Impacts and dynamics of volcanically-generated jökulhlaups, Eyjafjallajökull, Iceland


Jökulhlaup Hazard Management and Risk Mitigation:

Lessons from Eyjafjallajökull 2010

Date: Thursday the 25th of August 2011, 08:30-17:30 hrs
Location: Veðurstofa Íslands (Icelandic Meteorological Office), Bústaðavegur 7-9, 150 Reykjavík, Iceland

Workshop goals
The workshop was designed to share scientific knowledge about recent jökulhlaups from Eyjafjallajökull in connection with the 2010 eruption. In particular, initial results from a UK project based at Gígjökull, led by Newcastle University, UK and funded by the Natural Environment Research Council (NERC), were presented. Representatives from agencies tasked with managing and mitigating jökulhlaup hazards were encouraged to attend. Through a series of presentations and informal debates, the workshop aimed to explore scientific understanding of volcanically-induced floods in order to help to prioritise future research. There was also opportunity for informal networking and discussion. This report summarises the key elements of the workshop and draws together some information on the research project that formed the starting basis for wider discussion.

09:00 - Opening remarks and outline of agenda

The workshop was opened by Matthew Roberts of the Icelandic Meteorological Office (IMO). Matthew outlined the holistic nature of the research and monitoring work being carried out on Icelandic volcanically-generated jökulhlaups and briefly explained the format of the workshop. He emphasised that one of the key elements of the day would be the open discussion during the afternoon, from which it was hoped that recommendations for future research and hazard management could be drawn. He then handed over to Andy Russell (Newcastle University), who invited workshop participants to introduce themselves (for further details, see the end of this report).

09:15 - Jökulhlaups from Eyjafjallajökull: observations and insights
The workshop presentations began with a talk from Matthew Roberts (IMO) entitled ‘The April 2010 Eruption of Eyjafjallajökull Volcano: Glacial Flooding and Attendant Hazards.’ The presentation opened with an over-view of the April 2010 eruption of Eyjafjallajökull, explaining the three distinct phases of activity - subglacial volcanism, followed by phreato-magmatic eruption and then sustained lava flow generating chronic flooding. Matthew’s presentation concentrated on the first 60 hours of 39 days of eruptive activity, with emphasis on jökulhlaups generated by the volcanic activity. Key elements of the presentation centred on constraining the timing and intensity of flooding. The locations of IMO monitoring stations were illustrated and the pre-eruption configuration of Gígjökull and its proglacial lake were described; the steepness of the Gígjökull ice fall was stressed as a key control on events that followed. Jökulhlaup inundation occurred to the north and south of Eyjafjallajökull in a series of floods within the eruptive event. Early in the morning of April 14th a large jökulhlaup cascaded from Gígjökull creating an ice canyon as water emerged at elevation on the glacier after subglacial transit. Records obtained by flood gauging recorded a decrease in water temperature and an increase in electrical conductivity interpreted as flood water pushing out the water previously occupying the proglacial lake basin. The flood was characterised by a very rapid rise to peak discharge with rapid incision and widening of the flood channel. The mean flood velocity was estimated at 1.3 ms-1. Floodwater also discharged supraglacially southwards from the vicinity of the eruption site, cutting a trench in the ice 3km long and inundating Núpakotsdalur with a sediment-rich flow containing copious ice blocks. Subsequent increases in eruption intensity generated repeated jökulhlaups from Gígjökull that inundated the Markarfljót; these floods were characterised by periodic rises in discharge with intervals of very low runoff.

