Summit observations: emissions

A first target of EPL-RADIO is the proximal aerosol plume characterisation, i.e. at Etna summit. Figure 1a,b shows (hbox {PM}_1) and (hbox {PM}_{10}) (particulate matter with diameter less than 1 and (10,upmu hbox {m})) concentrations measured with the OPC at Etna summit (BNC and VOC craters) on 20/07/2017. Etna’s plume was directed towards the south-east direction. Two plume traverses were realised on that particular day. The (hbox {PM}_1) concentrations were around (100,upmu hbox {g/m}^3) for most of the traverse and reached up to (600,upmu hbox {g/m}^3) in what can be identified as the approximate center of the summit plume. The (hbox {PM}_{10}) concentration reached up to 8,000(,upmu hbox {g/m}^3) in a limited-sized location. Results from chamber testing have shown that these low-cost sensors can underestimate (hbox {PM}_1) mass concentration and overestimate (hbox {PM}_{10}) at these extreme values (see “Methods” section). However, the observed concentrations are of the same order of magnitude as for previous measurements at Etna’s summit20,21. At similar passive degassing conditions, Allen et al.20 observed mass concentrations between (sim 😉 1,400 and (200,upmu hbox {g/m}^3), in the coarse mode (diameter (hbox {D}_{p} > 3.5,upmu hbox {m})), and between (sim 150) and 4,500(,upmu hbox {g/m}^3), in the fine mode ((hbox {D}_{p} < 3.5,upmu hbox {m})), at varying distances from the near-summit to (sim ;10,hbox {km}) distance. Similarly, very high concentrations have been observed for other volcanoes, like the Popocatepetl volcano (total suspended particles (sim ;1,440,upmu hbox {g/m}^3))22 or the Soufrière Hills ((hbox {PM}_{10} = 100)(500,upmu hbox {g/m}^3))23. Our observations confirm (and exceed) extremely large PM values for Etna’s summit and for similar volcanoes. These extremely big PM sources (including the most dangerous small-sized particle matter) have been associated in the past to degraded air quality and connected pulmonary diseases of the neighbouring people22,23,24. These health risks are still to be systematically assessed for Etna area, including for the persistent passive degassing activity.

One can note that the (hbox {PM}_{1}) and (hbox {PM}_{10}) hot-spots are not localized exactly in the same plume section. Therefore, the sources of fine and coarse particles may be different. Moreover, the column integrated (hbox {SO}_{2}) concentrations were simultaneously measured along the same path; time series of (hbox {PM}_{1}) and (hbox {SO}_{2}) concentrations are shown in Fig. 1c,d. The (hbox {SO}_2) measurements show very limited correlations with both (hbox {PM}_1) and (hbox {PM}_{10}) concentrations (see a scatterplot of (hbox {PM}_1) and (hbox {SO}_2) concentrations in Fig. 1e). This evidence contrasts with previous evidences of a strong correlation of total particle column and (hbox {SO}_2) et Etna near-summit15. Even if during this measurement session the volcanic plume was mostly flattened to the ground due to the moderate wind at the summit, it has to be mentioned that part of the plume can be missed from the PM observations because it is located at higher altitudes. While the wind was not extreme, during the session, it is still possible that some of the coarser aerosol observations are moderately affected by surface particles re-mobilisation. Due to these limitations and the lack of complementary observations of atmospheric oxidants or ultra-fine particles, the precise origin of the observed aerosol is not possible. Nevertheless, these results do not show clear evidence of formation of secondary aerosols. Primary aerosol emissions were found dominating, with respect to secondary aerosol formation, at proximal locations, for other passive degassing plumes like for Masaya volcano in the past25.

Figure 1

Fine ((hbox {PM}_1), a) and coarse ((hbox {PM}_{10}), b) particle matter concentration measured with OPC near the central BNC and VOC craters (approximate craters position indicated in the maps) during a double plume traverse on 20/07/2017. The background maps of panels a and b are generated with Google Maps (, Map data (copyright 2020) Google). Fine particle matter concentration (c) and (hbox {SO}_2) time series (d) and scatter plot of simultaneous observations (e).

