def remove_products(name):
'''
For the dirty image, we don't need to keep the:
* residual
* flux
* model
If the image does not exist (i.e., no PSF due to flagging), remove all.
'''
removals = [".flux", ".residual", ".model"]
# Check if the image exists
if not os.path.exists(name + ".image"):
rmtables(name + ".*")
else:
for remove in removals:
rmtables(name + remove)
outframe = "LSRK"
veltype = "radio"
minpb = 0.1
weighting = 'natural'
robust = 0.0
restfreq = rest_freq
usescratch = False
interpolation = 'linear'
if do_concat:
ms_names = []
for f in listdir_fullpath(line_direc):
if f.endswith(".ms"):
ms_names.append(f)
rmtables(concat_ms)
concat(vis=ms_names, concatvis=concat_ms)
if dirty_cube_nosub:
# Check that the concatenated MS exists
if not os.path.exists(concat_ms):
raise IOError("Concatenated MS does not exist in the given directory.")
if not os.path.exists(dirty_image_direc):
os.mkdir(dirty_image_direc)
out_image = os.path.join(dirty_image_direc,
line_name + "_" + project_name + "_dirty_nocontsub")
def ft_beam(vis=None,
refant='Tile012',
clobber=True,
spectral_beam=False,
subcalibrator=False,
uvrange='>0.03klambda'):
"""
def ft_beam(vis=None,refant='Tile012',clobber=True,
spectral_beam=False,
subcalibrator=False,uvrange='>0.03klambda'):
# Reference antenna
refant='Tile012'
# Overwrite files
clobber=True
# Option to include the spectral index of the primary beam
spectral_beam=False
# Option to add more sources to the field
"""
# output calibration solution
caltable = re.sub('ms', 'cal', vis)
if vis is None or len(vis) == 0 or not os.path.exists(vis):
print 'Input visibility must be defined'
return None
# Get the frequency information of the measurement set
ms.open(vis)
rec = ms.getdata(['axis_info'])
df, f0 = (rec['axis_info']['freq_axis']['resolution'][len(
rec['axis_info']['freq_axis']['resolution']) / 2],
rec['axis_info']['freq_axis']['chan_freq'][len(
rec['axis_info']['freq_axis']['resolution']) / 2])
F = rec['axis_info']['freq_axis']['chan_freq'].squeeze() / 1e6
df = df[0] * len(rec['axis_info']['freq_axis']['resolution'])
f0 = f0[0]
rec_time = ms.getdata(['time'])
sectime = qa.quantity(rec_time['time'][0], unitname='s')
midfreq = f0
bandwidth = df
if isinstance(qa.time(sectime, form='fits'), list):
dateobs = qa.time(sectime, form='fits')[0]
else:
dateobs = qa.time(sectime, form='fits')
if spectral_beam:
# Start and end of the channels so we can make the spectral beam image
startfreq = f0 - df / 2
endfreq = f0 + df / 2
freq_array = [midfreq, startfreq, endfreq]
else:
freq_array = [midfreq]
# Get observation number directly from the measurement set
tb.open(vis + '/OBSERVATION')
obsnum = int(tb.getcol('MWA_GPS_TIME'))
tb.close
info = mwapy.get_observation_info.MWA_Observation(obsnum, db=db)
print 'Retrieved observation info for %d...\n%s\n' % (obsnum, info)
# Calibrator information
if info.calibration:
calibrator = info.calibrators
else:
# Observation wasn't scheduled properly so calibrator field is missing: try parsing the fieldname
# assuming it's something like 3C444_81
calibrator = info.filename.rsplit('_', 1)[0]
print 'Calibrator is %s...' % calibrator
# subcalibrators not yet improving the calibration, probably due to poor beam model
if subcalibrator and calibrator == 'PKS0408-65':
subcalibrator = 'PKS0410-75'
elif subcalibrator and calibrator == 'HerA':
subcalibrator = '3C353'
else:
subcalibrator = False
# Start models are 150MHz Jy/pixel fits files in a known directory
model = modeldir + calibrator + '.fits'
# With a corresponding spectral index map
spec_index = modeldir + calibrator + '_spec_index.fits'
if not os.path.exists(model):
print 'Could not find calibrator model %s' % model
return None
