INTERNATIONAL VOLCANIC ASH TASK FORCE (IVATF)

Transcription

1 IVATF/1-IP/19 22/7/10 INTERNATIONAL VOLCANIC ASH TASK FORCE (IVATF) FIRST MEETING Montréal, 27 to 30 July 2010 Agenda Item 3: Results of the EUR/NAT VATF Meeting (Plenary) 3.1: Evaluation of the Eyjafjallajokul eruption and lesson learned SUMMARY OF UK S RE-ANALYSIS OF PAST VOLCANIC ASH AIRCRAFT RELATED EVENTS USING VAAC LONDON NAME MODEL (Presented by United Kingdom) SUMMARY This paper provides a short summary of some work carried out by the Met Office in the UK at the request of Rolls Royce and Flybe to re-analyse some past volcanic ash aircraft related events using the NAME Model in order to identify the ash mass concentrations that have the potential to cause problems for jet engines. This is a piece of work that has been undertaken to assist in identifying suitable thresholds for ash mass concentrations that could provide improved guidance to aviation authorities and airlines during eruptions. 1. INTRODUCTION 1.1 Following the explosive eruption of the Eyjafjallajokull volcano in Iceland on 14 th April 2010 the Met Office in the UK came under significant pressure from the airlines and engine manufacturers to: a) validate the NAME dispersion model used by VAAC London to predict the dispersion of ash from the volcano, and b) provide advice to engine manufacturers about the mass concentrations of ash in the atmosphere that can cause jet engines problems. 1.2 One of the ways that the Met Office in the UK was requested to do this by Rolls Royce was to re-analyse previous volcanic eruptions that had caused aircraft specific problems. The incidents that were re-analysed were: (10 pages) IVATF.1.IP en.doc

2 IVATF/1-IP/ a) Galunggung eruption (Java) the eruption in June 1982 led to an all engine failure on a BA Boeing 747; b) Redoubt eruption (Alaska) the eruption in December 1989 led to power loss in all 4 engines of a KLM Boeing (damage to the aircraft was estimated to be $80 million); c) Soputan the eruption in May 1985 led to all 4 engines of a Qantas Airways Boeing 747 being replaced; d) Hekla (Iceland) the eruption in February 2000 led to a NASA DC-8 research aircraft suffering engine damage; e) Kasatochi (Aleutian Islands) the eruption in August 2008 led to a plume of volcanic ash across North America and disruption to flights in Alaska; f) Manam (Papua New Guinea) the eruption in July 2006 led to a twin engine flame out on a Gulfstream; and g) Pinatubo (Philippines) the eruption in June 1991 led to 16 aircraft encounters with volcanic ash which caused 10 engines to be damaged and replaced. 1.3 Flybe also asked the Met Office in the UK to look at an eruption of Sakurajima (Kyushu, Japan) on 14 May when flights were occurring in the area. 2. DISCUSSION 2.1 In all of the cases reanalysed, except the Soputan case where there is some uncertainty about the actual flight track, the NAME model was able to give good guidance about why the aircraft in question encountered ash and in the Flybe case why the aircraft did not encounter ash. 2.2 The re-analyse for each of the eruptions has to make some major assumptions about the source strength of the volcanoes and the height to which ash is ejected. This has the potential to lead to significant errors in the prediction of the position of the dispersing plume of ash and the mass concentrations in the plume. When there is little or no quality information about the source strength of the plume the re-analysis has had to use a number of different emission scenarios to find the one that best matches the aircraft observations. 2.3 A summary of 4 of these cases are presented here. For more information on these or the other events analysed contact VAAC London. The following sections give more detail on the Galunggung, Redoubt, Hekla, and Manam cases respectively. 2.4 Galunggung June Intermittent explosions began on 5 April The ten eruptive episodes that had occurred as of late June were separated by quiescent periods that ranged from three days (early in the eruption) to three weeks (before the June explosions). In total more than 30 large explosive eruptions occurred between April 1982 and January 1983, including ash clouds that rose more than 10 km above sea level. The June explosions dropped 8-18 cm of ash and lapilli on villages 7-10 km W of

