fnviron. Sci. Technol. 1995, 29, 330-337
Idustrial Emission C1Lidar Techlligues
Usina U
HANS E D N E R , * PAR R A G N A R S O N , A N D E V A WALLINDER Department of Physics, Lund Institute of Technology, P.O. Box 118, S-221 00 Lund, Sweden
A mobile DIAL (differential absorption lidar) remote sensing system has been employed in the monitoring of industrial pollutant emissions. Measurements of sulfur dioxide and mercury vapor were performed at nine different Swedish industrial plants within the framework of a control program commissioned by the Swedish Environmental Protection Agency. Total atmospheric fluxes of these species could be determined by combining wind data with a mapping of the concentration distribution downwind from the sources. The emission values obtained were compared with in situ measurements made by the companies themselves. The values from single point sources correlate well, whereas the DIAL system normally measures a higher total flux from several distributed sources including diffuse emissions. The results are used to discuss the applicability and limitation of the DIAL technique for remote surveillance of industrial emissions.
Introduction With the growing awareness of the serious environmental impact of certain industrial activity and more stringent regulations on emissions, the need for powerful measurement techniques for air pollutant emission is increasing. Optical remote sensing techniques are particularly advantageous, allowing large-area monitoring and avoiding sample extraction and preparation difficuities (1-3). Longpath absorption techniques, such as differential optical absorption spectroscopy (DOAS),have been very successful during recent years and are routinely used to monitor the air quality in many urban and industrial areas ( 4 , 5 ) . While DOAS can only give the mean pollutant concentration over a fixed path, three-dimensional measurements can be performed with the lidar (light detection and ranging) technique, by using pulsed lasers and detecting the backscattered light from molecules and aerosols in the atmosphere in a radar-like mode (6).If the wavelength of the laser is varied from an absorption line of a certain gas to a nearby wavelength position with less absorption, the detected changes in the backscattered light intensity can be used to evaluate the gas concentration profile along the laser beam. This technique, called differential absorption lidar (DIAL),is now in operation for many pollutants, and several mobile DIAL systems have been developed. The ability to remotely measure the flux of a certain gas from different sources is particularly useful. This is achieved by combining wind data with a mapping of the concentration distribution in avertical section downwind from the source. In this way, diffuse emissions can also be studied, and the total pollutant flux from all sources can be measured. In this paper, we will discuss the application of mobile DIAL for measurements of industrial emissions of mostly SOPand Hg performed at nine different Swedish companies, listed in Table 1,within the framework of a control program commisioned by the Swedish Environmental Protection Agency. The measurements were made during four short periods within 8 months at different places in Sweden. In the next section, the lidar equipment and measurement methodologywill be described, followed by some examples of the results from different industries, including comparisons with in situ measurements of the emissions. Finally, the applicability and limitation of lidar techniques for industrial emission control will be discussed.
Lidar Equipment and Measurement Methodology The mobile DIAL system used in these measurements is a flexible system designed for the detection ofvarious species. The system has previously been described in detail (7),but some of the subsystems have since been upgraded. An overview of the system is given in Figure 1. The laser, telescope, and all electronics are contained in aVolvo F610 truck with a specially designed cargo compartment measuring 6.0 x 2.3 x 2.1 m3, The truck is equipped with four sturdy supporting legs, which can be hydraulicallylowered to ensure high stabilityduring measurements requiringhigh directional accuracy in the optical system. Electric power in the field is supplied by a 20-kWdiesel generator installed in a small trailer towed by the truck. The important characteristics of the updated DIAL system are summarized in Table 2. The laser transmitter
