focal mechanisms of micro-earthquakes in the little carpathians -
DESCRIPTION
AIM second annual meeting September 29-30, 2011 Prague. Progseis, Ltd. Focal mechanisms of micro-earthquakes in the Little Carpathians - time-frequency identification of problematic input data for Isola MT inversions. Miriam Kristekov á Lucia Fojtíková - PowerPoint PPT PresentationTRANSCRIPT
Focal mechanisms of micro-earthquakes in the Little Carpathians
-
time-frequency identification of problematic input data
for Isola MT inversions
AIM second annual meeting
September 29-30, 2011 Prague
Miriam Kristeková Lucia Fojtíková
Geophysical Institute SAS, Bratislava, Slovakia
in cooperation with
industrial partner:
Juraj Sekereš Dagmar Sekerešová
Progseis, Ltd., Trnava, Slovakia
Progseis, Ltd.
Outline
Motivation of our work:
Why we pay attention to ISOLA method?
Need to investigate sensitivity of ISOLA method
to the choice of parameters of computation
Possibilities how to verify
whether proper parameters are used
orwhether obtained results are reliable
Use of TFA for identification of problematic input data (indication of necessity to improve velocity model used
or at least to omit problematic data)
Conlusions
For estimation of focal mechanisms:
different methods – different advantages/limitations
e.g.
FOCMEC (focal mechanisms, inversion from polarities of P-waves)
[Snoke 2003]
needs data from many stations
with precise location and
optimal coverage of focal sphere
AMT (moment tensor, inversion from amplitudes of P-waves)
[Vavryčuk 2009]
needs data from several (at least 7-8) stations
with suitable configuration
Motivation
Another possibility
ISOLA (moment tensor, waveform inversion) [Sokos & Zahradnik 2009]
allows to obtain solution for a small number of stations
(theoretically for the only one station)*
+ it could be really helpful when dense local network is not available
- method could be more sensitive to the data and parameters of computation
In general, in computations of focal mechanisms
there are uncertainties of results
caused
by input data (quality, configuration) or
by parameters of computations
(e.g. velocity model – for real complex geological structures)
By processing of large statistical datasets of events
and
by using different methods
it is possible to eliminate (at least partly) influence of these uncertaties
and to compensate lack of our knowledge
and
to obtain more reliable solutions
(e.g. [Fojtikova et al 2010])
However, what to do in cases
when large statistical dataset is not available,
and/or
it is not possible to use several independent methods?
Limited number of stations available
and their unsuitable configuration
quite common situation
when analyzing weak local events
in regions with moderate seismicity
Then using ISOLA method could be helpful
By using data from only few stations,
ISOLA method could be more sensitive to the quality of data
and to the choice of parameters of computation
moreover,
In some cases this could be the only one applicable method
without possibility to verify obtained result with other method
it is important and necessary
(especially for such a cases)
to investigate sensitivity of ISOLA
to the choice of parameters
(e.g. freq. range for inversion, velocity model used, etc.)
and
to be careful when verifying
whether the obtained result is reliable
In order to make inversion less sensitive
to unknown tiny details of the velocity model used
ISOLA uses S- and surface waves mainly
(lower frequency part of seismograms)
ISOLA has usually been used for regional events,
not for microearthquakes
(exception was [Fojtikova et al 2010])
=> different frequency range is analyzed
Therefore, the first natural question was:
what freq. range is suitable for waveform inversion of weak local events?
We investigated and selected suitable frequency ranges
(containing useful signal)
for ISOLA inversions of weak local events
using time-frequency analysis
and this work
was already presented at previous AIM meeting
As it was shown in previous presentations of our team
at this meeting
ISOLA method is
sensitive to the changes in the velocity model used
=> sufficient knowledge of velocity model is neccessary
Therefore
important issue when considering possibilities of the use
of the ISOLA method
is the following one:
How to identify cases when our knowledge of the model is insufficient
and
when computation could lead to the biased results?
MKNET
Micro-events in
Little Carpathians Mts. area(2001-2009, Ml>1)
S03
V08
V14
V05
V04
W05
W01
X04
Y01
X01
V06
Q05
Q13
Q15
Q16
R03
S02
T01
U01
V15
V17
X02
X03
Q12
V19
V03
V07
V09
Event V14
Event V14 was a special case in previous analyses
It seemed that the two different solutions of focal mechanisms
are equally probable
FOCMEC AMT
Solutions from ISOLA varied between these two types
and for the fixed velocity model
they strongly depended on the selection of stations
used for computation
level of agreement between real and synthetic seismogram was
similar for these different solutions,
so this criterion did not help to select more reliable solution...
This strong dependence on the selection of stations
was probably consequence
of not sufficient knowledge of the
velocity model in the area of interest
(Dobra Voda source zone has complicated local geological structure)
or by some artifacts in data
Therefore we started with more detailed analysis
of input data (seismograms) from individual stations
looking for some anomaly
We performed time-frequency analysis
using continuous wavelet transform (CWT).
DVO, D = 11.23 km
Z N E SMO, D = 2.67 km
KAT, D = 7.22 km
BUK, D = 5.26 km
60.0e-10
45.0e-10
1.4e-10
9.5.0e-10
19.0e-10 30.0e-10
4.6e-10
1.9e-10 5.6e-10
1.8e-10
4.0e-10
1.1e-10
SMO, BUK,
KAT Ecomp.
SMO, BUK, KAT
FOCMEC AMT
SMO, BUK, KAT
Aditionaly, we have noticed that
the cases when ISOLA results
resembled FOCMEC results
were cases with low number of
stations including KAT
Thank you for your attention