Distal and itinerant observations: small-scale variability

Getting new insights into the volcanic aerosol plume at larger proximal scale, i.e. under (sim 20,hbox {km}) from the degassing craters, is a further target of EPL-RADIO. Figure 2a shows a summary map of the MIIOM observations of the volcanic aerosols plume’s optical properties during C2 campaign. Aerosol optical depths (AOD) for Etna’s plume were derived at 320.5 nm (hereafter referred to as UV AOD) and 1,020.0 nm (hereafter referred to as NIR AOD), with the method described by Sellitto et al.19. Using UV and NIR AODs, the Ångström exponent (AE) and the Ångström (beta) coefficient are subsequently derived19. In general, shorter-wavelength AODs carry more information on the presence of smaller particles than longer-wavelength AODs26. The spectral variability of the AOD (and then of the extinction of radiation due to the presence of the aerosol) can be represented using the empirical Ångström law, as a function of AE and (beta):

$$begin{aligned} mathrm {AOD}(lambda )=beta lambda ^{-mathrm {AE}} end{aligned}$$


Starting from the fundamental theory of particles-radiation interaction (Mie theory26), AE and (beta) can be considered as optical proxies for particles mean size and burden/composition, in the aerosol layer. Bigger AEs are linked to smaller particles, on average, in the sampled plume. While this relationship of AE and average particle size is generally true, precise studies on the average particle size evolution by means of AE observations may be complicated by the possible evolution of the overall particle size distribution, and in particular of the number of its modes. Log-normal size distributions with two20 or three modes15 have been found at near-crater locations for passive degassing volcanoes. On the contrary, mono-modal size distributions, in the accumulation mode (typical mean particles size around 0.1–(0.5,upmu hbox {m})), have been found at distal locations27 . The (beta) parameter is representative of the AOD value at wavelength of (1.0,upmu hbox {m}). For the particle types in volcanic plumes and the typical size distributions discussed above, (beta) is representative of the coarse particles (mainly ash) in the plume. Thus, while its interpretation is more complex than for the AE, in our present study (beta) can be linked to the burden of the coarse aerosol in the layer, bigger (beta)s being representative of bigger burdens. Despite the intrinsic complexity and uncertainty in their interpretation, the two Ångström parameters are widely recognised as diagnostic tools for volcanic plumes morphological and microphysical properties and their inherent evolution processes28. The average values of AE and (beta) are estimated for each measurement session at each location, during C2, as listed in Table 2 and are shown on the map. Thus, Fig. 2a represents the small-scale variability and evolution with distance of the average size and coarse aerosol burden of a volcanic plume. During C2 campaign a clear image of progressively more dominant smaller particles, associated with a smaller burden of coarse particles, is apparent from crater to increasing distances. At the same time, the UV AOD has a different behavior during both campaigns C2 and C3, with slightly increasing values in the distal field, thus indicating a possible slight increase of the finer aerosol content in the plume. This result is consistent with previous airborne remote observations of Spinetti and Buongiorno16, that have shown a similar increasing trend, for increasing distance from the crater, of the AE and a near-constant to slightly increasing trend of the distal (starting from a few km downwind the craters) short-wavelength AOD (at about 400 nm), during passive degassing activity of Mount Etna volcano. A substantial variability of the plume properties is found at spatial scales that are smaller than typical grid-points of chemistry/transport models and pixels of satellite observations. This reveals that processes are at play which are not represented with current regional scale modelling and observations. To get further insights into the small-scale aerosol variability, in Fig. 2b,c we show the MIIOM AE (C2 campaign) and MIISP AE (C3 campaign), as a function of the distance from craters and of simultaneous measurements of (beta). The error bars in Fig. 2b,c represent the standard deviation of the average AE and (beta), thus representing their variability for each measurement session/location. Results shown in Fig. 2b,c confirm the picture drawn before for C2 campaign: progressively smaller particles and decreasing coarse aerosol burden are found along plume dispersion in the first (sim)10 km around Etna’s degassing craters. The correlation of the occurrence of smaller particles with increasing distance and with thinner plumes is significant (R(^2) correlation coefficients 0.90 and 0.71, respectively). This can be attributed to relatively quick sedimentation of coarser ash particles (bigger than a few micrometers) emitted by discontinuous ash puffs (observed during C3 campaign and on 20/07/2016 during C2 campaign) and, possibly, by sustained gas-to-particle conversion of (hbox {SO}_2) volcanic emissions to secondary SA and subsequent grow by condensation (up to a few hundreds nanometers). These two processes are expected to concur to the progressive modification of the size distribution towards smaller average sizes. During C2 campaign, volcanic activity was based on a prevalent passive degassing and (hbox {SO}_2) emission rates were significant yet very variable (average values of 2,240.5 ± 882.5 t/day, observed by the FLAME network). Nevertheless, the same picture is not found for C3 campaign. Even if smaller particles sizes and coarse aerosol burdens are found with increasing distance from the emitting craters, the regression lines have less pronounced slopes and much smaller correlations (0.21 and 0.33) than for C2 campaign. This may be linked to the larger amount of emitted coarse ash particles and/or a smaller signal of gas-to-particles conversion to SA. For C3 campaign, sporadic explosive activity is found, with significantly larger ash emissions and with smaller (hbox {SO}_2) emission rates (average values of 1,315.9 ± 254.2 t/day), which can be limiting factors to secondary SA in-plume production. A similar negative correlation between the volcanic aerosol amount and average size, stronger for ash-free that ash-bearing plumes, as shown in Fig. 2c, was found during past field campaigns at different volcanoes, including Mount Etna28,29, and attributed to ash sedimentation and in-plume SA formation.

