Environ. Sci. Technol. 2001, 35, 2309-2313
Field Evaluation of a Method for Estimating Gaseous Fluxes from Area Sources Using Open-Path Fourier Transform Infrared RAM A. HASHMONAY,* DAVID F. NATSCHKE, AND KEITH WAGONER ARCADIS G&M, Inc., P.O. Box 13109, Research Triangle Park, North Carolina 27709 D. BRUCE HARRIS AND EDGAR L. THOMPSON U.S. EPA, ORD, NRMRL, Mail Drop 61, Research Triangle Park, North Carolina 27711 MICHAEL G. YOST Department of Environmental Health, Box 357234, University of Washington, Seattle, Washington 98195
This paper describes results from the first field experiment designed to evaluate a new approach for quantifying gaseous fugitive emissions of area air pollution sources. The approach combines path-integrated concentration data acquired with any path-integrated optical remote sensing (PI-ORS) technique and computed tomography (CT) technique. In this study, an open-path Fourier transform infrared (OP-FTIR) instrument sampled path-integrated concentrations along five radial beam paths in a vertical plane downwind from the source. A meteorological station collected measurements of wind direction and wind speed. Nitrous oxide (N2O) was released from a controlled area source simulator. The innovative CT technique, which applies the smooth basis function minimization method to the beam data in conjunction with measured wind data, was used to estimate the total flux from the simulated area source. The new approach estimates consistently underestimated the true emission rates in unstable atmospheric conditions and agreed with the true emission rate in neutral atmospheric conditions. This approach is applicable to many types of industrial areas or volume sources, given the use of an adequate PI-ORS system.
Introduction This paper describes results from the first field experiment to validate a technique for determining fluxes from fugitive gaseous air pollution sources. The method is designed as an applicable path-integrated optical remote sensing (PI-ORS) monitoring approach for estimating the total emission rate directly from the measured concentration and wind data. Moreover, this approach is independent of dispersion model assumptions. Several methods have been developed and applied (1-9) in the past to estimate emission rates from 10.1021/es0017108 CCC: $20.00 Published on Web 04/25/2001
2001 American Chemical Society
fugitive sources such as landfills (3), coal mines (5, 6), or water treatment plants (7, 8) using PI-ORS technologies. All previous methodologies combine downwind path-integrated concentration (PIC) data, wind measurements, and plume dispersion modeling to retrieve the total emission rate. The ideal approach for measuring the flux from an upwind emission source would directly measure the integrated concentration across the entire crosswind vertical plane located downwind from the emission source. Multiplying by the averaged wind speed component, in the normal direction to the vertical plane (weighted average by heights if several wind monitors are mounted in different elevations), provides the flux flowing through this plane and, therefore, the upwind source’s emission rate. In most situations, it is impractical to directly measure the plane-integrated concentration using PI-ORS methods due to the complex beam configuration needed to cover the entire relevant vertical plane (2, 7, 8). In more recent methods (3, 4), an open-path Fourier transform infrared (OP-FTIR) instrument was scanned diagonally from a fixed ground level location to retroreflectors at different heights to partially cover the vertical plane. Dispersion modeling was then applied to interpolate and extrapolate the measured data and generate the planeintegrated concentrations. Differential absorption lidar (DIAL) could directly measure a spatially resolved map of the contaminant concentration (10) and, by integrating the concentration over the whole map, provide the required plane-integrated concentration. Cost, a limited number of potentially detectable target contaminants, and calibration difficulties, so far, have prevented the wide application of the DIAL technology. In earlier simulation studies (11-14), we suggested and tested the application of radial beam geometry and the smooth basis functions minimization (SBFM) (15) computed tomography (CT) approach to reconstruct the smoothed field of concentration in a plane. The new relevant beam geometry is slightly different from the previous slanted beam geometry methods; several ground level retroreflectors in alternating path lengths have been added. This allows us to directly fit a bivariate Gaussian function to the measured PIC, and a smoothed mass equivalent concentration map is generated for the vertical plane. Estimates of the emission flux are retrieved in the same way as previous methods suggest, which is by multiplying the relevant wind speed at each height level.