# Generate the primary beam
delays = info.delays
str_delays = ','.join(map(str, delays))
print 'Delays are: %s' % str_delays
# load in the model FITS file as a template for later
ftemplate = pyfits.open(model)
# do this for the start, middle, and end frequencies
for freq in freq_array:
freqstring = str(freq / 1.0e6) + 'MHz'
# We'll generate images in the local directory at the right frequency for this ms
outname = calibrator + '_' + freqstring
outnt2 = calibrator + '_' + freqstring + '_nt2'
# import model, edit header so make_beam generates the right beam in the right place
if os.path.exists(outname + '.fits') and clobber:
os.remove(outname + '.fits')
shutil.copy(model, outname + '.fits')
fp = pyfits.open(outname + '.fits', 'update')
fp[0].header['CRVAL3'] = freq
fp[0].header['CDELT3'] = bandwidth
fp[0].header['DATE-OBS'] = dateobs
fp.flush()
print 'Creating primary beam models...'
beamarray = make_beam.make_beam(outname + '.fits', delays=delays)
# delete the temporary model
os.remove(outname + '.fits')
beamimage = {}
for stokes in ['XX', 'YY']:
beamimage[
stokes] = calibrator + '_' + freqstring + '_beam' + stokes + '.fits'
# scale by the primary beam
# Correct way of doing this is to generate separate models for XX and YY
# Unfortunately, ft doesn't really understand cubes
# So instead we just use the XX model, and then scale the YY solution later
freq = midfreq
freqstring = str(freq / 1.0e6) + 'MHz'
outname = calibrator + '_' + freqstring
outnt2 = calibrator + '_' + freqstring + '_nt2'
# divide to make a ratio beam, so we know how to scale the YY solution later
fbeamX = pyfits.open(beamimage['XX'])
fbeamY = pyfits.open(beamimage['YY'])
ratiovalue = (fbeamX[0].data / fbeamY[0].data).mean()
print 'Found <XX/YY>=%.2f' % ratiovalue
# Models are at 150MHz
# Generate scaled image at correct frequency
if os.path.exists(outname + '.fits') and clobber:
os.remove(outname + '.fits')
# Hardcoded to use the XX beam in the model
fbeam = fbeamX
fmodel = pyfits.open(model)
fspec_index = pyfits.open(spec_index)
ftemplate[0].data = fbeam[0].data * fmodel[0].data / (
(150000000 / f0)**(fspec_index[0].data))
ftemplate[0].header['CRVAL3'] = freq
ftemplate[0].header['CDELT3'] = bandwidth
ftemplate[0].header['DATE-OBS'] = dateobs
ftemplate[0].header['CRVAL4'] = 1
ftemplate.writeto(outname + '.fits')
print 'Wrote scaled model to %s' % (outname + '.fits')
foutname = pyfits.open(outname + '.fits')
# Generate 2nd Taylor term
if os.path.exists(outnt2 + '.fits') and clobber:
os.remove(outnt2 + '.fits')
if spectral_beam:
# Generate spectral image of the beam
fcalstart = pyfits.open(calibrator + '_' + str(startfreq / 1.0e6) +
'MHz_beamXX.fits')
fcalend = pyfits.open(calibrator + '_' + str(endfreq / 1.0e6) +
'MHz_beamXX.fits')
ftemplate[0].data = (n.log(fcalstart[0].data / fcalend[0].data) /
n.log((f0 - df / 2) / (f0 + df / 2)))
beam_spec = '%s_%sMHz--%sMHz_beamXX.fits' % (
calibrator, str(startfreq / 1.0e6), str(endfreq / 1.0e6))
if os.path.exists(beam_spec):
os.remove(beam_spec)
ftemplate.writeto(beam_spec)
fbeam_spec = pyfits.open(beam_spec)
ftemplate[0].data = foutname[0].data * fbeam[0].data * (
fspec_index[0].data + fbeam_spec[0].data)
else:
ftemplate[
0].data = foutname[0].data * fbeam[0].data * fspec_index[0].data
ftemplate[0].header['DATE-OBS'] = dateobs
ftemplate.writeto(outnt2 + '.fits')
print 'Wrote scaled Taylor term to %s' % (outnt2 + '.fits')
# import as CASA images
if os.path.exists(outname + '.im') and clobber:
tasks.rmtables(outname + '.im')
if os.path.exists(outnt2 + '.im') and clobber:
tasks.rmtables(outnt2 + '.im')
tasks.importfits(outname + '.fits', outname + '.im')
tasks.importfits(outnt2 + '.fits', outnt2 + '.im')
print 'Fourier transforming model...'
