National usgs marine Geohazards Workshop Menlo Park March 1, 2011 Notes by Amy Draut Tom Brocher, welcoming remarks
Download 157.74 Kb.
|
- Navigate this page:
- Jason Chaytor
- March 2, 2011 Michelle Coombs
Steve Kirby, notes: Marine earthquake and tsunami hazard themes and trends for active margins (Where, how big, how often, and what impacts?)
Emile Okal: if I tell you there’s a 50% chance I will live to see a tsunami in LA, as of when I died, by the time I die, either you were right and one happened or you were wrong and it did not. How much beyond that kind of statement do people want to see? We live only once, so what is a meaningful way to represent probability? Hyndman: Need probabilities that are appropriate to the societal and economic application. Must deal with who will use the information. Craig Weaver: In OR, WA, they want to know for a school that’s in a tsunami zone, what is the probability it will get wet over a certain time frame? They often now prohibit critical structures from being built in high risk zones. If risk is only once in 500 years, they won’t do anything; these are usually poorer communities. If risk is once a year, they will act on that. In Grays Harbor, they moved hospital to on top of a hill based on probability of getting inundated. Guy Gelfenbaum: there’s a significant question as to whether or not Cascadia event would significantly shake urban areas. More research is needed in to how much shaking should prompt evacuations etc. Emile Okal: Sumatra latest case – tsunami earthquake with weak shaking (October 25); the message had been that you can’t evacuate every time you feel an earthquake, can’t cry wolf over M5 and other small events. At what level do you start worrying or tell people they should worry and evacuate? Educating people is difficult because two earthquakes may feel similar but have different mechanisms and danger. Amplitude and duration are both factors that should go into education. Craig Weaver: NOAA and their tsunami prediction work is really only interested in waves; some places will be damaged first from ground shaking, which will damage evacuation routes before tsunami waves even arrive. Really need more focus on interactions these problems. Jody Bourgeois: many other aspects of zoning, costs, etc. also affect how people plan their futures, that go into what probability assessments affect. It’s not just about loss of life. Much damage is done by smaller events, not just the bigger events. Kurile tsunami in 2006 (relatively small) did $10 million damage to Crescent City that was not anticipated. [Patty M.]: Christchurch, New Zealand: mainshock (7.1 strikeslip earthquake right on the coast) was weathered fairly well but aftershocks damaged infrastructure. Liquefaction severity was noteworthy. Emile Okal: In that case, soil gradations also liquefied, caused problems. Some of the buildings were retrofitted. Lucy Jones: there, their building codes are very similar to ours (New Zealand), and had much damage. Carolyn Ruppel: organization and focus of the theme list. Slides can be triggered. Framework geology that has static parts, and others are secondary effects. Sam Johnson: we want swath bathymetry everywhere, ideally, for a whole suite of tectonic geomorphology, not just fault morphology and activity. Roy Hyndman: Local tsunami modeling, nearshore bathymetry needed for accurate tsunami effect modeling. And importance of submarine landslides. How often are the main tsunami waves caused by landslides rather than earthquake-caused displacement? Are those just an exception? Emile Okal: it’s an exception that is common enough that it’s not 90% either way. It’s probably one in five (of big tsunamis). It’s not quite the norm, but it’s very common. It is a well-held idea that strikeslip earthquakes shouldn’t generate tsunamis because they don’t involve vertical motion. But normal mode theory predicts them to generate tsuamis just as efficiently as any other kind. Reconcile this because at tip the of a strikeslip fault there is vertical motion. Remember M8 earthquake south of New Zealand just a few days before the 2004 Sumatra tsunami that produced a (smallish) tsunami and has been largely forgotten because of the larger event a few days later. The tsunami source itself is always dipolar, with some stuff moving up and some stuff going down. In strikeslip source, the same kind of thing happens, though may be on smaller scales. But certainly some strikeslip earthquakes can produce tsunamis. Craig Weaver: 15% of tsunami warnings that are put out are caused by strikeslip earthquakes with underwater landslide component. So they are not insignificant. Bruce Jaffe: some sediment-transport models also are aimed at improved interpretation of sedimentary record to allow you to infer tsunami wave height and other wave characteristics. This