Volcanic hazard assessment of Tristan da Cunha

This report provides an assessment of the volcanic hazards of the Overseas Territory of Tristan da Cunha, which was carried out by the British Geological Survey on behalf of the Foreign and Commonwealth Office and the Department for International Development. The assessment is based upon fieldwork undertaken during a visit to the island in January and February 2001. Tristan is a large oceanic volcano, which rises more than 5500 metres from the sea floor to reach an altitude of 2060 metres above sea-level. It has the form of a shield volcano, the outer flanks of which have been truncated by very high sea cliffs formed by marine erosion. The volcano is composed mainly of basanitic and tephritic lavas, although tephri-phonolites and rare phonolites also occur. Parasitic scoria cones occur around the flanks of the volcano, and scoria deposits intercalated within the older lavas attest to localised explosive activity from parasitic cones throughout the volcano’s growth above sea level. New radiometric dates acquired during the assessment, indicate that the volcano grew above sea level (i.e. the top 2 km) between about 140,000 and 20,000 years ago, largely by the voluminous eruption of lavas. This appears to have culminated in outpourings of relatively fluid lavas from the summit region of the volcano during a phase of activity between about 35,000 and 20,000 years. Little activity has occurred on the upper flanks of the volcano since that period, except for the occasional eruption of parasitic scoria cones, the last of which occurred almost 11,000 years ago. The coastal strips, found in the northwest of the island around the Settlement and Patches areas, and in the south around Seal Bay Plateau and Stonyhill Point, are young constructional features. They each have a basement of relatively young lavas on which superficial alluvial fan deposits have accumulated. Young parasitic volcanic centres have erupted on the coastal strips over the past few thousand years, the most recent of these being the lava dome and flows that were extruded close to the Settlement in 1961-62. Radiocarbon dating of peat samples collected from the volcanic deposits of these areas indicates that there have been 7 such parasitic eruptions on the two coastal strips within the past 3,000 years approximately. The coastal strip around the Settlement is constructed in the landslide scar formed by a largescale sector collapse of the volcano’s flanks. The collapse produced a submarine debrisavalanche deposit on the seafloor to the northwest of the island, which is estimated to have a volume of about 150 cubic kilometres. The age of this major flank-collapse is uncertain, but it most probably occurred during the past few tens of thousands of years. There is some evidence that the youngest lavas infilling the landslip scar were erupted after the main rise in post-glacial sea levels had taken place, which would imply an age of less than about 6,000 to 8,000 years. The main volcanic hazards are considered to be those that would be associated with future eruptions on the coastal strips, should there be any. Based on past activity, such eruptions would likely produce parasitic scoria cones, and small lava domes or tholoids with associated lava flows. The hazardous effects of such eruptions are described in the report. On the basis of newly acquired radiometric dates, it is estimated that there is a 25% probability of an eruption occurring on one or other of the two coastal strips during the next 100 years. Assuming an equal chance of an occurrence on either coastal strip, the probability of an eruption occurring in the next 100 years on the strip around the Settlement or Patches area is halved to 12.5%; which is equivalent an eruption on average every 800 years. An eruption from the upper flanks or summit region of the volcano is considered to be highly unlikely. If, however, a parasitic eruption were to occur on the northwestern flank, this could pose a hazard to the Settlement area below. Based on available geological evidence however, the probability of such an eruption occurring is roughly estimated to be less than 1 in 60,000 per year. Some concerns have been expressed by members of the population regarding the residual fumarolic activity on the 1961-62 eruptive centre, and whether this might indicate the possibility of renewed volcanic activity from this centre in future. A detailed appraisal of the 1961-62 eruption is given in the report. The fumarolic activity is extremely weak and of very low temperature, and shows no obvious involvement of magmatic fluids. The activity is typical of that caused by the heating of meteoric water (infiltrating rain water) by residual heat within the old vent area, and as such should not be a cause for concern. The hazard of rockfalls from the cliffs behind the Settlement was investigated. The rockfalls result from the piecemeal disintegration of a well-jointed volcanic plug exposed in the cliffs. Currently this occurs sporadically, producing small blocks, which accumulate close the foot of the cliff. Under current conditions these rockfalls are not considered to be directly hazardous to the Settlement itself. More intensive rockfalls associated with volcanic tremors in 1961 produced much larger blocks, a few of which rolled down slope from the foot of the cliffs, but still stopped well short of the Settlement. Rockfalls of this size could occur again in this area in the event of future volcanic tremors, although the topography provides some protection and any large blocks are likely to roll towards the NNE, obliquely away from the Settlement. Almost every volcanic process is characterised by seismic signals that can be used to gain diagnostic information on the level of activity. The 1961-62 eruption was typical in this respect, by being preceded by several months of ground tremors that were felt by the population. It is highly likely that there would have been a more prolonged period of seismic activity prior to the felt tremors that preceded the 1961-62 eruption of Tristan. These earlier and weaker seismic signals would have been detectable had the island be monitored by seismometers. Seismic monitoring of Tristan is therefore strongly recommended, as this would maximise the warning time of an impending eruption. The report therefore provides technical specifications and preliminary cost estimates for the installation a seismic monitoring network on the island. In the event of a future eruption on Tristan, it may be necessary to evacuate the island. A preliminary plan is therefore provided in the report, and recommendations are made for equipment that should be procured to facilitate such an evacuation if it were to be enacted.