Moment Tensor 20041226005853

Comment

Thus solution is base on using a 200 second time window for this earthquake. A very long period filter was used - effoectively 200 to 250 seconds because of the filtering and the deconvolution. This inversion used the entire trace from P -30 to P + 2000 seconds. There is no attempt to model P, SH and SV phases, but rather the entire trace.

This example is also meant to test the ability of the Green's functions for a large earthquake.

Location

2004/12/26 00:58:53 3.30 95.98 30

Moment Tensor Comparison

The following compares this source inversion to the USGS Rapid Moment Tensor Solution and to the Harvard CMT solutions, if they are available.

SLU USGS HARVARD CMT

SLU Moment Tensor Solution 2004/12/26 00:58:53 Best Fitting Double Couple Mo = 5.19e+28 dyne-cm Mw = 8.41 Z = 40 km Plane Strike Dip Rake NP1 150 50 90 NP2 330 40 90 Principal Axes: Axis Value Plunge Azimuth T 5.19e+28 85 60 N 0.00e+00 -0 150 P -5.19e+28 5 240 Moment Tensor: (dyne-cm) Component Value Mxx -1.28e+28 Mxy -2.21e+28 Mxz 4.50e+27 Myy -3.83e+28 Myz 7.80e+27 Mzz 5.11e+28 -------------- #######--------------- ---############------------- ---################----------- -----##################----------- -----#####################---------- ------######################---------- --------#######################--------- --------########################-------- ---------############ #########--------- ----------########### T ##########-------- ----------########### ###########------- -----------########################------- -----------#######################------ ------------######################------ ----------####################----- P -----------###################---- -------------################---- --------------##############-- ----------------##########-- ------------------#### -------------- Harvard Convention Moment Tensor: R T F 5.11e+28 4.50e+27 -7.80e+27 4.50e+27 -1.28e+28 2.21e+28 -7.80e+27 2.21e+28 -3.83e+28

04/12/26 00:58:50.76 OFF W COAST OF NORTHERN SUMATRA Epicenter: 3.298 95.778 MW 8.2 USGS MOMENT TENSOR SOLUTION Depth 7 No. of sta: 31 Moment Tensor; Scale 10**21 Nm Mrr= 0.91 Mtt=-0.89 Mff=-0.02 Mrt= 1.78 Mrf=-1.55 Mtf= 0.47 Principal axes: T Val= 2.53 Plg=55 Azm= 50 N 0.09 8 308 P -2.61 34 213 Best Double Couple:Mo=2.6*10**21 NP1:Strike=274 Dip=13 Slip= 55 NP2: 130 79 98 ------- ----------------- ---##############---- --####################--- ##-########################-- #----########################-- -------############ ########- ---------########### T #########- -----------######### ########## -------------#################### ---------------################## ------------------############### -------------------############ -------- -----------######### ------- P --------------##### ----- ----------------- --------------------- ----------------- -------

ecember 26, 2004, OFF W COAST OF NORTHERN SUMATRA, MW=9.0 Meredith Nettles Goran Ekstrom CENTROID, MOMENT TENSOR SOLUTION HARVARD EVENT-FILE NAME M122604A DATA USED: GSN MANTLE WAVES: 73S,202C, T=300 CENTROID LOCATION: ORIGIN TIME 01:01: 9.0 0.3 LAT 3.09N 0.04;LON 94.26E 0.03 DEP 28.6 1.3;HALF-DURATION 95.0 MOMENT TENSOR; SCALE 10**29 D-CM MRR= 1.04 0.01; MTT=-0.43 0.01 MPP=-0.61 0.01; MRT= 2.98 0.16 MRP=-2.40 0.16; MTP= 0.43 0.00 PRINCIPAL AXES: 1.(T) VAL= 4.01;PLG=52;AZM= 36 2.(N) -0.12; 3; 130 3.(P) -3.89; 38; 222 BEST DOUBLE COUPLE:M0=4.0*10**29 NP1:STRIKE=329;DIP= 8;SLIP= 110 NP2:STRIKE=129;DIP=83;SLIP= 87 -#######--- #################-- #####################-- -########################-- ----############# #######-- ------############ T ########-- --------########## #########- -----------#####################- -------------###################- ---------------#################- ------------------##############- -------------------############ ------- ------------########- ------ P --------------#####- ----- ------------------- ----------------------- ------------------- -----------