3 - 3 - IVATF/1-IP/19 the volcano, destroying hundreds of houses. Indonesian newspapers reported that people a few km from the crater heard thundering sounds and saw glow over the volcano before the explosions began. Residents of Tasikmalaya (17 km ESE of the volcano) saw incandescent tephra ejection at 19:00 LT, 19:10 LT, and 19:30 LT on 24 th June Light ashfall began about midnight in Bandung, stopping by morning, but the city remained in semi-darkness into the afternoon. Images from the Japanese GMS and the NOAA 7 polar orbiter showed that the eruption cloud moved W then curved toward the S with its distal end reaching a point roughly 850 km S of Galunggung At 20:50 LT on 24 th June 1982, a British Airways jumbo jet with 240 persons on board, flying roughly 150 km WSW of Galunggung at ~11 km altitude, encountered an ash cloud that stalled all four of its engines and abraded its windshield and wing surfaces. St. Elmo's fire was noted in the cockpit as the jet flew through the ash cloud. The aircraft lost 7.5 km of altitude before the engines could be restarted, but it landed safely in Jakarta. (Source: Smithsonian Institute Global Volcanism Program Information on the start and stop times of the eruptive phase beginning on 24 June 1982 is inconsistent and limited. Hanstrum and Watson (1982) state that information from Indonesia indicated that the eruptions occurred at about 15:00 UTC which is inconsistent with the reported time of the aircraft encounter with the volcanic ash plume (13:50 UTC). However, they also report that the aircraft encounter occurred at approximated 15:00 UTC which is roughly an hour later than reported elsewhere. Satellite imagery at 18:00 UTC from the Japanese Geostationary Satellite GMS-2 (see Fig. 2, Hanstrum and Watson, 1983) shows the volcanic ash plume from the Mount Galunggung eruption suggests the volcano was still erupting at this time. The ash plume is not visible on satellite imagery at 12:00 UTC. Residents report, however, seeing emissions of ash at 19:00 LT, 19:10 LT and 19:30 LT (12:00 UTC, 12:10 UTC and 12:30 UTC) (see Smithsonian Institution Global Volcanism Program) Precise location information (i.e. latitude and longitude) of the aircraft encounter with the volcanic ash plume has not been found in the literature. Reported locations are 130 miles SE of Jakarta (Rolls Royce, private communication), 150 km WSW of Galunggung (Smithsonian Institution Global Volcanism Program km south of Jakarta (Johnson and Casadevall, 1994) and 230 km south of Jakarta (Hanstrom and Watson, 1983). All four reported locations (labelled Encounter 1 through to Encounter 4, respectively) are shown in figure 1. It can be seen that the distance between the reported locations is large relative to the distance from the volcano. The reported aircraft location given by Rolls Royce (Encounter 1) is situated very close to the volcano. A flight path taking the aircraft SSE down through Sumatra and across the western edge of Java is shown in Hanstrum and Watson (Fig. 1, 1983) and, not surprisingly, agrees well with the location of Encounter 4 (Hanstrum and Watson, 1983). The aircraft s altitude is consistently reported to have been ft The emission rate was determined using the VAFTAD table. Eruption heights between 12 km and km correspond to a VAFTAD threshold of Consequently the source term in NAME has been increased by a factor of 10 from the notional 1 g / 6 hr to ensure that the red-grey-black thresholds currently used in the European region correspond to 1 x 10-17, 1 x and 2 x contours on the graphic in Figure 1. Similarly for eruption heights between km and km, the source term in NAME has been increased by a factor of 100 from the notional 1 g / 6 hr to ensure the same contour levels / thresholds can be applied A number of NAME model runs were done varying the source strength and height to see the sensitivity of the predictions to the source strength. These are not shown here.