In the evening of April 15th, floodwater broke out of the surface of Gígjökull prompting the evacuation of population within the vicinity of the Markarfljót. This jökulhlaup was laden with sediment and ice fragments and was characterised by a viscous, smooth-surfaced, lobate flow front advancing at 4-4.5 ms-1 extending across the entire Markarfljót valley. This flow of ice and debris advanced as a hyperconcentrated frontal flood wave displacing river water generating a bore; it was followed by a more turbulent fluid flow body. Downstream measurements were used to constrain the timing of the flood duration; the jökulhlaup on the 15th April was swift with rapid transition of water downstream, the flood waning within an hour. The presentation speculated on the ice- and debris-rich nature of the flood on the 15th April; it is possible that ice was trapped in the downstream path of floodwater or that water was trapped and then released by a tephra dam. The nature of the flow prompted comparisons with other ice-laden flows from high elevations with steep floodwater flow paths (e.g. Öræfajökull in south-eastern Iceland, Mount Redoubt in Alaska and Nevado del Ruiz in the Columbian Andes).

The presentation described how alterations were made to Iceland’s ring road (Route 1) in attempts to deflect flows in order to preserve the bridge over the Markarfljót, where the initial jökulhlaup from Gígjökull reached peak discharge 5 hours later, damaging the road. Matthew also commented on the difficulties in estimating peak discharge for floods of this nature and highlighted the implications for design of structures given the difficulties of getting reliable estimates. The presentation concluded by contrasting the nature of the jökulhlaups on the 14th and 15th of April.

Several issues were the subject of wider discussion following this presentation:

  • The difficulties entailed in obtaining reliable measurements of peak discharge are still one of the biggest issues in flood research. Given the nature of these sorts of floods, this is a key challenge.
  • The problems inherent in predicting peak discharges have clear implications for the design of flood-resistant structures (e.g. bridges) and flood management structures (e.g. dykes systems, retaining walls).
  • Within-flood processes such as erosional scouring and sedimentation make flood stage recording difficult and absolute changes to river systems difficult to gauge.
  • The jökulhlaup on the 15th April was ice-rich; however, ice melt leaves little trace. This has implications for reconstruction of other similar events from the sedimentary record and the ensuing characterisation of floods from particular locations.
  • The mechanisms by which floodwater moves through and exits glaciers needs further research -for example, it is possible that collapse pits could have become outburst locations as hydraulic pressure rose during melting of glacier ice at Eyjafjallajökull. Subglacial and intraglacial floodwater pathways exert a key control on the locations at which floodwater bursts from the ice and the consequent sub-aerial routing of jökulhlaups.
  • The implications of the Eyjafjallajökull jökulhlaups for monitoring and prediction were discussed, with many participants emphasising that, very often, sending a person to make observations and relay those observations was a simple but effective way of getting real time information on flood parameters which could then feed into emergency management.

Following Matthew’s presentation, Andy Russell (Newcastle University) gave a brief outline of the NERC-funded project ‘Impacts and dynamics of volcanically-generated jökulhlaups, Eyjafjallajökull, Iceland’. He described the unique opportunity presented by the collection of pre-flood data at a variety of locations around Eyjafjallajökull, and described the fieldwork that occurred during and after the eruption, emphasising the need for data collection before subsequent erosion and deposition destroyed vital evidence that could be used to reconstruct flood characteristics. (Further details regarding the project and the project team members can be found at the end of this report.)

Andy then gave a presentation entitled ‘Impacts and dynamics of volcanically-generated jökulhlaups during the 2010 Eyjafjallajökull eruption, Iceland’. This talk examined the sedimentary impact of the April 14th and 15th jökulhlaups on both the Gígjökull basin and the Markarfljót river system. It was stressed that the floods on the 14th and 15th of April were part of a sequence of jökulhlaups at one location. Attention was once again drawn to the pre-flood landscape of the Gígjökull area, clarifying the relationship between the glacier and the lagoon below it. The presentation went on to describe the influx of water, ice and sediment into that system as a consequence of the flooding generated by the eruption in Eyjafjallajökull.

