The intervals marked as non-stationary PSD's for double
coincidence are actually periods for which the cross spectrum
between H1&L1 or H2&L1 is found to be non-stationary as
seen in the S3 stochastic background analysis. At this time
there is not a separate flag for single interferometers. If
either IFO has a non-stationary PSD, then both get marked.

The intervals are being provided as part of the S3DQ release 6
in case they prove useful for other analyses. One could imagine,
for example, studying the relevance of the L1 flags by
looking at intervals when both H1-L1 and H2-L1 show non-stationarity.
Or one could study the H1 flags by looking at intervals when the
H1-L1 cross-spectrum is non-stationary, but H2-L1 is not flagged.
Similarly for H2. Because of random overlap and periods when one
Hanford IFO is down, these studies would not distinguish completely
cleanly which IFO's are the culprits in the cross-spectrum
non-stationarities.

The deadtimes associated with these flags are substantial (5-25%,
depending on how they are used). It would be unwise to use them
without careful evaluation of their relevance. If other analyses
do find these flags to be promising, however, the stochastic group
will be asked to provide refined versions, based on single-IFO PSD's.

Note that, by constrution, these flags have been defined only for
intervals with double coincident science mode for L1 and one or
both of H1/H2.

K. Riles - June 21, 2005