeq2013.04.0129 Received 12 Apr. 2013. *Corresponding author ([email protected]).

Abbreviations: PM, particulate matter; PM10, particulate matter with a diameter of 10 micrometers or less.

1341

2011). Evidently, more gaseous and PM emission estimates are needed for CAFOs, particularly for open-lot beef cattle feedlots. Techniques appropriate for estimating emission rates from area sources include micrometeorological techniques, mass balance techniques, atmospheric dispersion models, and atmospheric tracers (National Research Council, 2003). Recently published pollutant emission rates for beef cattle feedlots were determined using atmospheric dispersion models such as WindTrax, a backward Lagrangian stochastic-based model (Flesch and Wilson, 2005), and AERMOD, a Gaussianbased and the current USEPA-preferred regulatory model applicable to primary pollutants and toxic air emissions (USEPA, 2005b). WindTrax has been used in quantifying emission rates of NH3 (Flesch et al., 2005), odor (Galvin et al., 2006), and PM10 (McGinn et al., 2010) from beef cattle feedlots and GHGs from dairy cattle facilities (Leytem et al., 2011), while AERMOD also has been applied in feedlot studies on NH3 (Faulkner et al., 2007) and PM10 (Bonifacio et al., 2012). Micrometeorological techniques have long been used to quantify emission rates of various gases such as fumigants from agricultural croplands, which, like beef cattle feedlots, are openarea sources. Although these techniques require complex and extensive instrumentation, they are the most direct, unobtrusive methods of measuring mass and energy transfer rates between the surface and the atmosphere (Ham and Baum, 2007). A commonly used micrometeorological method for determining emissions is the flux-gradient technique (Denmead et al., 1974; Kanemasu et al., 1979; Prueger and Kustas, 2005; Muller et al., 2009). The flux-gradient method has been used to estimate emissions for NH3 (Myles et al., 2011), HNO3 (Myles et al., 2011), ozone (Muller et al., 2009), SO2 (Myles et al., 2011), and pesticides (Prueger et al., 2005) for agricultural lands. This method also has been applied to cattle feedlots to quantify emissions of amines (Hutchinson et al., 1982), NH3 (Baek et al., 2006; Hutchinson et al., 1982), and H2S (Baek et al., 2006) and, on a small scale, has simulated a cattle pen to measure CH4 (Harper et al., 1999). Limitations associated with the use of micrometeorological techniques on feedlots, however, were not directly addressed in these studies. Micrometeorological techniques are based on certain key assumptions. One key assumption is the horizontal homogeneity of the source and, consequently, its emission rates. Although feedlots are nonideal locations and are made up of different types of surfaces such as pens (the largest surface area), unpaved roads, waste storage structures, and buildings, Baum et al. (2008) indicated that micrometeorological techniques can be applied at cattle feedlots. Another key assumption is that the source must be flat, and field measurements of surface roughness from a previous study (Baum et al., 2008) and the present study have suggested that the feedlot surface is relatively even. The source must also have adequate fetch (i.e., downwind distance) such that concentration measurements would not be affected by emissions from other sources. The conventional fetch/measurement height ratio is 100:1 (Wilson and Shum, 1992), but as summarized by Baum et al. (2008), lower minimum fetch/measurement height ratios of 15:1 to 75:1 could be applied based on several studies. The present study was designed to quantify PM10 emissions from an open-lot commercial beef cattle feedlot under a variety of meteorological and cattle pen moisture conditions. It is 1342

the first use of the flux-gradient technique for PM from cattle feedlots. Vertical profiles of PM10 concentrations and highresolution meteorological measurements were used to compute the concentration gradients and particle eddy diffusivity, respectively, required in the flux-gradient technique. Results of this work provide critical information for producers, conservation specialists, and regulators on the magnitude of PM10 emissions, derived using a micrometeorological technique, from a cattle feedlot typical of those in much of the western United States.

Materials and Methods Feedlot Description A 1.7- by 0.5-km commercial beef cattle feedlot was studied, with the longer length in the north–south direction (Fig. 1a). Based on a previous study (Baum et al., 2008), this feedlot may be considered relatively flat, with surface roughness of 4.1 ± 2.2 cm. The feedlot has a capacity of 30,000 cattle in a total pen area of approximately 0.59 km2 (59 ha) surrounded by agricultural croplands. Field monitoring, which included PM10 concentrations and micrometeorological measurements, was conducted continuously from May 2010 through September 2011; however, measurement data, which were also used in another study (Bonifacio et al., 2013), were incomplete in some months due to several instrumentation- and weatherrelated problems. Measurement data completeness was 87% for micrometeorological parameters and 46 to 61% for PM10 concentrations. In 2010, the average head capacity at the feedlot was 27,000; for the whole year, the estimated mean percentage of empty pens was only about 10%. Dust control methods for that year included a manure scraping frequency of two to three times per pen, equivalent to pen scraping of once every 4 to 6 mo, and water application on unpaved roads and alleys. In 2011, the average head capacity at the feedlot was lower at 25,000, so many more pens were empty (approximately 18%). In addition, manure scraping and water application practices changed in 2011. Pen surfaces were scraped more frequently (≥3 times per pen) than the previous year. More important, water was applied on pens rather than on unpaved roads and alleys to alleviate heat stress on the cattle.

