Lidar observations of Kasatochi volcano aerosols in the

Document Sample
Lidar observations of Kasatochi volcano aerosols in the Powered By Docstoc
					                   JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ???, XXXX, DOI:10.1029/,

1   Lidar observations of Kasatochi volcano aerosols in
2   the troposphere and stratosphere
               1            1                  2         2                   3
    L. Bitar, T. J. Duck, N. I. Kristiansen, A. Stohl, and S. Beauchamp

      S. Beauchamp, Meteorological Service of Canada, 16th Floor, Queen Square, 45 Alderney Dr.,

    Dartmouth, Nova Scotia, Canada, B2Y 2N6 (

      L. Bitar, Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Nova

    Scotia, Canada, B3H 3J5 (

      T. J. Duck, Department of Physics and Atmospheric Science, Dalhousie University, Halifax,

    Nova Scotia, Canada, B3H 3J5 (

      N. I. Kristiansen, Norwegian Institute for Air Research (NILU), P.O. Box 100, N-2027 Kjeller,

    Norway (

      A. Stohl, Norwegian Institute for Air Research (NILU), P.O. Box 100, N-2027 Kjeller, Norway


          Department of Physics and Atmospheric

    D R A F T                      December 2, 2009, 3:15pm                         D R A F T

 3   Abstract.        The eruption of Kasatochi volcano on 7–8 August 2008 injected

 4   material into the troposphere and lower stratosphere of the northern mid-

 5   latitudes during a period of low stratospheric aerosol background concentra-

 6   tions. Aerosols from the volcanic plume were detected with a lidar in Hal-

 7   ifax, Nova Scotia (44.64◦ N, 63.59◦ W) one week after the eruption and for the

 8   next four months thereafter. The volcanic origin of the plume is established

 9   using the FLEXPART Lagrangian particle transport model for both the strato-

10   sphere and troposphere. The stratospheric plume descended 47.1±2.8 m/day

11   on average as it dispersed, corresponding to a cooling rate of 0.60 ± 0.07

12   K/day. The descent rate was the same for the tropopause (within statisti-

13   cal uncertainties). The top of the plume remained steady at about 18 km al-

14   titude, and was likely sustained by vertical eddy diffusion from large-scale

15   horizontal mixing. The lower boundary of the plume descended with the tropopause.

16   The optical depth between 15–19 km altitude was relatively constant at 0.003

17   for 532 nm wavelength. Observations and modeling of Kasatochi aerosols in

     Science, Dalhousie University, Halifax, Nova

     Scotia, Canada.

           Norwegian Institute for Air Research

     (NILU), Kjeller, Norway.

           Meteorological Service of Canada,

     Dartmouth, Nova Scotia, Canada.

     DRAFT                           December 2, 2009, 3:15pm                         DRAFT

18   the middle and lower troposphere indicate they reached the ground. The vol-

19   canic contribution to surface PM2.5 did not exceed 5 µg/m3 at the measure-

20   ment site.

     D R A F T                     December 2, 2009, 3:15pm                        D R A F T

     1. Introduction

21     The Kasatochi volcano in the central Aleutian Islands of Alaska (52.17◦ N, 175.51◦ W)

22   erupted three times between 2201 UTC on 7 August and 0435 UTC on 8 August

23   2008, followed by a continuous ash-rich discharge for approximately 17 hours thereafter

24   [Waythomas et al., 2008]. The eruption injected a sulfur-rich plume into the lower strato-

25   sphere [Kristiansen et al., 2009; Prata et al., 2009; Karagulian et al., 2009], perturbing

26   its chemistry during a time of relatively low background aerosol content [Deshler , 2008].

27   Sulfur dioxide (SO2 ) in the eruption plume oxidized and condensed into sulfate aerosol,

28   which was distributed across the Northern Hemisphere together with other emissions that

29   were observed by many different measurement systems [Martinsson et al., 2009; Theys et

30   al., 2009; Prata et al., 2009; Karagulian et al., 2009; Bourassa et al., 2009; Hoffmann et

31   al., 2009]. A significant amount of SO2 was introduced into the troposphere [Kristiansen

32   et al., 2009], although this has received less attention.

33     The eruption plume was observed by the Dalhousie Raman Lidar in Halifax, Canada

34   (44.64◦ N, 63.59◦ W), 7445 km distance from the volcano, over a four-month period from

35   15 August – 4 December 2008. Anomalous increases of stratospheric aerosol content were

36   first detected on 15 August 2008 and observed thereafter whenever the meteorological

37   conditions permitted. Tropospheric aerosol layers measured by the lidar on 21–24 August

38   2008 were also from the Kasatochi eruption. The analysis and interpretation of measure-

39   ments from the lidar together with surface data and model simulations are presented in

40   this paper.

     D R A F T                      December 2, 2009, 3:15pm                        D R A F T

41     FLEXPART, a Lagrangian particle transport model [Stohl et al., 2005], was used to

42   establish the volcanic origin of observed aerosols. Comparisons of the stratospheric mea-

43   surements with FLEXPART are presented in a separate paper [Kristiansen et al., 2009],

44   and a similar approach was used here to confirm the tropospheric detections. Although

45   the consequences of volcanic aerosols for climate have been investigated [Kravitz et al.,

46   2009; McCormick et al., 1995; Robock , 2000, 2002], an assessment for the potential impact

47   on air quality is still needed. The Kasatochi plume is shown to have likely reached the

48   surface near Halifax, causing PM2.5 increases of up to 5 µg/m3 . The EPA standard for

49   air quality is 65 µg/m3 over 24 hours [U.S. EPA, 2004], and so the overall impact on

50   surface air quality was small, although potentially widespread.

