92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Wednesday, 25 January 2012: 9:15 AM
Airborne in-Situ and Lidar Observations of Volcanic Ash During the 2010 Eruption of Eyjafjallajökull
Room 357 (New Orleans Convention Center )
Ben Thomas Johnson, UK Met Office, Exeter, United Kingdom; and F. Marenco, K. Turbull, J. Haywood, A. J. Baran, R. Cotton, P. Brown, H. Webster, Z. Ulanowski, R. Burgess, P. Rosenburg, A. Wooley, E. Heese, and J. Dorsey

Poster PDF (6.9 MB)

During April-May 2010 the Facility for Airborne Atmospheric Measurements (FAAM) BAe-146 aircraft made a series of 12 flights targeting volcanic ash clouds around the UK from the eruption of Icelandic volcano Eyjafjallajökull. The spatial extent and vertical distribution of ash clouds were surveyed using an on-board Leosphere 355nm lidar. In-situ measurements, including an Cloud and Aerosol Spectrometer (CAS) were used to determine ash particle size distributions, optical properties and mass concentrations. The specific extinction coefficient derived from the in-situ measurements were then used to estimate ash mass concentration profiles and column loadings from lidar extinction retrievals. The in-situ and lidar-based estimates of ash mass have has been used to validate ash forecasts from the Met Office dispersion model (NAME), and constrain remote sensing retrievals from both ground-based and satellite sensors.

Ash clouds were observed over the UK region between altitudes of 2 – 8km, usually in vertically thin (0.5 – 2km deep) but horizontally extensive layers (100 – 850km). The lidar showed these layers to be very patchy and inhomogeneous with features on horizontal scales of 20 – 100km. The lidar and in-situ data from a Cloud and Aerosol Spectrometer showed peak ash concentrations of 200 – 2000 µg/m3 during a series of seven flights from 4 – 18 May. This was in part a consequence of the safety policy that led to the targeting of regions forecast to have concentrations of 200 – 2000 µg/m3. The lidar showed similar peak concentrations to the CAS, although on some occasions the lidar was able to remotely sense ash layers (from above) that, for safety reasons, the aircraft was not able to penetrate. Our analysis shows that ash mass concentrations derived either from the lidar or from the in-situ probes are subject to a factor of 2 uncertainty due to the sensitivity of optical measurements to assumptions regarding particle shape, refractive index and density. The mixing of ash with ice clouds and other aerosol adds addition uncertainty. An exceptional mass concentration of 5000µg/m-3 was observed over Scotland on 14 May, although this may be an overestimate due to contamination by ice. Whilst NAME predicted the appropriate magnitude ash concentrations on most occasions, errors in the timing and geographic spread of ash clouds were also apparent on some occasions. These errors have been traced to many uncertainties, including the volcanic source strength, the vertical profile of ash release, the initial particle size distribution and the meteorological conditions.

Aerosol size distributions within ash clouds showed a fine mode (0.1 – 0.6 µm) associated with sulphuric acid and/or sulphate, and a coarse mode (0.6 – 35µm) associated with ash. The ash mass was dominated by particles in the size range 1 – 10µm (volume-equivalent diameter), peaking at around 4um. The maximum observed diameter varied between 20µm and 35µm depending on distance downwind from the volcano. The dominance of 1 - 10um particles was confirmed by electron-microscope analysis of aerosol filter samples. The rapid fall in particle sizes above 10µm is not well represented by dispersion models when initiated with ash size distributions based on near-source deposits. This implies that different methods may be required to back-out the required source size distribution, and that a greater understanding of near-source aggregation and fall-out is needed.

Electron-microscope images and scattering patterns from the SID2-H (Small Ice Detector) probe showed a high degree of irregularity and variability in the shape of the ash particles. Therefore the CAS and lidar data was processed assuming a fractal polyhedral model for irregular particle shapes. This reduces the derived mass from in-situ and lidar measurements by 20 - 40% compared to the assumption of spheres. The derived mass was also sensitive to the refractive index. Increasing the refractive index from 1.52 + 0.0015i (based on mineral dust) to an upper value of 1.59 + 0.004i (based on in-situ estimates of Eyjafjallajökull ash from the DLR Falcon aircraft; Schumann et al., 2011, ACP) led to ~30% increases in ash mass. The 3-wavelength nephelometer indicated high aerosol scattering coefficients (up to 500 x 10-6 m-1) and low Ångstrom exponents (0.0 +/- 0.3) across visible wavelengths (450 – 700nm). The ratio of optical extinction to mass, i.e. the specific extinction coefficient (kext), was estimated to be around 0.6m2/g at visible wavelengths based on the typical CAS size distribution. This parameter has been used to derived ash mass concentrations from lidars (both on-board and ground-based), and column loadings from remote sensing retrievals of aerosol optical depth. Again, a factor of two uncertainty applies due to the uncertainties in particle properties. Our observations also show elevated levels of SO2 (up to 100ppbv) in all ash clouds. Although the ratio of SO2 to ash mass was highly variable (30 to 250ppbv per 1000µg/m3) SO2 is thus shown to be useful for discriminating between volcanic and non-volcanic aerosol layers.

Supplementary URL: