Download
report(1.51MB PDF)
CONTENTS
Foreword
Author's Note
Executive Summary
Introduction
• Climate Change
• Atmospheric
Hazards
• Geological Hazards
• Hydrological
Hazards
Sources & Further Reading
| Home Contact us Gallery Site Map Resource Centre Search | ||
 |
||
| About us People News Publications Education & Training Events | ||
|
Hazard & Risk Science Review 2007 5.Geological Hazards |
|
|
New Madrid, Cascadia and California quake risk revisited  Figure 8. Seismicity of the central and eastern United States, showing earthquakes with magnitudes greater than 3 in the National Earthquake Information Center PDE (1973 to Present) and significant U.S. earthquakes catalogues (1568–1989). The future threat of large, damaging earthquakes in the New Madrid Seismic Zone, on a scale of the 1811-12 events, remains a hotly disputed topic. Courtesy: American Geophysical Union. The possibility of a moderate to large, damaging earthquake, in the New Madrid Seismic Zone (NMSZ) (Figure 8) is driving research on strain accumulation rates in the region, and leading to continuing dispute as to the actual risk. In a paper published in the Journal of Geophysical Research, E. Calais10 of Purdue University, and colleagues, present the results of a deformation study undertaken over the last decade across the central and eastern United States. The authors report that GPS monitoring in the New Madrid region over this period reveals no sign of the elevated strain rates that might be expected if the crust was building towards an earthquake. In a second paper in Eos, Seth Stein86 of Northwestern University examines, in some detail, the nature of the dispute relating to earthquake risk in the NMSZ, which essentially comes down to what GPS surveys undertaken by different teams are saying about deformation rates in the region. While Calais and co-workers find no evidence of any significant deformation, another team, led by R. Smalley reported significant deformation in some parts of the NMSZ (see HRSR2006 at www.benfieldhrc.org). This, some have suggested, may reflect the way the GPS data are processed rather than true movements. Stein also reviews the implications of NMSZ deformation for earthquake risk; the deformation may well be zero, implying that the zone has shut down and that, therefore, there is no risk of future large earthquakes. Alternatively, if real movements are still occurring, they could be transitory effects related to the 1811-12 earthquakes, and therefore also not linked to the accumulation of strain preparatory to a future large earthquake. No doubt Smalley and his team would disagree, and this is an issue that we will undoubtedly be revisiting in future reviews. Switching to the Cascadia Subduction Zone (CSZ), which has in the past ruptured in megathrust earthquakes comparable in size to the December 26th, 2004, Sumatra event, the past provides a good guide to the present and future. In this regard, Alan Nelson58 of the USGS, and co-workers, present, in a paper published in Quaternary Research, a new catalogue of seismic events in the CSZ, going back some 7,000 years. Most importantly, Nelson and colleagues show that earthquakes on a range of scales occur on the CSZ, which does not always rupture along its entire 1100 km length in one go, as happened in the magnitude ~ 9 event in 1700 AD. They show that the northern section of the CSZ seems to break in long ruptures in the greatest earthquakes, while ruptures in the southern part of the zone vary more in terms of both rupture length and recurrence interval. The authors provide evidence for eight long ruptures in the last thousand years, giving some idea of the time-averaged return period of an event capable of triggering a significant tsunami. The timing of the next great CSZ earthquake continues to attract considerable interest, particularly in relation to the possibility of identifying any precursory signals. One possible sign, described in Quaternary Science Reviews by Ian Shennan and Sarah Hamilton80 of Durham University in the UK, is pre-seismic subsidence. From studies of coastal sediments, Shennan and Hamilton reveal evidence for such subsidence immediately (around a decade) prior to great earthquakes in Alaska over the last 3,300 years, including the 1964 magnitude 9.2 event. The authors do not come up with a definitive mechanism for such subsidence, but they do raise the possibility that a-seismic slip - so-called ‘silent earthquakes’ - may be responsible. It is worthy of note that such events have been reported, in recent years, for parts of the CSZ. Not withstanding the cost of damage caused by ground shaking associated with the next great CSZ earthquake, which have been estimated at between US$40 and US$60 billion, losses arising from the accompanying tsunami could be far greater. Modelling undertaken by Josef Cherniawsky14 of Canada’s Institute of Ocean Sciences, Fisheries and Oceans, and colleagues, provide some idea of the scale of tsunami run-up that can be expected. In a paper published in Pure and Applied Geophysics, Cherniawsky and co-authors predict that future tsunamis in the vicinity of Vancouver Island (British Columbia), sourced either by rupture of the whole 1,100 km long Cascadia Subduction Zone or just its northern segment, would have run-ups in the 5 – 8 m range, and as high as 16 m in places. (Figure 9) shows a series of snapshots from a future Cascadia tsunami simulation modelled by Steve Ward of the University of California, Santa Cruz.  