Methodology The proposed methodology uses a two-dimensional SBFMCT technique applied to PI-ORS data to reconstruct the crosswind-smoothed concentration map in a vertical plane. The plane-integrated concentration from a reconstructed mass equivalent concentration map, along with the averaged wind data, provides an estimate of the total flux from the upwind emission source. The key to this methodology is a rather simple and sparse PI-ORS beam geometry that allows reconstruction of smoothed concentration maps in a downwind vertical plane. The setup of the beam geometry and upwind area source in the field experiment is illustrated in Figure 1. This beam geometry includes five beam paths approximately in the crosswind vertical plane. Three beam paths are on the ground level with different path lengths, and the tower elevates two retroreflectors above the longest path. When using an OP-FTIR instrument, we suggest scanning among the beam paths for at least 20 min to compensate for the relatively slow data acquisition capabilities and to allow a buildup of an approximate Gaussian plume. VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2309
To compute the assumed PIC values, the basis function is integrated along the beam path’s direction and path-length. In our beam geometry, it is convenient to express the smooth basis function G in polar coordinates r and θ:
G(r,θ) )
A 2πσyσzx1 - F212
{
[
(r‚cosθ - my)2 1 exp 2(1 - F212) σy2
]}
2F12(r‚cosθ - my)(r‚sinθ - mz) (r‚sinθ - mz)2 + σy σz σ2 z
FIGURE 1. Field configuration and OP-FTIR beam geometry. The beam geometry is in a vertical plane 50 m downwind from a controlled emission area source. The beam geometry consists of five beam paths, three scanning the OP-FTIR device to ground level (2 m height) retroreflectors, and two slanted beam paths scanning to elevated (7 and 13 m height) retroreflectors mounted on a tower. Wind speed and direction data are collected and averaged over the same time interval. To develop a reliable time-averaged plume profile, it is desirable to get as many repeated measurements as possible at each beam path. Therefore, the sampling time at each retroreflector should be as short as practical limits allow. For example, we used 1 min averaging in this field study. This relatively short time will allow the PI-ORS device to aim back 4 times to the same retroreflector over the about 20minute interval. The PIC data are then averaged for each beam path prior to application of the CT reconstruction method. We used the SBFM reconstruction approach with a twodimensional smooth basis function (bivariate Gaussian) in order to reconstruct the smoothed mass equivalent concentration map. In the SBFM approach, a smooth basis function is assumed to describe the distribution of concentrations, and the search is for the unknown parameters of the basis function. Since our interest is in the plane-integrated concentration and not the exact map of concentrations in the plane, we fit only one smoothed basis function (one bivariate Gaussian) to reconstruct the smoothed map. However, this methodology does not assume that the true distribution of concentration in the vertical plane is a bivariate Gaussian. Earlier computational studies (13, 14) showed that one might fit a single bivariate Gaussian function to many kinds of skewed distributions and still retrieve a reasonably good estimate of the plane-integrated concentration. The fit of a single bivariate Gaussian function to a multiple mode distribution was also examined and found that the reconstructed plane-integrated concentration conserved well the test input plane-integrated concentration. In each iterative step of the SBFM-CT search procedure, the measured PIC values are compared with assumed PIC values which are calculated from the new set of parameters. 2310
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001
(1)
The bivariate Gaussian has six unknown independent parameters: A, a normalizing coefficient that adjusts for the peak value of the bivariate surface; F12, correlation coefficient which defines the direction of the distribution-independent variations in relation to the Cartesian directions y and z (F12 ) 0 means that the distribution variations overlap the Cartesian coordinates); my and mz, peak locations in Cartesian coordinates; and σy and σz, standard deviations in Cartesian coordinates. To fit the unknown parameters of the smooth basis function to the PIC data, an error function for minimization has to be defined. The sum of squared errors (SSE) function is defined in our study as
SSE(A, F12,my,mz,σy,σz) )
∑(PIC - ∫ G(r, θ , A, F ri
i
i
0
i
12,my,mz,σy,σz)dr)
2
(2)
where PIC represents the measured PIC values and the index i is for the different beams. The SSE function is minimized using an iterative minimization procedure, such as the Simplex (16) method, to solve for the unknown parameters. As mentioned earlier, our interest is in the plane-integrated concentration; therefore, we fit one bivariate Gaussian surface to match the volume under the underlying true concentration distribution surface. This volume is highly conserved in the fitting procedure, which emphasizes agreement over the five path integrals. Six independent beam paths are sufficient to determine one bivariate Gaussian that has six independent unknown parameters. Some reasonable assumptions also may be made when applying the SBFM-CT method to this problem, in order to reduce the number of unknown parameters to four; e.g., setting the correlation parameter F12 equal to zero. This assumes that the reconstructed bivariate Gaussian is limited only to changes in the vertical and crosswind directions. In this case eq 1 reduces into
G(r,θ) )
{[
]}
2 (r‚sinθ - mz)2 A 1 (r‚cosθ - my) exp 2 2πσyσz 2 σy σz2
(3)
One also can fix the peak location in the vertical direction to the ground level when ground level emissions are known to exist, as in our field experiment. However, in this methodology, there is no requirement to apply a priori information on the source location and configuration. Once the parameters of the function were found for a specific run, we calculated the concentration values for every 4 × 4 m square elementary unit in a vertical domain size of 250 × 24 m. We then integrated these values, incorporating wind speed data at each height level to compute the flux. In this stage, we converted the concentration values from parts per million by volume to grams per cubic meter, considering the molecular weight of N2O and ambient temperature. This
enables us to calculate directly the flux in grams per second using wind speed data in meters per second. In most of the previous methodologies, OP-FTIR instruments were used as the PI-ORS device, mainly due to its simultaneous chemical analysis capability. However, when only a few species are of interest, it might be more efficient to employ other laser-based PI-ORS technologies such as tunable diode laser (TDL) or path-integrated DIAL, both of which have longer ranges, better detection limits, and much faster scanning capabilities.
