Near Field Dust Exposure from Cotton Field Tilling and Harvesting

Transcription

1 Near Field Dust Exposure from Cotton Field Tilling and Harvesting April L. Hiscox* and David R. Miller The University of Connecticut Department of Natural Resources Management and Engineering 1376 Storrs Road U-4087 Storrs, CT USA Britt A. Holmén The University of Vermont Civil & Environmental Engineering Votey Bldg Room 213B Burlington, VT Wenli Yang Crocker Nuclear Laboratory One Shields Ave. University of California Davis, CA Junming Wang New Mexico State University Plant & Environmental Sciences Dept. 3Q Las Cruces, NM * Corresponding author: Phone: , Fax:

2 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. This paper presents measurements of near-field spatial concentrations of the dust plumes emitted during tilling and harvesting of an irrigated cotton field. Remote lidar measurements of plumes emitted from cotton field disking and harvesting are presented. 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 sensible heat convecting off the soil surface. On a windy day, disking the field showed TSP and PM 10 concentrations at the source itself of up to 176 μg/m 3 and 120 μg/m 3, respectively. These resulted in in-plume peak concentrations of about 1.22 μg/m 3 at 10 m downwind and 0.33 μg/m 3 at 100 m downwind. This paper demonstrates the integration of lidar images of aerosol plume cross sections to determine suspended aerosol concentrations at varying distances downwind of the tractor. Keywords Lidar, Fugitive Dust, Exposure, Stability, Dispersion, disking, cotton harvest 1. 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 1

3 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. There is a growing literature on the health effects from epidemiology studies of respirable dust aerosols from agriculture. 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 has demonstrated that cab technology and management practices can reduce exposure on the tractor significantly (Nieuwenhuijsen and coworkers 1998a, b). 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 best management practices for the reduction of emissions. The experimental field was a flood-irrigated cotton field at the New Mexico State University, Leyendecker Plant Science Research Farm, Las Cruces, NM, (Lat.32.2 º N, Long ºW). 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. 2

4 This paper reports on the near-ground time and space variations of aerosol dust concentrations during tilling and harvesting of the cotton field. Its purpose is to quantify the near-field spatial concentrations of the emitted dust plumes which potentially contribute to high intensity human exposure events. 2. Methods 2.1 Measurements The 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 PM 10. In addition, micrometeorological parameters (3-D wind components and temperature) were measured at a rate of 20 Hz with a 3 dimensional sonic anemometer (Model: CSAT3, Campbell Scientific, Logan, UT). The sonic anemometer measurements were used to calculate the stability parameter (z/l), where z in this case is 1.5 m and L is the Obukov length [L=-(ρ c p T u 3 * )/(k g H)], where ρ is the air density, c p is the specific heat of air, T is the air temperature, u * is the friction velocity, k is the von Karman constant (0.4), g is the acceleration due to gravity, and H is the sensible heat flux (Stull, 1988). 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 5-9 full threedimensional plume pictures for each pass. Further details of these instruments and the measurement procedures can be found in the companion paper (Holmén et al. 2006). This paper focuses on the plume dispersion and concentration as measured from the lidar scans for two different operations: spring disking and fall harvesting. 3

5 Dispersion The most usable way of quantifying plume dispersion is by the Gaussian plume dispersion parameters: σ y (cross wind) and σ z (vertical), where σ 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. (2006a) in the horizontal direction: 2 ( 0.5ΔY ) ( 2ln α ) σ y = [1] where, ΔY is the edge to edge distance (meters) of the plume perpendicular to the plume axis, and α is the ratio of maximum backscatter (B m ) to edge backscatter (B e ). 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 minutes), 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 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. 4

6 94 {insert Figure 1 about here} Concentrations/Calibration Plume concentrations are determined following the methods of Hiscox et al. (2006b). 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, that 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 three meters 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 the plumes as shown in Figure 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. {insert Figure 2 about here} Figure 3 displays the calibration line for daytime disking operations. The linear relationship confirms that a stronger source results in a larger, more concentrated plume. The calibration presented includes only passes that occurred under unstable conditions (Holmén et al., 2006). Passes that occurred under near neutral and stable conditions are not used in the calibration because an inverse relationship is seen due to 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 (Holmén et al., 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 Figure 5

7 , where in contrast to Figure 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 Results {insert Figure 3 about here} {insert Figure 4 about here} Cross-wind dispersion Horizontal dispersion parameters along the plume axis for disking 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 three meters above the ground. This location is taken to represent that of potential exposure by humans. For disking, 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 meters per meter of downwind distance. The measurements here show a linear growth of meters per meter of downwind distance. The lidar measurements also show a larger initial spread (7.6 meters) as opposed to the zero spread in the point source theory. This is due to the nature of these operations. The tractor is a moving source, and the disk path is 3m wide, so the larger area source resulted in a wider initial spread. {insert Table 1 about here} 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 meters downwind and then increases again. This trend can be seen in Figure 5 which shows a single lidar slice during the harvest, with two separate plume cells easily identifiable. It should be noted that 6

8 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. {insert Figure 5 about here} Plume Concentrations Plume concentrations with height are shown in Figure 6 for the unstable disking 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 dispersion parameters. {insert Figure 6 about here} 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 meters downwind for the 10 different disking passes, along with the average wind speed for each pass. The concentrations shown are the approximate concentrations of total suspended particulates that an individual 10 or 100 meters from the tractor would be exposed to. {insert Table 2 about here} 4. 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 Figure 1 where the portion of the plume farthest from the source is over 100 meters 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 disking and the harvesting day were similar (Holmén et al., 2006), 7

9 the differences between the operations were due to varying meteorological conditions rather than a difference in material. During the disking operation, much higher wind speeds were observed throughout the day; an average of 5.4 m/s compared to 1.6 m/s on the day of harvest. Both days were unstable with the average z/l = -1.2 for the disking day and for the harvesting day. But the very low wind speeds on the harvesting day made it extremely unstable, frequently approaching free convection. Wilczak and Tillman (1980) and others have shown that under unstable conditions sensible heat is convected away from the ground surface in convective cells with typical length and width dimensions less than a few hundreds and greater than about ten meters, 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 Figures 2 and 5 as a sequence of billows in the plumes. Under the moderately unstable, higher wind speed conditions of the disking day, the plumes in Figure 2 are arranged into about 5 convective cells over a 200 m length of plume, or about 1 every 40 meters. 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 meters long but are less elongated in the downwind direction and larger in the cross wind direction as shown in Figure 5. The higher wind speeds during disking 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 8

10 speed unstable case, but there is little vertical dispersion or vertical rise. This indicates plume fanning occurs on a short time scale (Figure 4). 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 us uncertain of the plume characteristics in these conditions. Plume concentrations were only available for the unstable passes of the disking 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 can not be applied in situations were the full plume cannot be scanned as was the case during stable conditions when a majority of the plume stayed under the three meter minimum height measurement of the lidar as noted above. Figure 3 shows peak TSP and PM 10 concentrations at the source itself of up to 176 μg/m 3 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 meters downwind and average of 0.24 μg/m 3 at 100 meters downwind. Based on the average plume width (± 3 σ y ) and tractor speed, at a fixed point 10 meters downwind of the tractor path a person at a single location would be exposed to a plume for approximately 30 seconds. This results in an average 30 second potential total dust exposure of 6.8 μg/m 3 for each pass of the tractor. At 100 meters downwind, this value decreases to 4.8 μg/m 3 over an exposure time of 50 seconds 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 minutes, 28 seconds at 10 meters downwind and 16 minutes, 42 seconds at 100 meters downwind, for an entire day of operations. The NIOSH REL 9

11 (National Institute of Occupational Safety and Health Relative Exposure Limit) for silica is 0.05 mg/m 3 for a 10 hour 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. 5. Conclusion Presented here are measurements of downwind plume dispersion and concentrations from agricultural generated dust. Clouds of dust generated from field disking were detectable at levels up to 0.3 μg/m 3 at distances greater than 100 meters from the source. Individual plumes were seen to have dispersion parameters of greater than 40 meters (disking) and greater than 60 meters (harvesting) at a downwind distance of 160 meters. 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. 10