The jökulhlaup on April 14th was characterised by turbulent fluidal flow and an exponential rising stage. The floodwaters caused 4m of lake level rise within the proglacial lake at Gígjökull and 15m of erosion at the lake spillway into the Markarfljót. Deposits from the flood were all located below well-defined wash limits and comprised poorly-sorted sands and gravels arranged as large scale bars. The flood on April 15th was an ice and sediment rich flow with high strength. It punched holes in flood defences and overlaid the deposits from the previous flow on the 14th April, exhibiting a distinctive lobate morphology. Seepage lines developed after deposition indicating de-watering from an ice-rich flow. The deposits were locally less extensive than the flood on April 14th, permitting the two deposits to be differentiated clearly.

The presentation stressed that some sediment analyses are still in progress, but it is clear that the April 14th and April 15th jökulhlaups had very different geomorphic and sedimentary impacts, reflecting their different rheologies. It was once again stressed that there were a series of floods within the eruptive event at one location. Erosion and deposition during this sequence of floods was complex given the number of events and this complexity has implications for the reconstruction of such events from the sedimentary record. Ice-rich deposits, such as those deposited during the April 15th jökulhlaup, have only been inferred from the sedimentary record and have not been observed during previous volcanic eruptions in Iceland. The variability of jökulhlaup characteristics and impacts resulting from the same volcanic eruption constitutes a challenge for the management of volcano-glacial hazards and for the interpretation of the sedimentary record.

Several issues were the subject of wider discussion following this presentation:

  • A sequence of jökulhlaups at one location presents challenges for management, particularly when floods have different characteristics.
  • The complex nature of erosion and deposition (given the number of floods from the volcanic episode) has implications for interpretation of the sedimentary record.
  • The retreat of glaciers frequently results in the formation of proglacial lake basins into which sediment is deposited; this sediment is remobilised in future events. This led participants to consider processes of medium and longer term landscape change as sediment is recycled.
  • The effects of temporary storage of sediment in lake basins were debated and the similarities between the 2010 jökulhlaups from Eyjafjallajökull and the 1996 jökulhlaup on Skeiðarársandur were commented upon.
  • Comparisons were made between the deposits from the Eyjafjallajökull jökulhlaup of April 15th and the deposits of the Kota fan and older Skógar fans in Iceland.

Stuart Dunning (University of Northumbria) gave a presentation entitled ‘Terrestrial Laser Scanning’. This focused on work that has been carried out in the Gígjökull basin as part of the NERC-funded project ‘Impacts and dynamics of volcanically-generated jökulhlaups, Eyjafjallajökull, Iceland’. Stuart gave workshop participants a brief overview of Terrestrial laser Scanning (TLS) as a survey tool and described some of the advantages and disadvantages of TLS. The ability to collect high-resolution data (e.g. 11,000 survey points per second), with a full 360° view and to generate such data rapidly is offset by issues relating to interference disrupting the effective operation of the scanner (e.g. raindrops causing bounce-back of the laser beam) and the effects of low incidence angles in the environment being surveyed (the laser returns best from high-angled surfaces). The presentation then proceeded to describe some of the results pertaining to the project. TLS survey data collected before and after the jökulhlaups into the Gígjökull basin are being used to generate Digital Elevation Models (DEMs) to quantify geomorphic change in the basin, particularly in relation to sediment accumulation and gradient changes. Initial results suggest that the jökulhlaups have resulted in 70m aggradation at the back of the basin and 15m incision at the edge. Results also suggest that there has been 200,000m3 of sediment deposition within the basin; what is unclear at present is how much of that infill consists of buried ice. Ground Penetrating Radar (GPR) work during August 2011 should shed further light on the nature of the infill in the Gígjökull basin.

Several issues were the subject of wider discussion following this presentation:

  • The scope for monitoring different sites using TLS was debated along with the identification of key points at which infill could be monitored as it evolves.
  • Estimates of sediment storage at different locations? Possibility of making better use of grain size data?
  • There was some deliberation on the possible use of TLS to monitor within-flood variability in the future.
  • The infilling of the basin has resulted in a surface with a steep gradient for transporting floodwaters during any further jökulhlaups at this site, or, depending on the erosive capabilities of the next flow, will the basin infill be a source of sediment supply to the flood?
  • The impact of sediment deposition from jökulhlaups was once again considered, particularly the drift of sediment from the sequence of events at Eyjafjallajökull west/south-west into the offshore zone.
  • Gígjökull ice margin exerted a strong control on routing the floodwater during the April 2010 jökulhlaups and the extent to which other glaciers exert a control on floodwater routing was discussed. This raised the point that coupled glacier and sandur systems need truly holistic assessment in order to fully understand floodwater routing, timing and magnitude and to be able to provide meaningful information on these variables to agencies managing such systems.

12:30 - Volcanically-generated jökulhlaups: examples from Iceland and New Zealand

Following lunch, Andy Russell (Newcastle University) gave a brief presentation on some of the modelling work that has been done as part of research on other systems. It is intended that data collected from survey and sedimentological work at Gígjökull and along the Markarfljót will be used to test the ability of a range of 2-D hydraulic models to predict the propagation of jökulhlaups with different flow types. This modelling work has not yet been completed, but Andy was able to describe some of the key stages envisaged in this stage if the research. Research will attempt to link sediment transport with the hydraulics of flooding in order to model erosion depth and landscape change and will contrast two different modelling approaches. Of particular interest will be to see how well the models can accommodate within-flow sediment interaction and the interaction between the flow and the bed. The presentation drew on work that has been done on floods from Mount Ruapehu in New Zealand. This system is highly confined (as opposed to the broad system of the Markarfljót).

This presentation closed the formal presentations which formed the basis for the wider discussions during the afternoon session.

13:00 - Discussion: Management of hydrological hazards at Eyjafjallajökull and wider implications for monitoring and managing rivers prone to jökulhlaups
The afternoon session of the workshop provided an opportunity for discussion of key issues triggered by the sharing of scientific knowledge of the jökulhlaups at Eyjafjallajökull during the previous sessions. This session drew on input from representatives from organisations tasked with managing jökulhlaup hazards and river system adjustment to jökulhlaups. The discussion broadened to highlight some wider implications and sought to identify research priorities reflecting input from the community of specialists responsible for managing the impact of jökulhlaups on population and infrastructure. This section of the report summarises the main elements of those discussions.

