Published online 20 February 2008
Published in J Environ Qual 37:551-556 (2008)
DOI: 10.2134/jeq2006.0408
© 2008 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
Near-Field Dust Exposure from Cotton Field Tilling and Harvesting
April L. Hiscoxa,*,
David R. Millera,
Britt A. Holménb,
Wenli Yangc and
Junming Wangd
a Natural Resources Management and Engineering, The Univ. of Connecticut, 1376 Storrs Rd. U-4087, Storrs, CT 06269
b School of Engineering, The Univ. of Vermont, Votey Bldg. Rm. 213B, Burlington, VT 05405
c Crocker Nuclear Lab., One Shields Ave., Univ. of California, Davis, CA 95616
d Plant and Environmental Sciences, New Mexico State Univ., Dep. 3Q, Las Cruces, NM 88003
* Corresponding author (april.hiscox{at}uconn.edu).
Received for publication September 26, 2006.
 |
ABSTRACT
|
|---|
The frequency and intensities of dust exposures in and near farm fields, which potentially contribute to high intensity human exposure events, are undocumented due to the transient nature of local dust plumes and the difficulties of making accurate concentration measurements. The objective of this study is to measure near-field spatial concentrations of the dust plumes emitted during tilling and harvesting of an irrigated cotton field outside of Las Cruces, NM (soil class: fine-loamy, mixed, superactive, thermic Typic Calciargid). A comparison of remote lidar measurements of plumes emitted from cotton field operations with in situ samplers shows a strong agreement between the two techniques: r2 = 0.79 for total suspended particulates (TSP) and r2 = 0.61 for particulate matter with diameter less than or equal to 10 µm (PM10). Plume movement was dependent on the short-term wind field and atmospheric stability. Horizontal spread rate of the plumes, determined from lidar measured Gaussian dispersion parameters, was less than expected by a factor of 7. Thus, in-plume downwind concentrations were higher than expected. Vertical dispersion was dependent on the rise of "cells" of warm air convecting off the soil surface. On a windy day, discing the field showed TSP and PM10 concentrations at the source itself of up to 176 µg m–3 and 120 µg m–3, respectively. These resulted in in-plume peak TSP concentrations of about 1.22 µg m–3 at 10 m downwind and 0.33 µg m–3 at 100 m downwind. The measured concentrations highlight a potential exposure risk to people in and around farming operations.
Abbreviations: BMP, best management practice • Lidar, light detection and ranging • PM10, particulate matter having an aerodynamic diameter less than or equal to 10 µm • TSP, total suspended particulates
|
INTRODUCTION
|
|---|
THE CONCERNS about dust emissions from agriculture fields can be grouped into two classes: "near-field" exposure of agriculture workers and adjacent land activities and "far-field" additions of dust aerosols to the general regional air pollution load. Near-field exposures tend to deliver high intensity, short event doses and far-field exposures are most often low intensity, chronic doses. The term "near-field" here denotes distances from the source generally smaller than several hundred meters where an individual plume can still be distinguished.
Kirkhorn and Schenker (2002) recently reviewed the work linking health effects to respirable aerosols in agricultural settings. But little information is available on actual exposure concentrations and exposure times for the agriculture field environment. In a recent review, Schenker (2000) noted that exposures to inorganic dust among farmers and farm workers in dry climates involved in activities that perturb the soil commonly results in exposures of 1 to 5 mg m–3 of respirable dust and greater than 20 mg m–3 of total dust. Most research has investigated the time-integrated exposures of tractor drivers and some have demonstrated that cab technology and management practices can reduce exposure on the tractor significantly (Nieuwenhuijsen and Schenker, 1998; Nieuwenhuijsen et al., 1998). Still, little is known about the frequency and intensities of doses received elsewhere in and near the fields due to the transient nature of local dust plumes and the difficulties of making accurate concentration measurements in dynamic plumes (Holmén et al., 2006). Thus, specific field-, crop-, and weather-related best management practices (BMPs) have not been defined.
