Using W phase for regional tsunami warning and rapid earthquake hazard assessment

Transcription

1 Using W phase for regional tsunami warning and rapid earthquake hazard assessment Luis Rivera 1 Hiroo Kanamori 2 1 Strasbourg University, Institut de Physique du Globe 2 Seismological Laboratory, Caltech 2nd International Workshop on Earthquake Early Warning Kyoto, April 21-22, DPRI, Kyoto University

2 W-phase Introduction We have recently developed a source inversion technique based on the waveform modeling of W-phase. W-phase is a very long period (200s-1000s) phase arriving right after the P. It was first recognized after the 1992 Nicaragua earthquake. The inversion technique was originally devised to work for large events (Mw> 7.5) with teleseismic data and it provideas a VLP characterization of the source (e.g. Tsunami earthquakes). We explore here the possibility of an application with regional data and with smaller magnitudes.

3 W-phase Example: Nicaragua, 1992 Vertical traces - integrated once ANMO, = 28 PAS, = 36 ALE, = s

4 W-phase Example: Nicaragua, 1992 Vertical traces - integrated once ANMO, = 28 PAS, = 36 ALE, = s

5 W-phase: Example: Nicaragua, 1992 Deconvolved + bp 200s-1000s Vertical displacement ( µ ) ANMO, = 28 PAS, = 36 ALE, = s

6 W-phase: Example: Nicaragua, 1992 Deconvolved + bp 200s-1000s Vertical displacement ( µ ) ANMO, = 28 PAS, = 36 ALE, = s

7 Time window W-phase time-window P, P + 15 s/

8 W-phase in the global context travel time [s] Global scale P+15 W-phase distance [degrees] P S Definition Time window: P, P+15 Bandpass:.001 Hz Hz Properties Fast group velocity: km/s Fairly insensitive to: shallow lateral heterogeneities source second order details Avoid large amplitude surface waves

9 Source retrieval from W-phase Time domain Point source (VLP data) Inversion: main features Need a preliminary source location: PDE, JMA Library of precomputed Green s functions Linear inversion > Moment tensor components PDE ( < 50 ): t min Grid search ( < 90 ): t min RT implementation: β-test at NEIC-USGS, (Gavin Hayes)

10 Example: global data Tokachi-Oki-2003 t min t min

11 Regional data Extension to regional data and lower magnitudes Target: Mw > 6.5 Data distribution: < 12 6 min. > 5 : high gain data, nonlinearity. Modifications: Time window: (P, P+15 ) inappropriate Frequency band: signal/noise ratio

12 W-phase time window 2500 Global scale 500 Regional scale P+15 12s 6min travel time [s] P+15 travel time [s] P 100 P distance [degrees] distance [degrees]

13 W-phase time window 2500 Global scale 500 Regional scale P+15 12s 6min travel time [s] P+15 travel time [s] P 100 P distance [degrees] distance [degrees]

14 Extension to lower magnitudes Acceleration noise spectrum at MAJO O A B C

15 Extension to lower magnitudes Acceleration noise spectrum at MAJO O A B C

16 Extension to lower magnitudes Acceleration noise spectrum at MAJO O A B C

17 NOP SHR KNP HSS HID URHKSR IMG KMU MMATMR 40 GJM IYG TYS KSN 40 Data: Japanese broadband network (F-net) Events: M jma > YSI YZK NRW IZH YTY NSK TGW ISI OKW SBR INN UMJ TSA FUK STM TKD TMC AMM SIB KYK TKO TAS SAG YAS WJM KZK HRO YMZ ASI SRN ONS TSK KNMTTO SGN TGA NAA FUJ KNY JIZ WTR NOK KIS KMT SBT KZS HJO AOG KSK OSW 25 STS 1 STS

18 F-net, , M jma > 6.7 (1/4)

19 F-net, , M jma > 6.7 (2/4)

20 F-net, , M jma > 6.7 (3/4)

21 F-net, , M jma > 6.7 (4/4)

22 Regional W phase focal mechanisms (t 0 + 6min) I A A A C D K A A A A A A A A D A A A A A 25 STS 1 STS A

23 Regional W-phase, example of fit: 2008 Iwate A ( Hz Hz, n = 4, W ) KZS KNP TGA NOP YAS NMR WTR NKG SHR ABU HJO YZK KIS NOK SAG AOG KMT ISI NRW YSI UMJ OKW TGW NSK TSA YTY INN SBR TMC IZH TKO STM SIB TAS FUK OSW mm sec

24 Moment Magnitude: gcmt - W-phase W phase Mw gcmt Mw

25 2003 Tokachi-oki: depth effect H = 45km δ = 29 M w = 8.02 H = 28km δ = 14 M w = 8.15 HVD(H = 28km) δ = 11 M w = 8.26

26 Conclusions We use F-net data in the range (5 < < 12 ) Time window: t P, t P + 180s Variable frequency band: (.00167Hz.005Hz) (.005Hz.010Hz) Moment tensor solution available at t 0 + 6min Can be done completely automatic and Provide a solution 6 min after the origin time.