National usgs marine Geohazards Workshop Menlo Park March 1, 2011 Notes by Amy Draut Tom Brocher, welcoming remarks

UCERF (Uniform CA earthquake rupture forecast) v.2. Time-dependent rupture analysis. Get probability of different earthquake magnitudes over different time scales. In CA, M6.7 has ~99.7% chance of being exceeded within 30 years. Also define participation probabilities for where earthquake of various magnitudes could happen. Some of the biggest, M7.7 and up, could be generated offshore. Coastal areas have greatest shaking hazard. Crucial issue is deciding where rupture starts and stops, whether jumps over to other faults or not. All of CA’s active faults are within 5 km of another one. Obviously, huge economic losses from M6.8, 7.0, 7.8 simulated rupture on Hayward Fault (hundreds of billions $, especially because coastal sediments don’t withstand shaking well). What is magnitude-frequency distribution of earthquakes for certain faults? This question governs how we forecast great earthquakes, tsunamis. Need to look at geometric relationships in detail.

PTHA is like PSHA but includes far-field sources and numerical propagation models. Determine source model, then propagation-inundation model, and aggregate probabilities. Case study done for Seaside, OR, flood insurance rate maps. Bayesian likelihood functions to combine empirical and computational probabilities (Parsons and Geist, 2009), compare with population centers.

Submarine landslide probability, increased interest from Nuclear Regulatory Commission work. Few dates makes it difficult. Monte Carlo simulations. Future research: how to statistically test the models? It’s like flipping a coin once, when you have one big event.
David Oglesby: Dynamic rupture models - tell you about rupture and vertical motions on main vs. splay faults, likelihood of rupture jumping segment boundaries. But only as good as input and assumptions that go into them. Need good info on state of stress in the region, and fault geometry.

Chris Goldfinger: had situation like that in Cascades where modeling to produce a known tsunami was easier when assuming splay fault, but how do you assess the likelihood of that?

Steve Hickman: any dynamic model is very sensitive to rheology assumptions. How can you constrain those better in model?

David Oglesby: very hard to do experiments with high temp, high pressure, high slip rates. Can model based on earthquakes that have already happened to get some constraints. Drilling on faults, more experiments will improve model reliability.

Carolyn Ruppel: for Tom, what data are needed from offshore? How do you account for faults you don’t know about?

Tom Parsons: Geodesy, though hard to do offshore. Fault geometry is extremely important, need case by case assessment which is much harder offshore, determining where faults connect at depth. Huge source of uncertainty. Need 3D seismic data to find stepover zones, $$ is a big concern.

Steve Kirby: When your predictions aren’t met by historical record, seek more sophistication/complexity in models.

Sean Gulick: for Eric, Nankai PTHA, did they use 1700 Cascadia type event?

Eric Geist: sources included everywhere from there around to Kamchatka, etc. Was a demonstration project. For Cascadia source, logic was that was most likely source for 1700. Wasn’t comprehensive but captured a lot of the details.

Sean Gulick: in a place where secondary landslides are a big factor, sounds like net effect of the two would make PTHA result very useful to communities.

Eric Geist: yes, and now deal earnestly with landslides because of NRC funding. Extremely complex to deal with landslide tsunamis probabilistically.

Chris Goldfinger: did tsunami assessment for the Cannon Beach region in Oregon (state-funded). Didn’t consider landslide triggered by earthquake. But had conundrum where ground motion typically disconnected from tsunami.

Eric Geist: generally time history of displacement doesn’t matter. But have to be careful in how to deal with static sources.

Peter Haeussler: splay faults important in southern Alaska, at least 3 splay faults ruptured during 1964 quake. For modelers, if geologists find all those splays and don’t know when/which rupture, what can we do to help modelers?

David Oglesby: anything geologists can tell modelers about splay activity timing helps, but just knowing they exist helps. The more info, the better.

Emile Okal: if you’d done this 25 years ago, wouldn’t even have known about the 1700 earthquake. How do you handle the Marianas?

Eric Geist: more convinced by statistically well defined studies, and the best we can tell statistically is it doesn’t matter which subduction zone, the slip rate matters but Gutenberg-Richter model does well. We don’t have enough data to test much statistically.

Steve Kirby: all subduction zones aren’t equal. West Pacific island arcs deserve attention to historical record and there are no M8s well described for 8000 km of island arcs. Far field modeling locked up by NOAA (resources). But near field modeling needs more source details, bathymetry. Is that a niche for USGS?

Eric Geist: that’s where we’ve been working so far. From 2010 Chile earthquake, and others, want finite fault models produced by NEIC being fed into local models. Using that combination worked well for Chile and elsewhere. Need better communication with PMEL for transferring fault models to hydrodynamic models. USGS has/does lots of sediment transport work now, constructing PTHA models. For the Seaside modeling, relied heavily on PMEL. We have the capability but not necessarily the capacity.

Peter Flemings: Sub-seafloor oil and gas containment. ONLY dealing with issues below the mudline.

How can we preserve and build the required expertise to respond to large oil venting in US waters? Is there a USGS role beyond spill response? Macondo well (BP oil spill 2010). Overpressure generation by disequilibrium compaction. Overpressure occurs when low permeability, high sedimentation rate and high compressibility combine to prevent complete dewatering. Nearly constant effective stress below “fluid retention depth” (FRD) with pore pressure and lithostatic stress gradients almost equal. For fine, low-perm seds, fluid pressure increases and FRD decreases with increasing sedimentation rate. FRD is at the seafloor in many places. Sustained flow and high pressure if aquifer breached; enormous compressibility, extent, and permeability. Dynamic fluid migration mechanisms: hydrofracturing, or shear failure (frictional failure), e.g., from normal faulting.

Drilling-induced blowout: fluid propagates upward through borehole, pore pressure follows hydrostatic gradient, creating mechanism for pore pressure to equal least principal stress, drive fracturing along the stress – fluids migrate vertically upward. Hydraulic fracture propagation. Can be arrested by stress contrasts, leak-off into permeable units to side (sands), or material contrasts. 2006 Lusi mud volcano in Indonesia was induced by drilling (?) failure, leading to blowout expected to last >25 years. Davies et al., 2006, 2010.

Natural blowouts: many natural examples, e.g., on Gulf of Mexico coast. Trap integrity and sea floor venting from “blown traps” such as salt diapirs. Will reservoir bleed out everything in it? Can be long-lived, 10s of 1000s of years. Fluid, sediment expulsion at seafloor. Negative impedance craters, free gas. Vents are directly above structural crests of reservoir sands. Reilly and Flemings, 2010. Reseach frontier: coupled stress and pressure perturbation. Salt moves like a fluid in subsurface, get bizarre deformation/movement when dealing with sub-salt drilling risk.

Outstanding questions: how does fluid migrate vertically? Along faults or by hydrofracture? When/how does a vent self-heal? Sediments can self-load system. Can we estimate volume, duration of flow? How and after how much release does it heal? Can we better predict in situ stress and fluid pressure in marine settings? Current models are rudimentary.

Science community has ignored: we don’t practice riser drilling (case with fluid circulate to contain pressure), except for a very few studies. But industry does riser drilling all the time. So we (USGS geoscientists) have a very thin knowledge base about how we drill those pressured systems. Our culture has not been part of that step, we don’t have engineering and geoscience knowledge that industry uses all the time. We don’t study multi-phase systems, or practice hydrocarbon trapping, pressure measurement etc. This may be changing with CO2 sequestration studies now.

Reviewed timeline of explosion, fire, top kill, capping, static kill successful in August, relief well in September. Final kill/cementing in November 2010. USGS well integrity team evaluated geo hazards of shutting in the well and under what conditions (hurricanes, etc.). Well kill and cementing team did tech review, oversight. Possible adverse effects of shut in: hydrofrac to sea floor. Drilling the well walks fine line of avoiding blowout (mud window), each open hole drilled to specific width with nested liners, very complicated. If oil flows up around nested liners instead of through main hole, can cause blow out to start.