330 ENVIRONMENTAL SCIENCE &TECHNOLOGY I VOL. 29, NO.
2,1995
0013-936X/95/0929-0330$09.00/0
0 1995 American
Chemical Society
TABLE 1
List of Industrial Plants Where Lidar Measurements Were Performed under Contract with the Swedish Environmental Protection Agency during 1992 date
industry
type
location
species
Feb 17-18 Apr 22 Apr 23-24 Apr 27-28 May 18-19 May 20 May 21-22 Oct 14-17 Oct 19-21
Sysav Helsingborgs Energi Kemira Kemi Stora Papyrus Eka Nobel Norsk Hydroplast Scanraff Boliden Ovako Steel
refuse incineration heating plant (coal) chemical plant paper pulp mill chloralkali plant chloralkali plant oil refinery metallurgical industry steel industry
Malm0 Helsingborg Helsingborg Nymblla Bohus Stenungsund Lysekil Skelleftehamm Hofors
NO2, Son, Hg S02, Hg
I
I
I
I
I
I
0
I
2
3
4
5
so2 so2 Hg Hg
so2
Son, Hg Hg
m
U FIGURE 1. Overview of the mobile DIAL system.
is a Nd:YAG-pumped dye laser with frequency doubling and/or mixing to the UV and IR regions. The dye laser is equipped with a dual-wavelength option, enablingthe laser to be fired alternately at two different preset wavelengths. The dye laser wavelength scale can be calibrated against known Ne lines, observedwith the optogalvanic signal from a Ne-filled, hollow-cathode lamp, illuminated by a small part of the laser beam. Normally, the absorption in a small gas cell Wed with the species under investigation is used for fine tuning the wavelengths. The outgoing laser beam is directed coaxially with a vertically mounted telescope and transmitted into the atmosphere via a large flat mirror in a retractable transmittinglreceiving dome on the roof. A quartz window seals the dome. Steppingmotors are used to turn the dome and to tilt the mirror. It is thus possible to steer the measurement direction 360" horizontally and within an angle of 45" vertically. An extra folding mirror can be installed on the roof to facilitate vertical measure-
ments. In order to protect the mirror coating, the diameter of the exiting laser beam is made six times larger with a beam expander. Servo motors and micrometer screws provide remote control of the final turning prism, through which the overlap between the laser transmission lobe and the telescope field of view is controlled. Two video cameras are used to control and supervise the measurement direction. A mechanical chopper can block the beam if desired. The computer-controlledchopper is automatically used during a lidar measurement to obtain the signal due to background light and preamplifier offset. An adjustable field stop is placed in the focus of the receiving Newtonian telescope, by means of which the field of view can be varied from 2 to 5 mrad. After passing through an interference filter, the light is detected by a photomultiplier tube. Due to the fact that the beam is emitted coaxially with the telescope, the near-field backscattered light is very strong. To prevent detector overload VOL. 29, NO. 2, 1995 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
331
TABLE 2
Data for the Mebiie DIAL System laser dye laser
wavelengths
Emitter Continuum YG682-20, Nd:YAG rep. rate 20 Hz, 6-9 ns pulse length 1200 rnJ at 1064 nm Continuum TDL6O frequency doubling/mixing 200-4500 nm, 2-150 mJ, 0.5 mrad dual-wavelength, alternate-switching NOz: 448.03 nm (on), 446.60 nrn (off) SOz: 300.02 nm, 299.30 nm Hg: 253.652 nrn, 253.665 nrn
Receiver Newtonian, 40 cm diameter, f/2.5 computer-controlled steering mirror interference filters N02: transmission 73%, bandwidth 5 nm SOZ: 38%, 5 nm Hg: 12%, 9 nm, or 30%, 45 nm photodetector Thorn-EM19816QA. S20 cathode gain IO7, ramped 2-lops LeCroy units 6102,2 x TR 8818, digitizer 2 x NM8103A, 8013 8-bit, 10-MHz computers 2 IBM-compatible PC, 386 telescope
and to reduce the dynamic range of the signal, the gain of the PMT is modulated. This is performed by changing the voltage difference in the dynode chain. The gain is very low in the beginning and does not reach its full value until after 2-lops (the rise time isvariable). Due to geometrical effects of the Newtonian telescope and possible nonlinearities in the detector, the signal from the first 200 m is normally not used in the evaluation sequence. The signal from the PMT is preamplified and captured with an 8-bit 100-MHztransient digitizer. Normally, 2000 channels, each 10 ns wide, are recorded, corresponding to a lidar measuring range of 3 krn. After the recording, the data are transferred via a GPIB interface to the computer where they are added into a 32-bit data array at arepetition rate of 20 Hz. A DIAL measurement cycle consists of eight shots at each wavelength, fired alternately, and finally two shots with the chopper closed. Several cycles are then averaged and stored on disk after background subtraction. During these campaigns, normally 50 cycles were averaged, corresponding