Micrometeorological Measurements A 5.3-m tower, equipped with micrometeorological and eddy covariance (EC) instrumentation, was installed to measure micrometeorological conditions at the feedlot. The tower was set up in a pen approximately 0.4 and 1.3 km away from the north and south edges of the feedlot, respectively (Fig. 1a). The EC instrumentation included a three-dimensional sonic anemometer (Campbell Scientific) for measuring the three orthogonal wind velocity components (ux, uy, and uz) and air temperature and an infrared hygrometer (LI-COR Biosciences) for measuring water vapor density. A datalogger (Campbell Scientific) was used to measure and record variances and covariances of ux, uy, and uz and sonic air temperature as 15-min averages. Friction velocity (u*), Monin–Obukhov length (L), and surface roughness (zo) were computed from these measurements using formulations presented by Flesch et al. (2004) and Baum et al. (2008). These Journal of Environmental Quality

height and the other at the 7.62-m height. A separate datalogger (Campbell Scientific) was used to record wind velocity measurements as 20-min and hourly averages.

Particulate Matter Concentrations Tapered element oscillating microbalance (TEOM) PM10 monitors (Series 1400a, Thermo Fisher Scientific; federal equivalent method designation no. EQPM-1090-079) were used to measure the PM10 mass concentration at three sites simultaneously: (i) within the feedlot, approximately 5.5 m north of the EC tower; (ii) 5 m away from the north edge of the feedlot; and (iii) 800 m away from the feedlot south edge. Selection of these sites was based on feedlot layout, power availability, and feedlot management approval. Cross-calibration of the TEOM PM10 monitors showed slight variations between reported concentrations; for PM10 concentrations of 120 mg m−3 or less, readings between any two monitors varied by 100 g each were placed in separate sealed plastic bags. These bags were placed in an icebox while in the field and then transferred to a freezer on arrival at the laboratory. Within 7 to 8 d after sampling, the moisture of the samples was determined using the ASTM D2216-98 oven-drying method (American Society for Testing Materials, 2002).

Data Screening Before PM10 flux calculation, hourly data points, each composed of PM10 concentration and micrometeorological measurements, were screened based on: (i) the corresponding fetches of sampling heights with computed net PM10 concentrations; (ii) the number of sampling heights with net PM10 concentrations; and (iii) the vertical profile of net PM10 concentrations. The first screening was based on the fetch of each sampling height. For the measured PM10 concentration to be representative of the PM emitted by the pens, its footprint should fall within the feedlot boundary. This was accomplished by computing the matching fetch of the footprint for each sampling height, using the expression

x=

æ z ÷öP çç u ÷ [1] D ç ÷ k2 ln ( F So ) çè L ÷ø -L

where x is the fetch (m), |L| is the absolute value of the Monin– Obukhov length (m), F/So is the desired normalized flux (i.e., the ratio of measured flux to total source flux), zu is the new length scale (m), k is the von Karman constant (0.4), and D and P are similarity constants (Hsieh et al., 2000). The new length scale, zu, was derived as (Hsieh et al., 2000) é æz ö z ù zu = zm êê ln ççç m ÷÷÷-1 + o úú [2] ç ÷ zm ûú ëê è zo ø where zm is the sampler measurement height. Both L and zo were derived from sonic anemometer measurements. Values for the similarity constants D and P, which are both based on the atmospheric stability, were summarized by Hsieh et al. (2000). The normalized flux, F/So, was set at 0.7 to retain more data points without losing data quality (Baum et al., 2008). The second screening was based on the number of measurement heights with measured net PM10 concentrations. Because the PM10 emission determination using the fluxgradient technique involved vertical concentration gradients, hourly data points with at least two sampling heights with net concentration data were considered in the analyses (National Research Council, 2003). After the first two screenings, the numbers of data points based on the number of sampling heights with concentration data were 1676 for data points with four net concentrations, 741 with three net concentrations, and 562 with two net concentrations; the total number of data points after the first two screenings was 2979. Lastly, data points were screened according to the net PM10 concentration vertical profile appropriate for the flux-gradient technique, as described by Ryden and McNeill (1984). First, the concentration should be approximately linear with the logarithm of height, and second, the concentration should be decreasing 1344

with increasing height. Linearity between the net PM10 concentration and the logarithm of height was verified using Pearson correlation. As practiced in biostatistics (Colton, 1974; Gherman and Mironiuc, 2012; Lehman et al., 2009) and other research areas (USDA, 2012), a Pearson correlation criterion of 0.75 was applied to indicate strong and robust linearity. Thus, hourly data points with Pearson correlation coefficients ≥0.75 were used in flux computation. After this last screening, the numbers of hourly data points remaining for each height were 1755, 1713, 1449, and 1086 for 2.0, 3.8, 5.3, and 7.6 m, respectively, with the total number of hourly data points for flux computation at 1817.

Flux-Gradient Technique Using the flux-gradient technique, PM10 emission from the pens, Qp, (mg m−2 s−1), was estimated as Q p =-K c

dc [3] dz

where Kc is the eddy diffusivity for the variable of interest, which was PM10 in the present study (m2 s−1), and dc/dz is the vertical concentration gradient for the variable (mg m−3 m−1) (Dyer, 1974; Sethuraman and Brown, 1976; Kanemasu et al., 1979; Meyers and Baldocchi, 2005; Prueger and Kustas, 2005). The vertical concentration gradient, dc/dz, for each data point was estimated from the net PM10 concentration data and their corresponding sampling heights. Suitability of the flux-gradient technique for PM emissions at the feedlot studied was verified by determining whether the transport of particles at the feedlot was governed by turbulent or eddy diffusion or not. Using the criterion presented by Lilly (1973), the ratio of particle relaxation time, t, and the Lagrangian time scale, TL, must be 0.02. Therefore, it could be presumed that the particles emitted at the feedlot followed turbulence and thus could be considered as passive. The eddy diffusivity for PM10, Kc, was derived from Km and the Schmidt number, Sc, as (Flesch et al., 2002; Prueger et al., 2005): Sc =

