Volcanic eruptions and their consequences such as magma flows, lahars, ash fall, glacial floods (jökulhlaup) etc. are a permanent threat to many regions of the Earth. Due to the growth of mankind and extension of populated areas, a rising toll of dead with devastating economic impacts have been caused by volcanic eruptions during the last century.
It is therefore essential to observe volcanic eruptions and their consequences, even in inaccessible regions independent of daylight, weather conditions etc., to make sure that advance warnings can be given and precautions taken to minimise the risks for man and infrastructure.
The first tool which could provide this information is spaceborne radar. The SAR system on board ERS-1 satellite (launched in 1991) has proved to be an excellent instrument for the observation of dynamic processes. The eruption under Vatnajökull glacier in South Iceland was the first opportunity for ERS-1/2 SAR to observe a volcanic eruption with a subsequent glacial flood and to provide geocoded image products and interpretations to the Icelandic authorities within a very short time. (Most image products and interpretations were made accessible via Internet:
http://www.dfd.dlr.de/HOT-TOPICS/volcano.
The eruption could be observed the first time by ERS-2 on 3 October 1996. Thanks to the satellite s orbit configuration, the visibility was again possible on 6 and 8 October. Fortunately the event took place at a time when ESA started checkout activity on ERS-1. This gave the unique opportunity to observe the eruption with ERS Tandem data to enable interferometric processing. Both satellites passed over the area again on 21, 22, 23 and 24 October. After the flood, ERS-2 data are available from 7 and 9 November. The data of all nine overflights, except for 3 October, where acquired by the DFD receiving station at Neustrelitz (Fig. 1).
Figure 1. Immediately after the acquisition, the data were
transferred to the DFD processing and archiving centre at
Oberpfaffenhofen. With an average delay of two days, the raw data
were processed using MSAR processor to make Single-Look Complex
(SLC) and Precision Image (PRI) products. To improve further the
interpretation results, PRI products were precisely
orthorectified and geocoded by GEOS system to make a Geocoded
Iceland is the biggest solely volcanic island on Earth, with a total surface of 103 000 km² and 260 000 inhabitants, located slightly south of the Arctic circle on top of the mid-Atlantic ridge, the border between the European and the North American continental plates. Roughly 11% of the island is covered with glaciers. The Vatnajökull glacier is Europe s largest, with an area of 8300 km² and an ice sheet up to 1000 m thick. Iceland has many periodically active volcanoes, most of which are covered by glaciers. About every two to three years, a volcano eruption is registered, which means a permanent threat to settled areas, transportation, infrastructures, power supplies, aviation, etc. (Fig. 2).
Figure 2. The geocoded terrain-corrected mosaic of ERS-2 data (descending) of 6 October 1996 shows a part of the neovolcanic zone in Iceland. The latest eruption
is located in a N-S oriented mountain ridge under the Vatnajõkull ice sheet. It is surrounded by the Bárdarbunga, Hamarinn and Grímsvõtn volcanoes.
Information on the chronological course of the eruption is derived from Einarsson et al. (1996); information on lake level and ice thickness is given by O. Sigurdsson (pers. comm.).
On 1 October,a 5 to 6-km long row of elongated depressions formed, each about 2 km wide and 100 m deep. In the early morning of 2 October, the eruption forced its way through the 450-600 m thick ice in this area. An eruption cloud of pyroclasts (volcanic ash and tephra), volcanic gases and steam rose to an altitude of 7000-8000 m. Domestic air traffic was heavily affected and many flights were cancelled. Dense cloud coverage, the eruption cloud itself, and the inaccessibility of the region made direct observation of the eruption nearly impossible. During the early phase of the eruption between 1 and 3 October, a large amount of ice was melted when magma of more than 1000 C and volcanic gases forced their way through the ice. The meltwater of an estimated volume of 2-3 km³ sought its way to Grímsvötn caldera, filling a lake covered by a 50- 250 m thick ice sheet.
Judging by the size of the depression and the fissure in the ice shown by the radar data, the amount of meltwater has been estimated at about 2.4 km³ at that time. Figure 3 shows that the total visible length of the fissure, which was active during the eruption, was 6 km. Only in the southern part was it possible for the eruption to break through the ice at two spots. These two openings are separated by an ice bridge. At the northern end of the eruption site, a 3500 m long and 2100 m wide depression was formed. Between the active fissure and the Grímsvötn caldera, a shallow slightly curved subsidence extends from the eruption site to the subglacial Grímsvötn lake. It was formed by meltwater flowing through a subglacial canyon into the caldera. The dark stripes originating at the eruption sites are layers of dry pyroclasts of small grain size.
Figure 3. Geocoded terrain-corrected ERS-2 data (descending)
of 6 October. This view extends about 12 km north-south and 6 km
east-west and shows the eruption area itself.
Figure 4 shows that the drainage of meltwater accumulated in Grímsvötn caldera is possible in three directions. It will most probably drain to the south under the outlet glacier of Skeidarár-jökull, down to Skeidarársandur. This endangers the ring road which connects all inhabitated areas around Iceland. The destruction of the bridges which span the glacial rivers would mean a detour of 800 km for the road supply of southeast Iceland. All aspects of daily life would be affected (food, fuel, medical care etc.). On the other hand, drainage to the west to the Tungnaá river system could endanger the dams and hydroelectrical power plants which are indispensable for the Icelandic economy, as well as the inhabitated areas at the lower courses of the rivers. Drainage of all the water into the Skaftá river is also possible, which would en-danger the village Kirkjubaejarklaustur and the ring road.
Figure 4. Geocoded terrain-corrected ERS-2 data (descending)
on 8 October, showing the possible directions of drainage.
In the latest stage of the eruption and after its end on 13 October, the horizontal dimensions of the depression were greatly enlarged (Fig. 5). One reason is that the fissure is much wider at its base than on the surface and that the neighbouring ice is now breaking down into that space. Another reason is that the 400-600 m high ice walls bordering the fissure were pressed into the fissure from both sides by the ice mass behind.
Figure 5. Geocoded terrain-corrected ERS-2 scene (descending) of 21
October (subset size: 30x38 km). At the northern end of the
eruption site, the depression which is traversed by numerous
crevasses, had widened. The average width was 400 m and the
length 3.6 km. The total length of the area affected by the
eruption was 10.5 km.
A 300 m long volcanic ridge had appeared in the fissure with a little water-filled crater of approx. 140 m diameter calculated from the SAR scene, at its northern end. This crater seems to be the top of the 7-8 km long active ridge that has reached a height of 1560 m, which is 350-400 m higher than the former highest point of the subglacial mountain ridge (Fig. 6).
Figure 6. Geocoded terrain-corrected ERS-2 scene (descending) of 21
Oct., Loki crater section (subset size: 6.5 7.5 km).
After the end of the eruption an enormous amount of water poured still into the subglacial lake of Grímsvötn caldera reaching a height of 1510 m on 24 October. All glacial floods since the start of regular measurements (1954) began at a water level of 1430 to 1450 m and the water found its way slowly through the ice.
After generating a 3-D volume model of Grímsvötn reservoir based on data given by Björnsson (1988) and Björnsson & Einarsson (1990), the volume could be calculated. The increase of the water level above 1430 m was 80 m which means an additional volume of water of at least 2.5 km³. The calculation of the total amount of water which could be drained from the reservoir during an expected 180 m drop of the lake level due to former post-jökulhlaup lake levels gave us 3.3-3.7 km³. In addition, a volume of 0.5-0.7 km³ water kept in the ice fissure in the area of the eruption had to be accounted. Thus a volume of approx. 4 km³ had to be expected.
The drainage of the water was expected latest when the 1505-1510 m level was reached. At this level, water pressure should lift the glacier ice off the ground in the southeast part of the caldera. That would cause the sudden runoff of the water under the glacier down to Skeidarársandur. Having the margin of 1510 m exceeded and the water level rising approx. 25 cm/day, the glacial flood (jökulhlaup) was expected any time.
In a section of 19x29 km, Figure 7 shows three main phenomena:
Figure 7. Interferogram of ERS-1 and ERS-2 scenes (descending) of 21-22
October.
This means that the eruption could be much bigger than previously estimated. The fact that no subsidence can be observed in that section can be explained as follows: After the eruption had forced its way through the ice at the location of its strongest activity and comparably low ice thickness, the volcanic gases and the steam could escape. The heat of the eruption alone did not suffice to weaken the stability of the ice sheet. The high water level causes high water pressure, which stabilises the subglacial excavation. It can be presumed that subsidence in that area will occur very slowly or as a result of a glacial flood which lowers the water level in Grímsvötn caldera.