Simultaneous side-by-side measurements of MIIOM and MIISP volcanic AODs, and UVS (hbox {SO}_2) column amounts, were carried out during both C2 and C3 campaigns, and are compared in Fig. 3. The UV AOD correlates much better than the NIR AOD to UVS (hbox {SO}_2) estimations (Fig. 3a,b). In general, the correlation decreases with wavelength and is only significant in the UV-to-shorter-visible range (Fig. 3c). As shorter-wavelengths AOD values are more strongly correlated with the presence of small-sized particles in the aerosol layer than longer-wavelengths AOD, and (hbox {SO}_2) is the main precursor of the tiny secondary SA, this is a strong indication of (hbox {SO}_2)-to-SA conversion processes at play. In addition, UV and NIR AODs correlate both better to (hbox {SO}_2) during C2 than C3 campaign. This gives a further indication that possible (hbox {SO}_2)-to-SA conversion is more marked during volcanic passive degassing conditions, with stronger (hbox {SO}_2) emissions, than in mildly explosive conditions, with weaker (hbox {SO}_2) emissions. New SA particle formation has been recently identified for Etna’s degassing plume, at larger spatial scales, by Sahyoun et al.30. It is important to mention that the presence of in-plume and background atmospheric aerosols in the line-of-sight of the UVS spectrometer might have an impact leading to either over- or under-estimating (hbox {SO}_2) column amounts measurements, by multiple scattering and effective path lenght reduction due to light dilution, respectively31. These effects are expected to be negligible for the observations carried out in this study. Multiple scattering by in-plume aerosols is expected to be important for aerosol extinctions significantly larger than those observed in our study. The light dilution effect depends on the background atmosphere and is not related to the plume AOD. In addition, for vertical observation geometry, this effect is largely spectrally independent. As we took care to carry out observations at as much as possible vertical geometry, this has a negligible effect on the results discussed above.