Km [6] Kc Journal of Environmental Quality

The eddy diffusivity for momentum, Km, was computed as ku z K m = * m [7] fm where k is the von Karman constant (0.4), u* is the friction velocity (m s−1), zm is the mean geometric height (m), and fm is a nondimensional correction parameter. The friction velocity was obtained from micrometeorological measurements. The mean geometric height for each hour was computed using sampling heights with acceptable computed fetches and measured net PM10 concentrations. The value of fm was calculated following the procedure of Flesch et al. (2002) and Prueger et al. (2005), as presented by Hogstrom (1996): -0.25 æ z ö [8] fm = çç1-19 m ÷÷÷ çè Lø

for unstable atmospheric conditions (L < 0), and z fm = 1 + 5.3 m [9] L for stable atmospheric conditions (L > 0). Calculation of Kc involved use of the parameter Sc. In previous particle transport studies, Sc = 0.70 was applied (Guo and Maghirang, 2012; Zhang et al., 2008), whereas in an area source dispersion model, Sc = 0.64 was used (Flesch et al., 2004), but it is unknown whether these values are applicable to feedlot PM emissions. Before calculation of PM10 emissions using the fluxgradient technique, an Sc value appropriate for PM emissions at the studied feedlot was first established. This was accomplished in accordance with the experiment of Flesch et al. (2002). Combining Eq. [3] and [6], Sc is given by

S c =-

K m dc [10] Q p dz

In approximating Sc, the value of Qp used in Eq. [10] was determined using the integrated horizontal flux technique (Flesch et al., 2002), which is another micrometeorological technique that involves the measurement of wind speed and concentration to profile the horizontal flux (i.e., the product of wind speed and concentration), integrated with respect to height to estimate the source’s flux (Wilson et al., 1983; Ryden and McNeill, 1984). Expression for the integrated horizontal flux technique is given by: Qp =

upper standard deviations for Sc for this 5-mo period are shown in Fig. 2. The overall median Sc was 0.63. Therefore, in calculating PM10 fluxes using the flux-gradient technique (Eq. [3]) for the whole measurement period, Sc = 0.63 was applied.

Data Analysis The data were analyzed using SAS (SAS Institute, 2004). Mean values are presented for parameters that followed a normal distribution (i.e., bell-shaped, symmetric distribution), such as air temperature and u*; on the other hand, median values are reported for parameters (e.g., PM10 concentrations, PM10 fluxes, Sc, and Kc) that did not follow a normal distribution, which means the data distribution was skewed and asymmetric. Outlier analysis and standard deviation (SD) calculation were based on the procedure of Schwertman et al. (2004) for nonnormal data sets. In determining the influence of micrometeorological parameters on dc/dz, Kc, and PM10 emissions, backward elimination was applied with a 5% significance level.

Results and Discussion Micrometeorological Conditions WindRose plots (WRPLOT View, Lakes Environmental) are provided to show wind speed and wind direction trends at 6-mo intervals (Fig. 3). For the first 6 mo of the measurement period, most of the hourly data points (85%) had wind coming from the south (135–225°); for the second 6 mo, 45 and 32% of the data points had wind from the north (0–45°, 315–360°) and the south, respectively; and for the last months, 19 and 59% of data points were from the north and the south, respectively. Overall, the wind came from the south 55% of the time and from the north 25% of the time; wind coming from the east and west had occurrences of 14 and 6%, respectively. Based on their fetches, footprints for the four sampling heights were within the feedlot boundary 98, 87, 76, and 62% of the time for the 2.0-, 3.81-, 5.34-, and 7.62-m heights, respectively. Median fetch values for the acceptable hourly data points (n = 1626, after outlier removal) for the four measurement heights were 84, 163, 189, and 209 m, respectively (Table 1); note that the fetch of the instrumentation tower from the feedlot boundary ranged from 210 to 1308 m. Overall values for micrometeorological parameters at the feedlot are summarized in Table 2. During the 17-mo

1 zp u x c dz [11] x ò0

where c is the concentration (mg m−3), uxc is the horizontal flux (mg m−2 s−1) and the overbar indicates time averaged, and zp is the highest measurement height (m) (Ryden and McNeill, 1984; Sullivan and Ajwa, 2011). To have a good flux profile, it is desirable to conduct measurements at more than two heights; however, due to limitations in anemometer availability, wind speed measurements were made at two heights only (3.8- and 7.6-m heights). The measurements needed in this step were conducted from May through September 2011. Two criteria were applied in screening the data points: (i) in terms of ux, 3.8-m height < 7.6-m height; and (ii) in terms of uxc, 3.8-m height < 7.6-m height. Based on 291 hourly data points, hourly medians and lower and www.agronomy.org • www.crops.org • www.soils.org

Fig. 2. Hourly median ≤10-mm particulate matter (PM10) Schmidt number, Sc, determined for May through September 2011 (n = 291). Error bars represent upper and lower standard deviations. 1345

measurement period, August was the hottest month in both 2010 (32 ± 5°C) and 2011 (27 ± 5°C); December was the coldest month in 2010 (−3 ± 6°C), and January was the coldest

in 2011 (−5 ± 7°C). The highest wind speeds were observed in July (5.5 ± 1.8 m s−1) and August (5.2 ± 1.6 m s−1) in 2010 and in May (5.6 ± 2.7 m s−1) and June (5.6 ± 2.3 m s−1) in

Fig. 3. Wind speed and wind direction distribution at the feedlot: (a) May through October 2010 (n = 811 hourly data points); (b) November 2010 through April 2011 (n = 832 hourly data points); and (c) May through September 2011 (n = 1296 hourly data points). Table 1. Fetch values (n = 1626) for four measurement heights as calculated using the procedure of Hsieh et al. (2000). Parameter

Feedlot†

2.0 m

3.81 m

5.34 m

7.62 m

Data points, no.

