*Contributed by Jay Patton, Rob Witter, Tom Brocher, Alex Grant, and Kevin Schmidt (originals downloadable here)
|Current and future project support|
All participants introduced themselves. Rob Witter and Jay Patton agreed to take notes (thanks!).
The project is thought highly of in Reston.
In FY17 the project received $70k ($40k from programs, $30k from the NW Regional Director). The same request was made for FY18.
Guy welcomed everyone and echoed Brian's assurance that the project was widely highly regarded at the Program directors' level and above. He noted that the excitement and momentum started at ground level with scientists, which is how things should work. This week a CMGP subduction zone science planning workshop led by Nathan Miller and Janet Watt, illustrates the CMGP's endorsement. The project also is a great example of how different offices can work together.
|Larger context of SZS within the USGS|
Steve addressed the challenge of how to move ahead in tough budget times. He noted there is lots of support at higher levels, within the USGS and in Congress, for the science described in the SZS Plan. Evidence of the latter is the fact that the presidential cut of 15% has largely been restored.
New initiatives are highly unlikely (e.g. we will be lucky if level funded). However, by reprogramming and reprioritizing new efforts may be started, particularly given that there is lots of support at the program and mission level.
Guidance about subduction zone science priorities will be coming out in June.
|Knudsen, Keith (remote); firstname.lastname@example.org|
|Uplift along the Olympic Peninsula|
Began showing Lucinda Leonard's figure of coseismic subsidence sites, and Sean Paul pointed out Olympic Peninsula has sites with documented subsidence. Brian looked at many air photos and now LiDAR, which is best obtained from the WA DNR Lidar portal. However, there are gaps in coverage, that WA DNR will fill this year (all the way to CA/OR border). Brian has recently found a dozen sites with evidence of uplift (coseismic or rapid non-coseismic?). He showed examples of these sites in detail, noting that some uplift is definitely of late Holocene age. An abundance of well-studied archeology is a promising way to constrain a chronology.
ACOE flew entire coast, but there are problematic results (needs to be reflown).
Brian hypothesizes that lots of the coast of Vancouver Island likely looks like this.
Steve Hickman asks what caused the uplift and Brian speculates that it is tectonic. More field data are needed to determine if it is coseismic.
|New Bayesian foraminiferal transfer function of Kemp et al.|
Nelson, Alan (remote)
Witter, Rob (presented)
Rob presented for Alan about a summary of new Bayesian transfer function work, described in a BSSA paper in press by Kemp et al. Some of the revised subsidence estimates from the 1700 AD earthquake differ quite significantly from those in Wang et al 2013. Previous studies used weighted averaging regressions.
Rob reviewed the transfer function approach, which relates the abundance of a certain proxy to paleoelevation. The steps involved include visiting a site, and collecting modern samples along transect from upland to tidal flat to form a modern training set. These are input to a statistical model. The model is used to take fossil forams from seed core and relate them to paleo elevation.
Examples from various sites were shown.
Rob emphasized 3 points made in the Kemp et al. paper, that explain why coseismic subsidence estimates may change using this new method:
(1) Expanded training set provides more analogs for fossil data. Compiled data from 19 sites. Improves on 8 sites in OR used for earlier work in OR.
(2) BTF describe forams than may not have linear relationship to elevation (what they see in data). For forams that can be explained by a unimodal distribution,the BTF does not improve results much. For other species (e.g., Trachamminita) the BTF better handles non-linear relations.
(3) BTF allows use of stratigraphic priors (secondary proxy, other indications of RSL change: diatoms, lithology, etc.). These could be use to handle mixing contamination.
New BTF indicate greater subsidence north of Alsea Bay, consistent with prior estimates and implying that slip estimates might be increased there, though models may be insensitive to these new results. Kemp et al. only speculate why they find lower subsidence.
|Recent paleoseismic research on last 2 ky in Humboldt Bay, CA|
Much of the work shown is from J Padgett's PhD research, funded by the USGS Earthquake Hazards Program. Harvey summarized (1) a revised and updated paleoseismo history (1800 yrs), (2) info about 4 CSZ earthquakes, (3) new 14C ages, (4) microfossil investigations and (5) upper plate faults.
Set stage with map of Kelsey 2001 showing upper plate faults, particularly the St George thrust, Mad river, and Little Salmon faults and a Humboldt Bay paleoseismic site.
Humbolt Bay is a low energy depositional site that meets the criteria for paleoseismic dating (i.e. Alan Nelson criteria from 1996). Looked at 3 sites, taking cores in transects. Found correlations between all 3 sites. Measured 21 new ages, 18 of which were useful. Ages from OxCal model:
These are same ages as Talb Creek (south slough in Oregon that Rob presented. They have also looked at estimates of coseismic subsidence, using transfer function data (but not a BTF).
They want to utilize BTF, incorporating diatom data from Eileen Hemphill-Haley.
Space time diagram from the Swiss Hall site provides evidence for Little Salmon Fault contraction and regional subsidence, and that each time LSF ruptured it happened in conjunction with CSZ. The chronology in Arcata Bay matches the Swiss Hall site.
Art Frankel asks how results compare with the turbidite record, and Harvey replies that they match offshore, but not sure which ones. Chris notes that T-5 is important, and agrees with onshore ages.
|Onshore subsidence/tsunami record, uncertainty, future directions|
La selle, Seanpaul
Seanpaul reviewed the new Cascadia subsidence plot from Kemp et al's new BTF study, suggesting there are sources of error in transfer function: too small training set, no priors, no post seismic deformation included (Simon Englehardt working on this, using experiments by displacing modern peat into different environment, presented at GSA in 2017 seattle). He notes that tsunami deposits can complement subsidence record.