USGS well integrity team analyzed in situ stress and pressure transient tests, well logs, wireline, drill cuttings and gas analysis, sidescan sonar surveys, seafloor bathymetry, 2D and 3D high-res seismic lines, etc. Close collaboration with OSTP, national labs, BOEMRE, Coast Guard, NOAA for most of the summer 2010. Turn-around for scientific analysis was only hours, with much media attention. Assumed if hydrofrac would go all the way to seafloor unimpeded if 18” shoe failed. Felt that leak was possible, additional analysis needed. BP thought the reservoir was highly depleted though, in contrast. Rapid response USGS modeling of simulated pressure from preliminary reservoir model, had good fit of model to measured well-head pressures during shut-in, to determine the well was not leaking anymore (this was in July). Macondo sands are elongated stacked channels, long skinny reservoir. Compared 2D seismic surveys after shut-in to 2D and 3D before, to be confident well was not seriously leaking. NOAA ships ran sea surface sonar coverage looking for gas bubbles coming up through mudline, would have been first sign of impending seep. No new ones seen, only known existing ones. USGS also advised on procedures for well kill and cementing required knowledge of in situ stress, hydrofrac processes. Need to improve methods for monitoring oil/gas leakage before hand nad upward migration in marine seds, better use of seismic surveys to constrain leakage, better understand fluid flow pathways to sea floor. Under what conditions can fluid flow pathways heal? What limits total hydrocarbon release during underground blowout? Can entire reservoir empty, or will it self-kill? Need better modeling of hydrologic, poroelastic, geomechanical reservoir response during blowouts to guide containment, well kill, cementing operations.
Peter Flemings: what scientific expertise should USGS maintain?

Reservoir geology, flow modeling, borehole geophysics, in situ stress, geomechanics and rheology of poorly consolidated marine sediments, pore pressure/fracture pressure prediction in multiphase (gas/oil/water) systems, which industry does all the time. Drilling operations, drilling fluid engineering, pressure balance and well design, well control techniques, downhole measurements, well bore hydraulics. Goals: lower turnaround time in communication? Don’t reinvent the wheel.

Not effective to learn during disaster, need to have core knowledge base to provide faster solutions. How do we augment/identify research practice in areas from which you can transfer skills? Especially in riser drilling and multiphase systems; in many areas, USGS knowledge is only one person deep. Related areas: CO2 sequestration, energy-related projects, hydrates, geothermal energy, IODP, ICDP research drilling – e.g., ocean margins, SAFOD.

Geographic areas at risk: any margin with high sedimentation rates where drilling goes on: Alaska – Beaufort Sea, Chukchi, Gulf of Mexico, North Atlantic margin (Canada).

Roland von Huene: how was MMS involved in this?

Steve Hickman: MMS assisted with wellbore design issues, they supervised procedures after side kill was done; they did final plug-and-abandon supervision. They also identified where natural seeps occurred, potential other blowout identification. They have experienced reservoir engineers.

Walter Mooney: MMS knew routines, knew where seeps were. Practical, but if they got new problem requiring math, physics, less useful.

Roland von Huene: MMS was portrayed as idiots by media. What is USGS role?

Steve Hickman: they had good people on the ground in Houston. USGS role was to consider what if breach to seafloor. Would reopening well shut that down? No consensus. USGS important to mechanics of fluid migration, hydrofracs, pore pressure prediction.

Steve Kirby: with more deepwater drilling, is potential for similar accidents greater with time?

Peter Flemings: wells are getting more complex, and in deeper water. Macondo is like space shuttle situation, come back to just a process that wasn’t checked/rechecked. We can expect it to be worse, as more severe environments.

Steve Kirby: should USGS have standing team maintaining knowledge, capabilities on this?

Peter Flemings: USGS has not stayed abreast of technological advances in industry in past 30 years, but we’ll have to in order to provide good government oversight. We have totally different worldview than industry, to protect public interest not to make profit. Being impartial is essential to playing useful role.

Patrick Muffler: spent career in geothermal energy. Attrition of USGS geothermal research program has caused problems. Having our knowledge base rest with so few USGS people leaves our capabilities very vulnerable.