This work is part of a larger study designed to combine remote lidar measurements of the dust cloud generated by working the field with in-field particulate measurements to determine BMPs for the reduction of emissions. This paper reports on the near-ground time and space variations of aerosol dust concentrations during tilling and harvesting of the cotton field. The objective of this study is to quantify the near-field spatial concentrations of the emitted dust plumes which potentially contribute to high intensity human exposure events.
|
Materials and Methods
|
|---|
Study Site
The experimental field was a flood-irrigated cotton field at the New Mexico State University, Leyendecker Plant Science Research Farm, Las Cruces, NM (32.2° N; 106.8° W). The field preparation and planting took place from late March to early April of 2005, and the harvesting operation took place in November of the same year. The field is a mixture of Armijo clay loam and Harkey loam soil types with taxonomic classification of fine-loamy, mixed, superactive, thermic Typic Calciargids (NRCS, 2005).
Measurements
Before the disking operation the heavy soil was in clods varying in size from about 1 to 20 cm diam. Soil moisture measurements were made in the field using a hand-held time domain reflectometer (TDR) (Campbell Scientific. Inc, Logan, UT) and additional samples were collected and analyzed in the laboratory. Soil moisture measurements during the discing operation varied across the clods from 0.10 m3 m–3 (at the surface of the clods) to 0.62 m3 m–3 (at the centers of the clods). During the harvest, the soil structure was more homogeneous and the soil moisture varied from 0.33 m3 m–3 (in a cotton row) to 0.94 m3 m–3 (between rows) for the harvesting operation. More specific details on soil moisture measurements and site conditions can be found in Holmén et al. (2006)
The aerosol sampling instruments used for the work presented here were the University of Connecticut elastic backscatter lidar and optical real-time particulate samplers (Model: GT-640A, Met One Instruments, Inc., Grants Pass, OR) for total suspended particulates (TSP) and particulate matter less than 10 µm in diameter (PM10). These optical samplers use a continuous flow of air at a rate of 4 L min–1. A forward light scattering detector averages the concentrations in the sample air to 1 min averages. The samplers were factory-calibrated for concentration mass vs. light scattering using Arizona Test/Road Dust standard (ISO standard) standard soil. The samplers were located in the center of the field two to five rows away from the operation tractor.
Micrometeorological parameters (3-D wind components and temperature) were measured at a rate of 20 Hz with a three-dimensional sonic anemometer (Model: CSAT3, Campbell Scientific, Logan, UT). The sonic anemometer measurements were used to calculate the stability parameter (z/L unitless), where z in this case is the measurement height of 1.5 m and L is the Obukov length [L = –(
cp T u*3)/(k g H)], where is the air density (kg m–3), cp is the specific heat of air (J kg–1 K–1), T is the air temperature (K), u* is the friction velocity (m s–1), k is the von Karman constant (0.4 unitless), g is the acceleration due to gravity (m s–2), and H is the sensible heat flux (W m–2) (Stull, 1988). The stability parameter indicates the overall state of mixing potential in the boundary layer. For the purpose of this study the measured atmospheric stability was grouped into two classes: unstable (–0.5 < z/L < 0) and dynamically stable (0 < z/L < 0.5).
Each pass of the tractor consisted of a single swath of the field. In-field measurements with the particulate samplers were made continuously through the pass and remote horizontal lidar scans above the field were made at increasing heights, resulting in five to nine full three-dimensional plume pictures for each pass. Each full 3-D scan was completed in less than 1 min. The lowest scan was approximately 3 m above the ground and the maximum measurement height ranged from 30 to 70 m depending on the overall atmospheric conditions. All lidar data is background subtracted and corrected for the range-squared attenuation effects before any of the analyses presented here were performed. Further details of these instruments and the measurement procedures can be found in Hiscox (2006). This article focuses on the plume dispersion and concentration as measured from the lidar scans for two different operations: spring discing and fall harvesting.
Dispersion
The most commonly used method of estimating plume dispersion is by the Gaussian plume dispersion parameters: 
y (cross wind) and z (vertical), where is the standard deviation of the plume concentration in the cross wind or vertical directions (Turner, 1994). Plumes are assumed to have a Gaussian distribution (Gifford, 1959) and the dispersion parameters are reported in units of meters. In this work, it is of interest to measure the horizontal spread of the plume over the field so the cross-wind dispersion parameters are measured from the lidar scans. The dispersion parameters are found applying the techniques of Hiscox et al. (2006b) in the horizontal direction:
 | [1] |
where, 
Y is the edge to edge distance (m) of the plume perpendicular to the plume axis and 
is the ratio of maximum backscatter (Bm) to edge backscatter (Be). The edge of the plume is defined as the contour of backscatter that first exceeds the background value. Background values are defined as the average lidar signal of a "no discernable plume" slice taken during each pass.
The direction of the plume axis is defined as the average wind direction for the measurement period of the tractor pass (4 to 7 min), and the cross-wind dispersion parameter is perpendicular to the plume axis. The plume centerline is defined by fitting a line to the horizontal movement of the plume peak (point of maximum lidar backscatter) in time. Using the horizontal lidar slices closest to the ground level, dispersion parameters are measured at different distances from a source point along the plume axis. The source point is defined as the intersection of the plume centerline and the tractor's path. Figure 1
is a single lidar slice annotated to show the method of selection for the values used.
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 1. Example lidar slice of a horizontal plume. The rectangle represents the field being worked and the large arrow indicates the tractor path. The dotted line indicates the plume axis direction and the average wind direction. The source point is at the intersection of the plume maximum line and the tractor path (indicated by the arrow moving left to right); in this example, that is also the point of maximum backscatter (Bm). Be is the edge backscatter and Y is the plume width as needed for application of Eq. [1].
|
|
It is noted here that the diffusion coefficients could be obtained from optimal fitting of the entire lidar scan across the plume using a host of data points. But the system used here has the advantage of simplicity and ease of understanding. A comparison of the two techniques to determine added reduction in uncertainty from optimal fitting remains to be done.
Concentrations/Calibration
Plume concentrations are determined following the methods of Hiscox et al. (2006a). A conversion factor from lidar backscatter to concentration in the air is found by comparing total backscatter in the plume slice with the concentration measured with the TSP sampler. The maximum per pass of both quantities is used, resulting in one point per pass. There are two assumptions used to make this comparison. First, the plume generated from the tractor at any given position along its path can be treated as a point source. Second, all the material generated reaches a height of approximately 3 m above the ground (the lowest scan of the lidar) and it is detected in the lidar scan. We are confident of both of these assumptions based on the three-dimensional rendering of plume isocontours generated by Voxler from Golden Software, as shown in Fig. 2
. The three scan sequences of the lidar show individual plumes generating from points along the tractor paths, with little interaction occurring at the lowest level.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 2. Three-dimensional rendering of combined lidar scans: (a) looking from the top down and (b) vertical cross-sectional contours of the plume. The three plumes are all from the same pass of the tractor. The first lidar sequence is the darkest and the last is the lightest taken within 1 min after the tractor had stopped at the edge of the field. The arrow indicates the approximate tractor path on the field.
|
|
The calibration line for daytime discing operations is displayed in Fig. 3
. The linear relationship confirms that a stronger source concentration results in a larger, more concentrated plume. The calibration presented only includes passes that occurred under unstable conditions (Hiscox, 2006). Passes that occurred under near-neutral and stable conditions are not used in the calibration because we believe there was inadequate lidar sampling of the plume. Under near-neutral and stable conditions there is less buoyant rise resulting in a wide spreading, low-hanging plume. For our measurements this means higher background concentrations (Hiscox, 2006) with the majority of the plume below the lidar field of view, making the calibration ineffective. An example of such a plume can be seen in Fig. 4
, where in contrast to Fig. 2, the maximum height of the plume is much lower in this case. Not enough points were available to generate a calibration for the harvesting operation due to high background levels originating in nearby orchards.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 3. Total suspended particulates (TSP) at the ground vs. total lidar backscatter approximately 3 m above the ground. The linear fit is forced to an intercept equal to the background TSP. The result is a slope of 0.0034 and an r2 value of 0.79. The particulate matter with diameter less than 10 µm (PM10) fit is 0.0025 with a zero intercept and an r2 of 0.61.