TABLE 1. Summary of PIC Data (ppm-m) N2O PIC at run no.
retro 1
retro 2
retro 3
retro 4
retro 5
10/15-1 10/15-2 10/15-3 10/19-1 10/19-2 10/19-3 10/19-4
17.2 28.4 28.8 8.4 2.0 3.7 4.1
53.1 53.5 64.5 220 178 208 191
57.9 48.1 66.1 235 179 244 330
58.6 35.7 45.6 57.2 75.7 74.2 56.6
44.5 38.2 35.2 51.9 56.6 48.4 76.4
Field Study Study Design and Methods. The objective of the field study was to evaluate the feasibility of the suggested beam geometry and the CT algorithm in order to estimate the total emission rate under realistic changing meteorological conditions. Figure 1 presents the experimental layout for this work, which was performed at the Oxford-Henderson Airport in North Carolina. A monostatic Midac OP-FTIR was used in this field study. This instrument is described in detail in Childers et al. (17). It has a liquid nitrogen cooled mercury-cadmium-telluride (MCT) detector, and the IR source, interferometer, and detector housed in one box. A remote retroreflector is needed to return the external beam back over the same path to the spectrometer. Optical paths in a two-dimensional vertical array were utilized, consisting of five PLX, Inc. 20-cube retroreflectors as shown in Figure 1. A 12 m scissors jack supported mirrors 3, 4, and 5. Mirrors 1 and 2 were mounted on tripods at a height of 2 m above the ground at different distances from the OP-FTIR instrument. The OP-FTIR instrument was mounted on a scanner, custom built by the modeling and electronics shop of EPA’s Air Pollution Prevention and Control Division. Horizontal and vertical rotations were performed under computer control through a pair of Yuasa stepper-controllers. Meteorological data were collected every second using a Climet, Inc. system consisting of a Campbell data logger, two Tacmet sensing heads, and a radio frequency (RF) modem. Wind direction, wind speed, temperature, and relative humidity data were collected with this system. The Tacmet heads were located at heights of 2 and 10 m. The 2 m head was located on a tripod in the vicinity of the scissors jack. The 10 m head was mounted on top of the scissors jack. This system data are stored internally in the data logger and then downloaded directly to a computer via the modem. The N2O supplied by Air Products was released upwind of the optical path. The gas tank was fitted with a regulator, electric heat tape, and a Variac variable-voltage transformer. Since the gas is in liquid form in the tanks, there was significant expansive cooling, and the tank valve and regulator were heated with electric tape to prevent freezing. The N2O was released from an area source located about 50 m upwind of the optical path. This area source, 8 × 23 m, was constructed as an “H” pattern using porous rubber hosing. This hosing is a consumer grade product manufactured from recycled shredded rubber and is available as 0.75 in. diameter by 50 ft rolls. N2O flow rate was measured with a selfcalibrated Gilson tapered-tube flowmeter. Weighing the gas cylinder before and after an hour run confirmed the calibration of the flowmeter. Data presented here were collected during 2 days: October 15 and 19, 1999. Seven successfully completed runs were executed during the 2 days (three on the 15th and four on the 19th) this field study took place. Four plume traverses were considered as a complete successful run. Each plume traverse was completed after monitoring events were acquired at each of the five beam paths. A monitoring event was an average of 18 scans of the OP-FTIR instrument, with
a spectral resolution of 0.5 cm-1. Therefore, the duration of a monitoring event was approximately 1 min, and the duration of each run was about 20 min in this study. In all runs, the emission rate was set up to a nominal rate of 60 SLPM (standard liters per minute). This yields an emission rate of 1.7 g/s in standard conditions. All seven runs consisted of a total of 28 plume traverses, or 140 monitoring events, all of which were analyzed and quantified for N2O PIC. Absorbance spectra were generated using synthetic background spectra as described in the OP-FTIR guidance document (18). Classical least squares (CLS) fit was performed using the Midac Autoquant software and Infrared Analysis, Inc. reference spectra. The spectral analysis region ranged from 2075.5 to 2223.6 cm-1 and included the spectral features due to the absorption of carbon monoxide, carbon dioxide, nitrous oxide, and water vapor. The quantification procedure followed the procedure