12 Acknowledgements This work was supported with funds from the U.S. Department of Agriculture NRI CSREES program under contract and the University of Connecticut, Storrs Agricultural Experiment Station. The authors are also grateful to staff at the Agricultural Experiment Station at New Mexico State University for their generous cooperation during the field experiments. Special thanks to student research assistant Kathleen Knight for data processing assistance. 7. References Gifford, F. A., Smoke plumes as quantitative air pollution indices. International Journal of Air Pollution, 2: Hiscox A. L., D.R. Miller, and C.J. Nappo, 2006a. On the use of lidar images of smoke plumes to measure dispersion parameters in the stable boundary layer. Journal of Atmospheric and Oceanic Technology, 23: Hiscox, A. L., D.R. Miller, C.J. Nappo, and J. Ross, 2006b. Dispersion of fine spray from aerial applications in stable atmospheric conditions. Transactions of the ASABE, in press Holmén B. A., D.R. Miller, A.L. Hiscox, W. Yang, J. Wang, T. Sammis, and R. Bottoms, Near-Source Particulate Emissions and Plume Dynamics from Agricultural Field Operations, Journal of Atmospheric Chemistry, under review

13 Nieuwenhuijsen, M. J. and M. B. Schenker a. Determinants of Personal Dust Exposure During Field Crop Operations in California Agriculture. American Industrial Hygiene Association Journal 59, Nieuwenhuijsen, M. J., H. Kruize, M. B. Schenker b. Exposure to Dust and its Particle Size Distribution in California Agriculture. American Industrial Hygiene Association Journal 59, Schenker, M., Exposures and health effects from inorganic agricultural dusts, Environmental Health Perspectives Supplements, 108, S4, Stull, R.B An Introduction to Boundary Layer Meteorology. Kluwer Academic Publishers, Boston Stull, R. B., Meteorology for Scientist and Engineers, Brooks/Cole, United States Turner, Workbook of Atmospheric Dispersion Estimates, Lewis Publishers, Boca Raton Wilczak, J. M., Tillman, J. E The Three-Dimensional Structure of Convection in the Atmospheric Surface Layer, Journal of Atmospheric Science, 37, Figure Captions Figure 1: Example lidar slice of a horizontal plume. The rectangle represents the field being work and the large arrow indicates the path of the tractor. The dotted line indicates the 12

14 plume axis direction and the average wind direction. The source point is at the intersection of the plume maximum line and the tractor path, in this example, that is also the point of maximum backscatter. B e is the edge backscatter, B m is the maximum backscatter and ΔY is the plume width as needed for application of equation Figure 2: Three dimensional rendering of combined lidar scans: (a) looking from the top down. (b) viewed from an oblique angle. The three plumes are all from the same pass of the tractor, the darkest is the first lidar sequence and the lightest is the last and is after the tractor had stopped at the edge of the field. The arrow indicates the approximate tractor path on the field Figure 3: TSP at the ground vs. total lidar backscatter approximately three meters above the ground. The linear fit is forced to an intercept equal to the background TSP. The result is a slope of and an R-squared value of The PM 10 fit is with a zero intercept and an R-squared of Figure 4: An example of a plume generated under stable conditions. (a) Top down view (b) Vertical cross sectional contours of the plume. Under these conditions the maximum plume height is less than 30 meters as opposed to Figure 2 where the plume exceeds 50 meters Figure 5: Lidar slice from harvesting day. Two intense plume cells can be seen where the dust is entrained in surface layer convective cells

15 Figure 6: Height above the ground versus concentration of material aloft for the ten unstable passes on the disking day. 14

16 Tables Table 1: Cross wind dispersion parameters at varying distances downwind of the operation Disking Harvest Downwind Distance (meters from source) Mean σ y (meters) Standard Deviation Downwind Distance (meters from source) Mean σ y (meters) Standard Deviation Table 2: Peak downwind concentrations and average wind speeds for disking operation unstable passes. Pass Concentration Concentration Average 10 meters downwind 100 meters downwind Wind Speed (m/s) (μg/m 3 ) (μg/m 3 ) Average

17 Figure 1, Hiscox, A.L., Near Field Dust Exposure from Cotton Field Tilling and Harvesting

18 Figure 2, A.L. Hiscox, Near Field Dust Exposure from Cotton Field Tilling and Harvesting

19 Figure 3, A.L. Hiscox, Near Field Dust Exposure From Cotton Field Tilling and Harvesting

20 Figure 4, A.L. Hiscox, Near Field Dust Exposure from Cotton Field Tilling and Harvesting

21 Figure 5, A.L. Hiscox, Near Field Dust Exposure from Cotton Field Tilling and Harvesting

22 Figure 6, A.L. Hiscox, Near Field Exposure from Cotton Field Tilling and Harvesting