I think you'll have a bit of luck if once you do:
cube = fits.getdata('UCM2317_v300_com_sum.FC_AMC.fits')
you then reshape it to be 3D. It may actually work to do something like:
cube.shape = 27, 328, 247
or some permutation of number that multiply to give you the same final cell
count. I am just guessing at what these dimensions are based on the
original shape. You may have to manually re-arrange the data if the order
gets messed up.
If you end up having luck loading it up, please let us know. We are very
new to having observational astronomers load up their data, so I can't
promise it will be a smooth process, but definitely fun to try out!
On Wed, Feb 20, 2013 at 6:58 AM, Peter Teuben <teuben(a)astro.umd.edu> wrote:
On 02/20/2013 06:24 AM, Kacper Kowalik wrote:
On 02/20/2013 11:09 AM, Francisco Ocaña González wrote:
I found a post about 3D data in YT ,http://astrobiased.com/3d-data-vis-in-yt.html , very
interesting. I am
working with spectral cubes from PMAS Spectrograph (at 3.5m CAHA
telescope). Thus, I tried to use it. I am newbie with python, but I was
able to install everything :-)
Great to hear that!
However I get an error message when loading the data into yt's data
structure (script http://paste.yt-project.org/show/3182/
ERROR: ValueError: total size of new array must be unchanged
astropy: [ERROR ] 2013-02-20 10:46:53,395 ValueError: total size of new
array must be unchanged
Traceback (most recent call last):
File "3d_viewer.py", line 9, in <module>
pf = load_uniform_grid(data, cube.shape, 9e16)
line 490, in load_uniform_grid
grid_dimensions = domain_dimensions.reshape(nprocs,3).astype("int32")
ValueError: total size of new array must be unchanged
That happens due to the fact that load_uniform_grid expects that data
has 3 dimensions, whereas the data from your fits has only 2. You can
"fix" that by artificially adding one dimension by e.g. using numpy
function "expand_dims". I've modified your script slightly:
I hope it'll work for you.
I was able to view the data just fine in e.g. ds9, there's nothing wrong
with it. ds9 even handles 3D cubes these days.
but to me it looks like this data is unsuitable for yt. Perhaps the 984
positions along the Y axis of this fits image can be re-arranged in an
RA-DEC grid, and create a real 3D cube, something like a 30 x 30 x 2000,
perhaps along the lines Kacper has tried out, but the fits file does not
contain this information how to do that, in fact the FITS header is
criminally close to being useless. There's no WCS to speak of.
More seriously, before interesting things can be seen, the data need to be
calibrated. lines extracted, fitted, and you could perhaps see some
interesting structures in doppler-velocity space. I would look at IRAF to
handle this, they have lots of tools to handle such spectra.
yt-users mailing list