described in Childers et al. (17). Each day, background spectra were collected for the five measurement paths. Background spectra were acquired before the N2O release started, and we quantified these spectra for N2O to evaluate the background concentration, which was subtracted from all the quantified concentrations for the same day. After background subtraction, PIC values in each beam path were averaged for each run. As described in the previous methodology section, we applied SBFM-CT with the Simplex minimization algorithm to the measured PIC values to solve for the bivariate Gaussian function unknown parameters. We assumed that the correlation factor was equal to 0 and that the peak height and the source were at ground level. The latter of the two assumptions is based on the fact that PIC data on the lower retroreflector were always the highest (Table 1). We did not apply any restrictions for the other four unknown parameters. As a first guess, we substituted the following values: the peak concentration was 5 times the PIC of the third retroreflector; crosswind peak location was 150 m from the OP-FTIR; the standard deviation of the plume’s dimension, in the crosswind direction, was 10 m; and the standard deviation of the plume’s dimension, in the vertical direction, was 2.6 m. We chose these values arbitrarily and applied them to all reconstructions executed in this study. Running the SBFM-CT procedure with a different set of first-guess parameters did not change the resulting reconstructed fluxes. The reconstructed emission rate was calculated by numerical integration of the reconstructed bivariate Gaussian function weighted by multiplication of the interpolated wind speed and the cosine of the wind shift angle at each relevant height.
Results and Discussion The 2 days were very different in atmospheric stability, as reflected in the collected wind data in Table 2. The standard deviations in wind directions, σθ, are much larger for 10/15/99, and are compatible with category A of the PasquillGifford stability categories (σθ > 25 degrees) (19). On 10/19/99, the standard deviations in wind direction are between 9 and 18 degrees, which are compatible with VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2311
TABLE 2. Summary of Wind Data wind speed at
wind direction at
run no.
2m [m/s]
10 m [m/s]
2m [deg]
10 m [deg]
σθ at 10 m [deg]
10/15-1 10/15-2 10/15-3 10/19-1 10/19-2 10/19-3 10/19-4
1.9 2.1 2.0 2.2 2.4 1.7 1.7
2.5 2.9 2.7 2.8 3.2 2.2 2.5
117.1 71.2 70.9 7.8 13.8 22.6 20.3
62.0 48.4 48.3 22.9 27.8 39.4 31.6
88.5 31.9 31.6 10.8 9.3 18.4 12.1
FIGURE 3. The average reconstructed plume and levels where the PIC values for the four runs were averaged prior to the reconstruction and the reconstructed plumes and levels for each of the four runs on 10/19/99. The estimated emission flux is shown in the title of each graph. FIGURE 2. The average reconstructed plume and levels where the PIC values for the three runs were averaged prior to the reconstruction and the reconstructed plumes and levels for each of the three runs on 10/15/99. The estimated emission flux is shown in the title of each graph. categories D and C. These categories (A-D) cover the range of the daytime atmospheric stability conditions (19). Table 1 presents the PIC of N2O after background subtraction and shows a significant difference in the N2O PIC values between the 2 days of the field study. In the calculation of the flux, these differences are compensated for by the vertical and cross-sectional dimensions of the plume in which the integration is performed, and by taking the normal component of the wind speed to the vertical plane. Figure 2 illustrates the reconstructed plume dimensions and levels for the three runs on 10/15/99. It also shows the average reconstructed plume where the PIC values for the three runs were averaged prior to the reconstruction. The estimated emission flux is shown in the title of each graph, and the averaged emission flux is 1.13 ( 0.107 g/s ((10%) for that day. This consistly underestimated the true released emission rate, which was 1.7 g/s. The calculated emission from the reconstructed plume of the averaged PIC was similar 2312
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001