Amy Draut: how much of this technological expertise can we hope to learn about in USGS, given that much of it is proprietary to industry? How much useful contribution can we expect to make?

Hickman and Flemings: technology is much less confidential than info specific to individual oil/gas fields. Technology and even pre-drilling seismics is not necessary off limits; BOEMRE (MMS) gets it. But we have to be very careful, and they signed non-disclosure agreements.

John Haines: programmatic questions, things happen at high level above us often.

Carolyn Ruppel: from managerial standpoint – USGS would never have access to state of the art tools, to even learn how to do certain kinds of seismic work that would be needed in this context. We don’t get IODP work or similar experience endorsed by the Survey. We also have a mandate to not work on oil/gas production type stuff. How can we keep up our expertise given those challenges and lack of money?

Peter Flemings: USGS should identify/augment research practice that develops that skill set as much as possible. Such as SAFOD; not necessarily from coastal and marine program. Agency-wide issue.

Steve Kirby: You mentioned multi-phase systems (oil/gas/water)? Lots of dissolved gas in reservoir, correct? That boiled out during pressure release?

Steve Hickman: we considered static multiphase system (gh) for understanding pressure distribution. But had many issues. In situ it was all liquid but very gas-charged. Was single phase in the reservoir, so relatively simple to model, but as it came up the pipe it passed through the bubble point and became two phases, non-uniform. There was a plan in place to determine whether gas was thermogenic or biogenic, and other tests.

Dave Scholl: USGS role began with Santa Barbara blowout. Back then, it was hard to get proprietary information, so we started building up core capability. Increased when oil embargo happened. We’re back to where we were decades ago now, with only about two people who are experts on this in USGS, especially in the Pacific area.

Walter Mooney: we asked for well logs, and were interpreting them and much other data. We need to send USGS people to SEG, etc. and other meetings, form a geohazards response team ready to go who can maintain oil spill issue expertise in well logs, seafloor bathymetry, acoustics, high res seismics, sidescan, drilling records and borehole completion.

Steve Hickman: USGS raised issue of blowout to seafloor, worst case scenario which BP had not been considering previously (if broach and the relief well hadn’t worked). That got their attention.

Bill Schwab: just assessing who we have, what they know, where they work, would help because it took some thinking just to put together a response team when this accident happened. Took several weeks to convince NOAA to go do the surveys they eventually did. We did not have good communication and response to our requests from upper management, and NOAA ran the show. Need better interagency cooperation, which happened late in the game in this case.

Steve Hickman agrees. Agencies started out working independently, but better communication needs to be worked into the process.

Peter Haeussler: what about Beaufort, Chukchi Seas? If oil development there, what is USGS role in the future there?

Steve Hickman: general issues presumably apply but there’s a disconnect because BOEMRE is charged with offshore assessments; some of expertise needed for that would be with them rather than us. They don’t have many people who know much about seismic imaging and other necessary mechanics. USGS/BOEMRE need to work better together. And, USGS should do more framework geology studies in those areas; geodetic areas vary entirely in terms of geomechanical stress, framework environment below the mudline.

Carolyn Ruppel: last summer, shallow water Beaufort program was the first USGS has done in 30 years. We are flying blind there at this point, have no modern mapping or interpretation of sediment there. Chukchi – trying to get US/Russian permit, but now know almost nothing there. Hope we can get up to speed there before something bad happens; we don’t even know sedimentology of that area or framework geology well. We have nowhere near enough USGS experts who know about drilling and geoscience in those areas.

Steve Kirby: is it reasonable to think about financial assessment from oil/gas industry to contribute to this?

Steve Hickman: that assessment should include hazards.

Carolyn Ruppel: a committee is looking at this now.

Peter Flemings: industry is going to do a great deal more exploration, production, development in Arctic in next 10-20 years, expect major increase.

Emile Okal: an emergency is not the time to learn. We have same problem with tsunami warning arena. Nobody in tsunami warning centers has lived through a real disaster except Sumatra (far away). Is there something in the case of the BP spill that could teach us lessons that we can learn in earthquake and tsunami community? We haven’t lived through major earthquake disaster in US recently (except Loma Prieta, Northridge – those did not have major loss of life compared to what we may experience in the future). No one is still alive from 1906 San Francisco. How can we learn from what you went through with BP oil spill?

