XCORR(1)
NAME
xcorr - strategy module to compute cross correlations for
time-distance analysis.
SYNOPSIS
xcorr [-cnvz] [-KFSM] [argument= value
DESCRIPTION
xcorr computes temporal cross correlations for time-distance
analysis. Pairs of points are selected by their position in
the solar image.
Setting the verbose option (-v) results in more information
being sent to the standard output. Unfortunately, this is
mostly just a program progress report.
Data filtering is controlled by arguments FPASS, FWIDTH, and
flags -F and -K. The -F option suppresses filtering of the
f-mode (note, the default F=0 means filter out the f-mode).
The -K flag will result in a correction of the instrumental
astigmatism and a flattening of the spatial frequency
response of the MDI instrument.
The -S and -M flags control the form of the output. Unless
the -S flag is set, the output will be sorted in order of
increasing separation between cross correlated points. It
is meaningless to set this flag when AZMUTH=0.0, as the out-
put will always be sorted in this case. Set the -M flag
when using the DRANGE option if you want the full list of
distances and directions to be output. This results in a
lot of extra FITS files.
ARGUMENTS
in=dataset
The input dataset name. The input dataset is expected
to be a cube such as output by strategy module
track_region (see man page) with map=cyleqa. That is,
each input dataset is a "data cube" with longitude
along the fastest axis, sin(latitude) along the next
axis, and time on the slowest axis. The dataset is
assumed to include in its ancillary data the time
between consecutive images T_STEP, the number of
degrees per pixel at map canter MAPSCALE, and the lati-
tude at map center in degrees HG_LAT. NO DEFAULT
out=dataset
The output dataset name. NO DEFAULT
DIRECTION=string
The target direction. Possible options are north or
west. In the first case, the pairs of pixels selected
for cross correlation have the same or nearly the same
longitude. In the second case, they have the same or
nearly the same latitude. Default: north
DTMAX=int
The maximum correlation time ("lag") to be included in
the output, as a multiple of the interval between
images. The default is the length of the input time
series. This default is not very intelligent; for
example, when the input data spans several days.
DMAX=real
The maximum distance between pairs of points which are
to be cross correlated, in degrees. By default this is
the largest possible value that can be computed given
the dimensions of the input cube.
DMIN=real
The minimum distance between pairs of points which are
to be cross correlated, in degrees. (Default 0.0; note
that the cross correlation of points separated by zero
distance is the autocorrelation.)
DRANGE=real
The range used in binning the output. If 0.0, then
there is one row in the output file for each combina-
tion of distance separation and orientation. When
DRANGE is set, the cross correlations are averaged in
distance. Note that setting DRANGE to a non-zero value
will cancel the -S flag (the output will be sorted
automatically). (Default: 0.0)
AZMUTH=real
The maximum deviation, in degrees, from the target
direction. The default is 0.0, which means that only
those pairs of points which lie on the same parallel of
latitude (if DIRECTION=west) or meridian of longitude
(if DIRECTION=north) will be cross correlated. When
AZMUTH is set to a larger value, more pairs of points
will be cross correlated. (Default: 0.0)
FPASS=real
The input data is passed through a high-pass temporal
filter with gaussian rolloff before cross correlating.
FPASS is the cutoff frequency for the default high-pass
filter, in Hz. (Default: 0.0017)
FWIDTH=real
The width of the gaussian rolloff on the high-pass tem-
poral filter, as a fraction of FPASS. setting
FWIDTH=0.0 results in a square cutoff. (Default: 0.1)
EXAMPLES
xcorr -v -K
in=prog:mdi,level:lev2,series:track_08h[3723],sel:[0]
out=prog:mdi,level:lev2,series:rxcor_08h[3723],sel:[0-255]
DMIN=0.0 DMAX=10.0 AZMUTH=10.0 DRANGE=0.25
In this case, assume the input is a 256x256x480 cube of data
(created by track_region with map=cyleqa). There will be
256 output files, one for each latitude in the input cube.
SEE ALSO
track_region
DIAGNOSTICS
soi_errno is returned and also set as the abortflg value in
the results keylist.
BUGS
The approximation (used in the f-mode filter) for the p1
ridge is a polynomial fit and is probably not very good for
l > 1500.
The filtering performed under the -K option is not very well
tested on other than MDI data and in any case is probably
only appropriate for other instruments.
AUTHOR
Peter Giles pgiles@solar.stanford.edu