Color Color diagrams from SAS 39 data

back up to ippToPspsSAS39BillsNotes

Ken suggested that I look at the full force mean psf magnitude measurments by making color color diagrams. I thought it might be useful to compare these to the stack photometry so I did plots from those data as well.

In the following notes I describe what I did and show the resulting diagrams.

Forced Photometry

IPP psps forced photometry takes the positions of the detections made in the stack photometry and does forced photometry on the single frame warp images. These results are injested into DVO and an averaging si done there.

I decided to use the "view" ForcedMeanObjectView which is a join of the tables ObjectThin and ForcedMeanObject.

select ForcedMeanObjectView.*, (ForcedMeanObjectView.qualityFlag & 1) as extended   
INTO mydb.[forced_1406] from ForcedMeanObjectView 
where projectionID = 1406 and (ForcedMeanObjectView.qualityFlag) & 16 = 16

I decided to cut on the qualityFlag requiring that the "good stack" bit be set (see table ObjectQualityFlags for the bit detinitions) indicating that something (DVO ?) decided that we had good stack measurements. In order to keep they query time and size of the results table managable I queried a subset of the survey using the projectionID column. This restricted the results to objects in projection cell 1406 (334 < ra < 338 0 < dec < 4)

This yielded 562,311 objects.

The ForcedMeanObjectView only has fluxes. Since this was my first set of queries with the new schema, I decided to not convert the fluxes to magnitudes in the query. I simply grabbed the objects and ran the results through various iterations of a ​STILTS program to convert to magnitudes.at I used this perl script to facitiate the conversions. Adding entries like

8.9-2.5*log10(gFPSFFlux) AS gFPSFMag

to the list of columns in the query would allow us to avoid the STILTS step. However the database is going to be busy and keeping the math done there to a minimum seems like a polite thing to do. (8.9 is used as the zero point since fluxes are in Janskys).

Star / Galaxy Separation

I looked at the star galaxy separation discussion and consulted with experts and it was suggested that for this analysis I use something simple. iKronMag - iPSFMag > 0 ==> star otherwise we say it's a galaxy. Here is a distribution from the forced measurments.

Note that the EXT bit (extended) from the quality flags (as measured by IPP) isn't a very good star galaxy separator.

"Stars"

Here are several color color diagrams for objects with !iKronMag - !iPSFMag > 0

"Non-stars"

And here are the same diagrams for objects with iKronMag-iPSFMag < 0

Stack Photometry

Here is a similar analysis using stack photometry.

SELECT *
 INTO mydb.[stack_1406] from stackObjectView
 WHERE projectionID = 1406

This view has psf and kron magnitudes so the STILTS (or math in the query) step was not required.

For reasons that I don't quite recall I chose -.1 as the star galaxy separator value versus 0 above. These plots do not look signifcantly different with this cut than they do with zero.

"Stars"

Here are several color color diagrams for objects with !iKronMag - !iPSFMag > 0

"Non-stars"

And here are the same diagrams for objects with iKronMag-iPSFMag < 0