Long-Term Paleoseismology in Cascadia: Probabilities, Clustering, and patterns of Energy Release

Transcription

1 Long-Term Paleoseismology in Cascadia: Probabilities, Clustering, and patterns of Energy Release Chris Goldfinger College of Earth, Ocean and Atmospheric Sciences, Oregon State University Active Tectonics Group, Ocean Admin Bldg 104, Corvallis OR C. Hans Nelson, Joel E. Johnson*, Steve Galer, Jeffrey Beeson, Bran Black, Ann E. Morey*, Julia Gutiérrez-Pastor, Eugene Karabanov**, Andrew T. Eriksson*, Rob Witter and George Priest s, Eulàlia Gràcia****, Kelin Wang***, Joseph Zhang S, Gita Dunhill, Jason Patton*, Randy Enkin***, Audrey Dallimore***, Tracy Vallier, and the Shipboard Scientific Parties (52 students, colleagues, technicians) Goldfinger, C., et al., 2012, Turbidite Event History: Methods and Implications for Holocene Paleoseismicity of the Cascadia Subduction Zone, USGS Professional Paper 1661-F, Reston, VA, U.S. Geological Survey, p. 184 p, 64 Figures. Goldfinger, C., Galer, S., Beeson, J.W., Hamilton, T.S, Black, B., Romsos, C., Patton, J., Nelson, C.H., Hausmann, R., Morey, A., 2016, The Importance of Site Selection, Sediment Supply, and Hydrodynamics: A Case Study of Submarine Paleoseismology on the Northern Cascadia margin, Washington USA, Marine Geology in press/online.

2 Discovery of the 1700 AD earthquake is a triumph of sleuthing! In 1984, Canadian Geologist John Adams was the first to suggest that direct evidence of Cascadia earthquakes existed in the form of turbidites, deep sea deposits sometimes triggered by earthquakes

3 Geologic evidence of great earthquakes is abundant in the form of trees killed by saline incursion, and the peat-bay mud couplets formed by each earthquake.

4

5 Cascadia submarine canyon systems traverse the locked zone, making them sensitive to ground shaking. They are, for the most part, isolated from river systems during high-stand conditions

6 What actually happens during the earthquake? Synchronous turbidity currents are triggered within a few minutes of each other along the length of the margin

7 Turbidites are easy to capture, but what do they mean?

8 So our primary criteria for distinguishing earthquakes are 1) Aerial extent 2) Synchroneity, and 3) Sedimentology. Synchronous means within a few minutes to hours at most 14 C dating gets us only to within a few decades at best, usually not that good. So how do we constrain relative timing to within a few hours? Turbidite Paleoseismology: Extending the earthquake record Cascadia Core Sites: 1999 = gray, 2002 = yellow Older existing cores = white Washington Channels defined by 12 days of multibeam survey, now un-classified! Cheat!

9 Detailed correlations are constructed from high-resolution physical property data collected from the cores, including magnetic susceptibility (high and low), gamma density, P-wave velocity, resistivity, and CT imagery.

10 In addition to the confluence test, we correlate turbidites between remote sites to establish continuity, and test for synchronous triggering. Correlations are made on the basis of grain-size/physical property fingerprints within a 14 C age framework

11 CT imagery is invaluable for understanding turbidite structure and defining stratigraphic boundaries in detail. This image breaks out the sand fraction, the silt fraction, and the hemipelagic clay by their respective CT density values. The CT can reveal such subtle features as a worm burrow which is apparently lined with material slightly more dense than its surroundings (biogenic clay)

12 Correlation is done using oil industry techniques such as stretching and squeezing ghost traces to examine correlations, and flattening the correlation diagram to event horizons. Correlations supported by numerous radiocarbon ages.

13

14

15

16 Linking Onshore and Offshore: Exploring inland turbidites and ground motions.

17 Inland Evidence

18 Puget sound: 90 km between sites

19 Cascadia: The Movie T19 This sequence shows the Cascadia Holocene earthquake sequence. The slides are timed at 1 sec ~ 200 years. Event pulses that correlate at all sites are shown by flashes of the locked zone in red. Event size shown by intensity of red shading

20 T18

21 T17a

22 T17

23 T16a

24 T16

25 T15a T15an

26 T15

27 T14a

28 T14 Crater Lake (Mt. Mazama) Goes off!! ~7625 BP

29 T13

30 T12a

31 T12

32 T11 Biggest Cascadia Earthquake! ~ 5900 BP

33 T10f

34 T10e

35 T10d

36 T10c

37 T10b

38 T10a

39 T10 End of a 1200 year Gap in the North ~ 4800 BP

40 T9c

41 T9b

42 T9a

43 T9

44 T8b

45 T8a

46 T8

47 T7a

48 T7

49 T6b Northern San Andreas Fault

50 T6a Northern San Andreas Fault

51 T6

52 T5c

53 T5b

54 T5a

55 T5 Another ~ 1000 year Gap (north only) Ends ~ 1500 BP

56 T4a

57 T4

58 T3a

59 T3

60 T2c

61 T2b

62 T2a

63 T2 The penultimate earthquake ~ 480 BP

64 T1 (AD 1700)

65 Rupture lengths from paleoseismic data, past 10,000 years. Segment boundaries are roughly compatible with ETS segment boundaries proposed by Brudzinski et al., 2007, though both sets of boundaries are quite crude.

66 For the northern margin, probabilities are relatively low, many intervals longer than 360 years are in the paleoseismic record. The reliability analysis suggests at 360 years, 25% of repeat times will have been exceeded. Conditional probability in 50 years is 14% (12-17%). Northern Margin Southern Margin (slight revision of repeat times and probabilities, in 2016 Marine Geology paper).

67 For the southern margin, if our interpretation is correct, 70-93% of repeat times will have been exceeded. Northern Margin Southern Margin Conditional probability in 50 years is 37% (32-42%). Portland is in between these extremes, with a recurrence of ~ 340 years, and 50 year probability of ~ 20%. (This is a slight increase, 2016 Marine Geology paper) Southern Margin

68 What about clustering? There seems to be a poorly developed clustering, suggested here. It certainly makes a difference whether the next expected event is part of a cluster or not, if clusters exist, and if the next event reflects a repeat of recent behavior. In cluster 50 year probabilities are ~ 25%, not in a cluster, ~ 2%. Clustering seems better developed in the latter half of the Holocene. If a repeat were to occur, a gap may be next.

69 High-resolution CHIRP profile showing ~ 150 km of correlated turbidites Rogue Canyon to Trinidad, southern Cascadia Earthquake clusters you can see? Goldfinger, C., Morey, A., Black, B., Beeson, J. and Patton, J., 2013, Spatially Limited Mud Turbidites on the Cascadia Margin: Segmented Earthquake Ruptures?, Nat. Hazards Earth Syst. Sci., 13, doi: /nhess

70 Long-Term Energy Cycling (the battery model) Longer records would help answer some of the obvious questions such as whether clustering is a long term feature, or if our short 10ka record is random. The instrument already exists, and the experiment has already been run. Goldfinger et al Superquakes and Super Cycles, SRL v. 84 no.1 p

71 Questions?