From:	SMTP%"ellsworth@usgs.gov"  7-MAR-2007 11:56:31.72
To:	djackson@ucla.edu, ykagan@ucla.edu
CC:	shaw@ldeo.columbia.edu, field@caltech.edu
Subj:	Centroid separation of global temporal doublets

Hi David and Yan,

I prepared the attached figure using the data you showed yesterday on the
magnitudes and centroid separations of large magnitude temporal doublets in
the Harvard CMT catalog.  The plot compares the observed centroid
separation (X-axis) with a model separation for events that just touch
without overlap (Y-axis).  The model assumes a mantle depth and 5 MPa
stress drop.  The cartoon of two circles illustrates the model. Plotted are
the magnitudes of the first event in the pair.  The diagonal line separates
overlapping ruptures (according to this simple model) from those with no
overlap.

The tendency in the data is for no or little overlap.  Only the 2001 M 8.5
/7.7 pair falls in the area of large overlap.  Of course, the errors in the
differential centroid locations have to be considered, and those are
probably impossible to pull out of the CMT catalog.  As we have discussed
(and Hurukawa and I illustrated) the better approach would be to
systematically measure the centroid separations.  A study worth doing.

Bill
(See attached file: K&J doublets.jpeg)