Data set

The following broadband stations passed the QC and were used for the source inversion. AAK ADK ANTO ARU BFO ESK FURI KEV KIEV KIV LSZ LVZ MAJO MSEY RAYN TATO TLY ULN YAK YSS

Grid search

All observed and Greens function waveforms are corrected to instrument response to ground velocity in meters/sec for the passband of 0.004 - 5 Hz. The traces were then lowpass filtered at 0.25 Hz and interpolated to a sample rate of 1 second.

For the grid search, the observed traces and Green's functions are read in an cut using the following commands

Phase    Gsac Command              Comment
 P    cut A -30 A       CUTH = 95+0.3*DEPTH  
SH    cut T1 -60 T1     CUTH = 95+0.6*DEPTH
SV    cut T0 -60 T0     CUTH = 95+0.6*DEPTH

where the 95 is a maximum filter duration, DEPTH is in km, and the CUTH in sec

The cut windows attempt to include the P, pP, sP, pS, S and sS arrivals. However, oen must be very careful about the fact that PP may be included in some distance ranges.

The waveforms are then bandpass filtered by the application of the following high- and low-pass stages (an optional microseism filter):

hp c 0.0020 2
lp c 0.0050 2
int

The traces were next integrated to ground displacment in meters. Finally the observed data are interpolated to ahve the same sampling at the Green's functions.
NOTE: this was done for speed. The proper sequence is to read traces, filter and then cut - gsac will be modified to introduce a command CUTWR to define the cut upon a write.

The source inversion is a multipass operation since a lower frequency filter band is used for larger earthquakes and since a search is made over depth. Up to three passed of the outer loop are made, after which the moment magnitude is determined and filter settings readjusted. The inner loop over depth samples all depths from 0 to 800 km with 5 km increments in depth to 50 km, followed by 10 km depth sampling for the remaining range.

The following filter ranges are used according to the moment magnitude Mw:

   FILTER_BAND   FH(s)  FL(s)
       1           60     12      Mw < 6.4
       2          100     20      6.4 < Mw <= 6.9
       3          120     40      Mw > 6.9

The map displays the distribution of stations used for this source inversion.

Location of the earthquake (yellow star) and great circle path from the epicenter to each station (red) [created using GMT (Wessel, P., and W. H. F. Smith, New version of Generic Mapping Tools released, EOS Trans. AGU, 76 329, 1995.)]

For this data set the favored solution is

WVFGRD96   40.0   150    50    90   8.41 0.5369

The following figures show the sensitivity of the goodness of fit parameter so source depth, the waveform comparison as a function of epicentral distance in degrees and the source to station azimuth

Depth Sensitivity

Goodness of fit as a function of source depth. The measure is 1 - SUM (o -p)² / SUM o². A value of 1.0 is the best fit. The best double couple mechanism for the solution depth is plotted above goodness of fit value to indicate how the mefhanism may change with depth.

Detailed Waveform Comparison

P-wave Z component

Comparison of the observed traces (red) and solution predicted traces (blue) ordered in terms of increasing epicentral distance. Each pair of traces is annotated with the station name, epicentral distance in degrees, source to station azimuth in degrees. Each pair of traces is plotted with the same scale and the peak amplitudes are indicated at the lect of each trace. Finally the time shift between the P-wave first arrival picked and the the theoretical P-wave first arrival in the predicted trace is indicated, with a positive sign indicating that the predicted trace has been shifted to the right by the given number of seconds. as a function of source to station azimuth in degrees (D). The purpose of this display is to highlight the azimuthal dependence on the first motion. The traces are annotated with the station name at the top.

SH-wave T component

SV-wave R component

Inversion Details

Output of wvfmtd96 for the best depth..

Last Changed Fri Mar 23 01:28:40 UTC 2007