Figure 9. A series of snapshots from a future Cascadia earthquake tsunami simulation modelled by Steve Ward of the University of California, Santa Cruz; an animation can be viewed at www.es.ucsc.edu/~ward/ movies_eqtsu_index.htm. Courtesy: Steve Ward. A second paper on the CSZ tsunami threat appears in the Journal Of Geology. The authors, Robert Schlichting76 and Curt Peterson of Portland State University present geological evidence for high velocity tsunami inundations of the coastlines of Washington State and Oregon as a result of past tsunamigenic earthquakes on the central part of the CSZ. Schlichting and Peterson report minimum inundations of between 0.3 and 1.3 km, with a mean inundation of 0.5 km, results that highlight the serious threat of a future tsunami to the coastal zone of British Columbia, and the western United States. Further south, probably the worst earthquake scenario for California involves a major quake on the Puente Hills Thrust Fault (PHT) beneath Los Angeles, for which economic loss estimates are on the order of US$270 billion (Figure 10). The structure, growth and characteristics of the PHT are examined in detail by Lorraine Leon51 of the University of Southern California, and colleagues, in a paper in the Journal of Geophysical Research. Leon and co-workers describe how a picture of this ‘blind’ (no surface expression) thrust fault has been constructed using borehole and seismic reflection data. They recognise evidence for three large (magnitude 7 +) earthquakes in the past 8,000 years, each associated with significant displacement (between 7.5 and 10 m in total), and which probably involved rupture of the whole of the PHT. The authors conclude that the future occurrence of such an event directly beneath Los Angeles metropolitan region would generate enormous damage, confirming its potential to be the worst seismic disaster in US history.  Figure 10. The Puente Hills Fault lies directly beneath Los Angeles. A future large earthquake on the fault has the potential to trigger the worst seismic disaster in US history. Courtesy: Andreas Plesch and John Shaw. New seismic studies on a future Istanbul earthquake Continued concern over the expected future large earthquake on the North Anatolian Fault immediately south of Istanbul, is maintaining research interest in this area. In a disturbing paper published in Earthquake Spectra, Jacob Griffiths28, and colleagues, of Purdue University, Indiana, note that the city’s building stock is in no better state than that affected by the Düce earthquake of 1999, which followed four months after the catastrophic Izmit quake. The Düce event took 600 lives and caused the damage or collapse of around 3,500 buildings out of a total 12,000. Istanbul is home to about 12 million people housed in around one million buildings. Griffiths and his co-researchers, determine, on the basis of vulnerability indices calibrated for Turkish construction, and compared with the Düce quake, that a future earthquake near Istanbul may cause around 250,000 buildings to collapse or be severely damaged. In a second paper, in the Bulletin of the Seismological Society of America, Mathilde Sørensen84 and co-workers at the GeoForschungsZentrum in Potsdam, Germany, examine the ground shaking that might result from the next Istanbul earthquake. The team simulate the strong ground motions caused by a scenario (magnitude 7.5) earthquake in the Marmara Sea to the south of the city, and investigate the effects of varying input parameters on the nature of the ground motion. They conclude that such an event will have a ‘significant’ impact on the city, with the largest ground motions occurring in the southern and south-eastern parts of Istanbul. Here, ground accelerations in bedrock are expected to reach 500 cm/sec2 with velocities above 50 cm/sec. Predicting earthquakes: are we any closer? The accurate prediction of an earthquake remains the much sought after “holy grail” of geophysics, although some researchers remain convinced that they have already found it. Panayotis Varotsos91, of the University of Athens, for example, maintains that what he calls seismic electrical signals (SES) are generated in the crust prior (hours to months) to an earthquake, and that these can be used to successfully predict earthquakes. In a new take on the theme, Varotsos announces, in a co-written paper in Applied