Problems with Petrosian magnitudes for deblended sources

Petrosian magnitudes in Y, H and K appear to be much too bright (yes, bright) when the source is flagged as deblended.

Summary

Don't use Petrosian magnitudes in Y, H and K for sources flagged as deblended.

The problem

I have selected sources in the DR3 lasYJHKsource table which have

• Detections in all four bands
• ppErrBits in each band either set to 0 (no problems) or 16 (deblended)

These are then plotted on colour-magnitude diagrams for all colours and for four different magnitudes: AperMag3 (2"), AperMag7 (8"), AperMag13 (24") and PetroMag.

After plotting the non-deblended sources on the colour-magnitude diagram, I have plotted sources

• flagged as deblended in the redder band only
• flagged as deblended in the bluer band only
• flagged as deblended in both bands

What you would expect

Consider a source deblended in the redder band but not deblended in the bluer band. This source has too much flux in the bluer band, so it should look bluer than usual (blue contours on plots).

And vice versa.

Sources deblended in both bands should lie in a similar region to the non-deblended sources.

This is illustrated for the H and K bands for AperMag13 (24" aperture). Blue contours show sources that one would expect to be bluer than usual, and red contours show sources that should be redder than usual. All looks fine so far.

• hkAperMag13.png:
  

What happens with Petrosian magnitudes

With Petrosian magnitudes (in Y, H and K; J seems okay), the opposite trend is seen, as shown in this plot for H and K Petrosian magnitudes.

• hkPetroMag.png:
  

All 24 plots are in the ZIP file: DeblendingPlotsx24.zip: DeblendingPlotsx24.zip

-- AnthonySmith - 20 Dec 2007

What next?

Over to you guys!