4 IVATF/1-IP/ Redoubt December During the Mount Redoubt eruption in Alaska in December 1989, a KLM Boeing flight lost power to all 4 engines after encountering an ash cloud. It eventually managed to restart all engines and descend to land at Anchorage. Damage to the aircraft was estimated to be $80 million The event consisted of 4 major eruptions, which have all been considered as separate source terms in the NAME modelling. On the day of the aircraft encounter (15 December 1989) there were 2 eruptive phases Information on the location and flight path of the KLM aircraft has been taken from Casadevall (1994) (see below) and roughly plotted on the NAME output Six-hourly average plots of ash concentration show that a high density plume of ash passed over the encounter region during this time (see Figure 2). The location of the first engine loss is just as the aircraft reaches the boundary of the very high (yellow) concentration ash. In this case the contour in figure 2 corresponds to the 2x10-4 g/m 3 (or the red threshold used on the European charts) mass concentration threshold. 2.6 Hekla 28 February Following the 26 February 2000 eruption of Hekla in Iceland a NASA DC-8 research aircraft suffered engine damage (as documented in Grindle and Burcham (2003) and other reports) that has been attributed to it encountering ash in the Hekla eruption plume on 28 February Based on data from the Smithsonian Global Volcanism Program and Lacasse et al (2004) the modelled eruption has been taken as lasting from 18:20 to 20:00UTC and extending from the summit elevation up to 12km. This represents the most vigorous (subplinian) phase of the eruption and the highest column height. Radar images in Lacasse et al (2004) show the eruption column weakening from ~20: The NASA DC-8 was on a direct route from Edwards AFB, California to Kiruna, Sweden when it encountered the ash plume to the north of Iceland. Information on the exact location of the encounter is not consistent between different data sources, but it appears to have been within the region N and 9.0-0W. The encounter lasted between approximately 05:08 and The flight was in total darkness due to the lack of moon that night, but instruments on the aircraft recorded the presence of aerosols and gases during this period that were indicative of a volcanic plume. The encounter height is variously quoted as: FL370, 11.3km (~FL370 using conversion of 3.28) and 10.4km (FL341) NAME model output (see figure 3) shows that in this encounter zone at this time (output is for 6hr mean concentration at 06:00UTC, the box shows the encounter region), levels of ash above 2x10-3 g/m 3 (corresponding to the contour on the graphic in Figure 3 and corresponding to the grey zone on the current European charts) would have been present. In the FL level (not shown here), levels of ash above this threshold were also present, but located slightly further south, due to slightly weaker winds at lower altitudes Using these simple approximations in NAME suggests that the aircraft flew through an ash plume that would have been in the new grey zone concentration.