51     In the stratosphere, the plume descent and dispersion are investigated. The plume

52   descended at the same rate as the tropopause (within statistical uncertainties), and an es-

53   timate for the net stratospheric cooling rate is obtained. The plume maintained a presence

54   near 18 km altitude despite the descent, and this is attributed to vertical diffusion from

55   large-scale horizontal eddy mixing. The optical depth of the plume was fairly constant

56   over a four-month period, although a greater number of exceptional events were observed

57   in the first two months.

58     We begin by describing the lidar and measurement inversion process, and discuss the

59   specifics of the model simulations. The stratospheric measurements and interpretation

60   are given next, followed by the tropospheric analysis.

     D R A F T                     December 2, 2009, 3:15pm                         D R A F T

     2. Instrumentation and Modeling

     2.1. Dalhousie Raman Lidar

61     The Dalhousie Raman Lidar measures vertical profiles of scattering from atmospheric

62   aerosols, clouds and molecules. It is a transportable instrument that has been used in

63   international measurement campaigns [Duck et al., 2007; McKendry et al., 2008]. Figure 1

64   gives a schematic diagram and Table 1 lists its specifications. The lidar employs a high-

65   energy Nd:YAG laser that emits pulses of 532 nm wavelength light at a repetition rate

66   of 20 Hz. The pulses are expanded and collimated to minimize beam divergence before

67   being transmitted upward into the atmosphere. The outgoing laser beam is pointed

68   into the fields-of-view of a 25 cm Newtonian telescope and a co-aligned 8 cm refractor

69   (usually limited to 3 cm aperture). The two telescopes allow for the backscattered light

70   to be collected separately from both the near and far ranges. This capability is used to

71   simultaneously obtain measurements in the boundary layer and free troposphere / lower

72   stratosphere.

73     Optical fibers guide the signals from each telescope into a polychromator, which uses

74   interference filters, collimating lenses, and dichroic mirrors to separate the returns into

75   elastic (532 nm) and nitrogen (N2 ) Raman-shifted (607 nm) wavelengths. To obtain high-

76   altitude elastic returns with low noise, a mechanical shutter (bow-tie chopper) is used to

77   block the intense low-altitude signals from reaching one detector.

78     Photomultiplier tubes are used for photon detection and the signals are recorded as a

79   function of altitude using fast counting computer electronics. Neutral density filters and

80   irises are placed in front of the photomultipliers to maintain signal intensities at reasonable

81   levels.

     D R A F T                       December 2, 2009, 3:15pm                           D R A F T

82      A radar is used to protect aircraft flying over the measurement site [Duck et al., 2005]. A

83    circuit automatically interrupts the laser if an aircraft is detected during lidar operations,

84    which results in occasional measurement gaps.

85      The lidar is operated during intensive summertime measurement campaigns. Since the

86    eruption of Kasatochi volcano on 7–8 August 2008 occurred near the end of the 2008

87    campaign, measurements were continued thereafter, but with less time coverage, until 4

88    December 2008. The lidar signals were measured with a vertical resolution of 9.6 m and

89    a temporal resolution of 15 s. Spatial and time averaging are used in the data processing

90    to improve signal-to-noise ratios.

91      Profiles of the aerosol extinction at 532 nm wavelength are retrieved by inversion of the

92    elastic signals using the Klett technique [Klett, 1981]. The Raman technique [Ansmann et

93    al., 1990] was not used because signals are insufficiently strong for such an analysis. A lidar

94    (extinction-to-backscatter cross-section) ratio of 40 sr was assumed for the stratospheric

95    aerosols, which falls in the mid-range of values observed for background conditions and

96    major volcanic eruptions [Jaeger and Hofmann, 1991; Jaeger et al., 1995; Jaeger and

97    Deshler , 2002]. A lidar ratio of 71 sr was used for tropospheric aerosols, which is the

98    average observed for sulfate aerosols of urban/industrial origin [Cattrall et al., 2005]. A

99    similar value of the lidar ratio was measured in the upper troposphere of Ny-˚lesund,

100   Spitsbergen for the Kasatochi plume [Hoffmann et al., 2009].

101     The Klett inversion technique requires initialization at an altitude of known extinction.

102   We visually identify a range of contiguous initialization altitudes in each measurement

103   by an apparent lack of structure. This range is taken to be clear air, with zero aerosol

104   extinction. Initialization regions both below and above an aerosol plume are used as

      D R A F T                      December 2, 2009, 3:15pm                           D R A F T

105   appropriate, and are mixed evenly in the data set presented here. The aerosol extinction

106   values given in the following sections represent the median of all possible retrievals, which

107   reduces noise and uncertainties in the results.

      2.2. FLEXPART

108     The Lagrangian particle dispersion model FLEXPART simulates the long-range trans-

109   port and dispersion of many particles released from a defined source, while considering

110   wet and dry deposition, radioactive decay, and convection along the transport path. A

111   description of the model is given by [Stohl et al., 2005]. Both backward and forward trajec-

112   tories can be calculated using wind, temperature, and pressure fields from meteorological

113   analyses.