may be useful for distinguishing between near- and far-field tsunamis and those caused by earthquakes or landslides. Also, about effects of subsidence and hazards associated with that – should that be on the theme list [refers to an early comment by Sam Johnson]? Sean Gulick: need local bathymetry to get inundation models right. In Haiti, and some other strikeslip systems, e.g., Jamaica, you can predict with large sediment volume and steep slopes, and be more targeted about producing landslides. Lucy Jones: This may belong with NOAA because they do the warnings, but it’s not just a matter of whether tsunami will be produced. We evacuate sometimes at very miniscule risk levels, assuming the only bad thing would be to miss a tsunami. For, e.g., nursing home patients etc., it is very difficult just doing the evacuation and poses risk to people. There’s more than just making sure the evacuation “on/off switch is always on”. What mechanisms, etc. will really be likely to generate a tsunami where a full-on evacuation is required? Steve Kirby: giving public opportunity to weigh the evidence when deciding evacuation requirements. Brian Atwater: that’s very complicated, and at the state/local end, society can get in all sorts of trouble encouraging people to think for themselves in cases like that instead of just telling people to evacuate. When some counties tell people to evacuate and others don’t, that opens up doors for problems. Lucy Jones: research could be done to estimate/predict size of tsunami or increase range of information that could be provided. Brian Atwater: but emergency managers want more simple directives of when to order evacuations. Emile Okal: Spent the night last year in Tahiti during the warning for 2010 Chile tsunami. Six hours before waves arrived, they had the model and were pretty confident in it. They tried to provide local people with estimate of wave height, (4 m), and went into the field later and found 3.7 m. Got calls from hospitals that people in decision-making places don’t want to make the decision. The hospital instead called them asking whether they should evacuate. They evacuated 68 islands in Polynesia, but only 2 boats were lost (whose owners did not take them out to sea). Probably were too cautious in recommending evacuations, but generally was a success because no injuries reported and only those two boats lost. We’ve been trained to be extra cautious in many parts of society. This question transcends science. Uri ten Brink: with local tsunamis or landslides, research is very important at telling whether the sediment type is such that it would move quickly or cause mass flow. We can contribute more than just saying there will or won’t be a tsunami. Steve Kirby: about priorities – ground motion prediction is now very sophisticated, accounts for local side effects. For given earthquake size on a particular fault, models reveal where construction and population are vulnerable. For a given offshore source, e.g., San Diego fault system that Holly discussed, can we use tsunami modeling to identify crucial potential landslide sources that would impact our most vulnerable coastlines? Are we there? Emile Okal: yes, it has been done. Steve Kirby: How widespread? Uri: you can model whatever you want, if your source is well characterized enough. This is being done, by NOAA and Geist and Parsons, focusing on targeted localities where we are worried about risk to populations or infrastructure. Steve Kirby: what can USGS do to prioritize pieces of the seafloor that potentially could produce tsunamis that affect vulnerable coastlines? How do we prioritize what part of the sea floor merits the most investigation of sediment record and so on? Eric Geist: you mean a disaggregation of a place like Los Angeles? To do that, need to know what sources to use in the first place. Roy Hyndman: that requires being very specific about who will use our products and what kind of information they need. We should make a careful list of who will use and who will pay for it (Bill Schwab). We increasingly have deep sea infrastructure, including in deep sea. Grand Banks earthquake broke every transatlantic cable; that would be a major impact in communications. Shallow sea – wind farms, e.g. Think about offshore infrastructure vulnerability, not just that of coastlines. Patty M.: 3D seismic velocity models for marine areas of subduction zones would be very useful. Deployment coming up offshore Cascadia will be good opportunity to improve that. [unknown speaker:] We need groundtruthed rock properties in faults, to know what are frictional properties, pore pressures. Without those real world properties, models aren’t that well constrained. Guy Gelfenbaum: prioritizing – social scientists would have a lot to say about where we do hazards research and how to use it.