tasks.ft(vis=vis,
model=[outname + '.im', outnt2 + '.im'],
nterms=2,
usescratch=True)
print 'Calibrating...'
tasks.bandpass(vis=vis, caltable=caltable, refant=refant, uvrange=uvrange)
print 'Scaling YY solutions by beam ratio...'
# Scale YY solution by the ratio
tb.open(caltable)
G = tb.getcol('CPARAM')
tb.close()
new_gains = n.empty(shape=G.shape, dtype=n.complex128)
# XX gains stay the same
new_gains[0, :, :] = G[0, :, :]
# YY gains are scaled
new_gains[1, :, :] = ratiovalue * G[1, :, :]
tb.open(caltable, nomodify=False)
tb.putcol('CPARAM', new_gains)
tb.putkeyword('MODEL', model)
tb.putkeyword('SPECINDX', spec_index)
tb.putkeyword('BMRATIO', ratiovalue)
try:
tb.putkeyword('MWAVER', mwapy.__version__)
except:
pass
tb.close()
print 'Created %s!' % caltable
return caltable
ms.close()
flagmanager(vis=visfile,
mode='save',
versionname='before_cont_flags')
# not clear why this is done in other imaging scripts, but it
# seems to achieve the wrong effect.
#initweights(vis=visfile, wtmode='weight', dowtsp=True)
flagdata(vis=visfile,
mode='manual',
spw=linechannels,
flagbackup=False)
rmtables(contvis)
os.system('rm -rf ' + contvis + '.flagversions')
tb.open(invis)
if 'CORRECTED_DATA' in tb.colnames():
datacolumn = 'corrected'
else:
datacolumn = 'data'
tb.close()
# Average the channels within spws
assert split(vis=visfile,
spw=",".join(map(str, spws)),
field=field,
outputvis=contvis,
width=widths,
ia.close()
# If there's nothing there, max == min
max_val = stats_dict["max"]
if max_val == 0:
casalog.post("No valid region in the given mask. Skipping clean.")
valid_mask = False
else:
casalog.post("Mask contains valid region. Proceeding with clean.")
if valid_mask:
clean(vis=vis, imagename=out_root + '.clean', field='M33*',
restfreq='1420.40575177MHz',
mode='channel', width=1, nchan=1, start=1,
cell='1.5arcsec', multiscale=[0, 4, 8, 20, 40, 80],
threshold='2.2mJy/beam', imagermode='mosaic', gain=0.1,
imsize=[4096, 4096], weighting='natural', robust=0.0, niter=200000,
pbcor=True, minpb=0.2, interpolation='linear', usescratch=False,
phasecenter='J2000 01h33m50.904 +30d39m35.79', veltype='radio',
modelimage=model, mask=mask)
# Only run if the image was actually produces
if os.path.exists(out_root + ".clean.image") and model is not None:
feather(imagename=out_root + ".clean.image.feathered",
highres=out_root + ".clean.image",
lowres=model)
# If something went awry, and the image wasn't produced, remove the
# remnants.
if not os.path.exists(out_root + ".clean.image"):
rmtables(out_root + ".clean.*")