|
|
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 4. An example of a plume generated under stable conditions. (a) top down view and (b) vertical cross-sectional contours of the plume. Under these conditions the maximum plume height is less than 30 m as opposed to Fig. 2 where the plume exceeds 50 m.
|
|
|
Results
|
|---|
Cross-Wind Dispersion
Horizontal dispersion parameters along the plume axis for discing and harvesting days are presented in Table 1
. The results presented are an average for the whole day, and are for the lowest lidar slices, or approximately 3 m above the ground. This location is taken to represent that of potential exposure by humans. For discing, the horizontal dispersion increased linearly with downwind distance, although the rate of increase was less than would be predicted by standard plume models for these conditions. The simplified version of Taylor's statistical theory for dispersion (Stull, 2000) predicts a linear growth of 0.337 m m–1 of downwind distance. The measurements here show a linear growth of 0.0438 m m–1 of downwind distance. The lidar measurements also show a larger initial spread (7.6 m) as opposed to the zero spread in the point source theory. The tractor is a moving source, and the disc path is 3 m wide, so the larger area source resulted in a wider initial spread.
The results for the harvesting day do not show the expected linear trend in horizontal spread. Dispersion starts off large, decreases slightly until about 80 m downwind, and then increases again. This trend can be seen in Fig. 5
, which shows a single lidar slice during the harvest, with two separate plume cells easily identifiable. It should be noted that only four passes were used to obtain the statistics on the harvesting day due to dust interference from a neighboring operation which resulted in uncertainty in many of the measurements.
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 5. Lidar slice from harvesting day. Two intense plume cells can be seen where the dust is entrained in surface layer convective cells.
|
|
Plume Concentrations
Plume TSP concentrations with height are shown in Fig. 6
for the unstable discing passes. Concentrations decrease with height as expected. As the plume rose, some material fell out and the remaining aerosols continued to disperse, as shown by the increasing values of dispersion parameters in Table 1.
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 6. Height above the ground vs. concentration of material aloft (total suspended particulates [TSP]) for the 10 unstable passes on the discing day.
|
|
The calibrated lidar measurements were also used to measure the in-plume concentrations near the ground at various distances downwind. Table 2
shows the peak concentrations 10 and 100 m downwind for the 10 different discing passes, along with the average wind speed for each pass. The concentrations shown are the approximate concentrations of TSP to which an individual 10 or 100 m from the tractor would be exposed. The uncertainty levels in Table 2 are found from the uncertainty level of the slope value from the linear regression of Fig. 3. The slope value of 0.0034 used as the calibration factor has and uncertainty of 0.0006 based on Eq. [6.18b] in Wilks (2006).
View this table:
[in this window]
[in a new window]
|
Table 2. Peak downwind TSP (total suspended particulate) concentrations and average wind speeds for discing operation unstable passes.
|
|
|
Discussion
|
|---|
Overall plume movement is dominated by the average wind direction. Small variations in wind speed and direction, however, can result in large departures from the dominant direction. This is seen in Fig. 1 where the portion of the plume farthest from the source is over 100 m away from the average wind predicted axis. Between the two operations investigated here, plume dynamics varied greatly. Because the measured particle size distributions on both the discing and the harvesting day were similar (Holmén et al., 2006), the differences between the operations were due to varying meteorological conditions rather than a difference in material or moisture content. During the discing operation, much higher wind speeds were observed throughout the day; an average of 5.4 m s–1 compared to 1.6 m s–1 on the day of harvest. Both days were unstable with the average z/L = –1.2 for the discing day and –2.49 for the harvesting day. However, on the harvesting day, very low wind speeds made the plume extremely unstable. Frequently, wind speeds decreased to levels at which free convection may have become a factor in plume dispersal. Wilczak and Tillman (1980) have shown that under unstable conditions sensible heat is moved away from the ground surface in "cells" with typical length and width dimensions less than 300 m and greater than 10 m respectively, depending on the height above the ground. In this field study, the dust is apparently entrained in these surface layer convective cells near the ground, advected downwind at the translation speed of the cells and diffused vertically as the cells rise and expand. The cells can be seen in the lidar plume visualizations in Fig. 2 and 5 as a sequence of billows in the plumes. Under the moderately unstable, higher wind speed conditions of the discing day, the plumes in Fig. 2 are arranged into about five convective cells over a 200 m length of plume, or about 1 every 40 m. The convective cells appear to be elongated in the mean wind direction as noted by (Wilczak and Tillman, 1980). On the more unstable harvest day, the cells are again about 40 m long but are less elongated in the downwind direction and larger in the cross-wind direction as shown in Fig. 5.