Physics Letters, a new type of signal with the capability of providing warning of an imminent quake. Unlike SES, this second type of signal takes the form of electric and magnetic pulses that occur within minutes of the start of an earthquake, and which are related by the authors to processes in the rock occurring immediately prior to fault rupture. Varotsos and his team note that, while such signals do not allow true earthquake prediction, they may be useful in relation to the development of real-time seismic early warning systems. It should be noted that the whole idea of using SES as a method for usefully predicting earthquakes is disregarded by many, if not most, seismologists, and that the existence and utility of these new electric and magnetic signals may receive similar treatment. Certainly no sign of either was detected prior to the 2004 Parkfield (Calfornia) earthquake. In expectation of a quake along this stressed segment of the San Andreas Fault System, the area was instrumented and monitored more than any other in the world, at least partly with the aim of seeing precursory signs. In the event, when the magnitude 6.0 quake eventually happened on September 28th 2004, no warning signs, whatsoever, were detected. In a paper published in the Journal of Geophysical Research, Stephen Park62 and colleagues, of the University of California report that, while electrical signals were detected during the earthquake, these were confined to within 1 km of the fault. They also note that there were no precursory signals, and – on this basis – seriously question the possibility that such signals occur and can be detected, particularly in relation to smaller and more distant earthquakes than the Parkfield event. A more promising approach to earthquake prediction is addressed by Piotr Shebalin79 of the Russian Academy of Sciences, in Tectonophysics, whose work on ‘earthquake chains’ continues to attract interest. Earthquake chains are clusters of moderate-seized quakes that extend over large distances and that are formed of statistically rare pairs of events that are close in space and time (‘neighbours’). In the Tectonophysics paper, Shebalin, ‘mass tests’ his earthquake chain methodology against a random earthquake catalogue, and makes the claim, as a result, that ‘in the vast majority of cases, large earthquakes are preceded by earthquake chains’. The earthquake chain hypothesis is taken seriously enough for the California Earthquake Prediction Evaluation Council to endorse, in 2004, a prediction by Shebalin’s colleague, California-based Vladimir Keilis-Borok, for a magnitude 6.4 or larger earthquake in southern California (see HRSR2004). Although the event did not materialise within the prescribed time-slot on this occasion, the methodology is still regarded as having some potential as a predictive tool. The Parkfield Earthquake Prediction Experiment (PEPE) and the seismic precursor issue It would be a considerable surprise if the forceful scraping of two gigantic rock masses against one another did not result in some detectable precursory signals, whether or not these can be used for useful prediction. In respect of aseismic deformation before earthquakes, an excellent review by Evelyn Roeloffs73 of the USGS is published in the Annual Review of Earth and Planetary Science. One of the many case studies the review examines is the aforementioned 2004 Parkfield earthquake, prior to which no deformation was detected, despite intensive monitoring of the fault segment. The absence of precursory signals, of any type, prior to the Parkfield event, is discussed in more detail in a special volume of the Bulletin of the Seismological Society of America devoted to the experiment. In the Introductory paper, Ruth Harris32 of the USGS and J Ramón Arrowsmith, of Arizona State University, examine the ramifications of the experiment’s failure to predict the 2004 event (Figure 11). They conclude that predicting the fine details of future earthquakes remains a challenge, and note that a probabilistic – rather than deterministic – approach is likely to continue to form the mainstay of forecasting future damaging behaviour from earthquakes. In the same volume, D. Jackson38 and Y. Kagan, of the University of California, examine future lessons from the Parkfield venture, in particular in relation to the 1985 prediction that an earthquake would occur in the area before 1993, which underpinned the development of the PEPE. The authors note that the 2004 event was not the realisation of the prediction, as the quake was late and also had different characteristics from those predicted two decades earlier. Jackson and Kagan explain that the original 1985 prediction, based upon the quasi-periodicity of past events, and the model that underpinned it, were poorly constrained, and too vague to meet the accepted definition of a prediction. Most importantly, they note that this questionable model continues to form the basis of seismic hazard assessment in the US and in many countries around the world. The authors conclude by recommending the utilisation of improved prediction models that allow for interactions between fault segments and uncertainties in various critical parameters.  