Figure 2

Summary of MII optical properties observations of Etna’s plume during C2 campaign (18–22/07/2016, a): AE (proportional to the size of the markers), (beta) (white–grey–black shade of the markers), UV AOD (height of the column during plume traverse). The different experiments are also individuated in (a): plume half-traverse and the two plume transects scans (TS1 and TS2). The location of the LiDAR station is indicated with a blue cross. The locations of FLAME stations are indicated with red triangles. The background map is generated with Google Maps (, Map data (copyright 2020) Google). AE as a function of distance (b) and of (beta) coefficient (c), for MIIOM measurements collected during C2 campaign (in red) and for MIISP measurements collected during C3 campaign (blue). Individual points and error bars are associated to average values and variability for observations sessions at fixed locations (see Table 1). Regression lines are also reported (red and blue dashed lines) with their Pearson (hbox {R}^2) coefficients.

Figure 3

Scatter plot of MIIOM UV (a) and NIR (b) AODs and UVS (hbox {SO}_2) simultaneous observations, for C2 [indigo (a), and yellow (b)] and C3 campaign [mauve (a), and orange (b)]. Regression lines and respective Pearson correlation coefficients R(^2) are also provided for each campaign and wavelength. Pearson correlation coefficients R(^2) for correlations of simultaneous and co-located UVS (hbox {SO}_2) with respect to MIIOM/MIISP AODs observations at different wavelengths (c), for C2 (blue bars) and C3 (green bars).

Besides a general short-term depiction of the geographical near-source variability of volcanic AOD for Etna, in Fig. 2a different experiments are identified: two plume transects (TS1 and TS2, see details in Table 2), i.e. rapid longitudinal plume scans, and a plume half-traverse, i.e. a rapid perpendicular plume scan from the plume centre to one periphery. For the plume half-traverse, additional indications of the measured UV AODs are given in the map. The dramatic change in average size and coarse aerosol burden of the plume during the transects (details in Table 2), in such a short time and small distance, can be readily attributed to the sedimentation of possible coarser ash particles and the rapid formation of secondary SA. Coarser particles are found during TS2 than TS1, due to the visible ash puffs during this measurement session. Some residual relatively-coarse ash particles might be present at distal TS2 location due to these more pronounced ash emissions, which are unlikely present during TS1-distal observation ((hbox {AE} = 1.90pm 0.67), typical of extremely small particles). The larger relative uncertainties at proximal ((sim 95%), TS1, and (sim 60%), TS2) than distal UV AOD ((sim 45%), TS1, and (sim 35%), TS2), see Table 2, point at more inhomogeneous plumes near the sources. This plumes tend to get more and more temporally homogeneous with increasing distance, as different particles types (like fine ash) are removed from the plume along dispersion. Larger variabilities of volcanic plumes’ optical properties have been associated with puffiness of the plume and to the presence of ash29. The short-term atmospheric processes smooth the plume, in terms of its properties. The plume half-traverse revealed the perpendicular structure of Etna’s plume, during a typical passive degassing situation. The burden is maximum at the plume core (UV AOD (sim ;0.2)) and steadily decreases towards the plume periphery (UV AOD (sim ;0.1), i.e. half the burden than at the plume core). Correspondingly, the particle mean size decreases from the core to the periphery. New SA particles formation is expected to be more effective at the plume periphery, i.e. in presence of lower concentrations of pre-existing aerosols that can act as condensational sink30.

Table 2 Details about TS1 and TS2 MIIOM transect scans during C2 campaign. Date, approximate time, observation session location, UV AOD, AE are reported.

Distal and itinerant observations: vertical distribution

The vertical distributions of volcanic aerosols is further studied using LiDAR observations from the fixed SLN station. Figure 4 shows typical aerosol vertical structures at passive degassing conditions (19/07/2016). Back-trajectories analyses (not shown here) show that this profile is mostly affected by local air masses, at all altitudes. No larger scale features, like desert dust transport events, are observed at all altitude ranges. Zenith- and crater-pointing observations allow a detailed three-dimensional characterisation of the passive degassing plume. These observations reveal an aerosol layer at low altitude over the station (from near-ground to (sim) 2 km, aerosol signal peaking at (sim) 1 km, zenith-pointing observations), possibly covering the whole line of sight, from station to crater (crater-pointing observations, aerosol signal over the whole line-of-sight and peaking at (sim) 5–7 km, i.e. at and near the crater itself). The core of the volcanic degassing plume is characterised by weakly depolarising aerosols ((sim 3) and 1%, for zenith- and crater-pointing observations, respectively), thus indicating the predominant presence of spherical droplets, e.g. SA. The weak signal-to-noise ratio after plume crossing, i.e. visible in the average crater-pointing observations, may point at partially absorbing particles near the crater area, possibly SA embedding a more absorbing fine ash core.

Figure 4

Time series of LiDAR observations from SLN station (blue cross in Fig. 2, 19/07/2016) of range-corrected signal (a) and depolarisation ratio (b) at 532 nm, for zenith-pointing (as a function of the altitude) and crater-pointing (as a function of the distance from station) geometries. For both geometries, the average backscattering coefficient vertical profile at 532 nm is also shown (c, d), along with the average depolarisation ratio for volcanic-((delta _{mathrm {v}})) and non-volcanic-affected ((delta _{mathrm {nv}})) identified vertical ranges. For crater-pointing observations, two average profiles and mean depolarisations are shown for different time intervals (08:42–09:12 local time, in blue, and 10:15–10:45 local time, in red).

On 19/07/2016, quasi-simultaneous MIIOM and LiDAR zenith observations have been realised at SLN station. This allows the direct comparison of the AODs obtained with the two instruments. The MIIOM UV AOD (at 320.5 nm) is (0.14pm 0.06), for the volcanic plume. Using a LiDAR ratio (LR) of 48 sr, as done previously, for volcanic plumes, at the same station32, we obtain a LiDAR VIS AOD (at 532 nm) of 0.06. This value, scaled using the quasi-simultaneous AE estimations made with MIIOM at SLN, produces a full-column LiDAR UV AOD (at 320.5 nm) of 0.16. While a limited impact of other local aerosol sources at the very lowest altitudes cannot be excluded, we attribute to the volcanic plume the aerosols observed up to 2 km over SLN station and obtain a LiDAR UV AOD of 0.12, for the volcanic plume. The consistency of quasi-simultaneous MIIOM photometric and LiDAR UV AODs, for the volcanic plume, validates our choice of LR and the LiDAR observations themselves.