The higher wind speeds during discing resulted in higher plume translational velocities and a rapid dispersal downwind. The light and variable winds during harvesting resulted in a less organized downwind dust plume structure, but more vertical dispersion. During stable conditions, the near-field horizontal spread is about the same as the light and variable wind speed unstable case, but there is little vertical dispersion or vertical rise. This indicates plume fanning occurs on a short time scale (Fig. 4). When vertical dispersion is small and horizontal dispersion is rapid, plume fanning occurs; this is typical of a stable boundary layer (Stull, 1988). Fanning is observed in the lowest lidar slices on all of the near-neutral and stable passes, but most of the plume is beneath the lowest slices leaving uncertain plume characteristics in these conditions.
Plume concentrations were only available for the unstable passes of the discing operations due to lidar calibration and field of view limitations. First, the calibration can only be performed if the plume of interest can be separated from the background aerosol load which was a periodic problem on the harvest day. Second, it cannot be applied in situations where the full plume cannot be scanned as was the case during stable conditions when a majority of the plume stayed under the 3 m minimum height measurement of the lidar as noted above.
Figure 3 shows peak TSP and PM10 concentrations at the source itself of up to 176 and 120 µg m–3, respectively. The TSP levels measured by the lidar downwind were much lower: average of 0.49 µg m–3 at 10 m downwind and average of 0.24 µg m–3 at 100 m downwind. Based on the average plume width (±3 y) and tractor speed, at a fixed point 10 m downwind of the tractor path a person at a single location would be exposed to a plume for approximately 30 s. This results in an average 30 s potential total dust exposure of 6.8 µg m–3 for each pass of the tractor. At 100 m downwind, this value decreases to 4.8 µg m–3 over an exposure time of 50 s for each pass of the tractor. Thus, the intensity of exposure farther downwind is less, but the total dose is about the same. For the 20 passes per day conducted across a field in this experiment, this exposure level would be experienced for 11 min and 28 s at 10 m downwind, and 16 min and 42 s at 100 m downwind, for an entire day of operations. The NIOSH REL (National Institute of Occupational Safety and Health Relative Exposure Limit) for silica is 0.05 mg m–3 for a 10 h time weighted average. At the source measurements of TSP concentrations exceed the REL. But, the short times of exposure downwind make the total dose orders of magnitude smaller than the NIOSH REL. It should also be noted that these measurements were performed on a 2.8-ha research farm field, so the duration and number of passes on a larger scale commercial operation would be longer and greater. In addition, interference from offsite plumes during the harvesting operation did saturate the lidar measurements at times indicating that in dense agricultural areas, such as the Mesilla Valley of New Mexico, the combinations of plumes from multiple sources could pose a greater health risk to agricultural workers.
|
Conclusions
|
|---|
Presented here are measurements of downwind plume dispersion and concentrations from agricultural generated dust. These measurements are made possible by calibrating the lidar with in situ plume sampling. A strong linear fit with an r2 value of 0.79 for TSP and 0.61 for PM10 allow this calibration to be successful. Clouds of dust generated from field discing were detectable at levels up to 0.3 µg m–3 at distances greater than 100 m from the source. Individual plumes were seen to have dispersion parameters of greater than 40 m (discing) and greater than 60 m (harvesting) at a downwind distance of 160 m. Wind and stability are the major factors controlling the movement of field-generated material.