(1.12 g/s) as expected. Generally, the reconstructed plume’s dimensions on the 15th are much larger than the reconstructed plume’s dimensions on the 19th, as should be expected for the unstable atmospheric conditions. It is obvious that the variability in wind direction for the first run should be much larger than in runs 2 and 3, since the reconstructed plume in run 1 is much larger. Thus, σθ in run 1 is 89 degrees compared to 32 degrees in runs 2 and 3 as in Table 2. Figure 3 illustrates the reconstructed plume dimensions and levels for the four runs on 10/19/99. It also shows the average reconstructed plume where the PIC values were averaged prior to the reconstruction. The estimated emission flux is shown in the title of each graph, and the averaged emission flux is 1.53 ( 0.194 g/s ((13%) for that day. The calculated emission from the reconstructed plume of the averaged PIC was similar (1.45 g/s) as shown in the first graph of Figure 3. This result, for a more stable (neutral) day, usually slightly underestimated the true release rate by 10-15%. We note that earlier computational studies (13, 14) showed a similar trend of underestimation of the simulated emission flux. Nevertheless, these results show robust emission
reconstruction data that are very consistent with atmospheric conditions. We found the CT SBFM approach for estimating the emission rate to be robust for different atmospheric stability conditions, as long as the average wind direction vector intersecting the emission source is pointing between the first retroreflector (#1) and the tower. As in the computational studies, the Simplex algorithm was found to be sufficient. First applying a one-dimensional SBFM approach (20) to the ground-level-segmented beam paths allowed us to substitute the resulted values as fixed parameters of the bivariate Gaussian function. This afforded more degrees of freedom in the two-dimensional SBFM solution but did not change the reconstructed flux values. Based on the experimental data of this validation study, this method can provide consistent estimates of emission flux, which are well correlated to atmospheric stability conditions. This technology provides fairly robust estimates of the total emission from many kinds of fugitive sources along with the selection of an adequate PI-ORS device. A laserbased monitoring system may provide estimates with better time resolution than the 20 min applied in this study. Such a system could be applied in a continuous monitoring mode to function also as an alarm for departure from normal working conditions of many types of area and volume emission sources. Reconstructed maps should be able to provide measures of the initial near field dispersion parameters. These parameters, as atmospheric stability condition indicators, along with the estimated flux, could be input into a dispersion model to calculate in near real time the downwind field of concentrations.
Acknowledgments The algorithm development was partially supported by the Consortium for Risk Evaluation with Stakeholder Participation (CRESP) through Department of Energy Cooperative Agreement #DE-FC01-95EW55084. This support does not constitute an endorsement by DOE of the views expressed in this article. We appreciate the support of Jimmy Brummitt and Joseph Mahaffey from the Vance-Granville Airport Authority in allowing these tests to be conducted at their facility and Neal E. Darnell and Michael Kellogg of the Airport Staff for advice and suggestions during field operations.
Literature Cited (1) Scotto, R. L.; Minnich, T. R.; Leo, M. R. A method for estimating VOC emission rates from area sources using remote optical sensing. In Proceedings of the EPA/AWMA International Symposium on the Measurement of Toxic and Related Air Pollution; Raleigh, NC, 1991; p 698. (2) Minnich, T. R.; Scotto, R. L.; Leo, M. R.; Sanders, B. C.; Perry, S. H.; Pritchett, T. H. A practical methodology using open-path FTIR spectroscopy to generate gaseous fugitive-source emission factors at industrial facilities. In Proceedings of the Optical Remote Sensing, Application to Environmental and Industrial Safety Problems; Houston, TX, SP-81, 1992; p 513. (3) Milton, M. J. T.; Partridge, R. H.; Goody, B. A. Minimum emission rates detectable from landfill sites using optical integrated-path techniques. In Proceedings of the A&WMA International Specialty Conference on Optical Sensing for Environmental and Process Monitoring; San Francisco, CA, VIP-55, 1995; p 393. (4) Piccot, S. D.; Masemore, S. S.; Lewis-Bevan, W.; Ringler, E. S.; Harris, D. B. Field assessment of a new method for estimating emission rates from volume sources using open-path FTIR Spectroscopy. J. Air Waste Manage. Assoc. 1996, 46, 159.