Steve Hickman: Federal gov’t got involved in BP spill at very high level. That may have set a precedent. Science played a huge role in decisions at high level. Chain of command, information input was received at high level. We were lucky to have critical thinkers in Ken Salazar, Marcia McNutt at high level who respected science in this administration and could interpret science/technology for the administration; the process worked well. Decisions were made in hours, overnight, although it took weeks/months to get all that process in place. Having good technical discussions earlier in the process would have helped.

Patrick Muffler: core competencies are also important in our senior managers (Marcia McNutt in BP oil spill was in a position to be very helpful). In other situations, senior managers have been bureaucrats without those core competencies.

Steve Hickman: yes, Marcia being in place had a huge impact.

Lucy Jones: How do you make sure you have those partnerships in place before hand? – by getting people used to using our information and products in non-crisis times, so they know to turn to use immediately when a disaster happens. Here’s a science need, here’s how our science fills it. This is justification for very basic research, making sure that connection is there and in place. On 1-2 weeks time frame after a disaster, we need to be more of a presence.

Science-based framework of a marine geohazard initiative: active marine margins +/- submarine slumps and landslides.

1. 
   3D structure of active margins.
   
   1. 
      Where are the faults and what are their 3D geometries?
      
   2. 
      Where are slides/slumps and what are their structures?
      
   3. 
      Where are submarine volcanic calderas and problematic volcanic flanks? Tephra chronologies.
      
   4. 
      Where are the sediment sources and sinks (depocenters) that could record event chronologies and size distributions?
      
2. 
   3D kinematics of active margins.
   
   1. 
      Fault slip rates (seismic and aseismic): marine observations and GPS
      
   2. 
      Earthquake chronologies (the dynamic part of fault slip):
      
      1. 
         Paleoseismic observations
         
      2. 
         Paleotsunamic observations
         
   3. 
      Slide-slump chronologies
      
   4. 
      Estimates of landslide velocities
      
3. 
   Dynamics of active margins: largely modeling.
   
   1. 
      Ground-motion forecasts for model earthquakes
      
   2. 
      Probabilistic ground motion forecasts, PSHA, and mapping
      
   3. 
      Tsunami modeling of individual eq sources and aggregate tsunami wave field forecasts, PTHA
      
   4. 
      System-level 3D fault rupture modeling of a margin with rheological and observational constraints. Stress transfer, temporary “barriers”, etc.
      
   5. 
      Slide initiation and evolution using dynamic hydrodynamic modeling and relevant slide “rheology”
      
4. 
   Geofluid containment in oil and gas sedimentary basins
   

It’s really hard to separate paleoseismic and paleotsunamic observations using the sedimentary record.

We shouldn’t use the term “active margins”. We also work on passive margins such as northern Alaska or Gulf of Mexico, and there’s nothing passive about passive margin. Strong feeling from group that we should not leave out “passive margins” from this discussion, especially of slope stability and what triggers slope instability.

Homa: uncomfortable having landslides broken up and put in with earthquakes.

Seafloor instability should be treated as a separate problem.

Carolyn Ruppel: this list doesn’t constitute national priorities; wants to revert to the earlier list in program as basis for discussing a national program.

Steve Kirby: not comfortable with us breaking out for discussions according to individual margins, since we are to discuss national program. Carolyn agrees and wasn’t suggesting that.

Craig Weaver: have an assessment of our needs in what we’d require to go forward, outreach, partnerships, data, communications.

This will all get discussed tomorrow but we want to start by putting the science discussions first.

Links to shoreline change from erosion? Bill Schwab: although that is a large hazard, it’s a large part of existing CMG program; not addressing it here because it’s so linked to much longer temporal processes. We want to keep focus on more instantaneous hazards.

Talk about geologic “framework” rather than “structure” since it encompasses structural geology, stratigraphy, crustal structure and composition, etc.
Finally decided on just:

1. 
   Earthquakes
   
2. 
   Tsunamis
   
3. 
   Landslides