He showed a map of the distribution of 1700 AD tsunami deposits, and summarized recent work in Japan for the 2011 Tohoku earthquake, in which slip offshore corresponds with volume of deposit. The tsunami group has used Tohoku to validate tsunami sediment transport modeling. They've tested the sensitivity to bed roughness and sediment supply, to place uncertainty on volume estimates. Results are highly uncertainty and so just qualitative. Next, they want to bring the modeling to CSZ. Rob asks how sensitive are results to the slip model, and Seanpaul replies that they have not tested other models yet here, but have done this for other earthquakes (e.g., Sedanka).
Detailed mapping of 1700 AD tsunami in the CSZ Salmon River estuary, OR by Wendy Grant and Alan Nelson is shown ("they mapped the heck out of it") and a simulated tsunami, in which the simulation includes a splay fault (simulated tsunami shows more deposition than observed deposits). Various questions ensued about the modeling.
Seanpaul noted that Kelin Wang’s elevation changes could be incorporated in the modeling but have not been yet, and Rob Witter suggests their impact could be significant. Seanpaul says the current model overpredicts the data data.
The group is revisiting Bradley Lake (12 deposits) and Floras Lake. Last fall they collected cores in Floras Lake and found nice tsunami deposits, but need to collect geophysical data next.
|OSL dating of tsunami deposits from Cannon Beach|
Nelson, Alan (remote)
Seanpaul La selle (presenter)
OSL work has been done with Shannon ??? in Denver, to date 1700 AD deposits from Cannon Beach, Salmon River, Alsea Bay, and Siuslaw River. At Cannon Beach they find sand beds 2 (~800 BP) and 3 (~1000 BP) in core 13 actually both date from 200-300 BP. This changes previous interpretations of Cannon Beach stratigraphy (the best stratigraphic inferences can be wrong, so we need more OSL!). The group is planning to do OSL work at Bradley Lake also.
|SAFRR activities, Powell Center Tsunami project, NTHMP|
The USGS Science Strategy emphasizes the need to 1) identify tsunami sources, 2) assess sources and hazards and model tsunami inundation, 3) improve understanding of how tsunamis are generated, and 4) evaluate inundation history.
Stephanie summarized the NTHMP program, which is a consortium of scientists from the states, NOAA, FEMA, and USGS. The 2016 NTHMP workshop (57 attendees) improved collaboration between USGS and NTHMP mapping and modeling sub-committee. The states and territories want a long list of things, with 6 broad issues emphasized: 1) regional workshops on tsunami source characterization, 2) a general USGS-NTHMP workshop on tsunami issues, 3) bathymetry, and 4) collaboration between USGS and NOAA tsunami, 5) USGS to assist with identifying and characterizing sources, and 6) a tsunami source database.
A USGS Powell Center project now focuses on tsunami source standardization for hazards mitigation in the US. The first workshop is April 9-13, focusing on what is needed for a suite of sources.
The relevant SAFFR project is a scenario assessing the impacts of a tsunami generated by an M 9 Alaska peninsula earthquake, particularly on the Long Beach Port. It was developed by working closely with stakeholders. One big lesson was that institutional memory gets lost over a few years, when people move out of their positions.
2 yrs ago a 'Science for a Risky World' workshop was held, addressing the question of how to ensure USGS hazards info is delivered and incorporporated in risk assessments. A new Plan on this topic is about to be published; it contains 23 recommendations, 17 case studies, and 6 potential project ideas.
|Submarine landslides, earthquake recurrence, and geodynamics in southern Cascadia|
Brink, Uri (remote)
Uri described his study, which was done for different purposes. They measured the fraction of the margin slope covered by scars in 10 active and passive margins around the world, identified landslides and measured the scar fraction (fraction of area mapped that is covered by landslides) and mean slopes. The distribution of slide scars, shows that slope stability increases with increasing frequency of earthquakes (shaking strengthens) and decreasing of sedimentation rate in a non-linear way (sedimentation weakening). However, in central and southern Oregon there are many more slide scars than expected, relative to the regression lines. They hypothesize that abundant landslides indicates recent steepening of the slope, which could be due to various possibilities: 1) subduction of an aseismic ridge. 2) a buried propagator wake on JdF plate, 3) plate bending, 4) differences in composition of the outer accretionary prism, 5) slip behavior of arc (creep and aseismic slip), 6) locking, and 7) rotation of the PNW causing a seaward-directed motion in southern OR coupled with Siltezia terrance being closest to shore.
Uri presented evidence for and against these explanations, but concludes we need still to use marine geodetic measurements, detailed mapping of outer shelf and slope, dating of landslides, dredging outcrops to identify clues for recent seaward vs. vertical movements.
Chris Goldfinger notes that the sediments offshore southern CSZ are not much different than those in northern CSZ, and asks the landslides were mapped. Uri acknowledges that they used old multibeam data, and both note that there are differences in landslides mapped by Uri's group and in Chris's publications.
Ralph Haugarud asks about the role of Astoria or Nitinat fans and Uri said that they assumed that looking at large enough areas would average over local effects.
|Techniques for reducing uncertainty in describing and dating marine event layers|
Jason summarized many of the considerations and pitfalls of submarine coring studies. These included 1) technical/instrumental uncertainties (lab analysis are generally well constrained and reportable, field collection/sampling/environmental uncertainties abound), 2) event correlation, 3) sample location, 4) age corrections, 5) dateable material, 6) sediment mixing /bioturbation, and 7) sedimentation rates. He suggests that to reduce uncertainties we can re-evaluate existing data using new or modified techniques, collect more data (field or lab) using same methods, and/or develop new acquisition techniques. The WHOI group is working on the latter. Chris Goldfinger notes that the French have a new system to core without coring deformation (called CINEMA) using accelerometers; Jason says that they are trying to develop this same technique at USGS, but it is moving slowly.
|Source to sink sediment dynamics: influence on record of past events|
Jenna presented an overview or framework for how to think about sedimenation as a system involving erosion (sources), a transfer zone, and accumulation (sinks). She highlighted linkages between fluvial systems, tectonic basics, fault interactions, seafloor morphology, offshore storage locations, and dispersal pathways are all fundamental factors in understanding along strike variation in tsunamigenic earthquake and landslide recurrence.