Figure 11. Despite the Parkfield segment of the San Andreas Fault System being more highly instrumented and monitored than any other in the world, no precursory signals at all were detected prior to the earthquake itself (shown). Courtesy: Arizona State University. Patterns of seismic activity in time and space The identification and constraining of patterns of seismic activity in time and space is important in probabilistic forecasting of future earthquakes. In a paper published in the Bulletin of the Seismological Society of America, Lynn Sykes88 and William Menke of Columbia University, New York, analyse the repeat times of large earthquakes in Japan, Alaska, California, Cascadia and Istanbul, in order better to constrain long-term prediction. The authors make maximumlikelihood estimates of intrinsic repeat time, and its normalised standard deviation, the coefficient of variation (COV). This is a key parameter in time-varying probability estimates, and is defined as the standard deviation of inter-event times between large events that rupture all or most of a given fault segment divided by the mean repeat time for that segment. Sykes and Menke find that the (COV) 36 tends to be small (0 to 0.25) for several active fault segments where deformation is relatively simple and the behaviour well known, and larger for multi-branched faulting. In relation to specific areas, the authors suggest that 30-year probability estimates for the most active fault segments in the San Francisco Bay area are probably incorrect, with both the Hayward Fault and, perhaps, the Peninsular segment of the San Andreas fault appearing to be advanced in their build-up of stress that will be released in future large earthquakes. They also note that the multi-branched nature of the faulting may explain why the Tokai (Tokyo) earthquake in Japan has not yet occurred, and that one of the highest COVs (0.37) may explain why the 1985 Parkfield prediction failed. In a fascinating statistical study, reported in the Journal of Geophysical Research, Anna Maria Lombardi52 and Warner Marzocchi, of Italy’s Istituto Nazionale di Geofisica e Vulcanologia, present evidence for a clustering in time of large earthquakes in different parts of the world. The authors demonstrate that, worldwide, magnitude (surface) 7.0 + events tend to cluster in both time and space, while, for some seismic regions, there is evidence for long-term fluctuations in the earthquake rate. Notwithstanding the mechanisms and processes that might govern such behaviour, Lombardi and Marzocchi suggest that, from a practical point of view, the idea that seismic zones are stationary (unchanging) systems – which is implicit in seismic hazard assessment – should be treated with some caution. Earthquake loss estimation and hazard analysis Coming up with accurate estimates of earthquake loss is notoriously difficult, primarily because the ultimate figure is dependent upon so many different parameters, which may, themselves, be estimates, often not of the best quality. In a paper published in the Bulletin of the Seismological Society of America, Tianqing Cao11 of the California Geological Survey, and Mark Petersen of the USGS, use the results of a Monte Carlo simulation to evaluate the ground-motion uncertainty of the 2002 update of the California probabilistic seismic hazard model. The resulting ground-motion distributions are then used to evaluate – for wood-frame buildings in Los Angeles - the contribution of the hazard model to the uncertainty in earthquake loss ratio, in other words the ratio of the expected loss to the total value of a structure. The authors found that the uncertainty of ground motion, measured by coefficient of variation (COV) is amplified when converting to COV of loss ratio. For example the COV of loss ratio is almost twice the COV for ground motion at a return period of 475 years. The amplification of COV is dependent in a nonlinear way on the mean ground-motion level. Currently, in probabilistic hazard analysis, uncertainty parameters are regarded as being independent of the return period. As loss ratios increase strongly with return period the uncertainty in terms of COV therefore decreases. This artificial decrease is due to knowledge limitations in relation to uncertainty parameters and will affect current annualized loss estimates. The importance of ground motion during earthquakes is also addressed by Julian Bommer5, of Imperial College London, and Norman Abrahamson of Pacific Gas and Electricity in California. In an important review paper, published in the Bulletin of the Seismological Society of America, Bommer and Abrahamson