114     FLEXPART modeling of the eruption plume transport was conducted in forward mode

115   using SO2 as a tracer. Backward-mode simulations for the transport of anthropogenic

116   SO2 were also conducted for comparison. Dry deposition and oxidation by OH radicals

117   are parametrized as removal processes. The simulations employed meteorological data

118   from the European Center for Medium-Range Weather Forecasts (ECMWF), which have

119   91 vertical levels and 0.5◦ × 0.5◦ horizontal resolution for the eastern North Pacific region

120   (1◦ ×1◦ globally). The volcanic source term for the simulations was taken from Kristiansen

121   et al. [2009], who used FLEXPART, satellite observations of SO2 during the first few days

122   after the eruption, and an inversion algorithm to determine an optimal emission height

123   profile.

124     SO2 is gradually converted into the sulfate aerosol observed by the lidar, and so com-

125   parisons between the FLEXPART SO2 simulations and aerosol optical depth and surface

126   PM2.5 measurements are qualitative. The model was run at 500 m vertical and 1 hour

      D R A F T                      December 2, 2009, 3:15pm                          D R A F T

127   temporal resolution in order to provide detailed structural comparisons with the lidar

128   measurements.

      3. Stratosphere

      3.1. Vertical distribution and descent

129     Figure 2 shows aerosol extinction contours measured by the lidar on 21–24 August

130   and selected periods between 4–12 September 2008 which reveal the presence of aerosol

131   layers high in the atmosphere above Halifax. On August 21–24, an aerosol layer persisted

132   throughout the 84-hour measurement period at a relatively stable altitude of about 18 km,

133   and decreased in intensity with time. A second layer appeared in the morning of August

134   23 near 17 km and was observed until the morning of August 24. One month following

135   the Kasatochi eruption, distinct high-altitude aerosol plumes were still observed during

136   September 4–12. On each day of the measurements presented in Figure 2, the aerosol

137   plumes varied in optical density with values of the aerosol extinction ranging from just

138   above background levels to a maximum of approximately 0.03 km−1 . The aerosol layers

139   were highly structured with vertical thickness less than 1 km, and remained confined

140   between 16–18 km altitude. Light aerosol loading can be seen just below the main plume

141   in each measurement, likely of the same origin.

142     Temperature profiles measured by radiosondes launched from the nearest weather sta-

143   tion (Yarmouth, approximately 300 km southwest of Halifax at 43.86◦ N, 66.10◦ W) were

144   used to gauge the tropopause height (Figure 3). During the measurement intervals of

145   Figure 2 the tropopause ranged from 12–16 km altitude, indicating that the aerosols were

146   within the lower stratosphere. Injection of aerosols past the tropopause is suggestive of a

147   powerful event such as a volcanic eruption. In a separate paper, the stratospheric aerosols

      D R A F T                     December 2, 2009, 3:15pm                         D R A F T

148   observed above Halifax were verified to be from the Kasatochi eruption by using com-

149   parisons with FLEXPART simulations [Kristiansen et al., 2009]. Here, we focus on the

150   evolution of the plume in the four months following the eruption, assuming no further

151   stratospheric aerosol injections.

152     Figure 3 displays the altitude of maximum aerosol extinction (hereafter referred to as

153   just the “plume altitude”) versus time, along with the height of the tropopause. The plume

154   altitude was determined by inspection of hourly-integrated aerosol extinction profiles from

155   August–October and three-hour integrated profiles from November, both with a vertical

156   resolution of 100 m. The measurements during November were integrated over three hours

157   in order to improve signal-to-noise ratios since the measured aerosol extinction dropped as

158   the aerosols dispersed over time. Only profiles with identifiable maxima above the aerosol

159   background were considered, and those that contained too much noise to distinguish

160   an aerosol layer were ignored. Gaps in the data were mostly a result of meteorological

161   conditions inappropriate for lidar operation.

162     As seen in Figure 3, prior to 4 September 2008 the plume altitude increased with

163   time. This is not evidence for ascent of the stratospheric air mass, but is instead due

164   to differential advection of the plume by the jet stream, which has maximum speed near

165   the tropopause. After September 4, the plume descended until the end of observations

166   in early December, and this is likely due to stratospheric subsidence from net radiative

167   cooling during the transition from summer to winter. Hourly fluctuations in the plume

168   altitude were evident, in some cases varying by as much as 1 km during a single day.

169     The plume descent rate was determined to be 47.1 ± 2.8 m/day using a linear least-

170   squares fit. Weekly average values of the plume altitude were used in the calculation

      D R A F T                      December 2, 2009, 3:15pm                        D R A F T

171   to remove bias due to the uneven distribution of data in the 16 weeks of measurements

172   considered. The fitting procedure was applied only to the data measured after September

173   4.

174        The tropopause altitude shown in Figure 3 was determined from radiosonde measure-

175   ments obtained at 0000 UTC and 1200 UTC. The tropopause altitude was comparatively

176   much more variable, but in general also descended with time. A linear least-squares fit

177   after September 4 gives a descent rate of 52.7 ± 6.5 m/day. Although more uncertain,

178   this value is statistically consistent with the descent rate for the plume. The correla-

179   tion between the decent of the aerosols and the tropopause is intriguing given that the

180   tropopause is a dynamical feature. In this case, it appears to have descended like a mate-

181   rial element as the stratosphere cooled, although there is no obvious physical mechanism

182   to explain this behavior.

