Back to Vatnajökull main article

Late on the evening of 30 September 1996 an eruption started beneath the Vatnajökull glacier in Iceland. Over the previous 24 hours a sequence of earthquakes had been recorded around the Bardarbunga caldera. Similar earthquakes had occurred beneath the volcano many times during the previous 22 years, but none of these prior large earthquakes had significant aftershocks or were followed by magmatic activity such as the one that occurred in September 1996.

Numerous earthquakes, including five with magnitude grater than three, were recorded in two hours. Shortly after 13:00 hours, Science Institute seismologists informed the Civil Defence authorities as well as the scientific community about this unusual seismic activity and the possibility of impending eruptive activity.

The eruption site was discovered early Tuesday morning (01 Oct.) from an aircraft. By that time two elongated, 1-2 km wide subsidence cauldrons had formed on the ice surface of Bardarbunga, on the northern flank of the neighbouring Grímsvötn volcano. The cauldron formation indicated that the glacier was being melted by an eruption on a 4 km long fissure beneath the glacier, which is 400-600 m thick here. The meltwater drained into the Grímsvötn caldera under the ice shelf of the lake. In less than 24 hours a third of a cubic km of water had been added to the lake.

By 02 October one of the active craters had melted its way through the glacier and a massive steam column rose from the cauldron up to an elevation of 10,000 metres.

This image, acquired by Kiruna station just four days after the eruption began, shows how the heat had broken through the surface of the ice. An irregular white line represents the steep slopes of a canyon formed by ice melting. At the top of this line the black streak towards the north shows meltwater on the top of the ice.

By 09 October the eruption was taking place on a 9 km long fissure and volcanic products piled up above the fissure, forming a mountain ridge which in places approached 200 m high. About half of the area of Vatnajökull was covered by a thin layer of ash.

This map shows in more detail the approximate area covered by the SAR images below. In the southeast corner is Mount Svianukar, at 1720 metres (5659 feet) one of the highest points in the glaciated area, while stretching away to the northwest lies the Grímsvötn caldera.

This SAR image was acquired by the European Space Agency's ERS-2 satellite on 01 September 1996, a month before the event. It shows the normal scene on the glacier. The valley, or crevasse, that lies between Mounts Svianukar and Grímsvötn is clearly visible in the southeast corner, while the ridge of Mount Grímsvötn is just visible at the surface of the ice to the northwest. Apart from these features and two other ice features in the west of the image there is no sign of anything but the smooth, wet surface of the snow-covered ice cap.

This image sequence shows the evolution over the area where the fissure appeared. The similarity of the image acquired on 21 October (by the ERS-1 satellite) with that of 7 November shows that by these dates the eruption was over. The dark streak of meltwater is no longer present. The trapped water is breaking out to the south-west of this point, causing catastrophic floods. Another image, acquired on 22 October by ERS-2, has been used to create interferograms to reveal the full scale of the topography changes.

The ERS radar satellites also have another capability: through a special technique, called interferometry, it is possible to detect land movement and build 3-dimensional images using a number of images. Thanks to the availability of two satellites, ERS-1 and ERS-2, with only one day difference (the interferometric results shown here have been obtained with the tandem pair acquired on the 21 and 22 of October), the event can be shown clearly and measured with great precision. For interferometry the data used also have the phases values, that in the standard product (PRI) is omitted.

Since the data was acquired only one day apart, a reasonably good coherence over the whole scene is present, with the exception of a pear-shaded area around the fissure. This area was still subject to strong vertical and horizontal movements between the acquisitions, which make the data acquired in successive days uncorrelated for the retrieval of any reliable phase information.

The images below show an area of approximately 36 km per 44 km around the eruption site.

Intensity	Coherence	Phases	Unwrapped Phases
This is obtained by summing the intensity values of the two images. Due to the very short timeframe between the two acquisitions, the result is quite similar to the intensity of one acquisition only.	This is obtained by comparing the two images in intensity and in phase. In this way it is possible to detect very small changes. The dark areas indicate where the changes occurred (low coherence).	This image shows the phase difference between the two images. Each grey cycle (from black to white, with values from 0 to 360 degrees) represents a height difference. Due to this phase repetition, several cycles are visible.	This image is obtained from the previous one by adding 360 degrees several times, in order to unwrap the phases. Dark blue indicates the lower part of the image, while red the higher ones. The phases values are converted in altitude values.

The area around the fissure can be easily detected as the low coherence pear-shaded region, at the bottom-centre part of the coherence image. This lack of coherence is translated in a completely noisy area in the interferometric phase image. The quality of the applied unwrapping is appreciated on the unwrapped phase map, since errors due to lack of coherence are not propagated outside the low coherence region.

The image below is obtained from the unwrapped one on a larger area with respect to the above images.