  • Participants agreed that constraining and predicting maximum discharge from jökulhlaups remains a key challenge and one that is critical for the effective management of river systems prone to flooding.
  • Representatives from the Icelandic Roads Authority, Vegagerðin, confirmed that from their point of view, fundamental information is required for management centred on accurate prediction of the floodwater routeway(s), the size of the flood, flow propagation time, the predicted flood extent and the likely sediment concentration of the flow.
  • On the basis of the experience of the Eyjafjallajökull flood of April 15th, workshop participants discussed the possibility of putting gaps in levees in order to try to better control floodwater routing and associated impacts.
  • There was some discussion of post-event adjustment to the impacts of flooding. It was suggested that a system such as the Markarfljót could be used to calculate the potential impact of a year’s post-flood meteorological events using data collected by IMO.
  • Workshop participants highlighted the importance of changes caused by retreating ice margins. The dynamics of ice marginal retreat is linked to floodwater routing. The inclusion of glaciological evaluation was identified as intrinsic to flood assessment. Several participants emphasised that it is crucial to treat the glacier and sandur as a coupled system. It was also stressed that jökulhlaup trigger mechanisms are key to understanding flood routeways and that topography plays a crucial part in determining an array of jökulhlaup characteristics.
  • The longer term impacts of sediment deposition were discussed at some length. There are profound implications for management, as river systems that have been subject to high levels of deposition during jökulhlaups have the potential for reduced capacity during the following event, causing overtopping far more quickly than perhaps anticipated. The time interval between floods is an important parameter in the conditioning of systems and therefore has consequences for modelling flood behaviour.
  • The evacuations of population within the Markarfljót area during the Eyjafjallajökull jökulhlaups of 2010 were considered to have been successful, along with the timely action of the authorities in preventing extensive damage to Route 1 during the event. Workshop participants turned their attention to the potential for larger jökulhlaups generated by volcanic activity, prompted by the recent flooding in the Múlakvísl and the associated destruction of the bridge (July 2011). It was conceded that there are some events (e.g. a full-scale eruption of Katla) for which structural mitigation measures are largely ineffective and the evacuation of population to safety is the only course of action.
  • Representatives from the Icelandic Roads Authority, Vegagerðin, highlighted that it would be useful to the management of road and related infrastructure to have more information on different types of flow. For example, it would be useful to be able to have more information on the behaviour of different flows, the density of material, the concentration of ice within the flow etc.
  • There was discussion about the significance of integrated real-time data in the management of jökulhlaups. Representatives from IMO highlighted the importance of seismic records in monitoring and predicting volcanically-generated jökulhlaups. It was pointed out that ice resonates as floodwater travels through it (creating so-called ‘ice quakes’) and combining information on this seismic activity with river stage information can be used to provide warnings. The discussion also covered the experimental use of permanent GPS to measure glacier uplift, the monitoring of meteorological conditions, electrical conductivity and gas monitoring (particularly hydrogen sulphide) in aiding the management of river systems prone to jökulhlaups.
  • There was then some debate about the wealth of information now available and who has access to the information. There is recognition that different stakeholders have different information needs and it is vital to supply the information in a useable format. Masses of data are generated and having one website for relaying reliable real-time information was regarded as a good way forward. Several people commented on the ‘hit-rate’ of the IMO website during the Eyjafjallajökull eruption and the fact that this was a highly trusted source of information on the event. There was also agreement that data on website traffic can feed into monitoring of hazard awareness.
  • Some river systems are subject to repeated jökulhlaups and could be regarded as ‘chronic’ hazard systems. The response rate for different systems is different and participants discussed the tricky issues surrounding the knowledge needed to confidently make decisions to give flood warnings and to evacuate.
  • There was agreement that advancing technology provides opportunities for hazard management. There was debate about raising awareness of flood hazards and about the use of technology in the alert and warning phases of hazard management. The use of SMS ‘push’ messaging as a means of alerting mobile populations (e.g. tourists) to the risk of flooding was briefly discussed.
    • Árni Snorrason, Director-General of IMO, addressed the workshop towards the end of the afternoon. As well as drawing on the events of the 2010 Eyjafjallajökull eruption, he made references to the 1999 flood from Sólheimajökull, the 1996 flood on Skeiðarársandur and the Múlakvísl flood of 2011 and considered some of the issues in monitoring and managing volcanically-generated jökulhlaups. These events have prompted the Icelandic government to provide funding for a comprehensive collaborative study involving researchers, monitoring and management agencies and other stakeholders. This will contribute to preparation for the EU Floods Directive.

Workshop participants and contact details

Veðurstofa Íslands:
Árni Snorrason
Bergur Einarsson
Esther H. Jensen
Jórunn Harðardóttir
Matthew J. Roberts
Sveinn Brynjolfsson
Þorsteinn Þorsteinsson
Gunnar Sigurðsson
Einar Pálsson
Rögnvaldur Gunnarsson
Jarðvísindadeild Háskóla Íslands:
Björn Oddsson
Málfríður Ómarsdóttir
Óskar Knudsen
Newcastle University:
Andrew J. Russell
Andrew R. G. Large
Anne-Sophie Meriaux
Staffordshire University:
Fiona S. Tweed
Timothy D. Harris
Alice Dewey
Kirsty Price
Helen Hargreaves
University of Northumbria:
Stuart Dunning

You are viewing the text version of this site.

To view the full version please install the Adobe Flash Player and ensure your web browser has JavaScript enabled.

Need help? check the requirements page.

Get Flash Player