Major submarine landslide events and findings: Grand Banks 1929 – broke submarine cables in series. Timing of breaks used to calculate velocity of “something” that came down and broke them – turbidity proceeding south from Newfoundland, left nice deposit easily imaged today. Newfoundland had on-land tsunami that killed people at same time. Smaller landslides failed on Newfoundland slope, generated tsunamis, coalesced into turbidity current. Alaska 1964 – most deaths were from submarine landslide-induced tsunamis. Fjords – Valdez, Whittier, Seward. Hurricane Camille, 1969 Gulf of Mexico – began to appreciate that hurricanes can cause submarine landslides, major damage to offshore drilling structures. Nice, France, 1979. Aseismic. Construction on Nice airport, triggered landslide and tsunami that killed people on beaches. Lawsuits – was it caused by construction? Papua New Guinea 1998, there were 2000 people killed onshore. Earthquake-induced submarine landslide caused the tsunami. Storegga Slide, offshore Norway, imaged headwall of slide. Largest gas field in Norway. Could gas be safely exploited? Very large landslides can recur in same place, related to sea level. GLORIA mapping, 1980s Hawaii – lots of large landslide deposits found. Now multibeam mapping is finding and refining more. Hazards to offshore and coastal infrastructure. Also major factor in canyon development and turbidity current generation. Major factor in development of continental margins. How we find them? Visual observation: shoreline disruption, damage to offshore structures, anomalous tsunamis. Geologic/geophysical: Characteristic surface morphology, characteristic disruption of bedding. Coring/dating: of slide material and overlying deposits – tells you age, chronology. Estimate ages of seismic horizons extrapolated from nearby borings or other means (e.g., assuming a sedimentation rate). Normark – study of large landslide offshore Los Angeles. Dated cored material (with Mary McGann). Got age of about 9 ka. Submarine landslides happen worldwide, not just in active margins. Best environments: Fjords and active river deltas, submarine canyons/fans (steepness, activity), volcanic islands, open continental slope. Some major ones offshore North Carolina. Resurrection Bay fjord, Alaska, offshore Seward has big landslide blocks attributed to 1964 earthquake. Mississippi delta, too, margin almost totally covered with landslides, and there are many oil fields there. Have to design platforms to accommodate landslides coming through because they are so common there. See Orange et al., 2003. Greene et al., 2002 mapped failures in Monterey Bay. Pre-conditioning factors: underlying structures (e.g., diapirs), weak layer development/intraformational deformation, sediment accumulation, groundwater flow, development of excess pore water or gas pressures. Then a trigger causes failure. Failures can recur near same location, like reloading/refiring a gun. Sedimentation fills a depocenter. Amount of fill reaches a critical amount (loading the gun…) ambient triggers, e.g., earthquakes, induce failure (fire the gun). Loading and firing can be related to glacial cycles. Triggers: sediment loading, erosion, waves, gas, gas hydrates, groundwater seepage (those 5 are climate-related), earthquakes, volcanoes, diapirism, human activity. Human activity such as construction offshore Nice; a sewer facility offshore Seattle caused a failure. Jason Chaytor: pressing research questions on submarine landslides. Have all the significant submarine landslides along US margins been identified? Probably not. Many mapped, but more work to be done. Why, and how frequently, do large landslides occur at the same location? Investigating initiation, propagation, evolution of failure. Sediment composition, physical properties? Sedimentation rates, biogenic components, pore pressures? Local variation along a particular section of a margin? Determine age of most recent significant landslides (hazard evaluation and driving forces) and sequences (the latter are critical for probabilistic analysis). What do remains of previous tsunamigenic landslides look like? Can we always recognize them, distinguish tsunamigenic from non-tsunamigenic? Scar characteristics: headwall depth, volume, area, slope of failure surface, rotational vs. translational etc. Deposit morphology: disintegrative vs. cohesive, etc. Goleta Slide offshore Santa Barbara, short runout, shallow water, 2 km3 in last phase of sliding. Compared with 60 km long, 150 km3 of material off Georges Bank, in deep water. Blocky debris. Which one’s tsunamigenic? How important are climate and sea level in determining frequency and magnitude of tsunamigenic submarine landslides? Homa’s paper, 2009. Many date to last glacial maximum, but we haven’t dated that many of them. How important is hydrate dissociation in causing submarine landslides? Hydrates dissociate if temperature rises or pressure drops (diverted currents, lowered sea level). Gas can increase pore pressure, contribute to cyclic loading. Grozic (2010) lab perspective. Examples of hydrate dissociation-induced failure have to be identified. Comment from Guy Gelfenbaum: After 2001 earthquake, NOAA in Puget Sound imaged in fjords, urban areas off Seattle/Tacoma with big port facilities. Deltas were very steep with small/intermediate failures. Do we have the tools to measure those slopes there and determine how likely they are to fail? Geotechnical knowledge? Homa Lee: there are tools used on land all the time to predict landslide risk. Rob Kayen worked in Puget Sound with Homa in 1990s. Not sure if anyone has assessed those deltas’ stability quantitatively. But they are certainly real hazards. It’s not a question of tools/capabilities (the technology exists) so much as money. Bill Schwab: biggest limitation is always getting deep samples – cores and borings. Homa Lee: Industry can do it. Peter Flemings: most important is understanding how near you are to failure. What is the in situ state, based on field observations and monitoring? That is where new advancements should come. Homa Lee: Yes, need to measure geologic strata properties and pore pressure and is hard to measure. Uri ten Brink: try to get at measuring ambient noise without needing to take cores. Not making much progress in understanding initiation and sediment mobility. With better tools now, could make progress we hadn’t made before. It’s unresolved whether we have equal probability through time of having landslides. Also, on gas hydrates: consensus is that dissociation of hydrates does not contribute much to landslides. Jason Chaytor: some people disagree with the last point, think gas hydrates are important. People are split on this. May matter more in shallow water than in deep water. This is a pressing research question. Tom Pratt: once a landslide goes, wouldn’t the next likely area be places right next to it? And, some important landslides happen in smallish lakes, not continental margins, but people still live near those places (Lake Washington). Carolyn Ruppel: gas hydrates project official position is that it’s not a big thing to invest time/money in. There’s only anecdotal evidence for slope failure and hydrate dissociation. Overall, it’s not thought to be very important and should not be a priority. Hard to study anyway, because if you go to where a failure already happened, the conditions have completely changed. It is probably only a minor overprinting factor compared to other much more important priorities on this list. Charlie Paull: the gas hydrate problem is intriguing but we have no substantial information one way or the other with tools that would allow it to be resolved. He doesn’t think the question is really answered yet so discussion could go on forever. [unknown speaker]: there’s a lot (or could be) of crossover between lab and field studies/problems. Could you see any fundamental difference in physical properties between tsunamigenic and non-tsunamigenic landslide that would have allowed you to predict that tsunamigenic potential ahead of time? Homa Lee: best bet would be if you could remotely sense high excess pore pressures. Otherwise, pessimistic about being able to predict tsunamigenic landslide by running a seismic line over it. We can get at what kinds of materials will disaggregate and flow, and density and material type. Jason Chaytor [in response to Steve K’s question about prioritizing along coastline]: in places we can see where/ how slide initiated on slide plane, then stopped. Why did it stop? Will it happen again on that slide surface? Homa Lee: giving Rob Kayen’s talk on local slope stability. Simplified regional models of slope failure have been developed within GIS, although some are not well ground-truthed. Existing models for slope stability on land can be used offshore if adequate property/environmental data exist or can be assumed (e.g. pore pressure), and those can be hard to measure or estimate. Landslides change, sometimes transforming into sediment flows and turbidity currents; transformation processes not well understood. Initial velocities of landslide blocks as they convert into flows are critical for tsunami prediction. To do it right, would need sediment stiffness, strength, density. Coring disturbance alters important sediment properties and limits their usefulness. Deep samples are expensive, too. Non-destructive, in situ analysis partly overcomes these limitations. Rob’s work: surface wave studies are used to assess shear stiffness indirectly. Gets liquefaction potential. Inversion algorithm to back out