1626

1626

1626 Fetch, m

1397

1104

Median Min. Max. SD‡   Lower range   Upper range

420 210 1308

84 7 401

163 12 1214

189 16 1281

209 21 1246

– –

33 29

85 104

87 176

86 141

† Feedlot fetch based on feedlot dimensions. ‡ Two values for standard deviations, for lower and upper ranges, because of nonnormality of distribution. 1346

Journal of Environmental Quality

Table 2. Micrometeorological parameters at the feedlot for May 2010 through September 2011. Parameter

Air temp.

Wind speed

Friction velocity

Mean Min.

°C 14 −24

m s−1 5.0 0.5

m s−1 0.40 0.06

Max. SD

39 14

28.9 2.3

3.00 0.18

Surface roughness cm 4.0§ 1.2 × 10−3 100 1.9/7.1¶

Sensible heat I†

II‡

—————— W m−2 —————— 66§ −10§ 0.03 772 71/158¶

−0.05 −1,425 16/8¶

† Heat flux direction from surface to atmosphere. ‡ Heat flux direction from atmosphere to surface. § Value based on median due to nonnormal distribution for surface roughness. ¶ Two values for standard deviations, lower/upper ranges, due to nonnormal distribution for surface roughness.

2011; consequently, these months also had the highest friction velocities (0.43 ± 0.17 m s−1). The lowest wind speeds were measured in October 2010 (2.5 ± 0.5 m s−1), which also had the lowest friction velocities (0.22 ± 0.04 m s−1). The overall median zo was 4.0 cm, which was comparable to the median value (3.6 cm) reported by Baum et al. (2008) for the same feedlot.

Particulate Matter Concentration and Vertical Concentration Gradient

840 mm (plus the pen surface water application) and 19 ± 10°C, 13 mm (but with snow and/or ice on the pen surface) and −4 ± 8°C, and 482 mm and 25 ± 10°C, respectively.

Particulate Matter Eddy Diffusivity With Sc = 0.63, the overall mean hourly eddy diffusivity for PM10, Kc, was 1.10 ± 0.60 m2 s−1 for the 17-mo period. A diurnal trend was observed for Kc. As shown in Fig. 6, Kc was highest from 1200 to 1500 h and lowest and relatively steady from 2000 to 0800 h; based on lower and upper SDs (i.e., error bars in Fig. 6), however, Kc did not vary significantly within the day. The hourly trend of the different micrometeorological parameters might explain the observed trend for Kc. Figure 7 shows the hourly trends of sensible heat, u*, air temperature, and zo for the entire measurement period. The first three parameters had the same trend as Kc: the parameter was highest in the mid-afternoon and was lowest and steady from evening to mid-morning. Atmospheric conditions during the day could help explain this trend. The presence of solar radiation in the afternoon results in higher heat flux and air temperature that make atmospheric conditions unstable (i.e., strong vertical dispersion), whereas the absence of solar radiation keeps the atmosphere stable (i.e., minimum vertical dispersion) (Turner, 1994). Thus, unstable atmospheric conditions could explain the high Kc values estimated for the afternoon period, and stable conditions could explain the low and stable Kc values at night (Lee and Larsen, 1997; Yadav et al., 2003; Alfieri et al., 2009). Compared with Kc and other micrometeorological parameters,

The PM10 concentrations measured at all four measurement heights exhibited diurnal trends, with the highest concentrations measured during the early evening period (1900–2100 h) and the lowest during early morning period (0200–0500 h) (Fig. 4). The overall hourly net PM10 concentrations were 110 ± 112/204 (lower /upper SD), 66 ± 72/126, 56 ± 59/117, and 57 ± 59/101 mg m−3 for the 2.0-, 3.8-, 5.3-, and 7.6-m heights, respectively. Instead of having the lowest concentration, the 7.6-m height had roughly the same overall value as the 5.3-m height. Further analysis determined that median concentrations for the 5.3- and 7.6-m heights were 73 ± 68/124 and 60 ± 59/105 mg m−3, respectively, when considering only points with measured concentrations for both heights (n = 1035). Using the overall concentration values, regression analysis between the logarithm of measurement height and PM10 concentration indicated a strong linear relationship (R2 = 0.85, Fig. 5), as expected (Ryden and McNeill, 1984). Calculation of the hourly vertical concentration gradient, dc/dz, showed that the PM10 concentration decreased by approximately 10 ± 16/10 mg m−3 for every 1 m increase in height. Within a day, the 1900 to 2100 h period had the largest PM10 concentration gradient, 20 ± 49/22 mg m−3 m−1, and the 0200 to 0500 h period had the lowest gradient, 4 ± 7/3 mg m−3 m−1. Also, the vertical concentration gradient for the mid-morning to late afternoon period (0900–1800 h) was 12 ± 13/10 mg m−3 m−1. Before implementation of the third screening criterion (i.e., decreasing concentration with increasing height), there were 228 hourly data points, which were eventually screened out, that had concentrations that increased with height. In terms of percentages, 65% of these data points was measured in March to September 2011, 30% in December 2010 to February 2011, and the remaining 5% in May to November 2010. Corresponding precipitation Fig. 4. Hourly median net ≤10-mm particulate matter (PM10) concentrations as measured from four heights for May through September 2011 (n = 1086–1755). amounts and air temperatures for these periods were Error bars represent upper standard deviation estimates. www.agronomy.org • www.crops.org • www.soils.org

1347

Fig. 5. Plot of ≤10-mm particulate matter (PM10) concentrations against the logarithm of the measurement height using overall net PM10 concentration values.

zo was essentially stable the whole day, ranging from 2.7 to 5.5 cm (Fig. 7). As a function of Kc, PM emission would be high during conditions with high u*, air temperature, and sensible heat.

Fig. 6. Hourly median ≤10-mm particulate matter (PM10) eddy diffusivity, Kc, calculated for May 2010 through September 2011 (n = 1626 hourly data points). Error bars represent upper and lower standard deviations.

time series with wind u*, air temperature, and measured zo (Fig. 8). Among these parameters, PM10 emission had a similar daily trend with that of air temperature (Fig. 7b). Relatively high air

Particulate Matter Emission During the 17-mo measurement period, hourly PM10 flux ranged from ?0 to 272 mg m−2 h−1, with an overall median value of 36 mg m−2 h−1. Based on days with at least 12 hourly fluxes (n = 44), the overall median daily PM10 flux was 1.81 g m−2 d−1, which was slightly higher but within range of those recently published for beef cattle feedlots (McGinn et al., 2010; Bonifacio et al., 2012). McGinn et al. (2010) reported values of 1.45 and 1.61 g m−2 d−1 for Australian feedlots based on inverse dispersion modeling with WindTrax. Bonifacio et al. (2012) reported median PM10 emissions of 1.10 and 1.60 g m−2 d−1 for Kansas feedlots based on a reverse dispersion modeling technique with AERMOD. Differences in emission estimation approaches and feedlot conditions and characteristics could help explain the difference among these PM10 fluxes. Water application on pens was done using a solidset sprinkler system at one of the previously studied feedlots, but at the feedlot evaluated, water was occasionally applied on either unpaved roads and alleys (2010) or pens (2011) using water trucks. Although the feedlot was comparable to the Kansas feedlots examined by Bonifacio et al. (2012) in terms of air temperature, wind speed, and wind direction, it received far less precipitation (420 mm in 2010, 152 mm in 2011) than the other feedlots (average of 622 mm). As noted in a previous study (Bonifacio et al., 2011), rainfall effects on lowering PM emission generally lasted from 3 to 7 d. In calculation of the overall PM10 emissions (1.81 g m−2 d−1), 78% of the days used (n = 44) were preceded by at least 7 d with no rainfall. The PM10 fluxes for these 44 d (with at least 12 hourly values) were plotted as 24-h averages in 1348

Fig. 7. Hourly trends of (a) sensible heat, (b) friction velocity, u*, (c) air temperature, and (d) surface roughness, zo, plotted with ≤10-mm particulate matter (PM10) eddy diffusivity, Kc, for the entire measurement period (n = 1626 hourly data points). Journal of Environmental Quality

temperatures might have contributed to large PM10 emissions at the feedlot from June through July 2010 (air temperature of 28–33°C, median PM10 emission of 85 mg m−2 h−1) and May 2011 (air temperature of 5–23°C, median PM10 emission of 100 mg m−2 h−1), as seen in Fig. 7b. Days with low air temperatures, however, also could have high PM10 emission rates. For example, 1 and 4 Nov. 2010, despite having 24-h average air temperatures of 4 and 2°C, respectively, had daily PM10 fluxes of 87 and 129 mg m−2 h−1, respectively. Another example was 31 Jan. 2011 that had below-freezing air temperatures (−12°C, 24-h average) but high PM10 emissions (66 mg m−2 h−1, 24-h average). The high PM emissions determined for periods with low air temperatures could be attributed to the absence of precipitation (e.g., rainfall, snow, or water application) for extended periods of time. The high emissions in November 2010 could be largely due to negligible precipitation (2.0 mm) in October 2011. Statistical analysis showed that the number of days without rain significantly affected (P = 0.02) the resulting PM10 emissions. Although PM10 emission was a function of u*, their respective daily trends were different. Statistics on the hourly PM10 fluxes at the feedlot are summarized in Table 3. During a day, PM10 emissions at Fig. 8. Daily ≤10-mm particulate matter (PM10) fluxes plotted with (a) friction velocity, u*, (b) air temperature, and (c) surface roughness, zo. Days considered were those with at least 12 hourly the feedlot remained relatively high data points (n = 44). and steady from 0900 to 2100 h. This conditions had significantly higher PM10 emissions, ranging trend was different from other cattle from 9 to 146  mg m−2 h−1, than cold conditions, which had feedlot studies: the highest PM10 emissions during a day were only 3 to 27 mg m−2 h−1 PM10 emissions. For the trend within observed in the afternoon (1000–1700 h) at two Kansas cattle a day, peak in PM10 emissions for warm conditions occurred feedlots (Bonifacio et al., 2012) and in the early evening at two during the early evening period, from 2000 to 2100 h (116 and Australian feedlots (McGinn et al., 2010). Still, similar to these 146 mg m−2 h−1, respectively). For cold conditions, on the other previous studies, an increase in PM10 emissions was observed hand, the highest estimated PM10 emissions (14–27 mg m−2 h−1) during the early evening period (2000–2100 h). The reasons for were from the afternoon, from 1100 to 1500 h; surprisingly, no these different PM10 emission trends could include differences increase in PM10 emissions was measured in the evening (20% had relatively the feedlot between warm (21 ± 10°C) and cold (−2 ± 8°C) 10 smaller PM emissions (Fig. 10). For the studied feedlot, pen conditions were observed (Fig. 9). Using median values, warm 10 www.agronomy.org • www.crops.org • www.soils.org

1349

Table 3. Hourly medians and standard deviations for ≤10-mm particulate matter (PM10) flux (n = 1626) as quantified by the flux-gradient technique. PM10 flux Time h 2400 0100 0200 0300 0400 0500 0600 0700 0800 0900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 Overall

Data points no. 74 75 60 61 58 53 57 58 69 76 66 63 67 66 78

Median

Standard deviation† Lower SD

Upper SD

——————————————— mg m−2 h−1 ——————————————— 23 20 51 21 21 32 19 17 27 14 15 25 10 7 24 8 6 22 14 14 31 26 19 47 35 33 38 56 64 90 59 66 117 68 49 67 62 55 105 72 71 65 72 60 86

73 69 81 86 79 71 53 70 63 1626

59 53 59 59 47 63 55 36 22 36

58 48 52 64 40 81 66 42 22 38

54 96 64 72 90 135 143 115 35 85

† Two values for standard deviations, for lower and upper ranges, because of nonnormality of the distribution (Schwertman et al., 2004).

surface conditions with water content >20% (23–50%, n = 5) had PM10 fluxes ranging from 3 to 14 mg m−2 h−1, with a median of 11 mg m−2 h−1, whereas conditions with water content of 20% or less (8–20%, n = 16) had higher PM10 fluxes that ranged from 7 to 40 mg m−2 h−1 and had a median of 15 mg m−2 h−1; this implies a reduction in PM10 emissions of up to 60% for pens with surface water content >20%.

Conclusions

on sources like cattle feedlots must be further improved to reduce uncertainties in flux estimates. Individual contributions of different sources, such as pens, unpaved roads, and feed mills, to the overall PM emission also need to be assessed. Because pens may differ in soil type, soil and manure depth, and number of cattle, the effects of pen characteristics on PM emission also must be studied. In addition, uncertainties associated with stable conditions must be addressed to produce more reliable emission estimates.

The flux-gradient technique was implemented to quantify PM10 emissions from an open-lot beef cattle feedlot. The PM10 emissions at the feedlot varied diurnally and seasonally, with emissions peaking at 2000 to 2100 h during warm conditions and at 1100 to 1500 h during cold conditions, which had significantly lower PM fluxes. High friction velocity, air temperature, and sensible heat flux, and also low surface roughness during warm conditions, were observed for the highPM-emission periods. Pen surface water content highly affected PM10 emission, and a content of at least 20% (wet basis) significantly reduced PM emissions. Further research is needed on the applicability of the flux-gradient technique in estimating feedlot emissions. Its implementation Fig. 9. Hourly median ≤10-mm particulate matter (PM10) fluxes under warm and cold conditions.

1350

Journal of Environmental Quality

Fig. 10. Effect of pen surface water content (wet based) on ≤10-mm particulate matter (PM10) emissions (n = 21 d).

Acknowledgments This study was supported by the USDA National Institute of Food and Agriculture (Project no. 2009-35112-3544), K-State Research and Extension (Contribution no. 13-261-J), and the USDA–ARS. Technical assistance provided by Darrell Oard and Howell Gonzales of Kansas State University; Dr. Li Guo, Dr. Orlando Aguilar, and Curtis Leiker, formerly of Kansas State University; Dr. Kenwood Scoggin of the USDA–ARS, Ames, IA; and Dr. Bernardo Predicala of Prairie Swine Centre Inc., Saskatoon, SK, Canada, is acknowledged. Cooperation of feedlot operators and KLA Environmental Services, Inc., is also acknowledged.

References Alfieri, J.G., P.D. Blanken, D. Smith, and J. Morgan. 2009. Concerning the measurement and magnitude of heat, water vapor, and carbon dioxide exchange from a semiarid grassland. J. Appl. Meteorol. Climatol. 48:982– 996. doi:10.1175/2008JAMC1873.1 American Society for Testing Materials. 2002. Method: D2216–98: Standard test method for laboratory determination of water (moisture) content of soil and rock by mass. Am. Soc. Test. Mater., West Conshohocken, PA. Baek, B.-H., R. Todd, N.A. Cole, and J.A. Koziel. 2006. Ammonia and hydrogen sulphide flux and dry deposition velocity estimates using vertical gradient method at a commercial beef cattle feedlot. Int. J. Global Environ. Issues 6:189–203. Baum, K.A., J.M. Ham, N.A. Brunsell, and P.I. Coyne. 2008. Surface boundary layer of cattle feedlots: Implications for air emissions measurement. Agric. For. Meteorol. 148:1882–1893. doi:10.1016/j.agrformet.2008.06.017 Bonifacio, H.F., R.G. Maghirang, B.W. Auvermann, E.B. Razote, J.P. Murphy, and J.P. Harner III. 2012. Particulate matter emission rates from beef cattle feedlots in Kansas: Reverse dispersion modeling. J. Air Waste Manage. Assoc. 62:350–361. doi:10.1080/10473289.2011.651557 Bonifacio, H.F., R.G. Maghirang, E.B. Razote, B.W. Auvermann, J.P. Harner III, J.P. Murphy, et al. 2011. Particulate control efficiency of a water sprinkler system at a beef cattle feedlot in Kansas. Trans. ASABE 54:295–304. Bonifacio, H.F., R.G. Maghirang, E.B. Razote, S.L. Trabue, and J.H. Prueger. 2013. Comparison of AERMOD and WindTrax dispersion models in determining PM10 emission rates from a beef cattle feedlot. J. Air Waste Manage. Assoc. 63:545–556. doi:10.1080/10962247.2013.768311 Colton, T. 1974. Statistics in medicine. Little Brown and Co., New York. Denmead, O.T., J.R. Simpson, and J.R. Freney. 1974. Ammonia flux into the atmosphere from a grazed pasture. Science 185:609–610. doi:10.1126/ science.185.4151.609 Dyer, A.J. 1974. A review of flux-profile relationships. CSIRO Div. of Atmos. Phys., Aspendale, VIC, Australia. Faulkner, W.B., J.J. Powell, J.M. Lange, B.W. Shaw, R.E. Lacey, and C.B. Parnell. 2007. Comparison of dispersion models for ammonia emissions from a ground-level area source. Trans. ASABE 50:2189–2197. Flesch, T.K., J.H. Prueger, and J.L. Hatfield. 2002. Turbulent Schmidt number from a tracer experiment. Agric. For. Meteorol. 111:299–307. doi:10.1016/S0168-1923(02)00025-4 Flesch, T.K., and J.D. Wilson. 2005. Estimating tracer emissions with a backward Lagrangian stochastic technique. In: J.L. Hatfield and J.M. Baker, editors, Micrometeorology in agricultural systems. Agron. Monogr. 47. ASA, www.agronomy.org • www.crops.org • www.soils.org

CSSA, SSSA, Madison, WI. p. 513–531. doi:10.2134/agronmonogr47. c22 Flesch, T.K., J.D. Wilson, L.A. Harper, and B.P. Crenna. 2005. Estimating gas emissions from a farm with an inverse-dispersion technique. Atmos. Environ. 39:4863–4874. doi:10.1016/j.atmosenv.2005.04.032 Flesch, T.K., J.D. Wilson, L.A. Harper, B.P. Crenna, and R.R. Sharpe. 2004. Deducing ground-to-air emissions from observed trace gas concentrations: A field trial. J. Appl. Meteorol. 43:487–502. doi:10.1175/1520-0450(2004)0432.0.CO;2 Galvin, G., C. Henry, D. Parker, R. Ormerod, P. D’Abreton, and M. Rhoades. 2006. Efficacy of a Lagrangian and a Gaussian model for back calculating emission rates from feedyard area sources. In: V.P. Aneja et al., editors, Proceedings Workshop on Agricultural Air Quality: State of the Science, Potomac, MD. 5–8 June 2006. North Carolina State Univ., Raleigh. p. 605–614. Gherman, C.D., and A.I. Mironiuc. 2012. Evaluation of serum adipokines in peripheral arterial occlusive disease. Mediators Inflamm. 2012:257808. doi:10.1155/2012/257808 Gonzales, H.B., R.G. Maghirang, J.D. Wilson, E.B. Razote, and L. Guo. 2011. Measuring cattle feedlot dust using laser diffraction analysis. Trans. ASABE 54:2319–2327. Guo, L., and R.G. Maghirang. 2012. Numerical simulation of airflow and particle collection by vegetative barriers. Eng. Appl. Comput. Fluid Mech. 6:110–122. Ham, J.M., and K.A. Baum. 2007. Measuring ammonia fluxes from cattle feedlots using time-averaged relaxed eddy accumulation. In: International Symposium on Air Quality and Waste Management for Agriculture, Broomfield, CO. 16–19 Sept. 2007. Am. Soc. Agric. Biol. Eng., St. Joseph, MI. Paper 701P0907cd. Harper, L.A., O.T. Denmead, J.R. Freney, and F.M. Byers. 1999. Direct measurements of methane emissions from grazing and feedlot cattle. J. Anim. Sci. 77:1392–1401. Hogstrom, U. 1996. Review of some basic characteristics of the atmospheric surface layer. Boundary-Layer Meteorol. 78:215–246. doi:10.1007/ BF00120937 Hsieh, C.I., G. Katul, and T. Chi. 2000. An approximate analytical model for footprint estimation of scalar fluxes in thermally stratified atmospheric flows. Adv. Water Resour. 23:765–772. doi:10.1016/S0309-1708(99)00042-1 Hutchinson, G.L., A.R. Mosier, and C.E. Andre. 1982. Ammonia and ammine emissions from a large cattle feedlot. J. Environ. Qual. 11:288–293. doi:10.2134/jeq1982.00472425001100020028x Kanemasu, E.T., M.L. Wesely, and B.B. Hicks. 1979. Techniques for calculating energy and mass fluxes. In: B.J. Barfield and J.F. Gerber, editors, Modification of the aerial environment of crops. ASAE Monogr. 2. Am. Soc. Agric. Eng., St. Joseph, MI. p. 156–182. Lee, H.N., and R.J. Larsen. 1997. Vertical diffusion in the lower atmosphere using aircraft measurements of 222Rn. J. Appl. Meteorol. 36:1262–1270. doi:10.1175/1520-0450(1997)0362.0.CO;2 Lehman, S.J., C.L. Schlett, F. Bamberg, H. Lee, P. Donnelly, L. Shturman, et al. 2009. Assessment of coronary plaque progression in coronary computed tomography angiography using a semiquantitative score. JACC Cardiovasc. Imaging 2:1262–1270. doi:10.1016/j.jcmg.2009.07.007 Leytem, A.B., R.S. Dungan, D.L. Bjorneberg, and A.C. Koehn. 2011. Emission of ammonia, methane, carbon dioxide, and nitrous oxide from dairy cattle housing and manure management systems. J. Environ. Qual. 40:1383– 1394. doi:10.2134/jeq2009.0515 Lilly, G.E. 1973. Effect of particle size on particle eddy diffusivity. Ind. Eng. Chem. Fundam. 12:268–275. doi:10.1021/i160047a002 Massman, W.J., and X. Lee. 2002. Eddy covariance flux corrections and uncertainties in long-term studies of carbon and energy exchanges. Agric. For. Meteorol. 113:121–144. doi:10.1016/S0168-1923(02)00105-3 McGinn, S.M., T.K. Flesch, D. Chen, B. Crenna, O.T. Denmead, T. Naylor, and D. Rowell. 2010. Coarse particulate matter emissions from cattle feedlots in Australia. J. Environ. Qual. 39:791–798. doi:10.2134/jeq2009.0240 Meyers, T.P., and D.D. Baldocchi. 2005. Current micrometeorological flux methodologies with applications in agriculture. In: J.L. Hatfield and J.M. Baker, editors, Micrometeorology in agricultural systems. Agron. Monogr. 47. ASA, CSSA, SSSA, Madison, WI. p. 381–396. doi:10.2134/ agronmonogr47.c16 Muller, J.B.A., M. Coyle, D. Fowler, M.W. Gallagher, E.G. Nemitz, and C.J. Percival. 2009. Comparison of ozone fluxes over grassland by gradient and eddy covariance technique. Atmos. Sci. Lett. 10:164–169. doi:10.1002/ asl.226 Myles, L., J. Kochendorfer, M.W. Heuer, and T.P. Meyers. 2011. Measurement of trace gas fluxes over an unfertilized agricultural field using the flux-gradient technique. J. Environ. Qual. 40:1359–1365. doi:10.2134/jeq2009.0386 1351

National Research Council. 2003. Air emissions from animal feeding operations: Current knowledge, future needs. Natl. Acad. Sci., Washington, DC. Prueger, J.H., T.J. Gish, L.L. McConnell, L.G. McKee, J.L. Hatfield, and W.P. Kustas. 2005. Solar radiation, relative humidity, and soil water effects on metolachlor volatilization. Environ. Sci. Technol. 39:5219–5226. doi:10.1021/es048341q Prueger, J.H., and W.P. Kustas. 2005. Aerodynamic methods for estimating turbulent fluxes. In: J.L. Hatfield and J.M. Baker, editors, Micrometeorology in agricultural systems. Agron. Monogr. 47. ASA, CSSA, SSSA, Madison, WI. p. 407–436. doi:10.2134/agronmonogr47.c18 Purdue Applied Meteorology Laboratory. 2009. Quality assurance project plan for the National Air Emissions Monitoring Study. Rev. 3. Purdue Univ., West Lafayette, IN. Ryden, J.C., and J.E. McNeill. 1984. Application of the micrometeorological mass balance method to the determination of ammonia loss from a grazed sward. J. Sci. Food Agric. 35:1297–1310. doi:10.1002/jsfa.2740351206 SAS Institute. 2004. SAS system for Windows, release 9.1.3. SAS Inst., Cary, NC. Schwertman, N.C., M.A. Owens, and R. Adnan. 2004. A simple more general boxplot method for identifying outliers. Comput. Stat. Data Anal. 47:165–174. doi:10.1016/j.csda.2003.10.012 Sethuraman, S., and R.M. Brown. 1976. Validity of the log-linear profile relationship over a rough terrain during stable conditions. Atmos. Sci. Div., Brookhaven Natl. Lab., Upton, NY. Sullivan, D.A., and H.A. Ajwa. 2011. Review of airborne emissions from agricultural fumigants: Design and uncertainty considerations for the use of the integrated horizontal flux method. J. Environ. Qual. 40:1366–1374. doi:10.2134/jeq2009.0245

1352

Turner, D.B. 1994. Workbook of atmospheric dispersion estimates: An introduction to dispersion modeling. 2nd ed. Lewis Publ., Boca Raton, FL. USDA. 2012. Conservation Reserve Program: Soil rental rates. USDA Farm Serv. Agency, Washington, DC. USEPA. 2001. Emissions from animal feeding operations, draft. USEPA, Research Triangle Park, NC. USEPA. 2005a. Animal feeding operations consent agreement and final order. OAR-2004-0237, FRL-7864-4. USEPA, Research Triangle Park, NC. USEPA. 2005b. Guideline on air quality models. Code of Federal Regulations, Part 51, Title 40. Fed. Regist. 70:216. USEPA. 2011. Call for information: Information related to the development of emission-estimating methodologies for animal feeding operations. Code of Federal Regulations, Part 60, Title 40. Fed. Regist. 76:3060. Wilson, J.D., V.R. Catchpoole, O.T. Denmead, and G.W. Thurtell. 1983. Verification of a simple micrometeorological method for estimating the rate of gaseous mass transfer from the ground to the atmosphere. Agric. Meteorol. 29:183–189. doi:10.1016/0002-1571(83)90065-1 Wilson, J.D., and W.K.N. Shum. 1992. A re-examination of the integrated horizontal flux method for estimating volatilisation from circular plots. Agric. For. Meteorol. 57:281–295. doi:10.1016/0168-1923(92)90124-M Yadav, A.K., S. Raman, and D.D.S. Niyogi. 2003. A note on the estimation of eddy diffusivity and dissipation length in low winds over a tropical urban terrain. Pure Appl. Geophys. 160:395–404. doi:10.1007/s00024-003-8785-4 Zhang, N., Z.C. Zheng, and R.G. Maghirang. 2008. Numerical simulation of smoke clearing with nanoparticle aggregates. Int. J. Numer. Methods Eng. 74:601–618. doi:10.1002/nme.2186

Journal of Environmental Quality