At 9:30 pm of 4 November, the expected jökulhlaup of Grímsvötn started. A continuous high-frequency tremor was registered at a nearby seismic station caused by the lifting and breaking of the ice. At about 8:30 am on 5 November, water reached the glaciers rim and the rivers in Skeidarársandur started to rise from 50 m³/s to a flow rate of 6000 m³/s at 10:00 am. Three hours later the water had flooded the sandur on its entire width of 35 km and length of 20 km and had already destroyed a 380 m long bridge across Gigjukvísl, cut the main powerline along the coast and the optical telecommunication cable which had been burried deep in the sediments of the sandur plain.
At 3 pm the amount of water was estimated at 25 000 m³/s, at 10:30 pm the jökulhlaup culminated. An estimated amount of at least 45 000 m³/s flowed along a 50 km subglacial path beneath the outlet glacier Skeidarárjökull down to the alluvial plane of Skeidarársandur, transporting ice blocks of more than 1000 tons which damaged severely the foundations of the bridges (Fig. 8). All the day of 6 November, the water continued flowing to the sea on a continuously lowering level. In the afternoon of 6 November, a cloud of steam rose to 5000 m altitude for about 20 mn caused by a small eruption in the fissure, possibly triggered by the lowering of water pressure. On 7 November, around noon, the amount of water in the glacial streams diminished to a normal level and the jökulhlaup had ended.
Figure 8. Geocoded terrain-corrected ERS-2 mosaic (descending) of 7 Nov.
shows the area affected by the jökulhlaup. The flooded areas
can be easily recognised in comparison to Figure 4, mainly due
to high reflectance caused by the iceblocks which are spread all
over the flooded sandur plain.
All the areas which were flooded during the jökulhlaup remained covered by ice blocks of all sizes with a weight of up to 2000 tons and a height of up to 10-15 m (Fig. 8). Two bridges and a 10 km-long section of the ringroad were totally destroyed during the flood and another three bridges and 10 km of the road were severely damaged. The periglacial runoff system had been changed severely and millions of tons of sediments had been eroded and moved. For example, the former 300 m wide riverbed, where the totally destroyed 380 m long Gigjukvísl bridge was situated, has now widened to 2 km.
Figure 9 shows the whole area affected by the eruption and the glacial flood. In the area of the erupting fissure the ice movements caused by the lowering water level can be recognised. In the Grímsvötn caldera the ice movements during the jökulhlaup can be observed which lowered the water level by approx. 180 m. Clearly an approx. 500 m wide and 200 m deep sub- sidence in the south of the caldera can be recognised. That is caused by the breaking of the roof of the subglacial channel, which was created by the water forcing its way from the subglacial reservoir under the glacier over a distance of 50 km with a difference in altitude of about 1400 m down to Skeidarársandur alluvial plain.
Figure 9. Geocoded terrain-corrected multi-temporal ERS-1/2 mosaic
(descending) of 21, 22 Oct. and 7 Nov. (BGR). This colour
composite shows the whole area which was affected by the eruption
and the following jökulhlaup.
A total amount of water which was drained to Skeidarársandur was estimated close to 4 km³ (O. Sigurds-son, pers. comm.). That fits fairly good with our calculation of total expected runoff volume.
All together the amount of damage is now estimated at 31 million US$ which means approximately 120 US$ for every inhabitant. It will need at least two years to rebuild the ringroad completely.
At any time another subaerial or subglacial volcano can erupt in Iceland.
Björnsson H, 1988: Hydrology of Ice Caps in Volcanic Regions. Vîsindafélag Islendinga, Societas Scientarium Islandica, Rit. XLV, Reykjavík, 133 pp.
Einarsson P, B Brandsdóttir, M T Gudmundsson & H Björnsson, 1996: A Chronological Account of the October Eruption of Vatnajökull Ice Cap. Internet:
http://www.rhi.hi.is/~mmh/gos/chrono2.html
Brandsdóttir B, H Bjömsson, M T Gudmundsson, H P Einarsson & F Pállson, 1996: Jökulhlaup Update. Internet: http://www.rhi.hi.is/~mmh/gos/vat-update.html
Vedurstofna Islands, 1996: Preliminary Interpretation based on Monitoring by the SIL Seismological Network. Internet: http://www.vedur.is/ja/loki/eruption.html
ESA EOQ 54