SIMULTANEOUS MEASUREMENTS OF ... - Springer Link

Environmental Monitoring and Assessment (2005) 104: 281–293 DOI: 10.1007/s10661-005-1616-6

c Springer 2005

SIMULTANEOUS MEASUREMENTS OF ATMOSPHERIC POLLUTANTS AND VISIBILITY WITH A LONG-PATH DOAS SYSTEM IN URBAN AREAS JEONG SOON LEE2 , YOUNG J. KIM1∗ , BONGJAE KUK3 , ANDREAS GEYER4,5 and ULRICH PLATT5 1

Air Quality Laboratory, Advanced Environmental Monitoring Research Center, Kwangju Institute of Science and Technology (K-JIST), 1 Oryongdong Buk-gu, Gwangju, Korea; 2 Satellite Technology Research Center, Korea Advanced Institute of Science and Technology (KAIST), Gusungdong, Yusunggu, Daejeon, Korea; 3 Space Center, Korea Aerospace Research Institute, Daejeon, Korea; 4 Department of Atmospheric Sciences, University of California at Los Angeles, 7129 Math Sciences Bldg., Los Angeles, California, U.S.A.; 5 Institut f¨ur Umweltphysik, Ruprecht-Karls University Heidelberg, Im Neuenheimer Feld 229, Heidelberg, Germany (∗ author for correspondence, e-mail: [email protected])

(Received 23 October 2003; accepted 4 May 2004)

Abstract. In this paper, the applicability of a Long-Path Differential Optical Absorption Spectroscopy (LP-DOAS) system was checked for the feasibility of the simultaneous measurement of trace gases (such as O3 , NO2 , SO2 , and HCHO) and atmospheric visibility (light extinction by aerosols) in Asian urban areas. Field studies show that an LP-DOAS system can simultaneously measure the key pollutants (such as O3 , NO2 , SO2 , and HCHO) at detection limits in the ppb/sub-ppb range as well as the Mie extinction coefficient with an uncertainty of ∼0.1 km−1 at time resolution of a few minutes. It is thus concluded that the use of LP-DOAS system is feasible for simultaneous measurement of gaseous pollutants as well as an atmospheric extinction coefficient which is tightly bound to fine particulate concentration. Keywords: DOAS, extinction coefficient, optical monitoring, spectroscopy, visibility

1. Introduction Elevation of both gaseous (O3 , NO2 , SO2 , and VOCs) and particulate pollutants has helped to gradually degrade air quality in many Asian cities. The cause of haze formation in Asian cities has often been ascribed to high concentrations of both primary and secondary particles under highly humid conditions (Kim et al., 2001; Park et al., 2002). In order to develop strategies to improve air quality in those areas, continuous monitoring of air pollutants (including particulate matter) with a wide spatial coverage can be considered as prerequisite. From the economic point of view, an instrument that possesses the capabilities of several individual conventional point monitors for major gaseous and particulate pollutants would be highly desirable.

282

JEONG SOON LEE ET AL.

The Differential Optical Absorption Spectroscopy (DOAS) is a well established technique that has been used extensively to measure a broad spectrum of atmospheric trace gases at reasonably low detection limits such as: O3 , NO2 , SO2 , HCHO, HONO, NO3 , aromatics, and halogen oxides (Platt et al., 1979; Platt and Perner, 1980; Plane and Nien, 1992; Volkamer et al., 1998; Geyer et al., 1999). It was also approved by Environmental Protection Agency (EPA, 1995) as an equivalent method for measurement of gaseous pollutants (SO2 , O3 , and NO2 ). Several research groups were hence able to conduct their measurements of tropospheric pollution based on commercial DOAS systems (Mathew et al., 2001; Virkkula, 1997; Kim and Kim, 2001; Kim, 2004). Special numerical filters (such as Sawitzky-Golay high pass filters) were often used to separate spectral features associated with absorption of atmospheric trace gases from those due to scattering, the optical instrument functions, and turbulence (Platt, 1994). The concentration of the absorbing trace gases can then be derived by comparing the resulting high passed absorption spectra to the reference spectra for the species of interest, which were also treated with the same filter (Stutz and Platt, 1996). In a conventional DOAS analysis, the spectral information related to aerosol scattering is often neglected; they cannot easily be separated from those associated with the instability of the lamp and the optical setup (Horbath and Noll, 1969; Notholt and Raes, 1990). Recently however, it was tested whether a specially designed LP-DOAS system can also be used to characterize aerosol properties in the measurements of the Mie extinction coefficient (Flentje et al., 1997; Mueller, 2001). For this purpose, changes in the transmitted intensity of the system (due to the instability of the lamp and the optical setup) can be corrected periodically by comparing the signals along two light paths of different length. However, the setup and maintenance of this LP-DOAS system is rather difficult, and it may not be well suited for long-term automatic monitoring tasks. In this paper, we discuss the applicability of a standard LP-DOAS system for the simultaneous measurements of several trace gases (NO2 , O3 , SO2 , and HCHO) and the atmospheric visibility in urban areas of Asia, which are generally characterized by high pollution and particle levels. In particular, we attempted to answer the question whether an LP-DOAS system can be an alternative method to conventional point monitors deployed in a modern air quality monitoring network.

2. Experimental Advantages of the DOAS technique include such as: (1) the ability of measuring several gases with one system, (2) high time resolution, and (3) low detection limits with its inherent calibration (Platt, 1994). The fundamental idea of DOAS is to separate the differential optical absorption of trace gases from the integrated extinction along an open light path in the atmosphere. Atmospheric extinction can

SIMULTANEOUS MEASUREMENTS OF ATMOSPHERIC POLLUTANTS AND VISIBILITY

283

be expressed by applying the Lambert–Beer’s law as follows: σi (λ) · ci + σRay (λ) + σMie (λ) · T (λ, L) I (λ, L) = I0 (λ) · exp −L · (1) where ci denotes the concentration of the trace gas i, σ i (λ) is the characteristic absorption cross section at an wavelength λ, and L is the length of the light path. I (λ, L) is the intensity measured after the light has passed through the atmosphere. The term Io (λ) expresses the incident intensity: this can be measured by assuming that there is no air in the measurement path. Because this is not realistic, Io (λ) is often determined by the intensity of the lamp, I L (λ) in modern Long-Path DOAS systems (LP-DOAS); this value can in fact be measured by an optics shortly before the light is transmitted into the atmosphere. This intensity is multiplied by a correction factor A(λ, L), which expresses the difference in the intensity changes along the two optical setups including instrument function and geometrical light loss. I L (λ) = I0 (λ) · A(λ)

(2)

In addition to extinction due to absorption by various atmospheric trace gases, light is also attenuated by air molecules (i.e., the Rayleigh scattering coefficient, σ Ray (λ)) and by atmospheric aerosols (i.e., the Mie scattering coefficient σ Mie (λ)). The spectrum of the light can also be affected by instrument functions (such as mirror reflectivity and/or detector sensitivity) and by atmospheric turbulence (Stutz and Platt, 1996). Atmospheric turbulence can be represented by the factor T (λ, L) in Equation (1). 2.1. D ESCRIPTION

OF THE KJIST LONG - PATH DOAS SYSTEM

This study was performed using an LP-DOAS system developed by the Advanced Monitoring Research Center (ADEMRC), Kwangju Institute of Science and Technology (KJIST), Kwangju, Korea. Details of this system were discussed previously by Lee et al. (2002). Figure 1 shows the optical setup of the DOAS system used during our experiments. The system specification is also summarized in Table I. A Cassegrain type coaxial telescope (an F-number of 4 and an effective focal length of 80 cm) is used as both transmitter and receiver of the LP-DOAS system, respectively. The main part of the telescope is a 200-mm parabolic mirror coated to enhance UV reflectance. A super-quiet 150 W Xenon arc lamp (arc size of 0.4 mm × 2 mm) is used as a light source. This lamp has the advantage over conventional lamps used in other DOAS systems in that the light arc is extremely stable and does change slowly with time. The light beam from the lamp (Figure 1a) is collimated by the stabilized transmitter telescope with a beam divergence of 1.2 mrad and sent out to a retro-reflector array (Figure 1c) which is placed at a distance of 750 m. The retro-reflector array consists of seven round corner cube prisms with 60 mm

284


TABLE I System configuration of the K-JIST LP- DOAS system Telescope

Lamp

[0cm]Fiber Spectrometer

Detector

Type F-number Focal length Divergence Type Arc size Power Type Size Type Focal length Slit size Grating Type Cooling method Measuring range Spectral resolution Size Full well depth

Schmidt–Cassegrain ∼4 80 cm 1.2 mrad Xenon arc Hamamatsu L2274 0.4 mm × 2 mm 150 W Multi-mode step index quartz fiber, NA = 0.16 200 µm Czerny–Turner 320 mm 100 µm 1200 grooves/mm 1024 pixel photodiode (Hamamatsu S3904) 2 Peltier stacks 200–1000 nm 0.27 nm 2.5 mm × 25.4 mm 1.2 × 108 electron

Figure 1. A schematic diagram of the KJIST LP-DOAS system: (1) Xe arc lamp and shutter, (2) main parabolic mirror, (3) retro-reflector array, (4) auxiliary reflecting mirror, (5) lamp shutter, (6) lamp mirror with a center lens, (7) reflecting fiber mirror and filter wheel, (8) fiber optic guide, (9) spectrometer with PDA detector, (10) computer, and (11) controller. Beam paths (a), (b), and (c) represent lamp intensity measurement mode, light receiving mode, and outgoing light transmitting mode, respectively.


285

diameter; this sends the beam back onto the main mirror of the DOAS telescope with a beam deviation less than 2 inches. The diameter of the retro-reflector array (20 cm) is considerably smaller than the image size of the lamp arc of 0.5 m × 2.5 m. The light reflected back into the telescope is focused onto a multi-mode quartz fiber whose other end serves as an entrance slit of a F-4 Czerny-Turner type spectrometer (Figure 1b). For the trace gas measurements described here, a holographic grating (1200 grooves mm−1 ) blazed at 300 nm was operated between 290 and 350 nm with a spectral resolution of ∼0.3 nm (FWHM) at 343 nm to measure the concentrations of NO2 , SO2 , O3 , and HCHO. For the measurement of atmospheric visibility, a conventionally ruled grating (1200 grooves mm−1 ) blazed at 500 nm was used to obtain a visible spectrum from 500 to 560 nm range with a spectral resolution of ∼0.25 nm (FWHM). The spectra are recorded by a 1024 pixel photodiode array cooled at −15 ◦ C to reduce the thermal noise. In order to measure the incident lamp light intensity, a 1-inch diameter quartz lens is installed in the center hole of the lamp mirror (Figure 1 (6)). The part of light that is emitted from the lamp is focused on the fiber via this lens as shown in the Figure 1a. A second shutter in front of the light source (Figure 1 (1)) is used to measure the diffusive background light, which is scattered into the telescope. The spectrometer was calibrated periodically using Hg and Ne emission line spectra to map the channels of the photodiode array to the corresponding wavelengths. Reference spectra of the absorbing gases to be measured were then obtained from literature-based absorption cross sections by convoluting with the mercury line spectra and interpolation according to the dispersion of the pixels of the photodiode array. In the case of NO2 and SO2 , absorption spectra measured through specially prepared gas cells were used as reference spectra during our measurement (Lee et al., 2002). Two spectral regions used for the evaluation in this experiment were set to cover; (1) 293–312 nm for SO2 , O3 and HCHO, and (2) 325–342 nm for NO2 . In order to determine the concentrations of absorbing trace gases based on the measured air spectra, a series of data processing steps were involved. First, structures arising from sources than our lamp (such as an electronic offset, dark current, and scattered solar light) were subtracted from the measured atmospheric spectra. The logarithm of the resulting spectrum was then numerically band-pass filtered by initial subtraction of a smoothed copy (500-times repeated triangular smoothing) of the spectrum and then low-pass filtering (5-times triangular smoothing). Through these filtering procedures, slow changing spectral features due to Rayleigh and Mie scattering were removed from the spectrum. The concentrations of NO2 , O3 , HCHO, and SO2 were then determined by non-linear least squares fitting of the reference spectra, which were treated by the same filters, to a DOAS-filtered atmospheric spectrum (Platt, 1994; Stutz and Platt, 1996; Lee et al., 2002). The detection limit, which is defined as the concentration corresponding to the minimum detectable optical density, was calculated by fitting the errors calculated by a procedure suggested by

286


Stutz and Platt (1996). The detection limits of NO2 , SO2 , O3 and HCHO determined on a normal day were estimated as 0.61, 0.06, 5.4, and 2.2 ppb, respectively. In the case of an abnormal day (e.g., usually low visibility), the detection limits were easily raised to several times due to debasement of signal-to-noise ratio by insufficient light intensity. 2.2. DOAS MEASUREMENT AT 550 nm

OF ATMOSPHERIC EXTINCTION AND VISIBILITY

Besides the acquisition of the concentrations data for the gaseous pollutants from differential absorptions in the UV region, the visible spectrum of a DOAS system can be used to monitor the atmospheric visibility and Mie extinction coefficient σMie (λ). Prior to the determination of σMie (λ) at 550 nm in Equation(1), several factors (light attenuation by Rayleigh scattering, gas-phase absorptions, lamp intensity fluctuations, and atmospheric turbulence) have to be considered. The extinction of light by molecular scattering can be expressed as σRay (λ) = 4.4 × 10−16 /λ4 . It is therefore straightforward to remove the Rayleigh scattering signature from a spectrum. With σRay (λ) ∼ = 0.01 km−1 at 550 nm, Rayleigh scattering is almost negligible, compared to the extinction by aerosols. O3 and NO2 are the main gaseous absorbers at 550 nm. Absorption signals from both gases were subtracted from our measured spectra based on the concentration data at the wavelength region around 340 nm. It should be noted that the contribution of gas-phase absorption to the extinction at 550 nm is very low (∼0.01 km−1 ). A possible variation of the lamp intensity was corrected by dividing atmospheric spectra (in Equation (1)) by lamp reference spectra (Equation (2)) measured subsequently (Figure 3). Assuming A(λ,L) being constant, I L (λ) can be considered as an indicator of the Xe lamp stability. Misalignment of the optical fiber can also be detected by this procedure. In our experiment, the intensity of the lamp reference spectrum fluctuated normally within less than 4 (±1 σ variability in Figure 2a) for the entire experimental period. After removing % aforementioned factors, Dext (λ, L) due to atmospheric extinction can be expressed as, Dext (λ, L) = ln{A(λ, L)−1 · exp(−L · σMie (λ)) · T (λ, L)}

(3)

Random thermal fluctuations and atmospheric wind can lead to refractive index variations which can perturb the propagation of the DOAS light beam as expressed by the factor T (λ, L) in Equation (3). Atmospheric turbulence can cause light loss by optical scintillation at the fiber, angle-of-arrival fluctuations, and beam spreading. The various light losses can be estimated using (for example) the equations given by Biswas and Lee (2000). With the setup of our LP-DOAS system, beam spreading can induce a loss of light (∼8%); this is assumed by a strong daytime turbulence structure parameter, Cn2 = 10−13 m−2/3 . Atmospheric turbulence should not therefore significantly affect the results of σMie (550 nm), and T (λ, L) can hence


287

Figure 2. Time series of hourly averages of (a) the intensities of lamp spectra, (b) atmospheric spectra measured, and (c) Mie extinction coefficient σMie (550 nm) at Kwangju. Vertical bars represent the range of intensity variation for (a) and (b) and the error range of the Mie extinction coefficient for (c).

be set to 1 in the following considerations. Note that T (λ, L) was applied in the Equation (6) to estimate the uncertainty of extinction coefficient. The Mie extinction coefficient, σMie (550 nm) can thus finally be expressed as: σMie (550 nm) = −

Dext (550 nm, L) − ln(A(550 nm, L)−1 ) L

(4)

The visual range V at 550 nm can then be calculated from the Mie extinction coefficient, according to the well-known Koschmieder relationship: V =

−ln(0.02) σMie

(5)

The correction factor, A(λ, L) was empirically determined in this study by comparing the DOAS results immediately after a precipitation event to visibility data measured by the Korea Meteorological Administration (KMA) by a conventional contrast method. The uncertainty of σMie (550 nm) was estimated from: (1) the relative fluctuations of the lamp intensity ILamp , (2) the relative intensity fluctuations caused by the instability of the optical adjustment I S , and (3) the variability of the atmospheric turbulence T such as: σext

1 = (ILamp )2 + (I S )2 + (T )2 . L

(6)

288


With the typical values of ILamp = 4%, T = 8%, and the IS = 2%, an uncertainty of the extinction coefficient of ∼0.1 km−1 can be obtained. 3. Results and Discussion 3.1. MEASUREMENTS

OF

A IR POLLUTANTS

BY DOAS

A field campaign was conducted on 15–19 August 2001 in Heidelberg, Germany, to intercompare two DOAS systems between the KJIST long-path DOAS system and a DOAS instrument (ChinaSky). The latter system was recently designed and developed by the Institut fuer Umweltphysik (IUP) at Heidelberg for long-term monitoring purposes in China (Yu et al., 2003). Heidelberg is a small city with a population of roughly 130 000 with no heavy industry. The two systems were installed at the top of the IUP building, which is located in the northwestern part of Heidelberg city. Both instruments were aligned to the same set of retro-reflectors attached to a tower to the east. The length between the telescopes and the reflectors was 1400 m at an altitude of 25 m above the ground level. During the 4-day intercomparison period, mixing ratios of SO2 , HCHO, NO2 , and O3 of 0.86 (±1.52), 6.9 (±2.1), 9.2 (±6.9), and 57 (±17) ppbv were measured by the KJIST LP-DOAS system, respectively. (The data in brackets refer to the standard deviation of the data.) The diurnal variations of NO2 and O3 , which tend to exhibit an inverse correlation between the two species, are evident in Figure 3. The time resolution of the data was about 3 min on an average. Figure 3 shows that

Figure 3. Time series of the 3 min average mixing ratios of SO2 , HCHO, NO2 , and O3 in Heidelberg, Germany (August 15–19, 2001). Vertical bars represent the variation range of the measured concentrations.


289

Figure 4. Intercomparison of the mixing ratios of (a) SO2 , (b) O3 , and (c) NO2 ; between the two systems:KJIST DOAS (K-DOAS) and ChinaSky DOAS (C-DOAS).

the (KJIST) LP-DOAS system is well suitable to monitor the concentration changes of the key pollutants (e.g., SO2 , HCHO, NO2 , and O3 ) with good time resolution even under relatively clean conditions. The results of a regression analysis of the data between the KJIST (K) and ChinaSky (C) DOAS systems are shown in Figure 4. Note that the same absorption crosssections were used in both analyses to minimize the bias of this comparison. Relationships between the two systems can be expressed as follows; NO2 (C-DOAS) = 0.93 × NO2 (K-DOAS) − 0.3, SO2 (C-DOAS) = 1.05 × SO2 (K-DOAS) − 0.0, and O3 (C-DOAS) = 1.12 × O3 (K-DOAS) – 23.0. Squared correlation coefficients (r 2 ) higher than 0.95 were obtained for all three compounds. The intercepts and slopes of the NO2 and SO2 correlations are in the range of the uncertainties of both instruments (Lee et al., 2002; Yu et al., 2003) and are probably caused by different systematic spectral structures. A large intercept was observed for the ozone measurements, which is suspected to arise from the limited spectrometer stray-light correction of the ChinaSky system; this problem is currently under investigation (Yu et al., 2003). As for the absolute quantity of gaseous species, several groups have reported the absolute reliability of the DOAS measurement technique and provided some numerical values for absolute differences between DOAS and conventional methods (Evangelisti et al., 1995; Virkkula, 1997; Mathew et al., 2001; Kim and Kim., 2001). By EPA (1995), DOAS method was approved as an equivalent one for the measurement of ambient concentrations of NO2 , SO2 , and O3 . 3.2. R ELATIVE

EXTINCTION AND VISIBILITY MEASUREMENT

The applicability of the KJIST LP-DOAS system to monitor atmospheric Mie extinction was tested during a field intensive held during 13–20 April 2002 in the campus of the KJIST, Kwangju, Korea (35.10 N, 126.53 E). Kwangju located in

290


Figure 5. Time series of various parameters: (a) visibility measured by the KJIST LP-DOAS system, (b) visibility measured by a conventional contrast method, (c) relative humidity, and (d) PM10 mass concentration.

southwestern part of Korea is the fifth largest city in Korea with a population of 1.4 million people. The Mie extinction coefficient was measured over a path of 1500 m. The time interval between each complete measurement cycle was about 10–12 min. Figure 2c shows the time series of hourly averaged extinction coefficient at 550 nm, σMie (550 nm), during this measurement period, calculated according to Equation (4). Visibility determined according to Equation (5) is plotted in Figure 5a. In the morning hours, the development of fog could be observed regularly by our DOAS system at this site. The mean extinction coefficient and visual range during the entire measurement period were 0.45 ± 0.30 km−1 and 8.2 ± 13.0 km, respectively. During a rainy day (15 April), the wash-out of aerosol particles led to a low extinction coefficient of 0.22 ± 0.13 km−1 . The visibility measured by DOAS on this day increased up to more than 30 km. In Figure 5, the visibility measured by DOAS (averaged over an hour) is plotted along with relative humidity (RH) and the PM10 particle mass concentration. In addition, visibility data of KMA measured by a conventional method (contrast measured by human eye with time interval of 3 h) are included for comparison in Figure 5b. Both visibility data set generally agree well except under high extinction conditions. This disagreement may be attributed to the known uncertainty of the human eye measurement method at low contrast conditions. An inverse-power relation between the optical visibility of the DOAS system and the aerosol mass concentration appears in Figure 5a and 5d. In the morning of 17 April, for example, PM10 increased up to 158 gµg m−3 due to a weak dust storm, and the Mie extinction coefficient increased to 0.58 ± 0.28 km. Shortly after the


291

Figure 6. Relationship between DOAS-based visibility and the PM10 mass concentration (RH < 75%).

storm, the visibility increased up to more than 30 km. Visibility data also show an anti-correlation with relative humidity Figure 5a and 5c. The observed fast growth of the atmospheric aerosols (at RH above 75%) suggests that the aerosols were hygroscopic during the measurement period and that their deliquescence point was around 75%. Figure 6 shows that visibility is exponentially correlated with the PM10 mass concentration as previously found in Kwangju. From the linear correlation of the Mie extinction coefficient and the PM10 mass concentration, the mass extinction coefficient (σM ) at RH > 75% was determined as σM = 5.7 m2 g−1 at 550 nm. This value agrees well with the coefficient of 5.4 m2 g−1 measured by transmissometer in Kwangju (Kim and Kim, 2003). 4. Conclusions A Long-Path Differential Optical Absorption Spectroscopy system was designed (KJIST LP-DOAS), and its performance was tested for both simultaneous measurements of O3 , NO2 , SO2 , and HCHO, (even under relatively clean conditions) and visibility (aerosol extinction under polluted conditions; for example in Asian urban areas). Real time measurements of the trace gases (with a time resolution 0.95) was found for O3 , NO2 , and SO2 . Atmospheric visibility was also examined with the KJIST LP-DOAS system by measuring the atmospheric Mie extinction coefficient at a wavelength of 550 ± 2.5 nm. An algorithm is presented here to derive the atmospheric extinction coefficient, σMie (550 nm) from a spectrum of the DOAS instrument. The atmospheric visibility determined by the LP-DOAS was compared to the PM10 aerosol

292


concentration measured by a beta gauge. During an 8-day measurement period in Kwangju, a mean aerosol extinction coefficient (0.45 ± 0.30 km−1 ) and visual range (8.2 ± 13.0 km) were observed. Our results show that an LP-DOAS system can measure atmospheric visibility in an Asian urban area with a relative uncertainty of ∼10% if a stable light source and optical setup is provided. It can thus be concluded that the LP-DOAS system can be recommanded as an alternative to conventional point monitors in an Asian air quality monitoring network. Acknowledgements This work was supported in part by the Korea Science and Engineering Foundation (KOSEF) through the Advanced Environmental Monitoring Research Center (ADEMRC) at the Kwangju Institute of Science and Technology (KJIST), by Korea Ministry of Environment through the G7 project and by Korea Ministry of Education & Human Resources Development through the Brain Korea 21 project. We appreciate Prof. K.-H. Kim for his valuable comments and advice throughout this work. References Biswas, A. and Lee, S.: 2000, Ground-to-Ground Optical Communications Demonstration, TMO progress report 42–141. Burrows, J. P., Dehn, A., Deters, B., Himmelmann, S., Richter, A., Voigt, S. and Orphal, J.: 1998, ‘Atmospheric remote-sensing reference data from Gome: Part 1. Temperature-dependent absorption cross section of NO2 in the 231–794 nm range’, J. Quant. Spctrosc. Transfer 60(6), 1025–1031. Evangelisti, F., Baroncelli, A., Bonasoni, P., Giovanelli, G. and Ravegnani, F.: 1995, ‘Differential optical absorption spectrometer for measurement of tropospheric pollutants’, Appl. Opt. 34, 2737– 2744. Flentje, H., Dubois, R., Heintzenberg, J. and Karbach, H.-J.: 1997, ‘Retrieval of aerosol properties from boundary layer extinction measurements with a DOAS system’, Geophys. Res. Lett. 24(16), 2019–2022. Geyer, A., Alicke, B., Mihelcic, D., Stutz, J. and Platt, U.: 1999, ‘Comparison of tropospheric NO3 radical measurements by differential optical absorption spectroscopy and matrix isolation spin resonance’, J. Geophys. Res. 104(D21), 26,097–26,105. Horbath, H. and Noll, K.E.: 1969, ‘The relationship between atmosphere light scattering and visibility’, Atmos. Environ., 3,543–3,552. Kim, K.-H.: 2004, ‘Comparison of BTX measurements using a commercial differential optical absorption spectroscopy and an on-line gas chromatograpy system’, Environ. Eng. Sci. 21(2), 181–194. Kim, K.-H. and Kim, M.-Y.: 2001, ‘Comparison of an open path differential optical absorption spectroscopy system and a conventional in situ monitoring system on the basis of long term measurements of SO2 , NO2 , and O3 ’, Atmos. Environ. 35, 4059–4072 Kim, Y.J., Kim, K.W. and Oh, S.J.: 2001, ‘Seasonal characteristics of haze observed by continuous visibility monitoring in the urban atmosphere of Kwangju, Korea’, Environ. Monit. Assess. 70, 35–46.


293

Kim, Y.J. and Kim, K.W.: 2003, Visibility impairment by the atmospheric fine particles in an urban area, International symposium ‘Blue sky in the 21st Century’, Korea. Lee, J.S., Kuk, B. and Kim, Y.J.: 2002, ‘Development of differential optical absorption spectroscopy (DOAS) system for detection of atmospheric species; NO2 , SO2 and O3 ’, J. Kor. Phys. Soc. 41(5), 693–698. Mathew, L., Wu, R.Tai. and Lo, J.-G.: 2001, ‘Measurements of sulfur dioxide and formaldehyde in Taipei using a differential optical absorption spectrometer’, J. Air & Waste Manage. Assoc. 51, 94–101. Mueller, T.: 2001, Spectral extinction measurements with LP-DOAS. 1. International DOAS workshop, Heidelberg, Germany, September 13–14. Notholt, J. and Raes, F.: 1990, ‘Test of in situ measurements of atmospheric aerosols and trace gases by long path transmission spectroscopy’, J. Aerosol Sci. 21(Suppl. 1), s193–s196. Park, S. S., Kim, Y. J. and Fung, K.: 2002, ‘PM2.5 carbon measurements in two urban areas: Seoul and Kwangju, Korea’, Atmos. Environ. 36(8), 1287–1298. Plane, J.M.C. and Nien, C.-F.: 1992, ‘Differential optical absorption spectrometer for measuring atmospheric trace gases’, Rev. Sci. Instrum. 63(3), 1867–1876. Platt, U.: 1994, Air Monitoring by Spectroscopic Techniques, Chemical Analysis Series, Vol. 127, Wiley, New York pp. 27–84. Platt, U., Perner, D. and P¨atz, H.W.: 1979, ‘Simultaneous measurement of atmospheric CH2 O, O3 and NO2 by differential optical absorption’, J. Geophys. Res. 84, 6329–6335. Platt, U. and Perner, D.: 1980, ‘Direct measurements of atmospheric CH2 O, HNO2 , O3 , NO2 , and SO2 by differential optical absorption in the near UV’, J. Geophys. Res. 85(C12), 7453–7458. Stutz, J. and Platt, U.: 1996, ‘Numerical analysis and estimation of the statistical error of differential optical absorption spectroscopy measurements with least-squares methods’, Appl. Opt. 30, 6041– 6053. Virkkula, A.: 1997, ‘Performance of a differential optical absorption spectrometer for surface O3 measurements in the Finish arctic’, Atmos. Environ. 31, 545–555. Volkamer, R., Etzkorn, T., Geyer, A. and Platt, U.: 1998, ‘Correction of the oxygen interference with UV spectroscopic (DOAS)’, Atmos. Environ. 32, 3731–3747. Yu, Y., Geyer, A., Yang, Z., Platt, U. and Galle, B.: 2003, The ChinaSky DOAS system: A new tool for air pollution monitoring in China. (in preparation).