The potential intensity of total dust exposure decreases with distance away from the tractor, but the time of exposure to a single dust plume increases. This project has shown that downwind exposures in the near field are likely to be intense but short-term events as the plumes move across the field. The human health effects of such time varying dosages are currently unknown.
|
ACKNOWLEDGMENTS
|
|---|
This work was supported with funds from the U.S. Dep. of Agriculture NRI CSREES program under contract 2004-35112-14230 and the Univ. of Connecticut, Storrs Agricultural Experiment Station. The authors are also grateful to staff at the Agricultural Experiment Station at New Mexico State Univ. for their generous cooperation during the field experiments. Special thanks to student research assistant Kathleen Knight for data processing assistance.
|
NOTES
|
|---|
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.
|
REFERENCES
|
|---|
- Gifford, F.A. 1959. Smoke plumes as quantitative air pollution indices. Int. J. Air Pollut.
2
:42–50.[Medline]
- Hiscox, A.L. 2006. Aerosol transport and dispersion measurements in the near surface boundary layer. Ph.D. diss. The University of Connecticut, Storrs.
- Hiscox, A.L., D.R. Miller, C.J. Nappo, Jr., and J.B. Ross. 2006a. Dispersion of fine spray from aerial applications in stable atmospheric conditions. Trans. ASABE
49
:1513–1520.
- Hiscox, A.L., C.J. Nappo, and D.R. Miller. 2006b. On the use of lidar images of smoke plumes to measure dispersion parameters in the stable boundary layer. J. Atmos. Oceanic Technol.
23
:1150–1154.[CrossRef]
- Holmén, B.A., D.R. Miller, A.L. Hiscox, W. Yang, J. Wang, T.W. Sammis, and R. Bottoms. 2006. Aerosol emissions from field planting operations. p. 169–172, In Proc. Workshop on Agricultural Air Quality: State of the Science. Potomac, MD, 5–8 June 2006.
- Kirkhorn, S.R., and M.B. Schenker. 2002. Current health effects of agricultural work: Respiratory disease, cancer, reproductive effects, musculoskeletal injuries, and pesticide-related illnesses. J. Agric. Saf. Health
8(2)
:199–214.
- Natural Resources Conservation Service. 2005. Soil Survey Geographic (SSURGO) Database for Dona Ana County Area, New Mexico. USDA-NRCS, Fort Worth, TX.
- Nieuwenhuijsen, M.J., H. Kruize, and M.B. Schenker. 1998. Exposure to dust and its particle size distribution in California agriculture. Am. Ind. Hyg. Assoc. J.
59
:34–38.[Medline]
- Nieuwenhuijsen, M.J., and M.B. Schenker. 1998. Determinants of personal dust exposure during field crop operations in California agriculture. Am. Ind. Hyg. Assoc. J.
59
:9–13.[Web of Science][Medline]
- Schenker, M. 2000. Exposures and health effects from inorganic agricultural dusts. Environ. Health Perspect.
Suppl. 108
(S4):661–664.[CrossRef][Medline]
- Stull, R.B. 1988. An introduction to boundary layer meteorology. Kluwer Academic, Boston.
- Stull, R.B. 2000. Meteorology for scientists and engineers. 2nd ed. Brooks/Cole, Pacific Grove, CA.
- Turner, D.B. 1994. Workbook of atmospheric dispersion estimates: An introduction to dispersion modeling. 2nd ed. Lewis Publ., Boca Raton, FL.
- Wilczak, J.M., and J.E. Tillman. 1980. The 3-dimensional structure of convection in the atmospheric surface layer. J. Atmos. Sci.
37
:2424–2443.[CrossRef]
- Wilks, D.S. 2006. Statistical methods in the atmospheric sciences. Academic Press/Elsevier Boston, MA.
TOP
ABSTRACT
INTRODUCTION