183        The potential temperatures corresponding to the plume altitudes in Figure 3 were de-

184   termined by using daily radiosonde data. As shown in Figure 4, a linear least-squares fit

185   to the potential temperature data yields a lower-stratospheric cooling rate of 0.60 ± 0.07

186   K/day. This is somewhat greater than expected from global circulation model simulations

187   (e.g., Hamilton et al. [1995]), and this may be due to differences between the dynamics in

188   model climatologies and the actual atmospheric conditions during our measurements. In

189   any event, the direct radiative impact of the aerosols would be expected to have a minimal

190   impact, given that the Pinatubo volcano eruption caused 0.01–0.05 K/day heating of the

191   stratosphere at mid-latitudes [Robock , 2000] for much larger optical depths.

192        The vertical dispersion of the Kasatochi eruption plume with time is shown in Figure 5,

193   which provides daily-average aerosol extinction profiles. The figure contains all profiles

      D R A F T                       December 2, 2009, 3:15pm                         D R A F T

194   obtained during the campaign, regardless of whether or not the plume was clearly in

195   evidence. The aerosol extinction varied from day to day with maximum values observed

196   on August 21 and September 11 and 23, which are displayed in more detail in Figure 2.

197   The vertical extent of the aerosols was variable with some days having two distinct layers

198   present. The plume descent is apparent and is consistent with the results depicted in

199   Figure 3. The plume is evident throughout the four-month measurement period.

200     There was little trend in the maximum altitude of the layer, which remained relatively

201   stable around 18 km until near the end of the four-month period of observations. This

202   result is consistent with the aerosol measurements for 40◦ and 50◦ N from the OSIRIS

203   satellite instrument [Bourassa et al., 2009]. A steady upper altitude for the plume is

204   interesting given that air in the stratosphere slowly descended from radiative cooling

205   during the transition from summer to winter, and is due to vertical eddy diffusion. The

206   characteristic length scale LD in diffusion problems is given by

207                                          LD =    4Kz t

208   where Kz is the eddy diffusivity and t is time. Substituting 5 km ascent over 12 weeks

209   yields a diffusivity Kz ≈ 0.9 m2 /s, and estimates using shorter intervals (Kz is quadratic

210   in LD ) yield lower values. The results are consistent with expectations “of a few times

211   10−1 ” for large-scale horizontal eddy mixing [Holton et al., 1995].

212     The base of the plume followed the descent of the tropopause so that by November,

213   the initially vertically thin aerosol plumes were distributed from approximately 12 to 18

214   km altitude. The influence of tropopause variability on the aerosol layer structure is

215   particularly evident in the measurements obtained beyond October (i.e., after day 54).

216   Martinsson et al. [2009] identified elevated sulfur and carbon particle concentrations in

      D R A F T                      December 2, 2009, 3:15pm                        D R A F T

217   the upper troposphere / lower stratosphere over Europe for three to four months after

218   the eruption that they attributed to the Kasatochi eruption. Our measurements suggest

219   that the continued source of upper tropospheric Kasatochi aerosols may have been the

220   lower stratosphere. The presumed mechanisms for tropospheric-stratospheric exchange

221   are tropopause folding (e.g., [Holton et al., 1995]) and eddy diffusion from meso- and

222   small-scale turbulence at the tropopause (e.g. Duck and Whiteway [2005]). The gradual

223   addition of aerosols to the upper troposphere by way of the stratosphere might cause an

224   indirect seasonal impact on radiative transfer through the modification of cloud optical

225   properties.

226     The measurements obtained during August to early October showed considerable vari-

227   ability in the intensity and time variation of the aerosol plume on hourly time scales, char-

228   acterized by narrow and distinct extinction maxima (e.g. Figure 2). At times, aerosols

229   were not consistently present during a single measurement. This scenario gradually gave

230   way to the constant presence of aerosols in the latter half of October and November, but

231   with reduced and more uniform intensity. This evolution is consistent with gradual and

232   extensive horizontal mixing as seen in satellite measurements.

      3.2. Optical depth

233     Figure 6 provides the temporal evolution of the daily aerosol optical depths at 532 nm

234   wavelength between 15–19 km altitude determined from the extinction data in Figure 5.

235   Three measurements before the plume’s arrival show near-zero aerosol optical depth, which

236   illustrates the overall sensitivity of our measurement technique when compared to the

237   plume data that follow.

      D R A F T                      December 2, 2009, 3:15pm                          D R A F T

238     The optical depth of Kasatochi aerosols between 15–19 km remained relatively constant

239   at about 0.003 throughout the observation period. This is surprising at first because the

240   intensity of the aerosol layers immediately following the eruption was much higher (e.g.,

241   Figure 2) than toward the end. However, as noted above there were initially extended

242   periods where the plume was absent, and these were included in the integrations. Thus,

243   Figure 6 indicates that the aerosol was reasonably conserved during the horizontal mixing

244   process.

245     Relatively stable stratospheric optical depth over the same period was also observed by

246   the OSIRIS satellite instrument at 45◦ N [Kravitz et al., 2009]. The optical depth measured

247   by OSIRIS was about 0.0055 at 750 nm wavelength, which is of similar magnitude to the

248   mean of 0.003 we observed at 532 nm wavelength. The OSIRIS optical depth measurement

249   used a lower bound in potential temperature of 380 K, which corresponds to about the

250   same 15 km lower altitude limit for our optical depth measurements. If the difference

251   in wavelengths is taken into account, a scaling factor of 0.8 results [Kravitz et al., 2009].