5 - 5 - IVATF/1-IP/ Manam 17 July During 2005 and 2006 there had been many eruptions at Manam, a volcanic island just off the north coast of Papua New Guinea. On 17 July 2006 continuous ash eruption was observed from the ground that reached around 3km, at a similar time a pilot report stated that an ash cloud from Manam was reaching altitudes of ~ 4.6 km. Ash was not visible on satellite imagery due to local cloud cover, making verification difficult On July , a Gulfstream II aircraft was engaged doing survey work over Papua New Guinea, flying at FL390 (~12 km asl) near 6.2S 144.2E on a SW to NE mapping pattern. The Gulfstream was flying in apparently clear air approximately 270 km to the south west of Manam volcano. At 05:18UTC (15:18 LT), the right-hand engine cut out, descent was initiated, and the other engine failed a minute later. At FL290 (~8.8 km asl), the right-hand engine was successfully restarted, and the lefthand engine was regained at FL240 (7.3km, approx 3km above the highest ground level in the area). The aircraft continued on reduced power to Port Moresby where it landed safely. (From Tupper et al (2007)). Engine examination pointed to volcanic ash as the only plausible candidate for the engine failure To account for the incident it has been proposed that there must have been a volcanic eruption column up to around FL390 (~11,890m asl) at some point prior to the engine failure. However, there is no other evidence from observations (ground or satellite) for an eruption of this magnitude. At the time of the incident a well-defined very dense ash cloud from Manam was seen to at least FL150 (4.5km) from another aircraft, but its exact extent (lateral and vertical) was obscured by cloud cover. It is plausible that the full eruption column could not be seen at this time due to cloud NAME modelling of an eruption column with a maximum height of ~4.6 km shows that ash from such a low-level eruption would have been transported to the north west along the coastline and not towards the encounter location At higher altitudes, winds were towards the south west. Tupper et al (2007) used satellite observations of the movement of upper level cirrus to calculate that it would have taken just under 3 hours for erupted material to have travelled from the Manam region to the encounter location. Their work suggests that a suitable eruption scenario to use would be a short-lived (e.g minute) eruption up to FL390 (~11,890m). An eruption lasting from 02:00 to 02:45UTC on 17/07/2006 has been used as a first approximation NAME model experiments with ERA-Interim met data suggest that this eruption start time does not provide enough time for the plume to be transported to the encounter location. Further tests have been conducted for an eruption up to FL390 from 01:30 to 02:00; 01:00 to 01:30; 00:30 to 01:00, and 00:00 to 00:30 (all UTC). These show that a longer transport time is needed for ash to reach the encounter location and hence the eruption must have been earlier The 00:00 to 00:30UTC eruption timing provides the best match for ash reaching the encounter location at the required time from an eruption column up to FL390. Consequently, this eruption period has been used The assumed eruption height corresponds to a VAFTAD ash threshold of 10-18, so the release rate has been increased by factor of 10 to account for this and ensure consistency with the operational red-grey-black output products. There are no data on mass release rate or ash deposition for this event so this value is highly speculative. There are also very large uncertainties in the eruption height, start time and duration.

6 IVATF/1-IP/ An eruption from 00:00 to 00:30UTC places ash above the red zone threshold of in the region and height of the aircraft s mapping in the period before the engine failure. In the FL level the ash dispersal is to the south-west, directly towards the aircraft (see NAME output in figure 4 showing the 6 hour mean ash concentration from 00:00 to 06:00UTC). The box on the image shows the region in which the aircraft was performing mapping traverses. By coincidence the traverse direction (NW to SE) is nearly perpendicular to the plume, meaning that the aircraft would have been crossing the whole of the plume on each traverse and was likely to be encountering mass concentrations higher than 2x10-3 g/m 3 (corresponding to the contour on the graphic in figure 4 and corresponding to the grey zone on the current European charts) 2.8 Summary The analysis that has been undertaken to re-look at historical eruptions using the NAME dispersion model provides very positive evidence that state of the art dispersion models can give very good guidance on mass concentrations of ash advected and dispersed away from volcanoes so long as the model can be initialised with accurate source strength information For future research and development to provide most benefit to predicting mass concentrations, it should focus on either: a) Improving measuring techniques for estimating the volcano mass emissions and ejection height, or b) Mitigating against the lack of knowledge of the emission strength by including confidence levels in mass concentration predictions by for example using ensemble modelling techniques to simulate both the uncertainties in the volcano emission strength, and the background meteorological forecasts (which at the moment are considered to be small compared with the source strength uncertainties). 3. ACTION BY THE IVATF 3.1 The IVATF is invited to note the contents of this paper.

7 - 7 - IVATF/1-IP/19 Figure 1: NAME predicted 6 hourly average ash plume between FL350 and FL550 from 12Z-18Z 24/06/1982 assuming an eruption to a height of 20km starting at 12:00 UTC on 24/06/1982.

8 IVATF/1-IP/ Figure 2 NAME model output for Redoubt case with flight track overlaid. The aircraft was moving from north to south.

9 - 9 - IVATF/1-IP/19 Figure 3: NAME model output for the Hekla Case. The rectangle indicates the area of operation of the DC-8.

10 IVATF/1-IP/ Figure 4 NAME model output for the Manam case. The box indicates the area of operation of the Gulfstream aircraft. END