Seismic source inversion in mining environments Simone Cesca

Cesca, 2011. Source inversion in mining environments

AIM 2nd annual meeting, 29-30.9.2011, Institute of Geophysics, Academy of Sciences of the Czech Republic, Praha

Seismic source inversion in mining environments

Simone Cesca

Institute of Geophysics, University of Hamburg

[email protected]

http://mine.zmaw.de



Outline

The MINE project

• background, structure and project aims

Seismic source inversion problem

• theory and methods

• point and kinematic source inversion using the Kiwi tools

Source inversion in mining environments

• an application to coal mining induced seismicity (Ruhr, Germany)



The MINE project, a quick overview

MINing Environment: continuous monitoring and simultaneous inversion

Funding, positions, timeline

• German Ministery of Education and Science (BMBF – Geotechnologien Programme)

• Junior Research project, 1 Leading Scientist + 3 PhDs

• 3+3 years duration (started 1.7.2010)

Aims

• Mining monitoring (seismicity, fracturing, stress perturbation) and imaging

Adapt full waveform techniques from classical seismology and to mining

Combine information from different data streams

Provide portable software tools to analyse mine datasets



Research lines

Detection and location using full waveforms

Signal characterization

Moment tensor inversion

Extended source parameters

Spatio-temporal patterns of seismicity

Stress inversion (seismicity patterns and focal mechanisms)

Local earthquake tomography

WP1

Detection, Location,

Characterization

WP3

Stress inversion

WP2

Source

Inversion

WP4

Tomography



Hamm,

Ruhr region

Courtesy M. Bischoff

Biscoff et al. (2010)



Pyhäsalmi

Deep metal mine in Europe

Collaboration with NORSAR (Kuhn, Oye, Roth, Nath)

3D velocity model

Multi-stream monitoring system

Seismic

Internal deformation

Thermal measurements

Others

Figure T. Mäki (2000)



Cerville

SALT

MARLS

LIMESTONE(DOLOMIE)

190 m

125 m

50 m

Courtesy P. Jousset



Source inversion in seismology



Source inversion in seismology, overview

Global CMT catalogue, shallow earthquakes, 1976-2005

Slip map

kinematic model

(Li et al. 2002)

Source: http://cgsweb.moeacgs.gov.tw/



Source inversion in seismology, what do we need?

Moment tensor / Kinematic sourceInversion routine

Plotting and result evaluation

Data access(waveform and

metadata)Data preprocessing

Green's functions(database)

Inversion method,Inv. Parameters

(e.g. BP, tapers, ...)





OpenSource

Running under Linux

Python + Fortran implementation http://kinherd.org

Source inversion in seismology using the Kiwi tools



Point source model, double couple & moment tensor

More complete point source model is represented by a moment tensor (MT)

MT = MTDC + MTCLVD + MTISO

Earthquakes is often well modeled in terms of shear cracks, using a point source representation (DC model).

after Hasegawa et al. (1989)



Extended source model, definitions

Extended sources may be reproduced by superposition of several point sources distributed along a planar (or bended) rupture surface. Each point source start radiating when reached by the rupture front. Radiation lasts for a given (rise) time.



The eikonal source model

Circular area, plus constraints

Rupture velocity scales with shear velocity

Despite its flexibility, the eikonal source model is described by only 13 parameters (considering constraints and earth model as fixed and known)



One earthquake, five solutions



Source model parameters and inversion priorities

Source model parameters (13)

Time Lat Lon Depth M0 Strike Dip Rake NucX NucY RuptV RiseTRad

General source description

Source location Radiation pattern Rupture process

Scale of source model

Point source Finite source

Information from data

Low frequencies High frequencies

Inversion priority

Step 1, 2 Step 3



Greece,

Shallow earthquakes

2003-2007

extended sources

Cesca et al. JGR 2010



Source inversion, natural and induced seismicity

Cesca et al. submitted



Source inversion in mining environment



Bischoff et al. 2010

Ruhr region

Coal mining induced seismicity monitored by Ruhr University since 1983

About 1000 events are recorded between 0.7<ML<3.3 every year

Hamm region (blue circle)