(5) Piccot, S. D.; Masemore, S. S.; Ringler, E. S.; Srinivasan, S.; Kirchgessner, D. A.; Herget, W. F. Validation of a method for estimating pollution emission rates from area sources using open-path FTIR spectroscopy and dispersion modeling techniques. J. Air Waste Manage. Assoc. 1994, 44, 271. (6) Kirchgessner, D. A.; Piccot, S. D.; Chadha, A. Estimation of methane emissions from a surface coal mine using open-path FTIR spectroscopy and modeling techniques. Chemosphere 1993, 26 (1-4), 23. (7) Simpson, O. A.; Kagan, R. H. Measurements of emissions at a chemical wastewater site with an open path remote Fourier transform interferometer. In Proceedings of the EPA/A&WMA International Symposium on the Measurement of Toxic and Related Air Pollution; Raleigh, NC, 1990; p 937. (8) Whitcraft, W. K.; Wood, K. N. Use of remote sensing to measure wastewater treatment plant emissions. In Proceedings of the 83rd Annual Meeting & Exhibition of the A&WMA; Pittsburgh, PA, 1990. (9) Hashmonay, R. A.; Yost, M. G.; Mamane, Y.; Benayahu, Y. Emission Rate Apportionment from Fugitive Sources Using Open-Path FTIR and Mathematical Inversion. Atmos. Environ. 1999, 33(5), 735. (10) Walmsley, H. L.; O’Connor, S. J. The use of differential absorption LIDAR to measure atmospheric emission rates at industrial facilities. In Proceedings of the A&WMA International Conference on Optical Sensing for Environmental and Process Monitoring; Dallas, TX, 1996; p 127. (11) Hashmonay, R. A.; Yost, M. G.; Wu, C. F. Computed tomography of air pollutants using radial scanning path-integrated optical remote sensing. Atmos. Environ. 1999, 33(2), 267. (12) Price, P. N. Pollutant tomography using integrated concentration data from nonintersecting optical paths. Atmos. Environ. 1999, 33(2), 275. (13) Hashmonay, R. A.; Yost, M. G. Innovative Approach for Estimating Gaseous Fugitive Fluxes Using Computed Tomography and Remote Optical Sensing Techniques. J. Air Waste Manage. Assoc. 1999, 49, 966. (14) Hashmonay, R. A.; Yost, M. G.; Harris, D. B.; Thompson, E. L., Jr. Simulation Study for Gaseous Fluxes from an Area Source Using Computed Tomography and Optical Remote Sensing. In Proceedings of SPIE Environmental Monitoring and Remediation Technologies Conference; Boston, MA, November 1998; p 405. (15) Drescher, A. C.; Gadgil, A. J.; Price, P. N.; Nazaroff, W. W. Novel Approach for Tomographic Reconstruction of Gas Concentration Distributions in Air: Use of Smooth Basis Functions and Simulated Annealing. Atmos. Environ. 1996, 30(6), 929. (16) Press: W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes in FORTRAN, 2nd ed.; Cambridge University Press: Cambridge, MA, 1992. (17) Childers J. W.; Thompson, E. L.; Harris, D. B.; Kirchgessner, D. A.; Clayton, M.; Natschke D. F.; Phillips, W. J., Muliti-Pollutant Concentration Measurements Around a Concentrated Swine Production Facility Using Open-Path FTIR Spectrometry. Atmos. Environ. 2001, in press. (18) Russwurm, G. M.; Childers, J. W. FT-IR Open-Path Monitoring Guidance Document, 3rd ed.; Submitted by ManTech Environmental Technology, Inc., under contract 68-D5-0049 to the U.S. EPA, Human Exposure and Atmospheric Sciences Division, National Exposure Research Laboratory: Research Triangle Park, NC, 1999. (19) Seinfeld, J. H. Air Pollution: Physical and Chemical Fundamentals; McGraw-Hill: New York, 1975. (20) Hashmonay, R. A.; Yost, M. G. Localizing Gaseous Fugitive Emission Sources by Combining Real Time Optical Remote Sensing and Wind Data. J. Air Waste Manage. Assoc. 1999, 49, 1374.
Received for review September 28, 2000. Revised manuscript received March 2, 2001. Accepted March 8, 2001. ES0017108
VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2313