shear waves. Uses these to improve engineering community’s estimates of liquefaction potential, applied to cities including Kobe. Plans to modify methods to use in marine environment and analyze liquefaction potential. Pressing research: improved simplified regional assessments; improved understanding of initial motion, for use in tsunami models. Better in situ, non-destructive property measurements to adequate sub-seafloor depths. Improved, more economical methods for estimating/measuring pore or gas pressure. Emile Okal: Landslides should only recur in same place if you haven’t repressed or used up the conditions for why it failed. Have we ever observed “aftershocks” of landslides, events that “finish the job” a few days or weeks after one “main” slide? Charlie Paull: in Storegga, there was one big slide (at 8200 ka) and then smaller ones, but don’t know how much time separated them – hours or decades. It’s several thousand km3. How “clean” these are is a fundamental question – whether subsequent, “aftershock” slides occur from pressure diffusion. It’s a kind of stress transfer as these systems propagate over great distances. Retrogradational, backstepping systems, domino effect. Sean Gulick: any active-source OBS experiments can get shear waves but the data are rarely used. Could data-mine those? Uri ten Brink: Trying to do exactly that with OBS data. Two methods: compliance of ocean sediment (depends on shear modulus), and cross-correlation of ambient noise. This is an area ripe for more technological advances, and to do more at shelf edge and in deep water. Steve Kirby: Has there been lots of oil/gas field development where infrastructure has been damaged by submarine landslides? Does the industry think about this problem much in terms of probability of big investments being compromised? Homa Lee: Industry considers landslides a hazard that they design against. They build to accommodate it. Yes, there is damage, especially during hurricanes. [unknown speaker]: does shaking produce a non-linear response? How important is that to slope stability? Homa: Rob would have to answer that [Rob’s not here]. Steve Kirby: if you had a long enough array (air gun, acoustic waves), presumably you could get shear wave velocity profiles in deep water.
USGS Volcano Hazards Program monitors volcanoes in Alaska, western continental US, Hawaii, northern Marianas. Volcanic hazards are varied in impact, tie scales, and precursory activity; also study geologic history for recurrence interval and scale, style of eruptions. Marine volcanic hazards include volcanic landslides and submarine and island eruptions. Examples of volcanigenic tsunami disasters: Oshima-Oshima Island (1741) collapse, Unzen 1792, Krakatau 1883, Ritter Island, PNG, 1888. This type of hazard has low recurrence interval so must look internationally, but US volcanoes certainly pose these hazards. Ocean-island landslides. Hawaii – Morgan et al, 2009. Debris-avalanche type landslides; longer than they are wide. Some slumps, too. The big ones on N side of Oahu/Molokai (Nu’anu and Wailau slides; Garcia et al., 2006) are some of the largest on Earth, up to 5000 km3, similar to Storegga dimensions. Catastrophic emplacement, 100 to 200 m/s. Recurrence interval ~100 ky, most likely during shield-building stage, when there’s lots of magmatic output. Satake et al., 2002 modeled tsunami from Hawaiian slides. Devastating locally, high trans-Pacific wave heights. Slower-deforming flanks, Kilauea. Onshore, 8 cm/yr motion, 7 cm/yr uplift on bench; slow slip events. Phillips et al., 2008. Less dangerous as debris avalanches, but seismicity of Kilauea is notable, including M7.5 Kalapana quake in 1975. Slow-moving slump feature might transition into debris avalanche? There are debris blocks at its toe. What are conditions/triggers for large scale flank collapse? Composition, shera strength, pore pressure, thermal pressurization, earthquakes, explosive eruptions, sea level change? What causes more gradual flank motions – seaward creep, aseismic slip, flank eqs? What are frezuencies, magnitudes, distributions of large volcanic landslides? Aleutians – volcanic landslides at convergent margins. Inherently unstable material makes up stratovolcanoes – magmatic injection, hydrothermal pressurization and alteration, regional tectonics. Mt. St. Helens is an example. Failures can be deep-seated, removing much of edifice. Or thin-skinned, driven by high eruptive rates (Etna, Stromboli). Or Bezymianny-style, accompanied by directed blast like Mt. St. Helens. Typically involve subaerial edifice but if directed right, can flow into sea. Smaller volume than ocean-island slides. In Aleutians, debris avalanches (not slumps). In central part of arc, where volcanoes sit along north edge b/c volcanism center has moved