examine why the well established methodology of Probabilistic Seismic Hazard Analysis (PSHA), now the most widely used approach for estimating seismic design loads, remains open to misinterpretation, and – often – leads to increased estimates of the seismic hazard. They conclude that the main problem, most notably in recent studies, resulted from the treatment of aleatory variability in the ground motion prediction equations, which has a significant influence on the calculated hazard. Studies undertaken in the 1970s and 1980s frequently resulted in lower ground motions, probably due either to neglecting ground motion variability or to its treatment in a manner that artificially reduced its influence on the hazard estimates. Most pertinently, the authors note that any new revival in the nuclear industry will involve the relicensing of many existing plants for which the original seismic hazard studies were performed in the 1970s and 1980s, an issue that is particularly relevant in view of the damaging impact of the July 2007 Niigata (Japan) earthquake on the Kashiwazaki-Kariwa Nuclear Power Plant. Seismic hazard micro-zonation for urban areas is the first step towards a seismic risk analysis and mitigation strategy. In a paper in Soil Dynamics and Earthquake Engineering, Roberto Romeo71 and Cesare Bisiccia of the University of Urbino, Italy, present a methodology designed to evaluate the local seismic response of an urbanised location and to undertake a cost-effective micro-zonation study, with a view to providing guidance in the choice of risk mitigation measures. The authors use, as a test site, the town of Pergola (population ~ 10,000) in central Italy. The product of the methodology is a micro-zonation that incorporates the severity of the hazard as modified by the local seismic response and the structural features of the building stock. The authors claim that the resulting micro-zonation is more informative than others derived using a more conventional approach, as it may be used in the adoption of risk mitigation measures, such as building retro-fitting, and in the formulation of emergency plans for civil protection purposes. Nilsun Hasancebi33 and Resat Ulusay, of Hacettepe University in Ankara, Turkey, also examine the urban micro-zonation issue, in a paper published in Engineering Geology. In relation to the town of Yenisehir in the Marmara Sea region of western Turkey, the authors examine ground shaking effects at a micro-zone level, focusing on seismic wave amplification and periods. Using data from new ground surveys and past investigations, they show that during a future earthquake, amplification of seismic waves will range from 1.6 to 5, according to the nature of the substrate. Wave periods ranged from 0.51 to 8 seconds, pose a particular threat to higher rise buildings. In many cases, seismic zonation is based upon expected peak ground accelerations during a future earthquake. Predicting such accelerations is addressed by Debbie Dupois19 of HEC Montreal and Joanna Flemming of Dalhousie University in Halifax, Canada, in a paper in Earthquake Engineering and Structural Dynamics. Dupois and Flemming use data from 23 large earthquakes in western North America between 1940 and 1980, to provide an improved model of peak accelerations. This, the authors indicate, will ensure that insurers are better prepared for impending losses, and that engineers have more accurate design specifications to work to. New geological studies highlight tsunami hazards across the world The devastating events in the Indian Ocean in December 2004 continue to drive tsunami research, with a range of new studies pointing up the global extent of tsunami hazard and risk. In an important paper in the Bulletin of the Seismological Society of America, and based upon the Indian Ocean tsunami, Eric Geist24 of the United States Geological Survey, and co-authors examine implications of the event for tsunami forecasting and modelling. By comparing different tsunami models with empirical data, they conclude that there is significant variation in tsunami run-up height, which they attribute to uncertainties in tsunami source parameters, such as fault rupture length, and to complexities in the fault rupture process. While the great majority of earthquake-generated tsunamis occur in the Pacific Basin, the events of December 26th 2004 highlighted the fact that they may occur in any ocean basin. Even the north-eastern Atlantic is not immune, as noted in a paper published in the Scottish Geographical Journal by Alastair Dawson16 of the University of Aberdeen, and colleagues. The authors report a sand layer within the peat of the Shetland Islands, which they attribute to emplacement by a tsunami between 1300 and 1570 years ago. The deposit is around 9 m above current sea level, and was most likely the result of a local offshore landslide. Still in UK waters, Ed Bryant7 of the University of