252   This leads to a satellite-derived aerosol optical depth of 0.0069 at 550 nm, which is a little

253   more than twice what we measured. The difference can be partly attributed to systematic

254   uncertainties in the lidar ratio used in the retrievals. Notwithstanding, Kasatochi induced

255   a much smaller perturbation to the stratosphere than the Pinatubo eruption, which yielded

256   optical depths in the stratosphere up to 0.2 [Ansmann et al., 1997].

257     The variability of the optical depth measurements given in Figure 6 changed consider-

258   ably with time. The first two months were characterized by high variability, with a greater

259   incidence of outliers with high optical depth. This is to be expected for a plume before it

260   is well-mixed with the environment. The persistence of high optical depth outliers up to

      D R A F T                      December 2, 2009, 3:15pm                           D R A F T

261   two months after the eruption follows from the e-folding time for the conversion of SO2 to

262   sulfate aerosol, which is about 30 days [Textor et al., 2004]. During the latter two months

263   the aerosol load is much more consistent, with outliers having low optical depth. This

264   is consistent with a well-mixed plume occasionally descending to below the optical depth

265   measurement altitudes (15–19 km).

266     Interpretation of the optical depth measurements is subject to our assumption of a con-

267   stant lidar ratio, which might have systematic error and change with time. Measurements

268   of Kasatochi aerosols in the upper troposphere immediately following the eruption indi-

269   cated variable size, which would lead to varying lidar ratios. We expect that horizontal

270   mixing should diminish this effect after a few weeks time. Jaeger and co-authors [Jaeger

271   and Hofmann, 1991; Jaeger et al., 1995; Jaeger and Deshler , 2002] showed that the li-

272   dar ratio for stratospheric aerosols varies on a seasonal time-scale and decreases in the

273   year following a major volcanic eruption. The retrievals presented here have been ana-

274   lyzed using the different lidar ratios measured by Jaeger and co-authors for stratospheric

275   aerosols, which range from approximately 20–60 sr. The maximum systematic error in the

276   retrieved extinction profile introduced by assuming an incorrect lidar ratio is 40%. Our

277   actual uncertainties are expected to be below that level given the differences in time-scale

278   from what Jaeger and co-authors considered.

      4. Troposphere

279     Figure 7 shows the tropospheric measurement corresponding to the August 21–24 data

280   presented in Figure 2. Aerosol layers were observed on August 22 through August 24,

281   extending up to 7 km in altitude. The onset of an intense low-altitude aerosol event was

282   observed on August 22, which descended from about 3 km altitude at 0000 UTC down

      D R A F T                     December 2, 2009, 3:15pm                         D R A F T

283   to ground level reaching the surface just after 1300 UTC. In the middle troposphere an

284   optically-thin “halo” of aerosols with an area of clear air at the center is apparent. On

285   August 24, a decrease in the surface aerosol extinction was observed, gradually extending

286   up to 1 km in altitude until 1200 UTC while enhanced aerosol extinction remained in the

287   overlying layer between approximately 1–1.5 km altitude.

288     Figure 8 shows the FLEXPART-simulated SO2 emissions from the Kasatochi eruption

289   above Halifax for August 21–24. The simulated SO2 concentrations, used as a proxy for

290   aerosol formation, showed a similar overall vertical distribution as the aerosols observed by

291   the lidar. The “halo” of aerosols was evident and in good spatial and temporal agreement

292   with the lidar measurement, which provides strong evidence that this feature originated

293   from the Kasatochi eruption. The intense lower-altitude event at 3 km altitude was also

294   captured by the model as well as its descent to the surface. This indicates a contribu-

295   tion from Kasatochi emissions to the surface aerosol burden, although there were likely

296   contributions from anthropogenic and other natural sources as well.

297     The timing difference in the onset of the simulated and measured aerosol events can

298   be attributed to the coarseness of the SO2 emissions inventory. The emission profile

299   represents the mean for the first two eruptive events, and the third eruption occurred six

300   hours after the first. Higher temporal resolution in the inventory could not be realistically

301   obtained in the retrieval process (see Kristiansen et al. [2009]). Thus, timing errors of up

302   to six hours are not unexpected.

303     In comparison to the stratospheric aerosol plumes observed during the same lidar mea-

304   surement, much more variability was apparent in the vertical structure of the tropospheric

305   aerosols throughout the measurement period. The aerosol plumes in the troposphere ap-

      D R A F T                      December 2, 2009, 3:15pm                          D R A F T

306   peared more diffuse and were distributed over a wider altitude range than those observed

307   in the lower stratosphere. This is a consequence of the lower static stability of the tro-

308   posphere compared to the stratosphere, which leads to material being detrained from the

309   eruptive column over deeper layers than in the stratosphere where injection occurs in

310   discrete layers. The aerosol extinction ranged in value from about 0.002 km−1 for the

311   lightest aerosol loads to 0.12 km−1 in the regions of densest aerosol content. The intensity

312   of aerosol extinction in the troposphere was more than that observed in the stratosphere,

313   with the greatest values detected below 3 km altitude. Kristiansen et al. [2009] showed

314   that a larger portion of the SO2 (about 60%) was emitted into the stratosphere; however,

315   as observed from Halifax, the aerosol intensity was greatest in the troposphere. This is

316   likely due to the relatively rapid formation of sulfate aerosols by SO2 oxidation in the

317   lower troposphere, and contributions in the troposphere from other aerosol sources.