north. Mainly travel to north. Don’t have evidence for syn-collapse eruptions. Range of scales. Most known only at reconnaissance level. Example: Kanaga. Waythomas et al. (2007); Cooombs et al. (2007). Runout length 25 km; submarine debris area 220 km2. Another: Gareloi, mafic stratal cone, frequently active. Submarine debris area 95 km2. Kiska: Large landslide, 196 km2, 59 km3 volume, assuming 300 m thickness. Block-rich proximal facies, fines distally. Displaced softer “upper series” sediment on northern shelf on shelf. These pose local, not trans-Pacific hazard because of smaller volume; could cause much damage to local, fishing-based communities in Bering. Runups 10s of m locally but dissipating quickly. Generally, Aleutians has less swath mapping coverage/quality than the Marianas. Augustine volcano, lower Cook Inlet, has 13 debris avalanches in Holocene, near population centers. Waythomas et al., 2006. Most recent debris avalanche in 1883, caused 2-3 m wave runup on Kenai Peninsula. Did slope-stability modeling for Augustine. Contributing factors earthquake shaking, weak rock, magmatic intrusion. During 2006 eruption there, NOAA would issue tsunami warning if shallow M>4.5 eq occurred, but didn’t happen. Submarine eruptions: volcanoes shallower than 200 m of greatest concern. Evidence exists for these in geologic record but mechanisms, hazards poorly understood. Surtsey-type eruptions in Tonga, March 2009. Hawaii has Loihi volcano off Big Island. It’s in 1000 m water depth, but b/c of much eruptive and earthquake activity they keep an eye on it. In May 2010, submarine South Sarigan seamount in Marianas produced big plume to 14-15 km above sea level, poorly monitored. Krakatau 1883, enigmatic, submarine explosive pyroclastic eruption and volcanic collapse caused tsunami. Subaerial eruptions can also emplace material into sea that causes tsunamis. Waythomas and Neal, 1998. USGS role: Monitoring – precursory unrest common in volcanic crises; current monitoring focuses on subaerial volcanoes. Assessment – better first-order data needed in most cases. Probabilistic approach to marine hazards. Fundamental processes of marine hazards not as well understood as subaerial – modeling how eruption plumes work undersea. Most vulnerable: Alaska, lots of shipping, infrastructure; 80,000 aircraft/yr, 30,000 passengers/day. Primary hazard is airborne ash. 7500 year round residents in Aleutians (more in summer for fishing). Hawaii also, very vulnerable to waves, and generates trans-pacific effects. Data needs: seafloor mapping. Good now for HI, CNMI, not as much for Alaska. Seismic profiling, drilling into deposits. Piston coring, distal drilling to get frequency/magnitudes of past events. Subaerial tephra framework not well developed in Alaska. Many of these are same as for other marine hazards. Subaerial tephras as stratigraphic markers – interface with tsunami studies. Better seismic monitoring instrumentation. Partnering with GeoPrisms, Earthscope, other USGS programs. Emile Okal: skeptical that debris flows could move 100-200 m/s underwater. We have to be careful extrapolating scaling laws into marine environment. Eric Geist agrees. Michelle cites these as examples of work that has been done, but agrees geologic info and constraints needed to inform models better. Uri ten Brink: slides are not coherent wave, even for shallow submarine eruptions tsunami potential should consider that. Augustine too shallow for OBS work, challenging to do marine work around volcanoes. Steve Kirby: Do we know of any active submarine calderas in Aleutians? And, does Augustine have associated tsunami deposits? Dave Scholl: yes, and it would have been a bad day when that erupted. Michelle: yes, one partially exposed subaerially (Davidoff island), unknown bathymetry around it. Unknown age and character. We haven’t found conclusive Augustine tsunami deposits. Looking for hydrothermal alteration (Mt. Rainier, Baker) could be useful tool in marine volcanoes. Peter Haeussler: what other historical examples of submarine volcanoes producing tsunami? How good is documentation for pyroclastics entering water being a tsunami source? Michelle: Krakatau is thought to have been, possibly (during pyroclastic eruption) but enigmatic as to exactly what produced that tsunami. Pyroclastics are not thought to be a well documented cause. Montserrat, maybe. Steve Kirby: how much would tephra record at sea help assess volcanic hazards for Aleutians, southern Alaska? Michelle: it would help a lot, although wind directions control where tephra blows so the tephra record could miss an eruption depending on where you sample. Dave Scholl: in 2009 IODP drilling in Bering will help a lot with that, but there is much more work to be done.
Download 157.74 Kb. Share with your friends:
|