Key characteristics of Cascadia include 1) the Columbia river dominant system in northern CSZ, 2) in southern CSZ there is large input from small mountainous rivers (Eel River largest sed budget per km2 the world) and smaller fluvial systems may not be connected to offshore systems, there are regions of landward vs. seaward vergence, and sectors with different morphologies (more on this later from Janet).
The CMGP will acquire new multibeam bathymetry, and have allocated $367,400 for this. Data will be acquired by NOAA (R/V Ranier, limited by seafloor depth) using a Kongsberg EM710 at midrange depths of 200-1600 m. Coincident backscatter and water column data will be collected too. These data should provide near complete hi-res coverage of the upper- to mid-slope when combined with existing data sets. Three areas have been selected.
Future work will involve geophysical surveys in the southern CSZ, to look at the deformational history and sediment accumulation. The intent is to partner with HSU and use the Coral Sea ship.
|CMGP plans going forward|
Nathan presented a summary of how the CMGP plans to address this initiative in CSZ, noting that the CSZ is its main focus, but they also are working in Alaska and Puerto Rico.
Their plans center around science questions and the products that feed into hazards assessments. Science questions address issues of recurrence, why the CSZ is aseismic, and rupture complexity. Products include a 3-D fault model, USGS QFF database update, a recurrence history, comprehensive bathymetry, sediment properties (site response, Vs30, hydrate map), hazard assessment inputs (rupture complexity, coastal uplift/subsidence, submarine landslides), event response plans, and drilling targets.
Data and analyses emphasize systematic 2-D MCS work, to constrain hi-res quaternary deformation (1m vertical resolution, with slip rates), with deep penetration for crustal and forearc structure. Also are reprocessing existing MCS (Lee, Ewing). Would like to see targeted 3-D MCS, and will encourage academic community to do this (can't fund this work directly), which would be particularly important for resolving how fluids transport through rocks.
Acquistion of comprehensive multi-beam bathymetry is an FY18 priority, in the 200-1600 m depth range, in partnership with NOAA. For depths >1600 m will need UNOLS vessels, or Ocean E___ Trust vessels.
Some discussion of the response to a CSZ and an Alaskan earthquake ensued. Guy Gelfenbaum and Steve Hickman noted that any response would be organized around addressing scientific questions.
|Core log correlations|
Joan describe how she is meeting the challenge from Goldfinger and Patton's Subaqueous Paleoseismology workshop at GSA to "try this". She has attempted to assess core log correlation, guided by these facts: correlation requires going from space/depth to time and correlation is used to infer synchroneity (same causal event), common tie-points are critical for correlations. A goal of the approach taken is to answer the questions are correlations objective, significant and reproducible?
She notes that few of the cores share independently dated tie-points, requiring assumptions that deposits are the same. This leads to some circularity, and correlations can then only demonstrate consistency.
Method borrows from techniques used for voice recognition (speaking at different rates), and she tested its preformance with synthetic data, in the presence of noise, and with missing tie-points?
Application to real cores from two channel systems and multiple channels lead to the conclusion that we need a way to do correlations with some independent, well constrained dates. Otherwise correlations are non-unique and not robust.
Lots of discussion ensued. Chris Goldfinger made a philosophical comment that others have tried to use computer correlations, but it "does not work". He notes that Schlumberger does this by hand because this perfoms well for them and people get fired if it does not work.
|Update on recent work: Sensitivities, Offshore, Lakes, linkage to the NSAF|
Chris summarized a melange of projects.
Compared onshore/offshore paleoseismology potential: Onshore land-level changes recording threshold can be better than 0.6 cm vertical using present understanding of transfer functions (Englehart et al., 2013). Generally the threshold is ~M8.2.
Offshore threshold combined triggering and recoding threshold estimated to be M 7.1 for Cascadia (Petrolia 1992), assuming available sediment (abundant), modest slopes, a depocenter, and PGA > tenths of a G. These conditions are met virtually everywhere on the continental slop with slopes > ~10 degrees. Basically, threshold is much lower for offshore than for onshore (offshore probably M 6, 125 times more sensitive than coastal subsidence). Bradley Lake good place to test this, where there appears to be inter-tsunami turbidites.
2017 MG paper: Analysis of new and existing cores lead to changed recurrence in northern margin, and latitudinal ranges of some segments. This includes 2015 Slipstream Slide (Hamilton et al., 2015), an independent data set, which matches Goldfinger work in northern CSZ.
Collected new cores at Oceanus Basin, Filling 100km data gap that controlled segment boundary. Hydrate Ridge Basin West (M9907-56PC) core data matched almost perfectly, despite being 90 km apart. The smaller SCSZ events extend into this 100km gap. Extending the strike length, increased magnitude and RI for Portland OR region.
Rob Witter asks what observations go into tie points and Chris says ghost tracing, and slip and slide to look for alternate fits, and if the fingerprint is unique, then the tie-point correlations will work.
Lakes and Ground Motions: Chris notes that lots of people doing this, especially Europeans in Europe, Chile, New Zealand, etc. His group is working now in Tarbook, Lelend, and Sawyer Lakes. The Portland water supply is helpful, so cored Bull Run Lake also. Most lakes are glacial scour lakes, small drainage areas (to eliminate climate signal). They collected multibeam and subsurface seismic (CHIRP), inventoried sidewall landslides, and used CHIRP data to track deposits around lake.
Lakes are 90% event beds of some type. Hemipelagic sedimentss are cm scale thick. Event beds, coarse, fining up, load casts, etc. interpreted potentially as earthquake records. Event beds have diatoms, but don’t look different from non-event bed diatoms. So, sources are probably from sidewalls.
An OxCAL P_Sequence model has been developed from Bull Run data. Southern segment events are not in the lake data.
Correlation between cores within Leland Lake are easy, though inter-event beds don’t correlate well. Major event beds correlate across Puget Sound and certain event beds are very unique (T-3 very unique offshore and WA lakes, latitudinal changes track).
Slope Stability: Analyzing shear vane and push cores to estimate what is minium ground motion required to destabilize slopes. But, want maximum estimates, and to do so reduced slope in model until slope stops failing.
Bayesian Analysis: Including multiple data sets: onshore, offshore, confluence test, correlation of down-core series, correlation of mass per event, hemipelagic thickness, seismic stratigraphy, etc. to derive a PDF showing how factors control sediment record.
Cascadia Creeping Section: Just completed mapping offshore structures in CSZ. Structure map more defensible, based on more modern data than earlier work (Chris’ PhD thesis). The creeping section is undeformed offshore, and north and south of creeping section are highly deformed. So, little permanent strain on short term (from GPS) and long term (from seismic stratigraphy).
Lake Merced Paleoseismology: Revisiting linkages between CSZ and NSAF. 75% of event ages overlap (11 of 15) so hypothesizing a stress linkage makes sense. They've found the 1906 turbidite in Lake Merced and used Chrono to get bomb carbon ages (1957 eq). The Lake Merced appears to match the northcoast section offshore, perhaps better than the peninsula section.
Scott Bennett asks if the lakes in the Puget Lowland are sensitive to upper plate (Seattle fault) earthquakes. Chris says they see T-1, T-2, T-3 in the lakes and between T-3 and T-4, there is an event, which could be a SF earthquake, but there are not many extra events. Brian Sherrod notes that there are some upper plate earthquakes in those time intervals, so that is a problem. Chris says these extra events are not robust deposits.
|Update on recent work|
|Compilation of offshore and onshore geologic map of Cascadia|
Ralph asks if an offshore/onshore map for all of Cascadia would be useful, how far north and south such a map should extend, and how much of the back-arc be included?
Jay Patton notes that the Mchlaughlin 2000 mapping was based on geomorphic expressions.
Chris Goldfinger says 'yes!', adding that he's been mapping offshore and can distinguish sediments from rocks.
Janet Watt notes that the CMGP offshore seafloor mapping project includes some of this already. Ralph says this is great for shallow areas.
Ray Wells says there is lots of structural data, which could highlight gaps in data.
Rick Blakey suggests looking for an offshore expression of the Yakima Fold and Thrust Belt.
Harvey thinks this would be especially interesting.
Various scales are suggested, and people note that the best mapping scale will vary according to bathy resolution and seismic resolution. Not a good idea to dumb down to least common denominator.
Janet Watt notes that most people don’t look at printed maps, other suggest it should be an ArcGIS product.
Watt, Janet, email@example.com
East, Amy, firstname.lastname@example.org
Johnson, Sam, email@example.com
Modern Earthquakes & Models
|Update on simulations|
Wirth, Erin (presented, remotely)
Frankel, Art (remote)
Erin described the simulated ground motions she and Art Frankel have been generating for the M9 project. This is a 4 year $3 million NSF funded effort at UW, in collaboration with USGS. They've completed >50 M9 simulations, which include 20+ sensitivity tests and 30 M9 logic tree simulations (tree includes rupture depth, hypocenter location, slip distribution, subevent location branches). Two manuscripts are in review at BSSA and synthetic seismograms will soon be available online.
Erin showed an animation showing seismographs for different sites. Crescent City shows large amplitude due to proximity of fault. Seattle show large amplitude due to basin effects. La Grande show lower amplitude than Seattle because it is not in the Seattle Basin (both are similar distance to megathrust). Generally, there is a wide range of possible ground motions (0-10 Hz), varying by a factor of 10 at individual sites, strong basin amplification, and duration of strong shaking 70 sec at coastal sites to 120 s at sites 200 km distance.
They studied the 2010 Maule and 2011 Tohoku events, which suggested strong ground motions originated from discrete, deeper areas of the fault. The CSZ simulations are characterizes in a similar manner, with background slip and patches of high stress drops (sub-events).
Simulations also generate vertical deformation estimates, which they've compared with coastal paleoseismology. They compare predictions of different downdip limits for locking: the tremor zone, 1 cm/yr locking contour, 1 cm/yr + fully locked. The 1 cm/ye locking contour produces best match.
Along-strike trends in subsidence may be the result of the location of strong-motion-generating subevents.
Erin emphasizes that the sub-events are non-unique. She notes that a 'sub-event' has 2 requirements: larger amount of slip and fast slip (high stress drop). The latter isn't needed to match paleogeodesy.
Art Frankel notes that they've been meeting with UW engineers, who are inputing ground motion data into building simulations. They find that buildings in Seattle will collapse in worst case scenarios (events where subevents lead to rupture directivity toward Seattle).
Joan summarized results of a study just published, titled “CSZ onshore-offshore site response, submarine sediment mobilization, and earthquake recurrence.” in JGR.
The study tested the hypothesis that the distribution of site-response affects the susceptibility of turbidite generation, and thus their distribution. Results show significant E-W frequency-dependent site response, with significant differences in the 0.02-1.0 Hz and 1-10 Hz passbands. However, it is unknown which frequency range is important to mobilize sediments. The spatial sampling is sparse, but the site response changes markedly when cross shelf and slope, so this is probably a significant control on ground shaking. Thus, one needs to be cautious to compare onshore shaking intensity with offshore shaking (possibly ~1 order of magnitude greater offshore). Site response variations correlate well with structure, with basins identified in Wells et al. (2003), and methane seeps.
Next steps would be to improve the resolution using data from two ‘focused arrays’ deployed at Grays Harbor and Cape Mendocino.
Chris Goldfinger commented that site response may be linked to turbidite story, but the story is about the frequency of turbidites. Joan responds noting that site response affects susceptibility to failure, so turbidites may occur more frequently depending on site response.
|Historic and recent Cascadia forearc and backarc crustal earthquakes|
Why are there so many EQs in Puget Sound?
