What would ALMA detect, in an Early Science configuration, looking at a star-forming region in the dwarf galaxy Leo T, in the HNC J=4-3 line? The CASA task simdata takes a FITS image (shown here) as input, regrids this to a given resolution (spatial and frequency), sets the pointing position and scales the flux density (in Jy/pixel). This is sampled by the chosen set of antennas and duration, with optional added noise, to produce a measurement set. Finally, these visibility data are imaged to produce a cube according to the specifications. The measurment sets with and without added noise are also provided for you to image with other settings.
W49, a star-forming region at 11 kpc, was observed using the JCMT by Roberts et al. 2011A&A...525A.107R. This simulation uses W49_HCN.FITS, an HCN J=4-3 image cube of four spectral channels spanning 25 km/s, with the systemic velocity in the second channel. The input pixel size is 7.5 arcsec (about half the JCMT beam), total image 135 arcsec by 150 arcsec. The peak is 87 K km/s and the JCMT efficiency is about 0.7. The rest frequency of HCN J=4-3 is 354.50547 GHz.
Leo T is a nearby dwarf galaxy with signs of ongoing star-formation (Ryan-Weber et al. 2008MNRAS.384..535R). It is at 0.4 Mpc, nominal position 09:34:53.5 +17:02:5. Note that the total extent of the galaxy is about 6 arcmin, whilst the ALMA primary beam FWHM at 354 GHz is only 0.25 arcmin, so a more accurate position would be needed in real life. The galaxy systemic velocity is 11 km/s (converted into the LSR frame, and assuming the radio convention) and the total HI velocity span is about 40 km/s.
The ALMA Early Science Extended Configuration antenna table lists 16 antennas, providing baselines between 36 and 400 m. This gives a beam size of about 0.45 arcsec at 354 GHz. This (and many other antenna tables) are incorporated into recent CASA installations (from 3.2 onwards). If you do not have the Early Science antenna tables, you can download prototypes from alma.cycle0.extended.cfg alma.cycle0.compact.cfg
The coarsest spectral configuration of ALMA is 4096 channels per 2 GHz bandwidth. Averaging every 16 channels would give a spectral resolution of 6.6 km/s, close to the original, so let's use channels centred at [6.6, 13.2, 19.4, 25.2] km/s. Hence, the lowest frequency channel should be at 354.47567 GHz with a channel width of 0.0078125 GHz. The total useable bandwidth corresponds to over 1500 km/s.
Leo T is about 36 times more distant than W49. Adjusting to convenient numbers, each input 7.5-arcsec pixel becomes 0.2 arcsec and the total size is about 4 arcsec, which is well within the primary beam of 15 arcsec (ignor mosaicing options). However, in real life, there might be other line and continuum sources within the field of view/bandwidth.
The peak brightness becomes 0.1 Jy/pixel. The input noise would then become 0.0002 Jy/pixel. Assuming that we want good visibility plane coverage, try an observing time of 8 hr on source. The ALMA Sensitivity Calculator gives a noise level of 0.0015 Jy/beam so we will add noise.
Start with the default inputs and provide the parameter values described above. This assumes that the input FITS file, W49_HCN.FITS is in the current working directory, but you can enter a path if the data are elsewhere.
python provides a way to locate the antenna tables included in CASA
repodir=os.getenv("CASAPATH").split(' ')[0]
antdir = repodir+"/data/alma/simmos/"
default simdata#Root name for outputs (delete or move any existing files of same name, or use a new name, each time you run simdata; or else set the overwrite parameter to True):
project = 'LeoT_HCNJ4-3'#Input model details:
skymodel = 'W49_HCN.FITS' modifymodel = True inbright = '0.1Jy/pixel' incell = '0.2arcsec' indirection = 'J2000 09h34m53.5 +17d02m52' incenter = '354.47567GHz' inwidth = '0.0078125GHz'#Observational details you need to specify:
predict = True totaltime = '8h' antennalist = antdir+"alma.cycle0.extended.cfg" # **see note below thermalnoise = 'tsys-atm'# Imaging. Select a cell size to give you more than 3 pixels across the synthesized beam.
image = True vis = '$project.noisy.ms' imsize = 256 cell = '0.15arcsec' niter = 500# Diagnostics: watch the output as not all are shown at once.
analyze = True showarray = True showuv = True showfidelity = False#Check and execute:
inp simdata go simdata
** Antenna table path If you have an old CASA installation or want to provide other antenna tables, or if you get an error message ... neither antennalist nor sdantlist exist', you can set the entire path manually e.g. (for my linux installation in my home area)
antpath = '/home/amsr/CASA/casapy-32.1.15198/data/alma/simmos/'Watch the messages in the logger and the display of diagnostic plots.
Use the CASA viewer to display the output cube, LeoT_HCNJ4-3.image. The plot below shows the output image using 400-m baselines in colour scale, with the beam at lower left in white. The magenta contours (at [50, 100, 200, 400...] mJy/beam) are for an image made using 150-m baselines, peak 1.4 Jy/beam, beam also shown in magenta. You could re-run clean yourself using LeoT_HCNJ4-3.noisy.ms as the input, changing parameters or doing an interactive clean. Also, try different inputs to simdata, e.g. for the LMC at 50-kpc distance; you will find that the larger size means that this ALMA configuration will resolve-out some of the extended emission. To avoid this, try using the ALMA Early Science Compact Configuration antenna table with baselines of 18 to 150 m (don't forget to change the output project name as well as the antenna table in the simdata inputs).