Wollongong, Australia, and Simon Haslett of Bath Spa University in England, hypothesize that a massive tsunami struck the western UK in the early seventeenth century. Writing in the Journal of Geology, the authors interpret coastal deposits and structures in the Bristol Channel region as evidence that a tsunami was responsible for widespread coastal flooding and around 2,000 deaths in 1607. It is worth noting that this interpretation is strongly contested, and a more likely alternative explanation may be a powerful storm surge. The tsunami threat in the Mediterranean is also highlighted by a number of new studies. Writing in Marine Geology, Giovanni Scicchitano77 and colleagues, at the University of Catania (Sicily), report the discovery of boulders scattered along the coastline of southeast Sicily, between 2 and 5 m above current sea level. These, they interpret as having been emplaced by tsunamis associated with three Sicilian earthquakes, occurring in 1169, 1693 and 1908. Still in the central Mediterranean, L. Graziani26 of the Istituto Nazionale di Geofisica e Vulcanologia in Rome, and colleagues, re-evaluate the tsunamis generated by a particularly violent sequence of earthquakes that struck southern Calabria (Italy) in 1783 and 1784. Writing in Natural Hazards and Earth System Sciences, the authors report that the most lethal tsunamis associated with the event were triggered by landslides associated with strong ground shaking. Further east, G. Papadopoulos61, and colleagues, of the National Observatory of Athens, present – in Natural Hazards and Earth System Sciences – an analysis of tsunami hazard in the eastern Aegean and south east Turkey. Here they report 18 tsunamis and a return period for strong tsunamis of 142 years. The latest event, in 1957, caused extensive damage to the city of Rhodes, and to Fethiye in Turkey. The potentially devastating tsunami expected from the next great Cascadia earthquake has already been discussed. Other locations along the east coast of North America, however, are also at risk from tsunamis from more localised sources. One of the most notable is Monterey Bay in southern California. Here, according to H. Greene27 of the Monterey Bay Aquarium Research Institute (MBARI), and co-workers, the threat is associated with offshore instability. Writing in Natural Hazards and Earth System Sciences, Greene and colleagues report the occurrence in the area of massive submarine landslides that have formed over the past 200,000 years, with the most recent dated at between 8,000 and 10,000 years. More than 25 more recent, smaller, slope failures are also recorded. The authors warn of the possibility of a future large failure generating a tsunami as high as 10 m or more, and triggered by an earthquake or storm event. Two new studies have addressed tsunami risk ‘down under’. Writing in Marine Geology, Dale Dominey-Howes17 of Macquarie University, Sydney, presents a preliminary tsunami catalogue for Australia, which contains entries for 57 events, most (44) of which were recorded on the New South Wales coastline. Dominey-Howes reports a maximum run-up for an historical event of around 6 m, associated with an earthquake in Indonesia in 1977 and affecting the coast of Western Australia. The largest run-up for an ancient event, for which the source is not known, is between 100 and 130 m and occurred on the New South Wales coastline between 8,000 and 9,000 years ago. Another tsunami, around 6,500 years ago, may have penetrated up to 10 km inland in the same area. It is likely that a number of Australian tsunamis were triggered by earthquakes off the coast of South America on the other side of the Pacific. New Zealand, too, is threatened by great earthquakes in this region, and the risk they present is modelled probabilistically by William Power70 and his colleagues, of New Zealand’s GNS Science. In the journal Pure and Applied Geophysics, Power and co-authors report that three tsunamis sourced in South America have had a significant, damaging impact on the New Zealand coastline in the last 1,000 years. Looking ahead, the authors show that the highest hazard from a future South Americansourced tsunami is encountered to the north of Gisborne, on the North Island, and around the Banks Peninsula, immediately south of Christchurch, on the South Island. Collapsing volcanoes as major tsunami sources While most tsunamis are triggered by great submarine earthquakes, volcanoes provide an additional source. Since the 16th century, in fact, at least 17 volcanogenic tsunamis have been recorded that, together, took close to 60,000 lives. One way in which a coastal or island volcano can trigger a tsunami is through the collapse of a part of its flanks, and this is an area of research that has spawned a flurry of papers over the past 12 months. The phenomenon of volcano lateral collapse and its potential for generating destructive and lethal tsunamis is summarised in a Special