Tom shows the tomographic image of the Olympic Peninsula (Calvert et al., 2011), noting that seismicity is concentrated in the Siletzia and very deep (compared to CA), and the Cresent Terrane rocks are largely aseismic. He suggests the Siletzia is the strongest portion of the crust in Puget Lowland so all strain localized here. Lots of deformation is strike-slip.
Tom points out that the Spencer Canyon scarp is coincident with strange long lived seismicity (Entiat seismicity cluster). Since 1976, it has had a constant rate of seismicity, all M<3.7, and is consistent with aftershocks of a M 6.7-7.2 earthquake, similar to that estimated for the 1872 event (published in BSSA). Using rate-state theory (Dietrich, 1994), the Entiat seismicity cluster implies a RI ~1600 yrs. Trenches studies show 2 eqs in 8ka. Other seismicity elsewhere may be related to other old earthquakes.
1936 M ~6 Milton-Freewater Earthquake
We are unsure of the epicenter of this event, but Tom speculates it could be on the Hite fault, not near the Tri-cities as previously estimated (on the Wallula fault, which also trends near the Hanford Nuclear Reservation). Tom looked at instrumental locations and seismic intensities (newspapers), and previous epicentral estimates (from the ISC-GEM and Woodward Clyde). His analyses suggest the ISC-GEM location is more correct.
Tom hypothesizes that the megathrust is affected by the Siletzia and because it is aseismic (totally locked) earthquakes occur in the overriding plate (reflecting ongoing plate convergence).
|Paleoseismic trench results from the Olympic Mtns and Vancouver Island|
Scott presented results of two studies of the Leech River fault (LRF) in BC, and the Canyon River fault (CRF) in WA. The Canyon River/Saddle Mtn fault (SMF) system flanks southern margin of the Olympic Peninsula. The LRF is in southern Vancouver isle, may link to faults to east.
The CRF/SMF is expressed with lots of topographic lineaments, for ~60km. The LiDAR map shows higher terraces and lower fluvial terraces are cut by fault. Scott summarized Zebra and Mosquito trench studies, focusing on Zebra benched trench. Results indicate either 1 earathquake at 6150+-840 yr BP or 2 at 8720+-1380 and ~500 yr BP; he prefers the single earthquake model. The slip rate estimated is from a 3.73 m offset left lateral, down to north. This is a 60km oblique strike-slip fault. More Lidar are needed to improve characterization of fault, but it could generate a M 7.2-7.5 earthquake that would lead to strong shaking in Puget Sound.
The LRF has an uphill facing scarp, with downslope profiles along interfluves and channels suggesting >1 earthquake, with 3 events implied at interfluve site and 2 events at channel site.
Regional implications: dextral LRF and sinistral on CRF, support escape tectonic model for the Olympic Peninsula. Wells et al 2017 suggest linkage between megathrust and upper plate. Coulomb static stress modeling is underway to test how upper plate stress evolves following a CSZ event.
14C uncertainties are too broad to tell if events are simultaneous.
|Gales Creek fault slip & posited relationships to the megathrust|
Ray is a month away from publishing 51 geologic maps with Ralph Haugarud, and has been working with Joana Redwine on the Gales Creek fault (GCF). It is the largest fault in NW OR, Holocene active, and may interact with the megathrust. The GCF is part of the bigger GCF-Mt. Angel system, which is 200 km long extending from the Cascades to Nehalem Bank across some urban areas, is active for least 150 km in the Quaternary, and hosted the 1993 M 5.7 Scots Mills earthquake. It coincides with big geophysical gradients, running NW west of Tualatin Basin, also related to Portland Hills fault system (PHFZ) in PDZ. Aeromag show 12 km dextral offset (Blakely 2000, wells 2018 in prep). The GCF may be part of a compressional step over system, including the Canby and PHFZ to east.
The GCF is well mapped in OR coast range, where it becomes a thrust fault. Some trenches were dug by the US Bureau of Rec and USGS. In bedrock, 12 km right lateral offset, with an avg geologic slip rate of 0.4 mm/yr (1/2 Ma).
The GCF is a high priority for the USBR since it strikes through their dam, which they want to rebuild much bigger. They are doing lots of work to characterize the fault. Evidence indicates it is Holocene active.
The GCF is collocated with gradients in tremor (Wells et al., 2017), perhaps due to fluid migration. Kao 2009 shows tremor avoids faults and earthquakes. The GCF may link to offshore faults.
Ray notes that high frequency sub-events may lie down dip of most coseismic slip (as in Tohoku). In Cascadia, these would correspond with gravity gradient, suggested by offshore central OR earthquakes (Trehu and Williams). These might be “structural knots.” He has been contributing to simulations of fault interaction with RSQsim (Dieterich, UCR).
|Upper plate response/interaction with the megathrust in Cascadia|
Janet asks how the upper plate responds to megathrust along and across strike variation in slip behavior, and affect tsunamigenesis. The CSZ benefits from lots of existing data and knowledge; seaward, landward, and mixed vergence (can compare with other fault systems, like Sumatra). However, we are lacking systematic characterization of morphology and structure, and subsurface imaging to look at long term behavior.
Her group's research plan seeks to characterize 1st order morphology and structural variation along margin with existing data. Specifically examining the structure of the marine forearc, to identify and map and characterize active structures; sources, volumes, and pathways of sediments delivered to shelf and slope; constrain Quaternary event deformation history.
Specifically they are planning to collect high-res multibeam and seismic reflection data, focusing on the SCSZ in the fall of 2018 and Grays Harbor in summer/fall 2019. These will be integrated with existing geophysical data to link structures across shoreline and across international boundaries.
|Geophysical framework of Doty fault|
Rick describe the Doty fault (DF) in SW WA, which crosses the landward portion of CSZ forearc, has had no known historic earthquakes, but is likely a major player. He magnetic maps, noting clear geologic influence. The Doty fault is a north dipping reverse fault, that brought Eocene and younger rocks to surface. The DF appears related to Mt St Helens seismic zone.
