Imaging and monitoring with industrial seismic noise.

Transcription

1 Imaging and monitoring with industrial seismic noise. M. Campillo also : Boston, May 2016

2 Passive imaging: Long range ()*&#"'!"#"$%"&'!!!!!!"! #! Source in A the signal recorded in B characterizes the propagation between A and B. Green function between A and B: G AB!"#"$%"&'!"#"$%"&'!!!!!!"! #! G AB can be reconstructed by the correlation of noise or «diffuse» (equipartitioned) fields recorded at A and B (C AB ) A way to provide new data with control on source location and origin time

3 Seismological coda waves Individual cross- dominate. Emergence of the Green ADer averaging over 100 EQsè (Paul and Campillo, AGU 2001; Campillo and Paul, Science, 2003)

4 Cross- of coda and noise records Green = virtual seismograms - demonstrated for the retrieval of surface waves (e.g. Paul and Campillo, 2001; Campillo and Paul, 2003; Shapiro and Campillo, 2004.) or body waves (e.g. Zhan et al., 2010 ; Poli et al., 2012). High resolu@on velocity map of California obtained from ambient noise (Rayleigh) (Shapiro, Campillo, Stehly and Ritzwoller, Science 2005) Earth s mantle discon1nui1es from ambient noise ( phase transi1on è (P,T)) Body waves (Poli et al., 2012) Poli, Campillo, Pedersen. Science 2012 Reflec@ons at the base of the crust A (Poli et al., 2012) 60 time [s] P410P P660P AK135 Data final model B Transition Zone Lower Mantle depth [km] Upper Mantle Z- Z noise correla@ons 600 Z comp. actual earthquake Large N sensor array =è N2/2 correla@ons Vp [km/sec]

5 Smaller scale, industrial environment mine: various sources of noise tunnels (sca\ering) view Plan view Results from Olivier, Brenguier, Campillo, Lynch and Roux, 2015 GEOPHYSICS, VOL. 80, NO. 3 (MAY- JUNE 2015); P. KS11 KS25

6 Numerical in presence of the tunnels vs Diffusion Actual event vs Diffusion

7 Nature of the noise: example of a 5s record impacts of a hammer drill microseismic event mul@ple sources incl. (pumps, fans, etc.)

8 (ZZ) windows for body wave Removing of sources

9 distance Noise blind stacking Noise stacking

10 Convergence of the ZZ S wave velocity

11 Sca\ering from noise Diffusion approx.

12 tomography from noise Checkboard test

13 Inves1ga1on of coseismic and postseismic processes using in situ measurements of seismic velocity varia1ons in an underground mine Results from Olivier, Brenguier, Campillo, Roux, Shapiro and Lynch, 2015 Geophysical Research LeCers Volume 42, Issue 21, pages , 11 NOV 2015 DOI: /2015GL h\p://onlinelibrary.wiley.com/doi/ /2015gl065975/full#grl fig- 0001

14 Inves1ga1on of coseismic and postseismic processes using in situ measurements of seismic velocity varia1ons in an underground mine 12 day stack t (days) τ (s) Hourly correla@on func@ons for 12 days window (sca\ered waves) for the delay analysisè

15 a small change of seismic speed: coda waves Comparing a trace with a reference under the assump@on of an homogeneous change The doublet method: moving window cross spectral analysis of the delays Rela@ve velocity change: dv V (t) = dτ τ (t)

16 Temporal of the seismic velocity measured from all involving a par@cular sensor (4 hour window) The is larger than the one deduced from detected seismicity

17 Comparison of velocity changes and volumetric stress changes Instaneous velocity drop change

18 Velocity change due to blast and Fast dynamics Slow dynamics Change of baseline due to stress change Instantaneous change change

19 Conclusions: Passive (noise based) imaging is possible in industrial environment like mines. It requires a careful analysis of the noise proper@es Body waves are retrieved and could be used for imaging Time dependent elas@c proper@es can be inferred giving new clues on the geomechanical evolu@on

20

21 Measuring slight changes of seismic velocity using coda waves (long Numerical in a sca\ering medium 2D spectral elements, anisotropic intensity of sources Comparison of correla@ons with Green func@on Colombi, Chaput, Hillers et al., 2014 in press

22 Effect of sca\ering (single source) Beam forming Colombi, Chaput, Hillers et al., 2014 in press

23 Measure of the bias induced by a strong anisotropy of the wave field (delay with respect to the Green func@on) Blue: delay Red: rela@ve delay Fluctua@ons of dt/t of the order of 10-4 Colombi, Chaput, Hillers et al., 2014

24 of coda waves as the sum of of numerous paths For a single path: t f =l f /V l f We have to compute the contributions of paths with first scatterers at all distances l f and all azimuths θ

25 We have to consider that the distribution of distance between scattering events is exponential: where l is the mean free path < l f > = l t f = l f /V We make use of valid for l f > λ We remove the singularity for l f to 0 in a conservative way ( δt t ) and define: ratio of max average δt over δt of the average l f (=l the mean free path) Finally we consider all directions:

26 Numerical è Japan (same change of anisotropy) è

27

28