Publication of the Geological Society of London by Bill McGuire56 of the BUHRC. McGuire critically evaluates, in particular, the ongoing debate relating to the feasibility of large volcano collapses generating ‘mega-tsunamis’, defined as being more than 100 m in height at source and remaining destructive at oceanic distances. The ability for pointsources, such as a collapsing volcano (as opposed to a linear source, such as an submarine earthquake rupture), to trigger large and persistent tsunamis remains a matter for discussion. One piece of evidence for what is sometimes known as the ‘giant-wave hypothesis’ is the occurrence of coral and other marine debris at high elevations on the flanks of ocean island volcanoes. On the Hawaiian island of Lanai, such deposits have been interpreted both as evidence of raised shorelines, and as having been emplaced by waves with minimum run-ups of more than 170 m. In a recent important paper in the International Journal of Earth Sciences, J. M. Webster95, of Australia’s James Cook University, and co-workers, show that uplift rates of Lanai have been low for at least several hundred thousand years. They conclude, therefore, that marine deposits high on the flanks must have been emplaced as a result of a giant wave associated with the lateral collapse of a volcano within the Hawaiian archipelago. Evidence for a similar giant wave is provided by enigmatic marine deposits on the island of Gran Canaria (Canary Islands), and described in Marine Geology by Francisco Pérez-Torrado64 of the Universidad de Las Palmas, and colleagues. Here, rounded boulders and marine fossils, at elevations of between 41 and 188 m above current sea level, are interpreted as having been emplaced by a giant tsunami, or series of tsunamis, emplaced by a major lateral collapse on a neighbouring island, possibly Tenerife. (Figure 12).  Figure 12. Tsunami deposits between 41 and 188 m above current sea level on Gran Canaria, are interpreted as having been emplaced as a result of a major volcano lateral collapse on a neighbouring island. The 22 x 11 km landslide block top left on the sea floor north of the Canary Island of Fuerteventura is too old to be a candidate, but indicates the enormous scale of previous collapses. Courtesy: Juan Acosta. The more closely the sea floor is mapped around coastal and island volcanoes, the more evidence comes to light of past collapses. In a paper in Earth & Planetary Science Letters, Michelle Coombs13 of the Alaska Volcano Observatory, and co-authors, provide detailed evidence, in the form of submarine imagery, for numerous massive collapses at volcanoes in the so-called Aleutian Volcanic Arc to the west of Alaska. Most importantly, the authors note that future collapses could generate tsunamis that threaten the west coast of Alaska and Aleutian Island communities, alongside important north Pacific shipping routes between Asia and North America. Recent evidence for tsunami-generation from collapsing volcanoes is also provided for the Mediterranean. Here, Maria Teresa Pareschi63, and colleagues, of Pisa’s Istituto Nazionale de Geofisica e Volcanologia reveal, in a paper in Geophysical Research Letters, that a large collapse on the east flank of Sicily’s Mount Etna volcano, around 8,000 years ago, triggered a major tsunami in the eastern Mediterranean that may even have led to the depopulation of coastal areas in what is now Israel. A tsunami threat from collapsing volcanoes persists in the Mediterranean, with two landslides on the flanks of the island volcano of Stromboli triggering locally destructive tsunamis in 2002. The future stability of this volcano, which triggered a major regional tsunami around 5,000 years ago, is addressed in a paper by S. Falsaperla23, and coauthors, at the Istituto Nazionale de Geofisica e Volcanologia in Catania, Sicily. Future volcanic threats and their potential consequences As another year goes by without a major eruption in a developed country, and no large insured loss arising from volcanic activity, so market interest in the volcanic threat remains low. Nevertheless, at least 1,500 (and possibly as many as 3,000) volcanoes have the potential to erupt in the future, many located where they can cause major damage and huge losses. At the top end of the scale are the volcanic ‘super-eruptions’, which have the potential to bring about a so-called volcanic winter; a global freeze lasting for several years. Steve Self78, of the UK’s Open University, presents a comprehensive review of these events, the last of which occurred in New Zealand around 26,000 years ago, in the Royal Society’s Philosophical Transactions. Self points out that such an event, occurring today, would cause major disruption of our global society for months to years, bringing a high and sustained cost to global financial markets. Zeroing in more closely on the specific impacts of explosive volcanic eruptions, Robin Spence85 of