318     Figure 9 compares the average aerosol extinction profile measured between August 22 at

319   1200 UTC and August 23 at 1200 UTC with the average simulated SO2 profile during the

320   same time interval. The vertical resolution of the FLEXPART profile is 500 m whereas for

321   the lidar profile it is 50 m. The simulated SO2 profile reproduces the form of the measured

322   aerosol extinction profile, and captures both the surface aerosols as well as those in the free

323   troposphere. The fact that the upper layer appears relatively stronger in the FLEXPART

324   simulation can be explained by the fact that the model simulates SO2 whose removal rate

325   decreases with altitude. On the other hand, the aerosol extinction is a measure of the

326   secondary aerosol product whose formation rate decreases with altitude.

327     Additional tropospheric lidar detections of Kasatochi aerosols in August were confirmed

328   by FLEXPART (not shown). The model indicated that the first arrival of volcanic aerosols

      D R A F T                      December 2, 2009, 3:15pm                           D R A F T

329   over Halifax was between 0200–0600 UTC on 14 August, where two aerosol layers were

330   observed just below the tropopause at about 7 km and 8 km. Optically-thin aerosol

331   layers originating from the Kasatochi eruption were also observed in the free troposphere

332   between 4–7 km in altitude on 15 August 2008. On these days, the aerosols remained in

333   the free troposphere and did not reach the boundary layer. The tropospheric detections

334   of volcanic aerosols produced by Kasatochi were dispersed much more rapidly than those

335   observed in the lower stratosphere. This made their identification more difficult given

336   the abundance of aerosols in the troposphere from other sources and more rapid removal

337   processes, and thus later detections of tropospheric volcanic aerosols cannot be confirmed.

338     Figure 10 shows surface PM2.5 measurements during August 21–24 from Kejimkujik Na-

339   tional Park (44.4◦ N, 65.2◦ W) near Halifax together with the simulated SO2 from FLEX-

340   PART for both volcanic and anthropogenic sources. The PM2.5 increases during August

341   22, and dissipates on August 23, in agreement with the lowest altitude of the lidar’s aerosol

342   extinction measurement given in Figure 7. FLEXPART indicates that the aerosol content

343   during August 21 and 22 is due equally to the Kasatochi and anthropogenic emissions,

344   whereas the peak values observed on August 23 are dominated by anthropogenic sources.

345   The maximum value for PM2.5 during the period of potential Kasatochi surface influence

346   was 5 µg/m3 , which we take to be the maximum possible impact from the volcanic emis-

347   sions at Kejimkujik. Although this value is small compared to the EPA standard for air

348   quality (65 µg/m3 over 24 hours), there was potentially a very large area impacted with

349   varying intensities.

      D R A F T                      December 2, 2009, 3:15pm                          D R A F T

      5. Summary and Conclusions

350     Lidar measurements obtained from Halifax for four months following the eruption of the

351   Kasatochi volcano on 7–8 August 2008 were used to characterize the vertical structure and

352   evolution of the resulting volcanic aerosols. Kasatochi aerosols were first detected above

353   Halifax one week following the eruption, and were observed both in the troposphere and

354   lower stratosphere until 4 December 2008. The stratospheric aerosols developed as a

355   thin structured plume near 18 km altitude and gradually dispersed over the four month

356   observation period. Tropospheric aerosols were only observed definitely in the second

357   week after the eruption.

358     The stratospheric aerosol maximum descended with time in correlation with the

359   tropopause altitude during the transition from summer to winter. The descent corre-

360   sponded to a lower stratospheric cooling rate of 0.60 ± 0.07 K/day. The top of the plume

361   persisted at 18 km altitude, and was likely sustained there by vertical diffusion from large-

362   scale horizontal eddy mixing. The bottom of the plume reached the tropopause, and likely

363   provided an aerosol source for the upper troposphere, leading to a possible impact on cloud

364   properties and radiation during the four months or more following the eruption. Even

365   though the plume was dispersed the aerosol optical depth remained relatively stable, an

366   observation that is in agreement with similar measurements by OSIRIS.

367     In comparison to the stratospheric observations, the tropospheric aerosols observed on

368   August 21–24 were much more variable. Mixing with aerosols from different sources

369   diluted the tropospheric aerosols and removal processes made it difficult to observe them

370   after a short time interval had passed. Tropospheric aerosols originating from Kasatochi

      D R A F T                      December 2, 2009, 3:15pm                         D R A F T

371   likely reached the surface near Halifax. The maximum contribution to surface PM2.5 was

372   5 µg/m3 , which is considered small.

373     Acknowledgments. This study was supported by the Canadian Foundation for Cli-

374   mate and Atmospheric Science (CFCAS) and the Natural Sciences and Engineering Re-

375   search Council (NSERC) of Canada using equipment funded by the Canadian Foundation

376   for Innovation (CFI) and the Nova Scotia Research and Innovation Trust (NSRIT). Rob

377   Harris, Ben Bougher, and Marshall Hawkins helped operate the lidar during the summer-

378   time 2008 measurement campaign.


379   Ansmann, A., M. Riebesell, and C. Weitkamp (1990), Measurement of atmospheric aerosol ex-

380     tinction profiles with a Raman lidar, Opt. Lett., 15, 746–748.

381   Ansmann, A., I. Mattis, U. Wandinger, F. Wagner, J. Reichardt, and T. Deshler (1997), Evolution

382     of the Pinatubo Aerosol: Raman lidar observations of particle optical depth, effective radius,

383     mass, and surface area over Central Europe at 53.4◦ N, J. Atmos. Sci., 54, 2630–2641.