Mag anomalies, reduced to pole, new survey, oriented EW, 400 m apart, elev about 200 m above terrain. All igneous rocks show up as distinct anomalies. CF and Spirit lake show up. CRB low amplitude anom. DF (and other faults) show up as linear features. The DF does not appear to project straight to coast, and probably trends northwest.
Magnetic contacts overlain on a gravity map show that the DF bounds southern margin of large crustal uplift to north. It suggests the DF dips to north and units form syncline.
There may be a relationship between the DF, tremor density and crustal structure. There is a big decrease in tremor coincident with the surface trace of the DF, but at 40km depth, so what is connection with tremor? Ray Wells attributes this to structural complexity.
Rick notes that USGS mag surveys have been collected over 2 decades in the PNW. These data are useful for mapping faults not well seen at surface, and may be usable to map across the shoreline.
Discussion ensued about what seismicity is telling us about major faults, and Art Frankel notes that the faults are producing M 6.5-7.5 eqs but M 2-4 events are probably on smaller faults. He and others note that many fault (e,g, the San Andreas) are aseismic.
Work on Doty fault (no figures)
Sherrod, Brian (presented)
Lydia is working this season along the Chehalis River. WA now has program to deal with flood hazards and is installing some flood retention dams east of I-5. The DF part of seismic hazard assessment for these dams. WA legislature passed a budget and has funded this work. Lydia will map terrace surfaces based on LiDAR, do stream geomorph work, and cosmogenic age analyses. Ages will give uplift rate along margin. Bill Stephenson collected a high-res seismic line across DF, which is being processed now. The field data show 2 main faults.
|Little Salmon fault|
Harvey notes that the LSF differs from the other faults we have been discussing wrt its depth above the megathrust (~20 rather than 40 km), the geology. It will be imaged offshore this summer and he's currently looking at LiDAR.
LSF chronology suggests it may correspond with CSZ earthquakes. A number of trenches have been studied, with strain apparent on all 3 strands. They hope to determine if the LSF rupture during each CSZ, what is likely this is major strand of CSZ.
|Geochronology work in OR and WA, EOS and review paper|
An informal 'Cascadia Earthquake Landslide Working Group' has been formed, and met on June 12-13 for a workshop at the UofO. About 30 people participated, from UO, DOGAMI, WA DNR, the UW, USGS, and elsewhere. The Group is focused on two key questions; what is the signature of earthquakes in landscape, and do CSZ events cause landslides? Two primary research priorities have been identified and work is underway, described below. A project update is in review at EOS, titled “Hunting for the legacy of landslides from Cascadia’s great earthquakes”. There also was a GSA 2017 annual meeting session on SZ coseismic landslides.
Though there are >10,000 landslides in western Oregon, no landslide has been linked to 1700 AD yet. 14C precision does not provide annular precision for landslides, but tree ring studies from trees drowned in landslide-dammed lakes may, by matching a drowned tree's rings with a regional reference chronology. Jon showned examples studied to date, from Klickitat and Wasson lakes, in which event dates of 1751 and 1819 were obtained. Additional examples from Yellow, Burhard, and Little Lobster lakes are underway.
2. Landslide deposits + lidar roughness
A lower precision analysis permits evaluation of more landslides, using digital elevation data and the assumption that topography evolves following a diffusion model. This has been applied to the Oso landslide (Booth et al., 2017), in which spectral analysis quantified the change in the landscape over time, creating a calibration curve. If confident that the roughness evolves consistently across landslides, one could analyze thousands of landslides and beat down uncertainty on ages. The rocks in the current study area are comprised of the Tyee Formation (Eocene turbidites), where >3,000 landslides have been mapped. 14C ages are being derived to develop calibration curve, with accompanying tree ring analyses in a couple places (to help develop calibration curve).
Jon also describe the work of Alex Grant, who is working on a model of slide Dynamics (deformation), using ground motion simulations to see how landscape responds to simulated M9 and M8 CSZ, M 6.8Nisqully, and M 7.0 Seattle faultearthquakes. Models simulate translational (shallow tabular) or rotational (deeper) landslides in the Puget Lowland, particularly where there may be basin effects. Using these results to look at . Take ground motions, do Newmark model (acceleration history to get displacement).
Lots of discussion ensued, about the impacts of the local geology and rainfall, which several participants noted might overprint any earthquake-triggered landslides. Other questions arose regarding inferenes of magnitude from landslide area, moment, volume, etc. Jon noted that Tohoku has provided a useful case study to examine the validity of many assumptions.
Cerovski-Darriau,Corina (remote), firstname.lastname@example.org
Mechanics & Models
|3D understanding of fluid flow through forearc and seafloor seepage imaging|
Jared describe a the study of Costa Rica by Bangs et al., 2005 to show what we could learn about fault systems and fluid pathways from doing this in CSZ. He also described the study of Edwards et al. 2018 in which they derived detailed maps of the megathrust off Tohoku; they found long grooves (regions of higher amplitudes), 113-729 m high and width 27-58 m, with corrugations oblique from plate convergence, 11-18º and inferred to relate to concentrations of fluids. Changes in the corrugation patterns may indicate drainage pathways, etc.
He and others also are working offshore the Hikurangi margin, on data from a 3-D R/V Langseth cruise just completed.
|Incoming plate structure and hydration: Measurements and implications|
Structure plays a key role in megathrust behavior, as do fluids. Frictional properties seaward of trench also are important. These factors have been explored in the Aleutians, where the Shumagin Gap and Semidi segments have been compared, having inferred hydrated and dry upper mantles, respectively. Tomograms along these profiles (Shillington, 2015) show hydrated upper mantle have bending moment normal faults, that may be fluid pathways (there are more plate bending faults in Shumagin gap relative to Semidi segtment). Bending faults exist in Cascadia (Han et al, 2016). Bending faults also likely exist at the Middle America Trench, Costa Rica (Ranero 2003).