the Department of Architecture at the University of Cambridge, and co-researchers, presents – in a paper in Natural Hazards & Earth System Sciences – a GIS-based model capable of determining building damage and fatality and serious injury rates. The paper focuses on the Soufrière volcano on the Caribbean island of Guadeloupe, but has a general application to a range of eruption scenarios at any explosive volcano for which appropriate population and building portfolio data are available. A number of studies published in the last twelve months examine aspects of volcanic hazard and risk at specific volcanoes. In relation to the threat presented by Vesuvius (now quiet for 63 years) to the Naples region of Italy, Antonella Bertagnini4 of Pisa’s Istituto Nazionale de Geofisica e Volcanologia, and colleagues, writing in Geophysical Research Letters, determine the length of early warning that might presage the next eruption. Based upon geophysical precursors to the 1631 eruption, the most lethal in thousands of years, the authors report that the warning time prior to the next eruption may be as short as 2 – 3 weeks. This would require very rapid and well organised evacuation of the 600,000 or so inhabitants of the high risk zona rossa immediately adjacent to the volcano, if serious loss of life is to be avoided. Still in the Naples region, Francesca Bellucci3 and Giuseppe Rolandi of the Dipartimento di Geofisica e Vulcanologia, Università di Napoli, and Judy Woo and Chris Kilburn of the BUHRC present, in a Special Publication of the Geological Society of London, an assessment of the volcanic threat from the Campi Flegrei volcano. Located to the west of Naples, this low-lying caldera volcano last erupted in 1538 and has generated gigantic explosive eruptions in the past. Based upon studies of ground swelling and subsidence in the area since Roman times, including episodes of rapid uplift in the 1970s and 1980s, the authors propose that the volcano will be characterised by an elevated possibility of an eruption in 80 – 90 years. On the other side of the world, the New Zealand city of Auckland is also exposed to the volcanic threat, although of a rather different kind. Auckland is located in a so-called monogenetic volcanic field, within which a small, relatively short-lived volcano can ‘pop up’ with little notice (Figure 13). The Auckland Volcanic Field consists of 49 volcanic centres, the last of which (Rangitoto Island) erupted less than 800 years ago. Writing in the Journal of Volcanology & Geothermal Research, Bruce Houghton36 of the University of Hawaii, and colleagues, examine the likely impact of such an event. They conclude that despite the expected small size and intensity of a future eruption, the high density of buildings and lifelines would ensure that the effects were severe. Complete destruction across an area of 30 – 100 hectares could be expected within hours of the start of the eruption, resulting in losses of between US$130 and 900 million. Additional damage could be expected further afield, due to falling ash, as the eruption continued. The warning time for such an event could also be extremely short, as indicated by Steven Sherburn81 of GNS Science in Wairakei (NZ) and co-workers. In a paper published in the New Zealand Journal of Geology and Geophysics, the authors suggest that the current seismic monitoring network may provide as little as a few weeks to a few days notice of magma approaching the surface prior to an eruption within the city environs.  Figure 13. The Mount Eden cinder cone, part of the Auckland Volcanic Field, with the city’s Central Business in the background. Courtesy: Wikipedia. Volcanic eruptions are always heralded by precursory warning signs, including increases in seismicity and ground surface deformation, providing opportunities for successful prediction, days to weeks ahead. In a paper published in the Journal of Volcanology & Geothermal Research, Olivier Jaquet39 of Colenco Power Engineering (Baden, Switzerland) and colleagues, present a new probabilistic framework for eruption forecasting, based upon activity of the Soufriere Hills volcano on the Caribbean island of Montserrat. The methodology uses statistical concepts to identify precursors and forecast specific volcanic events, and has been used to make probabilistic predictions about the onset of lava dome growth at the Soufriere Hills volcano. «back to top« |
Geological Hazards New Madrid, Cascadia and California quake risk revisited New seismic studies on a future Istanbul earthquake Predicting earthquakes: are we any closer? The Parkfield Earthquake Prediction Experiment (PEPE) and the seismic precursor issue Patterns of seismic activity in time and space Earthquake loss estimation and hazard analysis New geological studies highlight tsunami hazards across the world Collapsing volcanoes as major tsunami sources Future volcanic threats and their potential consequences |