384   Bourassa, A. E., D. A. Degenstein, B.J. Elash, and E. J. Llewellyn (2009), Evolution of the strato-

385     spheric aerosol enhancement following the Kasatochi eruption: Odin-OSIRIS measurements,

386     J. Geophys. Res., special issue, submitted 2009.

387   Cattrall, C., J. Reagan, K. Thome, and O. Dubovik (2005), Variability of aerosol and spec-

388     tral lidar and backscatter extinction ratios of key aerosol types derived from selected Aerosol

389     Robotic Network locations, J. Geophys. Res., 110, D10S11, doi:10.1029/2004JD005124.

390   Deshler, T. (2008), A review of global stratospheric aerosol: Measurements, importance, life

391     cycle, and local stratospheric aerosol, Atmos. Res., 90, 223–232.

      D R A F T                         December 2, 2009, 3:15pm                             D R A F T

392   Duck, T. J., B. Firanski, F. D. Lind, and D. Sipler (2005), Aircraft-protection radar for use with

393     atmospheric lidars, Appl. Opt., 44, 4937–4945.

394   Duck, T. J., and J. A. Whiteway (2005), The spectrum of waves and turbulence at the tropopause,

395     Geophys. Res. Lett., 32, L07801, doi: 10.1029/2004GL021189.

396   Duck, T. J., B. J. Firanski, D. B. Millet, A. H. Goldstein, J. Allan, R. Holzinger, D. R. Worsnop,

397     A. B. White, A. Stohl, C. S. Dickinson, and A. van Donkelaar (2007), Transport of forest

398     fire emissions from Alaska and the Yukon Territory to Nova Scotia during summer 2004, J.

399     Geophys. Res., 112, D10S44, doi: 10.1029/2006JD007716.

400   Hamilton, K., R. J. Wilson, J. D. Mahlman, and L. J. Umscheid (1995), Climatology of the

401     SKYHI troposphere-stratosphere-mesosphere general circulation model, J. Atmos. Sci., 52,

402     5–43.

403   Hoffmann, A., C. Ritter, M. Stock, M. Maturilli, and R. Neuber (2009), Lidar measurements of

404     the Kasatochi aerosol plume in August and September 2008 in Ny-˚lesund, Spitsbergen, J.

405     Geophys. Res., special issue, submitted 2009.

406   Holton, J. R., P. Haynes, M. E. McIntyre, A. R. Douglass, R. B. Rood, and L. Pfister (1995),

407     Stratosphere-Troposphere Exchange, Rev. Geophys., 33, 403–439.

408   Jaeger, H. and D. Hofmann (1991), Midlatitude lidar backscatter to mass, area, and extinction

409     conversion model based on in situ aerosol measurements from 1980 to 1987, Appl. Opt., 21,

410     127–138.

411   Jaeger, H., T. Deshler, and D. J. Hofmann (1995), Midlatitude lidar backscatter conversions

412     based on ballooneborne aerosol measurements, Geophys. Res. Lett., 22, 1729–1732.

413   Jaeger, H. and T. Deshler (2002), Lidar backscatter to extinction, mass and area conversions

414     for stratospheric aerosols based on midlatitude ballooneborne size distribution measurements,

      D R A F T                         December 2, 2009, 3:15pm                            D R A F T

415     Geophys. Res. Lett., 29, doi:10.1029/2002GL015609.

416   Karagulian, F., L. Clarisse, P. F. Coheur, A.J. Prata, D. Hurtmans, and C. Clerbaux (2009),

417     Detection of SO2 , ash and H2 SO4 using the IASI sounder, J. Geophys. Res., accepted.

418   Klett, J. D. (1981), Stable analytical inversion solution for processing lidar returns, Appl. Opt.,

419     20, 211–220.

420   Kravitz, B., A. Robock, A. Bourassa, and G. Stenchikov, Negligible climatic effects from the 2008

421     Okmok and Kasatochi volcanic eruptions, J. Geophys. Res., special issue, submitted 2009.

422   Kristiansen, N. I., A. Stohl, A. J. Prata, A. Richter, S. Eckhardt, P. Seibert, A. Hoffmann,

423     C. Ritter, L. Bitar, T. J. Duck, and K. Stebel (2009), Remote Sensing and inverse transport

424     modeling of the Kasatochi eruption SO2 cloud, J. Geophys. Res., special issue, submitted 2009.

425   Martinsson, B. G., C. A. M. Brenninkmeijer, S. A. Carn, M. Hermann, K.-P. Heue, P. F. J.

426     van Velthoven, and A. Zahn (2009), Influence of the 2008 Kasatochi volcanic eruption on

427     sulfurous and carbonaceous aerosol constituents in the lower stratosphere, Geophys. Res. Lett.,

428     36, L12813, doi:10.1029/2009GL038735.

429   McCormick, M. P., L. W. Thomason, and C. R. Trepte (1995), Atmospheric effects of the Mt

430     Pinatubo eruption, Nature, 373, 399–404.

431   McKendry, I. G., A. M. Macdonald, W. R. Leaitch, A. van Donkelaar, Q. Zhang, T. Duck, and

432     R. V. Martin (2008), Trans-Pacific dust events observed at Whistler, British Columbia during

433     INTEX-B, Atmos. Chem. Phys., 8, 6297–6307.

434   Prata, A. J., G. Gangale, L. Clarisse,and F. Karagulian (2009), Ash and sulphur dioxide in

435     the 2008 eruptions of Okmok and Kasatochi - insights from high spectral resolution satellite

436     measurements, J. Geophys. Res., submitted.