>7000 events in 2006-2007 (14 months)

913 events 0.0<ML<2.0

DC inversion

MT inversion

Kinematic inversion and rupture modeling



Seismicity follows longwall mining,

Epicenters are spreaded over an area of about 2x2km

6 broadband stations (5 Guralp CMG, pink; 1 Trillium 40, purple)

9 short-period (Mark L-4C-3D, orange)

3 subsurface stations (yellow)

We work here at 0.5-2Hz or 1-4Hz, only BB stations are used



Seismicity follows longwall mining

Additional clusters

Average depth above mining level

Bimodal frequency-magnitude distribution



Inversion strategy

Depth M0 Strike Dip RakeTime Lat LonTime Lat Lon NucX NucY RuptV RiseTRad



Time Lat Lon Depth M0 Strike Dip Rake NucX NucY RuptV RiseTRadTime Lat Lon Depth M0

Step 1, Focal mechanism (DC and full moment tensor)

amplitude spectra inversion, whole waveform

Depth M0 Strike Dip Rake

1



Inversion strategy








1

Step 2, Polarity control / Centroid location

time domain inversion, whole waveforms

Depth M0 Strike Dip RakeTime Lat Lon

2



Inversion strategy








1

Step 2, Polarity control / Centroid location

time domain inversion, whole waveforms

Depth M0 Strike Dip RakeTime Lat Lon

2

Step 3, Kinematic model

amplitude spectra or time domain inversion, including high freqencies

NucX NucY RuptV RiseTRadTime Lat Lon Depth M0 Strike Dip Rake

3





DC inversion results overview

Successful inversion for 578 (over 913)

Magnitude range, Mw 0.3-1.8

Very similar mechanisms

Normal faults (80%) or oblique-normal

One steep plane, one sub-horizontal

Different strike are observed, strike angles are related to mining geometry





Wehling-Benatelli (2011)

Courtesy D. Becker

Waveform similarity analysis and

cluster analysis (relocated events)

Consistent focal mechanisms for

Major clusters





Moment tensor / extended source parameters

Full MT solutions significant for more than 100 events

Non-DC terms results are still ambiguous

Possible inversion artefact rather than source features

Preliminar kinematic inversion for 24 largest events (Ml > 1.0)

Kinematic model is significant for 8 events, only

In almost all cases (7), the vertical rupture plane is preferred



Conclusions

Full waveform moment tensor inversion successfully applied to mining induced seismicity at local scale (<2km) for low magnitude events (at now, down to Mw 0.3).

DC and MT focal mechanisms were successfully obtained for 587 selected events

Results are in good agreement with reference, when available (about 100 events), based on first polarities and S wave polarization.

Better results are obtained for a layered model and frequency range 0.5-2Hz

Focal mechanisms are characterized by similar ruptures. Normal faulting with one steep fault plane. In general, striking angles are linked to the mining geometry.

Non-DC resolution to be judged

Preliminar kinematic modeling for largest events (Ml≥1.0) point to a similar rupture mechanism along sub-vertical planes.



Thanks to:

A. T. Şen, Prof. Dr. T. Dahm, Dr. S. Heimann, F. Grigoli, S. Maghsoudi, A. Rohr,

M. Bischoff, T. Meier, S. Wehling-Benatelli

BMBF project MINE

GEOTECHNOLGIEN programme

The Kiwi tools are currently used at:

University of Hamburg, University of Potsdam, BGR Hannover, GFZ Potsdam,

University of Coimbra, Aristotle University of Thessaloniki ,Ruhr University Bochum

Further info on software and applications:

http://mine.zmaw.de

http://kinherd.org

Cesca et al., JGR 2010

Cesca et al., J. Seismol. 2010

Cesca et al., J. Seismol., submitted



Thanks for your attention

Documents

Seismic source inversion in mining environments Simone Cesca