Can we estimate amount of water, the source of water, and how it is distributed? Geodynamic models suggest a downward pumping of sea water along fault zones, leading to hydration.
Upper mantle seismic anisotropy at the MAT> systematic variation in wavespeed (km/s) relative to depth below Moho; this needs to be accounted for in models. There appears to be little thermal control on oceanic mantle hydration, though the mantle probably is dry in all but coldest SZs.
The WHOI group is building a small OBS that could be rapidly deployed and at a lower cost. Several examples illustrate the need for this. Following the 2013 M 7.3 Craig AK (Queen Charlotte fault) earthquake they mobilized UTIG OBSs, which took 3 months 3 weeks from mainshock to deploy. Following the 2006 Mt Augistine eruption months a deployment was launched but missed the main events.
Ideally OBSs could be deployed from an aircraft, small boat, or drone? If don’t need to level the sensor, then can make smaller sensors.
|Update on Cascadia Geodesy|
Fred describe some of the result in his 2017 paper with Eileen. They used a subsampled GPS velocity field for the entire western US. (Evans, 2015) in a new way. Instead of using a McCaffrey block model, they used a block+viscoelastic model that reduces to a time-independent block model in the high viscosity limit; viscocity is implementated as a correction to the observed interseismic velocity field, such that faults with a large Savage parameter and late in the ongoing eq cycle will have low contemporary strain rates.
The resulting strain field has strong WSW-ENE extension, so the velocity field will have more WW-ENE contraction. The epistemic uncertainty in locking on slab interface is quite large when using only land based data (Pollitz and Evans, 2017). The model could be improved using Schmalzle 2014 data could and seafloor GPS data. Dave Chadwell at Scripps been working on the latter, establishing monuments. One long lasting site is NP1 evaluated using ROV Jason in June 2014 (OTIC funded).
Examples from Peru (Gannon et al., 2005) and Japan show the importance of offshore GPS and what state-of-the-art networks look like, with GPS-A sites (Fujimoto, 2014) and S-NET sites (operated by NIED) in Japan as examples of the latter.
|Southern Cascadia geodesy|
Tide gage data suggest that Humboldt Bay is uplifiting at a rate of ~4 mm/yr. Jay suggests that this is due to upper-plate fault slip, particularly the Little Salmon fault. His group is examining GPS, tide gauge (have deployed a new one) and leveling data to better constrain this.
Most people from outside usgs, use methods based on single types of organisms, constrained from subtidal to uplands, but these cannot sample the entire gradient. This is a major limitation that a Bayesian approach overcomes. It may possible to do better with models that incorporate as many proxies as possible. Ralph: if use all organisms, how to deal with uncertainty? Brian: by careful separation of training and testing samples.
Joan: Do we need to re-evaluate older estimates? Brian: not many sites treated systematically with transfer functions and these needs to be done at all sites (throw out all older data). Brian and Rob: no person in house to do this work; USGS needs to hire a paleoecologist. Guy: mostly agree, but we can work with academics to get this done? Brian: I think so, but we need more people in general. Chris: need consistent program to do every site along the coast.
Compare paleo slip models derived from transfer function studies with geodesy and structures. Joan: challenge is to evaluate what these mean anything, beginning with a unified definition of sub event (high slip patch, high stress drop, how high?). Chris: if these boundaries/patches are persistent (?) then we could identify them. Joan: Art Frankel subevents have specific definition.Lots of discussion ensued about the persistence, or lack of, subevents.
Unified models of CSZ rupture needed, especially for tsunami inundation assessments.
Dating methods; 14C, OSL, other radiometric (U series for carbonate samples)
Larger multiproxy data sets:
Rob: Need to integrate many different datasets. Joan: one approach is to use megamodels, as SCEC is trying to do to build long term with short term forecasts. Steve: need to focus efforts. It has been hard for SCEC to focus. Ray: unified model rupture needed, perhaps beginning with the 1700 event? Joan: one approach focus on 1700 first, then move on. Rob: yes, seems to bias on MRE, but might not be most representative. Art: focus on looking at a specific mud turbidite, look onshore and at lakes, and try to reconcile offshore and onshore. How to resolve M 8 events. Frustrated that cannot get more information about M 8 events. Chris: well within reach. Janet: could do better at merging onshore/offshore to evaluate heterogeneity.
Art: for GMPE modeling, we need a 3-D model of Vs and Vp, Q attenuation, and to span offshore and onshore. Joan: at least 1 effort, workshop this spring in Corvallis, to develop community velocity model. We need to make sure to contribute to this. Janet: discussed this also at recent marine meeting.
Janet: need to focus on updating QFF database. Scott: don’t want to bias to just last event. Looking at pace time diagram in USGS PP. prior to T-10 fewer sites. Longer record southern OR 6-7ka record. 3-4 ka record in norther OR. Vancouver Island shorter record. Rob: this is because limitations between sea level and recordability. Rob: greatest value to look at first four earthquakes onshore, then combine with offshore and lake data using a large Bayesian model. Could relate to shaking variability, heterogeneous deformation, etc. Steve: always wonder why only look at 1700, but need to do this first, but not enough. Like what Art mentioned. Where are southern margin events. Onshore results? Important for seismic hazard map. Brian: have you run M 8’s in simulator? Were most affects offshore? Art: have run sub-events, but not M 8s as megathrust events (yet). Subevents produce essentially same results, but not really. We should start doing M 8’s. Chris: correlative to steve’s comment: worked on 1992 cape Mendocino earthquake offshore. If look at M estimates (e.g. PP 1661-F), not well constrained. Some events could be M 7s, and that is important result. Rob: if can trigger from M 7, why not more turbidites? Joan: aren’t smaller events more muddy? Chris: yes, depends spatially. Art: don’t you have 3 events post 1850? Chris: yes, 2. Rob: why not more turbidites. Chris: from Guttenberg Richter: yes. But GR probably not apply (don’t fill up with M 7s). Brian: re lakes in puget lowland. There are 20-25 large crustal earthquakes, but only 1 event (so far) in these lakes. Chris: need to work on that. Rob: how reduce uncertainty in SCSZ and are these data incorporated into seismic hazard map. Look where they agree. SCSZ more events. So, coastal record is filtered (maybe M 8.5 lower threshold for coseismic subsidence and tsunami), but can compare with offshore record. Some places eg. Bradley lake, more events. Chris: Bradley lake in sensitivity is in-between. Chris: If date something on flat spot on 14curve, get many ages, oxcal can make pdf very narrow. So, can mix and match broad and narrow response curves, may converge.