to a measurement time of less than 1 min for one lidar direction. Averticallidarscan usually consisted of 10- 15 directions where the measurement time for each direction was divided in several repetitions. While system control and on-line data handling are controlled by the system computer, data processing and evaluation can be performed on a separate computer. A typical evaluation consists of three phases. First a Gaussian smoothing function of a few channels' width is applied to the raw lidar signals. Then the ratio of the lidar signal at the two wavelengths is calculated. A running average function of adjustable width is finally used to evaluate the concentration as a function of distance. Several concentration profiles can be averaged over a chosen time period. Data from horizontal or vertical scans are presented as a 2D plot with the concentration value indicated by a gray scale. The data from a vertical scan can also be transferred to a single vertical or horizontal profile with a projection algorithm. In measurements of a pollutant flux from an industrial plant, vertical lidar scans are made downwind from all the 332 1 ENVIRONMENTAL SCIENCE &TECHNOLOGY / VOL. 29, NO. 2, 1995
sources. The lidar system should be placed in such a position that the plume can be scanned at an angle of about 90" to the wind direction. Normally, angles between 50" and 130" are adequate, whereas other angles make the flux determination very sensitive to an accurate measurement of the wind direction and to small wind shifts during the lidar measurements. Lidar scans are normally not made very close to the outlet from a smokestack as the plume velocity may be affected by the total gas flux inside the smokestack. On the other hand, if this gas flux is known, the pollutant flux can in some situations be determined by measuring the concentration at the outlet. Another limiting factor on measurements in the near-field is often that the attenuation of the laser beam is too high, either due to high pollutant concentration or due to particles and condensed water vapor in the plume. Better conditions are normally obtained at some distance from the source, of the order of some hundred meters. This distance is dependent on the initial concentration and the atmospheric conditions. The measurement directions should, if possible, be chosen so that the highest attenuation due to the plume is less than 75%. The possible influence of an inflow of a pollutant to the plant can be checked by performing a vertical lidar scan upwind of the plant. The flux is calculated from a vertical lidar scan by first integrating the measured concentrations over the area of the vertical plane mapped out by the scan. This integrated value is then multiplied by the wind speed perpendicular to the plane at the height of the plume's center of gravity. If the plume is distributed over a large height interval, it is necessary to integrate the concentration in different height intervals separately and to multiply by the wind speed at the appropriate height before summing the total flux. The accuracy of the wind measurements is very important for the flux determination. The lidar system is equipped with awind monitor mounted 5 m above ground, but additional monitors at other heights are also used. In the measurements presented here, normally two vane anemometers at two different heights were employed. If necessary, the wind speed at other heights was calculated with an empirical power law function. If the position of the source of the pollutant is known, the wind direction is revealed by the position of the measured plume. Sometimes,visible plumes in or near the measurement area can be used to determine the wind field. This is done by following and timing structures in the plume moving between two lines of sight from the lidar system.
Examples of Resuhs Here the results of some of the measurements of SOz and Hg emission will be presented. Due to low NO2 emissions from the plants investigated, this species was only monitored at one location. Normally, the most abundant NO, component in industrial emissions is NO, which was not a measurement parameter in these campaigns. However, NO can also be monitored with the DIAL technique, but with a shorter effective range (8, 9). SO2 was monitored at six different kinds of industries. At several of these plants, the flux values deduced from the lidar measurements could be compared with in situ measurements. Especially interesting studies of the precision of the lidar flux values could be made in situations where only one or a few well-controlled sources were monitored. Direct comparisons could then be made with the flux determined from in situ SO2 monitors (normally
>m&!&l? BW-700
-
5W 600
150
ww-SW
300-4w 200 300 100 200
-
U
0-100
100
-z v
E0, ..
50
. .
I"
..
..P
i
I
.
L
.
0 500
460
6 1
Distance (m) FIGURE 2. Oimibnfion of
SO2
in a vol(icaI plane naar a sulfuric acid plam obtained from a lidar scan
TABLE 3
Comparison of Results of SO2 Flux Determinations from Remote Lidar and in Situ Measurement Techniques lidar' Kemira. April 24 197 184 160 166 156 175 1 2 0
11-12 13-14 14-15 15-16 16-17 mean
20-21 14-15 15-16 mean excl. diffuseb 15-16 11-12 13-14 mean
137 139 157 147 +15 127
Scanraff, May 21-22 (Cracker Emission) 37 33 38 36 +5
in situ' 142 154 156 161 162 154
128 111 119 125 30 40 40 37
'Flux in kglh. 'Excluding diffuse emissions, measured separately to 20 kglh.
UV absorption instruments, calibrated once a week with a standard gas mixture) and measurements of the gas flow inside the smokestack Table 3 lists the results of such comparisons at three plants. These measurements were performed with the lidar system parked 500-1000 m from the main source. The numbers shown are hourly mean values together with the mean value over the whole measurement period. The estimated uncertainty of the mean lidar value is also indicated. Most of this uncertainty isdue totheaccuracyinthewinddeterminationsandwind fluctuations during the measurements. As can be seen, there is some scatter in the hourlyvalues,whereas the mean values show a good correlation. It should be noted that
the lidar flux values at Nymolla include some diffuse emissions, which werenot detected by the insitumonitors. This diffuse emission, due to leaks and ventilation, was emitted from some of the plant buildings at a lower altitude, and it could be determined separately on some occasions to be about 20 kg/h. Thus, this flux should be subtracted from the lidar values for the total emission for a correct comparison of the two results. In conclusion, the lidar technique can give quite a reliable value of the flux, at least if the measurements are averaged over some hours. The averaging tends to decrease the influence of fluctuations in the wind field and/or rapid changes in the emission values. Asigniiicant advantageofthelidartechniqueisitsability to measure the fluxfrom several sources, extended sources, and diffuse emissions. An example of a measurement of diffuseS02emissioosisshowninFigure2.The figureshows the results of avertical scan from the ground upwards near a sulfuric acid plant at Kemira. The SO2 concentration in the vertical plane scanned by the lidar is displayed. In the upperpart of the figure, the concentrated plume from a tall smokestack can be seen, but at lower heights some diffuse emission is also detected. Here the gray scale is chosen to show the distribution of the diffuse emission and is not appropriate for the higher concentrations in the main plume. The contribution of this diffuse emission normally not accounted for by the plant was on average about 10% of the total flux. The measurements at this plant could be comparedwithapreviouslidarcampaign 11yearsago (10). The comparison revealed that the total SO2 flux today is only about 40% of the previous emission and also that the diffuse emission part had decreased from a former value of 30% of the total flux. Figure 3 shows an example of SO2lidar measurements at the Boliden metallurgical plant, which contains several different sources. The result of a vertical scan downwind from all the sources is displayed in panel a of the figure. As can be seen, at least three plumes contribute to the total flux hut it is difficult to separate them at this position. Panel b shows another type of representation that can be made from such a measurement. Here the concentrations VOL. 29. NO. 2.1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY
m
333
> 700pglm3 600 - 700 500 - 600 400 - 500 300 - 400
200 300 100-200 ~
1200.
0-100
600
a
800
1000 Distance (m)
are vertically integrated, and the resulting vertical column content is copied onto a map of the industrial area. The direction of the scan is indicated by the horizontal axis. Togetherwith the wind direction, this type offigure can be 334 m ENVIRONMENTAL SCIENCE b TECHNOLOGY I VOL. 29, NO. 2,1995
1200
useful in tracking down the sources, especially if measurementsaremadeinvariousdirectionsduringdifferentwind conditions. A comparison was made of the total SO2 flux deduced from the lidar measurements and from in situ
I
I
13
14
15
16
17
Mean
nme (hours)
I Lldar measurements In slhr measursments:mcontfnuous,
FIGURE 4. Comparison of hourly SOt emission values determined from lidarrneasurementsandinsilumearuremento at a metallurgical plant.
measurements made by the company. The latter value is calculated by summing up emissionvaluesfromsidifferent locations, which are continuouslymonitored. Furthermore, manual measurements are made irregularly at some additional points, but the results from these are only reported as monthly mean values. Figure 4 gives hourly mean values recorded during one afternoon in October, where the result from the manual measurements has been recalculated from the October mean value. The lidar gives on average nearly 50% higher values for the total SO2 emission. This discrepancy may be partly explained by the fact that the manually measured sources do not have acontinuousemissionbut areofamoreintermittentnature. The difference is probably also due to diffuse emissions, which are not registered in the measurements made by the company. Mercuryistheonlyatmosphericpollutantthat is present predominantly in atomic form, and this leads to a large increase in detection sensitivity over molecular species when using optical spectroscopic techniques (11). Within this project, mercury was monitored at sixdifferent plants. One of the major Hg emission sources is the chloralkali industry, where a flowing cathode of metallic mercury is often used in an electrolyticcell. Figure 5 shows the results ofalidarmercury measurement at suchaplant, EkaNobel. The largest emission source here is the main electrolytic cell building from which mercury is ventilated through openings along the roof. The spreading Hg plume from this building can be seen in the figure, although part of the plume may be undetected since the lower angles in the vertical scan were limited by a nearby building. Closer to the lidar system, a second smaller plume can also be seen, which was not expected. The source of this was found to be a smaller cell house used intermittently where some construction work was carried out during the measurements. This activity obviously released a large amount of mercury and contributed substantially to the total Hg flux. Estimates of the Hg flux to the atmosphere were made by the company only at the main cell house. This estimate involved measuring the concentration inside the building and determining the aimow from different subsections. A
comparison with these data shows that the lidar measurements gave a total flux of 3 1 g/h, which was about twice as high as the company's own estimate. A similar measurement at Norsk Hydroplast gave even larger differences; 8.1 and 1.9 g/h, respectively, although no 'extra' plume was detected here. This substantialdifferenceissurprisingsince it is not likely that the lidar overestimates the emission. At Eka Nobel, part of the plume was possibly undetected as mentioned. Furthermore, the lidar technique only detects atomic mercury and not mercury compounds and mercury bound to particles. In emissions from chloralkali plants, the atomic form has been estimated to be 50-90% of the total mercury emission (12). Earlier lidar measurements of mercury concentration levels near a chloralkali plant and other sources have shown avery good correlation with point monitors based on an amalgamation technique (1315).
One situation where it is difficult to determine the flux by a combination of vertical lidar scans and wind data is when there are large and rapid fluctuations in the emission ona timescalemuchshorterthanthescan time. Examples of this type are the mercury emission from crematories and steel melting plants using scrap metal. Here the mercury normally escapes during a few minutes when the temperature of the furnace has reached a certain value. Anothermeasurement strategy is then to onlyuseonelidar measurement direction just above the smokestack with a hightimeresolution. Ofcourse, thegasflowmust beknown inorderto determinethemercuryflux. Figure6showsthe results of such a measurement at the Ovako Steel plant. The mercury content in the flue gases was continuously monitored during 10 h with a time resolution of 0.5 min. As can be seen, the concentration varies substantially. The mean mercury flux during two days was determined to 0.6 g/h. However, due to the dependence upon the type and the source of the scrap metal used, it is difficult to dra7.v far-reaching conclusions from such a short time period. No comparative measurements were made here.
Discussion The data presented here show that the lidar technique can be averyuseful method ofmonitoringindustrial emissions. During the campaigns, the mobile lidar system has been operated very efficiently, often with measurements at different industries during subsequent days without long setup times. AU the measurements were performedwithin the framework of a control program commissioned by the Swedish Environmental Protection Agency during announced visits to different plants. However, since lidar is a remote sensing technique capable of detecting plumes at distances of up to a few kilometers, it can also be employed successfully in unannounced inspections. Although not demonstrated here, a mobile lidar system can also be used to study the transport and dispersion of pollutants further away from the source, and thus provide information for atmospheric dispersion models. Another measurement strategy can then be to only measure vertically, while the system is transported under the spreadingplume. Measurements ofthis kind were recently performed with the system in studies of the SO2 flux from some Italian volcanoes (161. The main advantage of the lidar technique is the possibility of rapidly measuring the total emission from severalsources, includingdiffuse emission, at an industrial site without any preknowledge of the exact position of the VOL. 29. NO. 2.1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY. 335
FIGURE 5. Distribution of mercury in a vertical plane scanned by the DIAL system at a chloralkali plam The horizoml direction for the vertical scan is indicated by the distance axis.
9
11
13 15 Time (hour)
17
19
FIGURE 6. Time-resolved lidar measurements of the mercury concentration in the flue gases from a Steel industry furnace.
different sources. This ability is quite unique and often makes lidar measurements more cost-effective than the numerous measurements needed with point monitors or long-pathabsorption techniques to achieve the same result. The accuracy of the fluxdeterminations has been evaluated by comparisons with known emissions from localized sources. These comparisons show that lidarcan determine the mean flux over some hours with an accuracy of +lo% during favorable conditions. The accuracy can in some cases be worse than this since it depends on range, wind conditions, distribution of the plume, and topography of the measurement area. Thus, it is difficultto give a general 336. ENVIRONMENTAL SCIENCE &TECHNOLOGY i VOL. 29. NO. 2.1995
accuracy for emission values determined by alidarsystem, but the sensitivity and accuracy can be estimated for each individual measurement. The limiting factor is often the accuracy in the wind determinations. An improvement of the present system would be to use some kind of wind profiler on the lidar site to continuously monitor the wind field. The sensitivity can be checked by performingseveral scans upwind of the industrial area if there is a negligible inflow ofthe pollutant. Typically,thesensitivitywasabout 1 kg/h for SOz flux and about 0.1 g/h for Hg flux during these measurements. A drawback with the lidar technique in emission monitoring is the dependence on the weather situation. As with all optical remote sensing techniques, lidar measurements may be hampered by low atmospheric visibility due to fog or heavy rain and snowfall. The wind conditions are, as mentioned, very important in the flux determinations. Typically,we have found that the mean wind speed should be higher than 3 m/s during a lidar scan for acceptable results. The wind direction is also a limiting factor, and wind changes can necessitate changes in the position of the lidar system. The presence and access to different sites with a clear field of view around an industrial plant are of course also important in this context. All of these external factors can limit the available monitoring time but seldom completely prevent measurements. However, the time dottedtoalidarcampaignshouldbelongenoughtoendure bad weather conditions.
The applicability of the lidar technique is also dependent on the number of species that can be measured. During these campaigns SO2, NOz, and Hg were monitored, but the present mobile DIAL system is also capable of detecting other species such as NO, 0 3 , C12,CH20,and some aromatic hydrocarbons such as benzene, toluene, and xylene in the UV and visible spectral regions. An improvement in the monitoring capabilities can be made if the usable wavelength region is extended into the IR region. This enables measurements of other industrially important gases; for example, HC1, CO, and several hydrocarbons. Especially interesting is the possibility of monitoring total fluxes of volatile hydrocarbons that are emitted fugitively by oil refineries and where today other monitoring methods are inadequate (17). Some drawbacks of DIAL in the IR region are the low backscattering, which makes range-resolved measurements more difficult, and more spectral interferences due to atmospheric C 0 2and H20. Monitoring alarge number of species places a demand on rapid switching between different wavelengths. In this context, the recent development of tunable all-solid-state laser systems is very interesting for the next generation of DIAL systems.
Acknowledgments This work was supported by the Swedish Environmental Protection Agency (SW and the Swedish Space Board. Helpful assistance duringthe field campaignswas provided by Rolf Svedberg, Borje Borgstrom, Anders Melin, and Goran Blomstrom at S N V . Support and encouragementfrom Sune Svanberg at our department are also gratefully acknowledged.
literature Cited (1) Killinger, D. K., Mooradian, A., Eds. Optical and Laser Remote Sensing, Springer-Verlag: Berlin, 1983. (2) Proceedings, Optical Remote SensingApplications to Environmental and Industrial Safety Problems; Houston, TX, April 6-8,
1992; Air & Waste Management Association: Pittsburgh, 1992. (3) Sigrist, M., Ed. Air Monitoring by Spectroscopic Techniques; Chemical Analysis Series 127; Wiley: New York, 1994. (4) Viswanathan, S.; Sands, M. D. Environ. Sci. Technol. 1992, 26, 650-652. (5) Edner, H.; Ragnarson, P.; Sp&mare, S.; Svanberg, S. Appl. Opt. 1993, 32, 327-333. (6) Measures, R. M. Laser RemoteSensing; Wiley-Interscience: New York, 1984. (7) Edner, H.; Fredriksson, K.; Sunesson, A.; Svanberg, S.; Uneus, L.; Wendt, W. Appl. Opt. 1987,26, 4330-4338. (8) Edner, H.; Sunesson, A.; Svanberg, S. Opt. Lett. 1988, 13, 704706. (9) Kolsch, H. J.; Rairoux, P.; Wolf, J. P.; Woste, L.App1. Phys. B 1992, 54, 89-94. (10) Egeback, A. L.; Fredriksson, K. A.; Hertz, H. M. Appl. Opt. 1984, 23, 722-729. (11) Edner, H.; Faris, G. W.; Sunesson, A.; Svanberg, S . Appl. Opt. 1989,28, 921-930. (12) Lindqvist, O., Ed. Water, Air, Soil Pollut. 1991, 55, 1-261. (13) Ferrara, R.; Maserti, B. E.; Edner, H.; Ragnarson, P.; Svanberg, S.; Wallinder, E. A m o s . Environ. 1992, 26A, 1253-1258. (14) Edner, H.; Ragnarson, P.; Svanberg, S.; Wallinder, E.; De Liso, A.; Ferrara, R.; Maserti, B. E. I. Geophys. Res. 1992, 97, 37793786. (15) Edner, H.; Ragnarson, P.; Svanberg, S.; Wallinder, E.; Ferrara, R.; Maserti, B. E.; Bargagli, R. Sci. Total Environ. 1993, 133, 1-15. (16) Edner, H.; Ragnarson, P.; Svanberg, S.; Wallinder, E.; Ferrara, R.; Cioni, R.; Raco, B.; Raddeucci, G. Total fluxes of sulfur dioxide
from the Italian volcanoes Etna, Stromboli and Vulcano measured by differential absorption lidar and passive differential optical absorption spectroscopy. J. Geophys. Res. 1994, 99, 18827- 18838. (17) Woods, P. T.;Partridge, R. H.; Milton, M. J.; Jolliffe,B. W.; Swann,
N. R. Proceedings, Optical Remote Sensing-Applications to Environmental and Industrial Safety Problems; Houston, TX, April 6-8, 1992; Air & Waste Management Association: Pittsburgh, 1992; pp 3-29.
Received for review March 17, 1994. Revised manuscript received September 12, 1994. Accepted November 1 1 , 1994.@ ES940173M ~
@
Abstract published inAdvanceACSAbstracts, December 15,1994.
VOL. 29, NO. 2, 1995 /ENVIRONMENTAL SCIENCE &TECHNOLOGY
337