437   Robock, A. (2000), Volcanic eruptions and climate, Rev. Geophys., 38, 191–219.

      D R A F T                         December 2, 2009, 3:15pm                             D R A F T

438   Robock, A. (2002), Pinatubo eruption: The climatic aftermath, Science, 295, 1242–1244.

439   Stohl, A., C. Forster, A. Frank, P. Seibert, and G. Wotawa (2005), Technical note: The La-

440     grangian particle dispersion model FLEXPART version 2.6, Atmos. Chem. Phys., 5, 2461–

441     2474.

442   Textor, C., H.-F. Graf, C. Timmreck, and A. Robock (2004), Emissions from volcanoes, Chapter

443     7 of Emissions of Chemical Compounds and Aerosols in the Atmosphere, C. Granier, P. Artaxo,

444     and C. Reeves, 269–303.

445   Theys, N., M. van Roozendael, B. Dils, F. Hendrick, N. Hao, and M. De Maziere (2009), First

446     satellite detection of volcanic bromine monoxide emission after the Kasatochi eruption, Geo-

447     phys. Res. Lett., 36, L03809, doi:10.1029/2008GL036552.

448   U.S. EPA (2004), Air quality criteria for particulate matter (final report, Oct 2004), U.S. Envi-

449     ronmental Protection Agency, Washington, DC, EPA 600/P-99/002aF-bF.

450   Waythomas, C. F., S. G. Prejean, and D. J. Schneider (2008), Small volcano, big eruption,

451     scientists rescued just in time: US Department of the Interior online publication People, Land,

452     and Water preprint.

      D R A F T                        December 2, 2009, 3:15pm                             D R A F T

Figure 1. A schematic diagram of the Dalhousie Raman Lidar System. The laser transmitter

employs a diverging mirror (DM) and collimating mirror (CM) as a beam expander, and an

actuator-controlled steering mirror (SM) directs the beam upward. Two telescopes (the Newto-

nian T1 and and actuator-controlled refractor T2) are used in the receiver. Light is coupled into

optical fibers using mirrors (M) and lenses (L). The polychromator makes use of long-pass (LP)

filters and interference filters (IF) to perform wavelength selection, and neutral density (ND)

filters to limit the signal strength. A beam splitter (BS) channels 99% of the primary 532 nm

channel through a chopper into the high-altitude portion of the receiver. Photomultiplier tubes

(PMTs, with wavelengths marked in nm) and fast-counting computer electronics are used for

signal detection. A pulse generator governs the instrument timing, and a radar is used to protect

aircraft flying overhead via the master interlock.

Figure 2. Stratospheric aerosol extinction (532 nm wavelength) measured above Halifax on 21–

24 August, and 4, 8–12 September 2008. The data show the temporal evolution of high-altitude

aerosol plumes originating from the 7–8 August 2008 eruption of Kasatochi volcano. Gaps in the

measurements are mostly due to interference from clouds below, which can completely attenuate

the laser beam.

Figure 3.     The altitude of maximum aerosol extinction (“plume altitude”) and that of the

local tropopause during August through November 2008. The black straight line marks the

mean descent of the Kasatochi plume altitude and the grey line marks the mean descent of the


Figure 4. The potential temperatures θ corresponding to the daily-mean plume altitudes from

the data in Figure 3.

D R A F T                        December 2, 2009, 3:15pm                             D R A F T

Figure 5.     The daily-averaged aerosol extinction profiles (532 nm wavelength) showing the

vertical dispersion of the Kasatochi plume. The tropopause height is overlayed as a white line

for comparison. Note the nonlinear time scale, varying from days to weeks. Interference from

cirrus clouds in the upper troposphere is greyed out at the bottom of the profiles.

Figure 6. The daily aerosol optical depths (532 nm wavelength) measured between 15–19 km

altitude in the four months following the eruption of Kasatochi.

Figure 7.    Tropospheric aerosol extinction (532 nm wavelength) measured above Halifax on

21–24 August 2008.

Figure 8.     FLEXPART simulation of Kasatochi SO2 emissions appearing above Halifax on

21–24 August 2008.

Figure 9.      Average profiles of observed aerosol extinction and simulated SO2 for Halifax

between 1200 UTC 22 August to 12 UTC 23 August 2008.

Figure 10. Measured PM2.5 and simulated SO2 for Kejimkujik National Park on 21–24 August

2008. The simulated SO2 is broken down into Kasatochi emissions and anthropogenic (other)


D R A F T                        December 2, 2009, 3:15pm                            D R A F T

Table 1.    Dalhousie Raman Lidar system specifications.


                   Laser Model                     Continuum Powerlite
                                                   Precision II 8020
                   Transmitted Wavelength          532 nm
                   Pulse Repetition Frequency      20 Hz
                   Pulse Duration                  8 ns
                   Pulse Energy                    550 mJ


                   Far Field Telescope Diameter    25 cm
                   Near Field Telescope Diameter   8 cm
                   Photomultiplier Tubes           Hamamatsu R7400P
                   Counter Board Model             FAST ComTec P7888
                   Counter Speed                   1 GHz

                   Data Acquisition

                   Elastic Wavelength              532 nm
                   Molecular Wavelength            607 nm
                   Range Resolution                9.6 m
                   Temporal Resolution             15 s

D R A F T                        December 2, 2009, 3:15pm                     D R A F T

Shared By: