1. The University of Reading Helen Dacre The Eyjafjallajökull eruption: How well were the volcanic ash clouds predicted? Helen Dacre and Alan Grant Robin Hogan, Dave Thomson, Ben Devenish, Jim Haywood, Franco Marenco, Ben Johnson, Albert Ansmann, Ina Mattis and Lieven Clarisse

2. The University of Reading Helen Dacre Motivation

3. The University of Reading Helen Dacre EUROCONTROL report from 14 - 20 April: 75% of European airspace closed 100,000 flights cancelled 10 million passenger journeys affected 7000 flights cancelled up to 18 May Motivation LevelConcentration (mg/m 3 ) High> 4 Medium0.2 - 2 Low< 0.2

4. The University of Reading Helen Dacre Operational volcanic ash transport and dispersion (VATD) models Quantitative model predictions  Source parameter uncertainty  Meteorological input uncertainty Future Work Talk Outline

5. The University of Reading Helen Dacre Operational VATD Modelling Meteorology (Every 3 hours) 40 km resolution Global UM Eruption Source Parameters Height of eruption plume Vertical distribution of emitted mass Rate at which ash is emitted Ash size distribution Transport 3D wind fields Turbulent mixing Sedimentation Wet and dry deposition Ash concentrations (6-hourly averages) FL000-200 FL200-350 FL350-550 INPUT MODELOUTPUT

6. The University of Reading Helen Dacre Eyjafjallajökull Source Parameters H

7. The University of Reading Helen Dacre Column Integrated Mass Concentration 14-18 th April

8. The University of Reading Helen Dacre Fine-ash Fraction? Mastin et al. (2009)

9. COMPARISON WITH GROUND BASED LIDAR

10. The University of Reading Helen Dacre IASI Volcanic Ash Product 12UTC 16 th April 00UTC 16 th April Leipzig 10UTC 16 th April22UTC 15 th April L. Clarisse Leipzig

11. The University of Reading Helen Dacre 1.5% Scaling to Observed Concentration at Leipzig A. Ansmann I. Mattis

12. The University of Reading Helen Dacre MODIS Visible Image 12UTC 16 th April 12:24UTC 16 th April10:44UTC 16 th April Chilbolton (Hogan et al. 2011)

13. The University of Reading Helen Dacre Scaling to Observed Concentration at Chilbolton (Hogan et al. 2011)

14. The University of Reading Helen Dacre 5-minute time series of plume height from the Icelandic radar (data from Petersen and Arason) Eruption Plume Height Data Missing scan Cloud obscured Mountain obscured

15. The University of Reading Helen Dacre Plume Height Reconstruction 4%3%

16. COMPARISON WITH AIRCRAFT LIDAR

17. The University of Reading Helen Dacre

18. The University of Reading Helen Dacre 5 th May

19. The University of Reading Helen Dacre 14 th May

20. The University of Reading Helen Dacre

21. The University of Reading Helen Dacre 17 th 1.6% 14 th 1.3 - 2.5%5 th 7.5%

22. The University of Reading Helen Dacre Peak Concentration and Layer Width 2% 5% Ash layer width = integrated column mass/max concentraton

23. COMPARISON WITH AIRBORNE SPECTROMETERS

24. The University of Reading Helen Dacre Fine Ash Particle Size Distribution 14 th 2.1%

25. The University of Reading Helen Dacre Fine Ash Particle Size Distribution 5 th 10.6%14 th 2.1%17 th 3.1%

26. The University of Reading Helen Dacre Fine Ash Fraction 3.5%

27. The University of Reading Helen Dacre It is possible to identify the ash layers detected with the ground based and airborne lidars with layers in the NAME simulations Observed ash layers are thinner than teh simulated layers and at lower altitude Horizontal and vertical structure of the simulated ash clouds are sensitive to assumptions about the profile of the ash emissions – no best profile but for weak activity a uniform profile may be best but for greater activity a concentrated profile better Quantitative comparison suggests that only about 3.5% of the erupted mass was in ash particles small enough to allow long-range transport It is necessary to represent the large, short-term fluctuations in plume height accurately Summary

28. The University of Reading Helen Dacre NAME did a reasonable job of capturing the horizontal structure of the ash cloud subject to possible timing and positioning errors that occur due to meteorology NAME underestimates maximum concentrations by a factor of about 2.5 OR NAME overestimates layer with by a factor of 2.5 Default particle size distribution in NAME contains too many 10-30µm diameter particles Summary

29. The University of Reading Helen Dacre Operational volcanic ash dispersion modelling Model input uncertainty  Eruption plume height, vertical distribution  Peak concentrations  Fine ash fraction  Particle size distribution Model/observation comparisons  Satellites  Lidars (ground and aircraft based)  In-situ particle measurements Will we do better next time? Future work Talk Outline

30. The University of Reading Helen Dacre NAME dispersion model Input  Eruption location  Eruption start time and duration  Eruption height, vertical distribution  Eruption rate (fine ash fraction)  Particle size distribution, density  Sedimentation velocity  Meteorology Output  Ash concentration  Mean travel time Operational Volcanic Ash Modelling

31. The University of Reading Helen Dacre Synoptic Analysis at 00UTC on 16 th April

32. The University of Reading Helen Dacre Modis AQUA visible image at 13:23 UTC 12UTC 16 th April

33. The University of Reading Helen Dacre IASI Measured Volcanic Ash 22UTC 14 th April 10UTC 16 th April 10UTC 15 th April 22UTC 15 th April

34. The University of Reading Helen Dacre

35. The University of Reading Helen Dacre