Adam Ginsburg's blog

Articles by Adam Ginsburg (Adam.G.Ginsburg@gmail.com)

• Sensitivity Plots using astroquery NRAO tool

  I wanted to see what the approximate depth achieved by (repeated) VLA observations of a few sources was, at least theoretically. This is relatively easy to do:

  from astropy import units as u
  from astroquery.nrao import Nrao
  result = Nrao().query(coordinates=coordinates.SkyCoord.from_name('Sgr B2(M)'),
                        radius=2*u.arcmin)
  
  import pylab as pl
  
  pl.figure(1)
  pl.loglog(result['Observing Freq']/1e3, result['TOS'], '.')
  pl.xlabel("Observed Frequency (GHz)")
  pl.ylabel("Time-on-Source (s)")
  
  pl.figure(2)
  pl.loglog(result['Observing Freq']/1e3, result['TOS']*result['Bandwidth'], '.')
  pl.xlabel("Observed Frequency (GHz)")
  pl.ylabel("Relative depth (Hz s)")
  

  

---

• Export to FITS with MIRIAD

  To convert from MIRIAD map files to FITS, use:

  miriad% fits in=h2comos1.clr out=h2comos1.fits op=xyout
  

  (see the CLASS conversion post)

---

• Position-Velocity Diagrams

  tl;dr version: IPython notebook here, code available on github

  Current state of affairs

  Position-velocity diagrams with arbitrary position axes have been a common tool for studying spectra for a long time, probably because the "native" format of long-slit spectra on CCDs is effectively a position-velocity diagram.

  It's becoming more common now to have data cubes (PPV) than 1D or 2D spectra. But PV diagrams still have utility. Unfortunately, there aren't very good tools to produce them.

  There are plenty of tools that can slice along any of the native axes of a data cube. If you take your data cube in RA/Dec space, extracting along lines of constant RA or Dec is trivial (GAIA can do it natively, but it's easy to slice a cube in IDL or python).

  However, if you wanted to slice along lines of constant GLON or GLAT, you'd have to reproject the cube to use this approach. There are nice tools that implement cube rotation: CASA has an imrebin recipe and STARLINK has a rotation tool (though I can't find it now...).

  However, this approach is terribly inefficient if you want slices at multiple different angles, and it's outright awful if you want to track a filament.

  An alternative approach

  The alternative is to interpolate between pixels along a specified axis. This can be done very efficiently in python using scipy's map_coordinates, which by default does a cubic-spline interpolation at the specified points.

  I've started to build a tool to extract position-velocity diagrams on this basis. It's fully functional, but doesn't do anything interactive just yet. pvextractor is available on github.

  There is a complete example demonstrating its use here.

  In short, though, it allows you to describe paths like this:

  

  and extract them into position-velocity diagrams like this:

  

  Postscript: Figshare

  (note: postscript as in "P.S.", not ".ps")

  In order to create the example document (here), I needed a small data cube that could easily be downloaded. Dataverse hosts the COMPLETE survey, but they make you jump through a javascript hoop to download their data, which means that it is not possible to write an easily reproducible script. Too bad, since I would have used them except for this minor interface issue.

  Instead, I've found that figshare has fixed up their interface since I last used them. It was easy enough that I decided to upload my IRAS 05358+3543 data cubes and had no trouble at all. I also posted an update to the arxiv page for the article indictating that the data are freely shared. I'll do my best to share the data for the rest of my papers too, though it will take some time and I'll hit data limits for W5 pretty quickly.

---

« Page 7 / 7