Download presentation
Presentation is loading. Please wait.
Published byAmber Jane Armstrong Modified over 9 years ago
1
1 Cythera M6.7 earthquake (January 8, 2006) in southern Aegean: uneasy retrieval of the upward rupture propagation J. Zahradnik, J. Jansky, V. Plicka, E. Sokos Charles University in Prague University of Patras
2
2 EMSC Diverse centroid position Unclear aftershock pattern Unclear fault Low DC%: ETH 60%, Mednet 56%
3
3 Inconsistent hypocenter and centroid moment solution nodal planes, and centroid in the middle
4
4 Teleseismic records Kikuchi-Kanamori method pP: depth P: complexity P pP
5
5 Bottom trace = synthetics (K & K) simple event and complex event
6
6 Regional records and EGF method apparent source time functions prove complexity plane 2 strike ~70° plane 1 strike ~200°... Neighborhood Algorithm provides two slip patches (similar to M. Vallée)
7
7 Lower misfit identifies the fault plane: strike ~70°
8
8 Relocation 30 teleseismic stations, pP-P: depth 90 km 21 regional stations (Greece + Italy), P and S Wadati diagram: Vp/Vs=1.75 Optimization of RMS: Vp/Vs=1.75 Relocation of regional data with first approximation of depth = 90 km and Vp/Vs=1.75 with various azimuthal and epic.distance weighting schemes
9
9 Free depth: This is uncertainty of mainshock location, not the aftershocks !
10
10 First approximation of depth 90 km:
11
11 Optimized Vp/Vs ratio:
12
12 hypo1 We relocated hypocenter 15 km South, 10 km East and 25 km below EMSC. EMSC this study
13
13 ISOLA code (Fortran & Matlab) multiple point-source moment tensors Free on web: Full waveform modeling of regional records
14
14 Iterative deconvolution (Kikuchi and Kanamori) modified for regional records Moment tensor (deviatoric, or DC-constrained) at each trial space-time position by minimization of the L2 waveform misfit (least squares) Optimum space-time position of subevents by maximization of the waveform correlation (grid search)
15
15 Free BB waveform data (Internet) Our LTK station soon on Orfeus, too.
16
16 Waveform modeling for f<0.1 Hz enables the source study
17
17 Hierarchic grid search of centroid f < 0.1 Hz search 1 search 2, etc. EMSC EMSC epic. is just the coordinate origin
18
18 Accurate centroid location needed for usable DC% search 1 search 2
19
19 Getting more accurate centroid makes DC% to converge search 2 search 3… DC% 10-30 only !
20
20
21
21 We found centroid 25 km East of EMSC epicenter and the DC% has converged to 10-30%. Does it imply that the source is actually strongly non-DC ? Not !
22
22
23
23 DC-constrained solution is an equivalent model Note different optimal source position. deviatoric DC-constrained M EMSC and Mednet M
24
24 Can we better justify our centroid position and MT ? Remember the inconsistency for Mednet centroid and EMSC hypocenter:
25
25 Our CMT is fully consistent with our relocation. Far from being trivial!5 … and it identifies the fault plane as the “red” nodal plane, strike ~ 80°
26
26 The EMSC hypocenter is also in the fault plane o5
27
27 The BB first-motion polarities are consistent with the CMT solutiono5 Red: this study Black: others
28
28 Where’s complexity found in EGF analysis and teleseismic modeling ? For f < 0.1 Hz, in addition to stable subevent 1 (1.1e19 Nm) the waveforms clearly indicate subevent 2, 6-sec later, comparable size ! (1.1e19 Nm) ? Sub 2 ? Solution for sub2 is not unique. Sub 1 M
29
29 1 2 Seeking sub 2 in the fault plane of sub 1: DC-constrainedpo5 X depth 85 km X X…EMSC X…this study depth 72 km depth 60 km
30
30 A double-event interpretation: Subevent 1: 1.10e19 Nm strike, dip, rake: (84, 64, 121)=(209 40 43) Subevent 2 (6 sec later): 0.87e19 Nm strike, dip, rake: (61, 86, 52)=(326, 38, 174) 1 2 EMSC this study Depths Sub 1: 60 km Sub 2: 76 km Hyp.: 85 km
31
31 Possible explanation of the apparently large non-DC: Summing up MT of these two 100% DC events provides a non-DC solution strike, dip, rake: (82, 70, 94) 1.6e19 Nm, DC%=57 near to long-period Mednet CMT strike, dip, rake: (81, 67, 139) 1.4e19 Nm, DC%=56 But Mednet centroid is too far… 1 2 M
32
32 Can we identify fault plane of subevent 2 ? x + 1 2
33
33 Can we identify fault plane of subevent 2 ? 5 strike 61° strike 326° 1 1 2 2 Nodal plane with strike 326° passes through the hypoc. !
34
34 hypo5 Nodal plane with strike 326° passes through the hypoc. !
35
35 hypo5 Hypothesis: both patches (on different fault planes) nucleated close to the same point, and ruptured upward, sub 2 being delayed with respect to sub 1. Depths Sub 1: 60 km Sub 2: 76 km Hyp.: 85 km 1 1 2 2 common hypoc.
36
36 1 2 X X X…EMSC X…this study Another possibility: fixed DC focal mechanism (that of sub 1). It moves sub 2 close to sub 1. Depth 60 km depth 69 km depth 72 km
37
37
38
38 1 2 Another possibility: fixed DC focal mechanism (that of sub 1) moves sub 2 close to sub 1 now we do not need the left segment … but how to explain low DC % and why the 6-sec delay ? depth 60 km depth 69 km
39
39 Interpretation I: Fixed mechanism Varred= 52% Interpretation II: DC-constrained Varred=64% !! strike, dip, rake: 84° 64°, 121° (for both) 84°, 64°, 121° 329°, 36°, 179° x x hypocenter depth 85 km (this study) 72 km 60 km 85 km 69 and 60 km
40
40 Methodical lesson and Cythera model Relocation and CMT inversion in same model enabled identification of the fault plane (strike ~80°) of the main patch. Hierarchic space-time grid search lead to convergence of the DC% to a low value. 100% DC-constrained solution provided a double-event model and explained the low DC% as only apparent non-DC. Rupture started at depth 85 km. Most stable slip patch was centered 35 km apart, at depth 60 km. Second large patch was delayed by 6 sec. Position and mechanism not unique. Possibly on a different fault plane. http://geo.mff.cuni.cz
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.