Cube
Analysis
In the following sections we walk through some other routines and
procedures which allow you to inspect the reduced data cube in various
ways.
Collapsing
to a 2D image
The first thing we can do is to collapse the data cube into a single 2D
image, producing what is sometimes termed a "white light" image. This
can be done using the KAPPA
command collapse,
> collapse in=tempcube_base axis=vrad estimator=mean out=tempcube2d
The image above shows the resultant "white light" image from a cube which has had its baselines corrected for. The axis to collapse the cube along is chosen with the axis parameter.
If you know what the axis is called then you may give that as your
response (in this case we know that the spectral axis is called VRAD - it's
case insensitive also!). If you do not know the axis name then provide
the axis
number, as in this example:
> collapse in=tempcube_base axis=3 estimator=mean out=tempcube2d
If you enter the wrong axis by mistake, then a most useful and polite
error message informs you of your mistake and also provides you with a
list of valid options:
AXIS - The axis to be collapsed /'VRAD'/ > redshift
!! 'redshift' is not a valid option for parameter AXIS.
! Valid options are: 'RA','Dec','VRAD'
! Options may also be specified by index in the range 1 to 3.
!! Please supply a new value for parameter AXIS.
AXIS - The axis to be collapsed /'VRAD'/ >
By default the mean
is used to calculate the final 2D image. Other options include median, mode and a wmean which
calculates the weighted mean based on the associated variance (and only
works if a variance array is present). The estimator
parameter is a a 'sticky' one, which means that it remembers it's last
value (stored in the adam
directory). So to be sure of what is entered, it is best to always
specify it or to use reset
on the
command line (although this will set all of your parameters to their
defaults). Look at
the help on the estimator
parameter for the many other options that are available, some of which
are described later in this document.
One can also collapse the cube along a specified range. If
the range is known (maybe from inspecting the cubes with GAIA)
then they can be specified with the low and high parameters:
> collapse in=tempcube_wcs axis=vrad estimator=mean low=-10.0 high=10.0 out=tempcube2d
If you are not sure about the range that you want to
collapse your cube along, then you may want to plot out a spectrum for
inspection.
Plotting
a spectrum within a cube
To do this, use the display
command to show your 2D image after collapse. To
find the position where you would like to extract and plot a spectrum,
use the cursor
command:
> cursor showpixel
The showpixel
parameter will ensure that the pixel indices as well as their WCS
values
are shown. An example return on the terminal screen after clicking on
the image would be:
> cursor showpixel
DEVICE - Name of graphics device /@xwin/ >
Use the cursor to select positions to be reported.
To select a position press the space bar or left mouse button
To forget the previous position press "d" or the middle mouse button
To quit press "." or the right mouse button
Picture comment: KAPPA_DISPLAY, name: DATA, reporting: SKY co-ordinates
RA (hh:mm:ss.s) Dec (ddd:mm:ss)
3:25:37.0 30:45:09
(-3.4 2.3)
The pixel indices are shown in parantheses. Note that floating point
pixel coordinates will be interpreted as integer pixel coordinates as
long as the NDF does not contain AXIS structures (created, perhaps,
using the setaxis
command in KAPPA).
To plot the spectrum:
> linplot 'tempcube(-3.4,2.3,)' device=xw \\
You can now visually inspect a spectrum from the cube on a xwindows
display to select your velocity limits.
Of course, this can be done in a more interactive and convenient way
from within GAIA.
Smoothing
data
There are various routines available which can smooth data and users
are encouraged to explore them to see which suits their needs best. In KAPPA, the most
common routines for data smoothing are block and gausmooth.
Each of the these routines can smooth the whole data cube, but their default behaviour is to do
so by iterating through each 2D plane.
Although this is not a true 3D smooth, the effect is still to smooth
the spectrum as the following image shows.
The red line is the original spectrum and the blue line shows the same
spectrum after the cube was subject to a 3 pixel gausmooth. The plane on which data is smoothed is controlled by the axes parameter. The default is axes=[1,2] which means that smoothing is successively performed in each parallel plane which includes these two axes.
The spectrum was smoothed using the command:
> gausmooth tempcube fwhm=3 axes=[1,2] out=tempcube_sm \\
Moment
analysis
It is also possible to use the collapse
command to look at the principal moments in your data cube. The image
below shows the result of using the Iwc estimator
in the collapse
command (the white light image collapsed over the same velocity
intervals is overlaid as blue contours):
> collapse tempcube_base estimator=Iwc axis=vrad low=-25 high=25 tempcube_Iwc
This calculates the value of,  , the intensity
weighted velocity in this case, for all pixels in the image. Using the
estimator Iwd
will return an image of the intensity weighted dispersion in the image.
The image shows how the northern lobe is blueshifted relative to the
southern, fainter stream of material.
Creating
channel maps
Channel maps 'salami-slice' the data along a particular axis and
present the integrated intensity for each slice. The most common reason
for creating channel maps is as a method of visually inspecting the
distribution of radiating gas as a function of velocity. To create a
channel map (seen in the image below), one can use the KAPPA command chanmap,
> chanmap in=cube axis=3 low=-20 high=20 nchan=6 shape=3 estimator=mean out=chan
Use the display command to then plot the channel maps out:
Here we have chosen 6 channels (nchan)
in total to be determined between -20 and +20 km/s along the velocity
axis (3). The widths of the channels are determined by the software and
are chosen to be uniform (as best possible). The shape
parameter ensures that there will be 3 panels plotted horizontally with
the rest of the channels distributed vertically. If you wish the
software to decide how to best display the channels, then enter a null ! character for
the shape command. For each slice, the mean is used to calculate the
intensity in that channel. The same estimator
choices that are available to the collapse
command are available to chanmap.
...ooh,
that looks like SPECX!
Finally, we have the return of an old favourite - well, almost. The clinplot
routine in KAPPA
allows users to display spectra on a defined grid, very similar to the grid-spectra
command in SPECX.
The image below was created by first displaying an image, choosing the
region which to display by specifying sections of the ndf, and then
overlaying the grid of spectra with the clinplot
command:
> display 'tempcube_Integ(-10:0,-10:0)' device=xw lut=~/lutgaia \\
> clinplot 'cube(-10:0,-10:0,-15:25)' noclear specstyle='colour(curves)=blue'
device=xw
Other parameters such as lmode,
ytop
and ybot
give you control of how the spectra are plotted.
The lut
parameter in the display
command allows a colour look-up table to be used for the image. Pixel
coordinates have been used to section the ndf so that only the region
of interest is displayed, and in the clinplot
command, a section on the third axis is used to select a velocity range
(defined in pixel coordinates not velocities).
The noclear
parameter in the clinplot
prevents the previous display from being erased. When this parameter is
used, the line graphics are first aligned (using the WCS information in
the axis frames of both images) and then plotted over the image.
<Back to
Contents>
|
