CYBERSKA

 A Cyberinfrastructure platform to meet the needs of data intensive radio astronomy on route to the SKA

GALFACTS Technical Team

GALFACTS Technical Team

GALFACTS project technical development team
Parent Group: GALFACTS Consortium

Navigation

No pages created yet

  • Public

Automated Source Finding - SAD

Last updated 3385 days ago by Samuel George Comments (10)

Having looked at the C-field 13 Jul 2010 Release with Duchamp and encountering a few issues I thought it would be prudent to run a similar analysis with the AIPS implemenation of SAD.

In the Stokes I image, some 489 sources were found down to a detection threshold of 1K (as with Duchamp a number of obvious false detections were made and is a concern however the imprints of the beams on the data will be detected by these algorithms).

Comparison of the sources detected with the NVSS

With Duchamp an offset was found, though we are concerned with the way implementation of Duchamp, still an offset is clearly seen when overlaying the Stokes I contours of NVSS images. There appears to be an offset when comparing the NVSS positions with those determined from SAD (around 40" in both RA and Dec, the beam is ~ 180", so well within the beam). There appears to be linear relationship in the source fluxes (note: GALFACTS fluxes are in K). 469 of the sources are matched within the beam.

image

image

Source offsets (circle represents radius 90"): PDF Version (old version)
Source fluxes: PDF Version (old version)


Comparison of the sources detected with the GB6

The same comparison as above was completed but with the GB6 catalogue, some 353 sources are matched within the beam. A similar offset in source positions is seen:

image

Source offsets: PDF Version

 

Polarization properties of the sources

Some very strange behaviours with frequency (very rapid changes) and over a 5% change across the band.

image

image

image

image

 

plots of Q/I vs U/I for all sources found with SAD.

and plots of Q/I, U/I with frequency for all sources found with SAD.

 

Comments

  • Russ Taylor 3391 days ago

    Looks interesting Sam.  Looks like a simple single correction for the field will correct the positions to within 10 arcseconds or so.  

    Can you fit a straight line to the NVSS/GALFACTS correlation for NVSS fluxes about about 100 mJy?  The slope is the average K/Jy for the image (are you using the integrated flux density for the NVSS?).

    Finally.  Do you have polarization information from the NVSS?    If you have Q and U you could in principle plot the average NVSS Q and U values on the Q vs U plots, say as a red dot.  If the world was working the best possible the red dot would be in the middle of the GALFACTS cluster.  If there is a rotation between GALFACTS and NVSS then not.

  • Samuel George 3391 days ago

    Indeed does look like a small correction, but I'm wondering if this is an issue with the raw pointing that we should correct before imaging? 

    I'll add the line and make sure I'm using the integrated flux density.

    I thought I'd already included a red dot on the Q/U plots with the NVSS values in the older plots with Duchamp, maybe I didn't update the plots on the page in the end (or I just did it for the average over 50 channel plots). When I produce these I'll make sure that's included.

  • Russ Taylor 3391 days ago

    Good point re the pointing.  If all tracks are affected the same way then a single correction in the image is fine.  If different tracks have different errors then we need to make the correction to the raw data.  Might be difficult to sort this out in the latter case.

    It looks like the delta delta is larger in the Duschamp results.  Do you understand what is going on?

     

  • Samuel George 3390 days ago

    Can you fit a straight line to the NVSS/GALFACTS correlation for NVSS fluxes about about 100 mJy?  The slope is the average K/Jy for the image (are you using the integrated flux density for the NVSS?).

     

    image

    Fit is for all NVSS sources > 100 mJy (fitted in log space): gradient: 0.850361153552 intercept 0.974868154036

    PDF VERSION

    and this is for the peak, not integrated. Will change this when I get chance.

  • Samuel George 3390 days ago

    It looks like the delta delta is larger in the Duschamp results.  Do you understand what is going on?

    I think so - I think I found an issue with the comparsion of this, I've not had chance to look at it properly yet.

     

  • Samuel George 3390 days ago

    I've now included the plots of Q/I vs U/I and with frequency in the main page - I've not implemeted the red dot here, yet as this had already run - I had implemented it previously for plots of the calibration sources (for the calibration for the C field).

  • Russ Taylor 3390 days ago

    The fitting should be done in linear space and using integrated flux density for the NVSS and peak brightness for the GALFACTS. 

  • Russ Taylor 3390 days ago

    Also would you please plot the NVSS versus GALFACTS positions with equal scales on the RA and DEC axes?  It is important to know if the scatter is larger in one direction.

  • Samuel George 3388 days ago

    Also would you please plot the NVSS versus GALFACTS positions with equal scales on the RA and DEC axes?

    Done, and I've updated the plots in the page / the pdf to download.

  • Samuel George 3385 days ago

    I've update the flux plot so it uses the integrated flux.

     

    image

    Linear Fit: gradient:  0.852 intercept: 0.947