Joan: if we were to have great eq tomorrow, perhaps it would be useful to think beforehand, what would be do to get the most out of it? Scott: think about data sets that are preserved and not preserved. Brian: The EHP has plans for each region how to respond; this year Pasadena next year Seattle. Steve: USGS leads for these tabletops include Keith Knudsen for nothern CA, Susan Hough southern CA, Brian for the PNW. In this meeting, the need to integrate coastal marine program with earthquake program. Joan: focus scientific or logistics? Steve: all: media, field response, office management. Brian: compiled contact information with people and their tasks. Harvey: these are both logistic and scientific, they are largely urban focused. Imagine nice ground ruptures, will be public safety initial, not science. Rob: considering seismicity rates, the next earthquake will likely be in Alaska. What science questions answer in Alaska and apply to CSZ? What would it take to get people to get to Alaska? M8 M9? Brian: Getting non Alaska people integrated into Alaska plan maybe up to Alaska. Rob: Peter has developed network of boats. Brian: in airplane after Nisqually, saw sed plume in Bay. Reported at meeting in Menlo park, get ship redirected to collect data. Identify these people, put on call down list. Let them know you will be coming to them. Nathan: for aftershocks, we could not get out there on the scale of a week. Steve: the only increase in funding at USGS has been following large earthquakes.
Staisch, Lydia (Brian presented)
Lydia is working on the database, which includes both onshore and offshore data. The onshore portion is almost ready. Once Chris publishes his offshore data, it will also be incorporated into QFF database. The entire database will include information about faults, seismic traces, potential field data, and more.
|Legacy data reprocessing|
Maureen summarized efforts to reprocess legacy seismic reflection data, of which there are many. Legacy data are sometimes the best or only available. Some legacy data was poorly processed, so can gain value by reprocessing. Legacy crustal-scale data are particularly valuable because it is difficult to collect new airgun data (permitting) and also expensive.
Reprocessing includes repicking velocities, applying new deconvolution methods, migrating data to enhance steeply dipping surfaces, depth migration, and prestack migration. Currently they are reprocessed data from Alaska.
Comparisons between Lee data before and after reprocessing show collapsing of the “ring” so can see data on shelf better, clarification of the BSR and shallow structures are also more well imaged.
For Cascadia there are lots of legacy data in NAMSS. Lee data also exist. For Cascadia, next steps are to identify high priority surveys for reprocessing (completed), develop workflow for basic processing for Lee and or Ewing data (in progress), complete basic and advance processing, look into lower priority processing industry datasets and or older academic govt surveys,identify data gaps and targets for new crustal scale and high res acquisition, interpret legacy data on a regional scale following reprocessing.
Powell Center Proposal
|Update on submitted proposal|
Staisch,Lydia (Jon presented)
Lydia was main person to put this together, along with Rob Witter and Janet Watt. The 2017 proposal was rejected with the feedback that there were no clear plans for addressing uncertainties highlighted, “and as important as Cascadia earthquakes are to the region, the proposal does not provide enough detail to clarify how the results would be transferrable to improving earthquake risk in general.” and “another consideration is that there aren’t any planners/managers/human behavior experts in the proposed working groups leading to the question; how will the new conclusions reached from this synthesis group be effectively communicated to the public?”
The new proposal, submitted at the end of January, requests support for 4 workshops, each 5 days. These questions would be addressed at each meeting.
|Review of concept, summary of 2-pagers|
Brian noted that this is the 2nd workshop for this recurrence task. The Landslide group had workshop last year and have put together white paper that is in reviewand serves as a good example of what we should be doing. He reiterated his request for 2 page summaries on the various topic areas, which would be used as bones for a white paper.
Now, we need somebody to grab banner and lead effort! This will be worked out over next 1-2 weeks.
Review External Activities
|Activities outside USGS|
Joan noted that a number of workshops and other activities are posted on this CDI SZ website. She encouraged everyone to get involved in these!
|Spring workshop - responses from pre-GSA inquiry|
The group discussed having an open workshop similar to the one planned before the GSA, which did not happen.
Brian advocates having it in Seattle with one goal of bringing in UW faculty, students, and dean, to integrate UW into this work. He has started the effort by working with Emily Roland. This is timely as the Dean of Oceanography is excited.
Chris Goldfinger suggests incorporating OSU too as they recently got state funded ship days and could use this time to help on project. Janet Watt notes that HSU also is supportive, so we should include them too. Joan Gomberg notes that UO Amanda Thomas has large number of onshore sensors.
Minimally, Brian and Janet want part of a workshop (1 day) to have a narrow focus to plan offshore work and bring in UW collaboration, the latter to plan for using the R/V Thompson and R/V Barnes.
After much discussion, the concensus seemed to be to expand the workshop to 2 days, focusing on land/sea interactions 1 day and ship/offshore/university integration on day 2. Janet notes that the Coastal Marine program is putting major effort into this program, and needs to build momentum to bring everyone along with project.
Chris Goldfinger suggests having an annual EHP meeting for the PNW. Brian says this is a great idea, which Steve Hickman echoes.