Radiative transfer modelling: direct impact on the regional climate

The near-source impact of the volcanic aerosol plume on the shortwave radiative balance is then estimated using the UVSPEC radiative transfer model. The measured average volcanic aerosols extinction profile of the AMPLE LiDAR at SLN station (19/07/2016), and validated by co-located photometric observations, is used as a representative near-source plume at typical passive volcanic degassing conditions. Even though here we use precise vertically-resolved estimations of the aerosol extinction of Etna’s plume, hypotheses are necessary for other optical properties not accessible from the synergy of LiDAR and photometer measurements. The absorption and scattering properties of the aerosol layer is based on the hypothesis of a predominance of tiny, highly-reflective SA10. The single scattering albedo (SSA), an optical proxy of the aerosol absorption, is set at typical values for high-reflective sulphate aerosols. Values around 1.00 have been reported for these particles in the shortwave spectral range33. As the uncertainty on this assumption is relatively high—the plume can contain a fraction of more absorbing ash particles, even if this is unlikely (see previous discussion on LiDAR and MIIOM observations, and volcanic activity type)—four groups of simulations are performed using SSA ranging from 0.97 to 1.00, with 0.01 SSA increments. The angular distribution of the plume-scattered radiation, can be modeled by scattering moments based on Heyney-Greenstein functions, corresponding to a given asymmetry parameter g (the intensity-weighted average of the cosine of the scattering angle, that can be obtained with the Mie theory34). In the shortwave spectral range, considering the expected mean size of the plume’s particles, a typical value of the asymmetry parameter, for very weakly-absorbing particles like freshly nucleated/condensed SA, is 0.5035. This value has been used as a reference for our simulations. Nevertheless, the uncertainty on this assumption can be large and so simulations for two additional values of g, 0.70 and 0.85, are performed. Based on these considerations, 12 simulations are performed, with four values of SSA and three values of g. It must be noted that the spectral variability of both SSA and g are not taken into account in this study. For all simulations, the surface reflectivity, which is an important parameter determining the local shortwave radiative balance, is set at 0.1 (wavelength-independent), which is a typical value of vegetated surfaces, like in the volcano near-surroundings. Figure 5a shows the clear-sky equinox-equivalent daily average top of the atmosphere (TOA) and surface radiative forcing and their ratio (called f ratio) as a function of the SSA, and their SSA-averaged values, for a 0.50 asymmetry parameter g. The TOA radiative forcing of the passive degassing plume is near-independent on the absorption properties of the volcanic aerosols, and has an average value of (sim ;-4.5) W/m(^2), thus indicating a consistent cooling of the local climate system. The surface radiative forcing of the passive degassing plume is more dependent on the absorption properties of the volcanic aerosols, and has an average value of (sim ;-7) W/m(^2). Correspondingly, the f ratio depends quite strongly on the absorption properties of the plume and has an average value of (sim) 1.45. The average TOA and surface radiative forcing and f ratio, as a function of the SSA, for simulations with the three values of g, are shown in Fig 5b. The same radiative forcing parameters, as a function of g, for simulations with the four values SSA, are shown in Fig 5c. From the comparison of these two latter panels, it can be seen that the parameter that brings the largest uncertainty is the asymmetry parameter g. A larger cooling effect is found for smaller values of g, which means smaller particles and then larger scattering back to space. For larger particles, values of f ratio up to near 3.0 are found, which means a consistent energy stuck into the atmosphere, that can lead to local heating of plume-occupied air masses. The effect of SSA is less strong but a clear trend of larger values of the surface radiative forcing and f ratio for smaller SSA, which means more absorbing particles, is found. The TOA radiative forcing is near-independent on the SSA. For more absorbing particles, values of f ratio up to about 2.0 are found.

These estimations of the radiative forcing can be compared with more distal estimations of the radiative forcing for Etna’s emissions during explosive eruptions. The radiative forcing efficiency (radiative forcing per AOD unit) has been estimated, at TOA and surface: (1) for a SA-dominated plume, at (sim 300,hbox {km}) downwind Etna for a moderate eruption (25/10/2013), at (sim) 40–50 and 50–(70,hbox {W}/(hbox {m}^2hbox {AOD}))11, and (2) for an ash-containing plume, at (sim 400,hbox {km}) downwind Etna for a relatively strong eruption (03/12/2015), at (sim ;112) and (145,hbox {W}/(hbox {m}^2hbox {AOD}))36. These values, scaled at the AOD observed in this case, lead to the forcings, at TOA and surface, of (sim ;-5)–6 and (-6)(8,hbox {W/m}^2), for the SA-dominated, and (sim) (-13) and (-17,hbox {W/m}^2), for the ash-dominated plume. Our estimations, for a proximal passive degassing plume, are more in the magnitude range of purely SA plumes. The proximal radiative forcing efficiency of Mount Etna’s passive degassing plume is comparable to the very distal (some hundreds km) efficiency of explosive SA-dominated plumes, though with a factor ten smaller AOD37.

Figure 5

Equinox-equivalent daily average TOA (blue dots) and surface (blue squares) radiative forcing, and their f ratio (red crosses), as a function of the single scattering albedo assumptions, for asymmetry parameter (hbox {g}=0.50). Their mean values (by averaging all estimations with the different single scattering albedos, for (hbox {g}=0.50)) is also shown with 1-standard-deviation error bars (filled blue dot, blue square and red cross) (a). Equinox-equivalent daily average TOA and surface radiative forcing and f ratio, averaged over all values of g and as a function of SSA (b), and averaged over all values of SSA and as a function of g (c).

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *