Effect of UV scattering on SO₂ emission rate measurements

1. Introduction

[2] The correlation spectrometer (COSPEC) developed in the 1960s enabled monitoring of the SO₂ emission rate, which is now one of the most important indicators of volcanic activity [e.g., Stoiber et al., 1983; Casadevall et al., 1994; Hirabayashi et al., 1995; Young et al., 1998]. The COSPEC measures absorption of scattered ultraviolet (UV) light by SO₂ molecules and calculates the amount of SO₂ along the light path (column concentration) from the absorption intensity. If there is no gain or loss of the UV light after absorption by SO₂, the amount of SO₂ in the plume can be obtained precisely from the measured absorption intensity. However, the UV light for the COSPEC measurement comes from sunlight scattered in the sky. This scattering occurs throughout the atmosphere and UV scattering between the plume and the instrument can cause significant attenuation of the measured absorbance. Moffat and Millan [1971] and Millan [1980] evaluated the effect of the UV scattering on the COSPEC measurements and concluded the scattering can cause significant attenuation of the absorbance in particular at grater distance from the plume. Generally, the SO₂ emission rate is measured by two techniques, the panning method and the traverse method. In the panning method (also called as stationary or scanning method), an instrument is mounted on a tripod which is placed beneath or to the side of the plume, and scans the plume vertically or horizontally. In the traverse method, an instrument is mounted on a car, a ship, an airplane or a human, and scans the plume by traversing beneath the plume. The distance of the measuring instrument from the plume is commonly larger in the panning method than in the traverse method. Therefore, it has been recommended to adopt the traverse method rather than the panning method [Stoiber et al., 1983]. For example, Sutton et al. [2001] reported that the SO₂ emission rates measured with the traverse method were larger than those measured contemporaneously with the panning methods. Application of the panning method, however, is necessary in some cases because of limited road access, and therefore quantitative evaluation of the error associated with the panning method is necessary. McGee [1992] performed the airborne traverse method with COSPEC at Mt. St. Helens at various distances beneath the plume and reported that the estimated SO₂ emission rates were the largest by the traverse at the bottom of the plume and decreased with the distance from the plume. Their results indicate that a significant reduction of the SO₂ column concentration can occur due to the UV scattering even with the traverse method. These theoretical and field studies indicate the necessity of a quantitative understanding of the attenuation of the absorbance due to the scattering for reliable estimation of the SO₂ emission rate not only for the panning method but also for the traverse method.

[3] Recently, differential optical absorption spectroscopy (DOAS) was applied for the SO₂ emission rate measurement by the use of a miniature UV spectrometer [Galle et al., 2002; Horton et al., 2006]. In contrast to the COSPEC, the new instruments (mini-DOAS, FLYSPEC) are not only small and relatively cheap, but also can obtain the absorption spectra [Elias et al., 2006]. In the theory of molecular scattering, grater scattering occurs at the shorter wavelength bands; therefore the analyses of the absorption spectra will help the understanding of the UV scattering effect. In order to clarify the effect of the UV scattering, we conducted a simultaneous measurement of the SO₂ emission rate by the panning method using the compact UV spectrometer system from multiple stations with different distances from the plume. The variation of the absorbance attenuation was thus clarified as a function of the distance from the plume and the wavelength band.

2. Methodology

[4] We performed the first field campaign that scans the same cross-section of volcanic plume from several points of a different place at the same time in October 6, 2004, at Aso volcano, Japan. The measurement day had fine weather, and visibility was over 30 km. Humidity and wind direction at noon were about 70% and NNE, respectively at the Aso meteorological station (Figure 1). The plume direction was also determined by visual observation as south-southwestward from the crater. Similarly, the altitude of the plume center was estimated as about 1300 – 1400 m although the lowest part was hugging ground. Three compact UV spectrometer systems were prepared for this field campaign. Our compact UV spectrometer system is based on USB2000 ultraviolet spectrometer (Ocean Optics Inc.), and has a rotating mirror unit and a stepper motor. This system configuration is similar to the measuring configuration reported by Edmonds et al. [2003]. Instruments were placed linearly at three points where horizontal distances from the volcanic plume are 640 m (St.1), 1660 m (St.2) and 2580 m (St.3). Viewing directions of three telescopes were fixed in a same range of azimuth direction to scan the same cross-section of the plume (Figure 1). The operation elevation angles of the telescope mirrors are different depending on the geometric configuration between the volcanic plume and the measuring point. The field-of-view of the telescope is 6.7 mrad, and widths of the field-of-view are 4 m (St.1), 11 m (St.2) and 17 m (St.3), respectively. Sampling intervals are variable at each measuring point because of the various properties of each instrument (Table 1). For these reasons, absorbance and column concentrations of three measuring points cannot be obtained for the same area and with the same time, and sampling intervals of the SO₂ profile are various. The intensity of the UV scattering also changes with amount of the fog, rain, or cloud [Moffat and Millan, 1971]. In order to reduce these unstable factors, measurements were performed on a fine day without any cloud between 11:00 and 12:00 when the solar elevation is high (49–52°). At each measuring point, 2000–4000 absorbance spectra were obtained and 40–80 SO₂ profiles of the plume were scanned. Spectra of five standard cells with different column concentrations (100–2000 ppmm) were obtained at the beginning and the end of the measurements at each measuring point. For calculation of the SO₂ column concentration, peak and trough height of several SO₂ peaks between 303 and 320 nm are independently used in our evaluation method (Figure 2).

3. Effect of UV Scattering

[5] The standard spectrum of the 900 ppmm range is compared with the plume spectrum with the SO₂ column concentration estimated at 900 ppmm range in 315 nm band (W3) obtained at each measurement point (Figure 2). The results indicate that absorbance of the spectrum on the short wavelength band decreases with the distance from the plume. The plume spectrum obtained at St.1, the closest measuring point, closely agree with the standard spectrum for wavelength band range (e.g., in Figure 2, estimates absorbance of W1, W2 and W3 bands to be column concentrations of 912, 952 and 962 ppmm, respectively). In contrast, the plume spectrum at St.3, the most distant point, is quite different from the standard spectrum (e.g., in Figure 2, estimates absorbance of W1, W2 and W3 bands to be column concentrations of 422, 744 and 961 ppmm, respectively); the absorbance decreases with decrease in the wavelength band. Although molecular scattering can explain the wavelength band dependence of the attenuation and the observed attenuation at shorter wavelength bands, these effects were much larger than expected from molecular scattering theory. While the theoretical understanding of the absorbance attenuation accompanying UV scattering is beyond the scope of this study, these results suggest that the scattering becomes more significant at shorter wavelength bands.

[6] We plotted the SO₂ column concentration calculated using absorbance of three different absorption wavelength bands (W1: 309 nm, W2: 313 nm, W3: 315 nm) in order to evaluate of the UV scattering effect under various conditions (Figure 3). If the UV scattering is negligible, the column concentrations obtained under various conditions should be the same, giving a one-to-one correspondence. However, the plots show a variety of trends depending on the column concentrations, wavelength band and distance, as follows: 1) The agreement of the concentrations obtained with three different wavelength bands is generally good at low concentration range, 2) the concentration obtained with the shorter wavelength band becomes lower than that obtained with the longer wavelength band, 3) the concentration where the disagreement starts to appear decreases with increasing distance from the measuring point to the plume, 4) the disagreement becomes larger at shorter wavelength bands, and 5) scatter of the data is larger at longer wavelength bands. These results clearly show that UV light scattering causes significant attenuation of the absorbance at short wavelength bands, especially with high column concentrations (i.e., large amount of absorbance). The results also reveal that attenuation depends not only on wavelength band and distance but also on the SO₂ column concentration.

[7] In order to evaluate the effect of the UV scattering on the SO₂ emission rate, the SO₂ emission rates averaged over one hour were calculated for the three wavelength bands at each station (Figure 4). Since the scanning measurements of the plume were performed almost simultaneously, the error due to plume speed is eliminated and the difference in the calculated emission rates can be attributed to the effect of the UV scattering. The error on the calculated SO₂ emission rate was estimated from the difference between the zero line and the baseline. Some errors were associated with the W3 band results because of baseline drift, whereas negligible errors were calculated for other bands. Figure 4 indicates that a lower SO₂ emission rate is calculated at shorter wavelength bands at each station, and the SO₂ emission rate calculated at each wavelength band decreases with increasing distance. The emission rate of W3 band is almost constant regardless of the distance although they are associated with some errors because of the low sensitivity. The SO₂ emission rate obtained with W1 band decreases with the distance more rapidly than that obtained with W2 band. For example, emission rates obtained with W1 band decrease about 50% from St.1 to St.3, whereas emission rates obtained with W2 band decrease about 30%. When the scattering effect is large, a larger attenuation of the absorbance will be observed at shorter wavelength bands (Figure 2c). On the contrary when the scattering effect is not important, there will be little difference among the column concentrations obtained at different wavelength bands (Figure 2a). Since there is no difference between emission rates obtained with different wavelength bands at St.1, the emission rate obtained at St.1 is likely not significantly affected by the scattering effect.

4. Discussion

[8] The SO₂ emission rate, which is one of the important indicators of volcanic activity, has several error factors. The largest error factor is the wind speed determination [Stoiber et al., 1983]. However, the wind speed (same as plume speed) will be better constrained by multiple spectrometer measurements [McGonigle et al., 2005; Williams-Jones et al., 2006]. One of other important error factors is the UV scattering. We evaluated the effect of the UV scattering by conducting a simultaneous measurement of the SO₂ emission rate by the panning method using a compact UV spectrometer system from various distances. As a result, two important phenomena with regard to the UV scattering were revealed. First, the attenuation of the absorbance (also the SO₂ column concentration and the SO₂ emission rate) becomes large with increasing distance from the volcanic plume. This relation between distance and absorbance is similar to the results by Moffat and Millan [1971]. Secondly, the attenuation by the UV scattering is larger at shorter wavelength bands than at the longer ones. Because of these effects, SO₂ column concentrations and SO₂ emission rates obtained at short wavelength bands are likely underestimated compared to those obtained at long wavelength bands. In this field campaign, the attenuation of the emission rate is not significant with any wavelength bands at 0.6 km distance from the plume, but attenuations ranged from 35 to 50% at shorter wavelength bands at 2.6 km distance (Figure 4). However, factors controlling the scattering intensity are not only the distance from a plume. The scattering intensity might be variable depending on number, size, or type of aerosols and atmospheric conditions (such as humidity and temperature). Therefore, we cannot propose a quantitative model for correction of the UV scattering intensity based only on the results of this study. For quantification of scattering intensity, it is necessary to perform measurement under various conditions comparing the difference in attenuation among different wavelength bands.

[9] Sutton et al. [2001] observed that SO₂ emission rates measured with the panning method are lower than those with the traverse method. Since the COSPEC uses the absorbance of 300–315 nm [Millan and Hoff, 1978], the column concentration may be underestimated according to the strong scattering effect at the short wavelength band. On the other hand, McGonigle et al. [2002, 2003] found that the SO₂ emission rates measured using the panning method and the traverse method are similar, when the measuring point of the panning was set beneath the plume. In their case, the effect of the UV scattering may have appeared similar, since configurations of volcanic plume and the measuring point were similar for both methods. However, the UV scattering can also affect the traverse method, and the SO₂ emission rate estimated with the traverse method can include significant errors [e.g., McGee, 1992; Weibring et al., 2002]. Furthermore, the UV scattering not only causes the attenuation of the absorbance, but also occurs in the plume resulting in an overestimation of the column concentration [Millan, 1980]. Therefore, quantitative evaluation of the scattering effect is important regardless of the applied methods.

[10] It was revealed by this study that the larger effect of the UV scattering appears at the shorter wavelength bands and with increasing distance from a plume. When the UV scattering is significant, the SO₂ column amount and the emission rate measured by the panning method can be greatly underestimated. For accurate measurements with the panning method, we recommended that it be performed near the plume using the absorbance of the longer wavelength bands. However, since we cannot perform the measurement always in the optimum conditions (because of the topography, volcanic activity, etc.), in order to evaluate the effect of the UV scattering, we recommend comparing standard and plume spectra (Figure 2) or emission rates obtained at different wavelength bands (Figure 4). If the standard and plume spectra have the same form or emission rates obtained at different wavelength bands are equal, the effect of the UV scattering can be considered insignificant. Furthermore, we need further detailed experiments to obtain a technique for correction of the data affected by the UV scattering. Quantification of the UV scattering effect on measurements of SO₂ emission rates is important since the effect of the UV scattering can appear not only on the results of the panning method but also on the traverse method.