Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Code Block
languagesql
select o.objID, o.raMean, o.decMean, o.raMeanErr, o.decMeanErr, 
   o.raStack, o.decStack, o.raStackErr, o.decStackErr, o.epochMean, 
   o.nDetections, o.ng, o.nr, o.ni, o.nz, o.ny, o.objInfoFlag, o.qualityFlag,
   m.gMeanPSFMag, m.rMeanPSFMag, m.iMeanPSFMag, m.zMeanPSFMag, m.yMeanPSFMag,
   m.gMeanPSFMagErr, m.rMeanPSFMagErr, m.iMeanPSFMagErr, m.zMeanPSFMagErr, m.yMeanPSFMagErr,
   m.gMeanKronMag, m.rMeanKronMag, m.iMeanKronMag, m.zMeanKronMag, m.yMeanKronMag,
   m.gFlags, m.rFlags, m.iFlags, m.zFlags, m.yFlags 
   into mydb.MyTable_HDF_3deg_extended from fGetNearbyObjEq(169.20,62.30,180.0) nb
inner join ObjectThin o on o.objid=nb.objid and o.nDetections>3
inner join MeanObject m on o.objid=m.objid and o.uniquePspsOBid=m.uniquePspsOBid and m.rMeanPSFMag<21.0

...

The same region of the sky contains 86,685 Gaia DR2 sources, of which 77448 have a valid five-parameter astrometric solution, and 68638 have also a valid G magnitude.  For the first matching stage, we do not require valid magnitude or five-parameter astrometry; the direct positional match with a generous 10" radius finds 82,398 matches, with the vast majority well within 1":

Image Modified

In this comparison, it is useful to distinguish point sources from extended sources.  The latter wuill likely be treated differently in Gaia vs. Pan_STARRS, and their astrometric and photometric accuracy is likely poorer.  One way to distinguish point from extended sources is to use the difference between Kron and PSF magnitudes; another is encoded in the QF_OBJ_EXT bit (bit 1) from the qualityFlag parameter in ObjectThin (extracted from the query in Step 1):

ExtendedFlag = (qualityFlag AND 1)

(this is a bitwise AND).  The ExtendedFlag thus defined correlates well with the Kron vs. PSF magnitude difference (which presumably is part of the definition):

Image Modified

Red dots in this plot have ExtendedFlag == true.  In the following, we use ExtendedFlag to identify extended objects.

...

The match distance distribution shows clearly that "true" matches likely have match distances << 1".  This plot shows that PanSTARRS r-band PSF magnitudes and Gaia G band magnitudes match very well for ``true'' point source matches (separation < 0.2"):

Image Modified

The plot shows two populations of point sources: a narrow band with r-G ~ 0, containing a majority of the matches, and a broader distribution at positive r-G values (brighter in G than in r).  Extended sources are mostly brighter in PanSTARRS r, as expected since Gaia, with its higher angular resolution, would only measure the core of the object.  (We will neglect extended sources henceforth.)  It turns out that the second population of point sources, at positive r-G, is constituted of very red stars, for which the extremely broad G passband is dominated by red photons.  This becomes clear if the r-G magnitude difference is plotted against PanSTARRS colors, e.g., r-i:

Image Modified

where the region of positive r-G (Gaia brighter) corresponds with positive r-i (red stars).  The narrowness of the sequence testifies to the quality of Gaia and PanSTARRS magnitudes. 

We can take one additional step and attempt to define a color transformation that maps Gaia magnitudes (G, B, and R) into PanSTARRS r.  A degree 4 polynomial results in a very good match, albeit with large scatter at faint magnitudes, where Gaia B and R magnitudes lose precision:

Image Modified

It would also be possible to define a combination of PanSTARRS magnitudes to map into Gaia G, but the above plots clealy show that magnitudes are well matched, so this has not been pursued further. 

...

Finally, we compare the astrometric match between Gaia and PanSTARRS using both mean object quantities and the average of detection positions.  Mean object astrometry is obtained from the DR2 table ObjectThin and has been adjusted to Gaia.  Detection positions are taken from the DR2 Detection table, and have not been adjusted to Gaia.  Accordingly, if we assign to each matched object a ``mean detection'' position corresponding to the average of its detections (as indicated by objID), this position is a true indicator of the underlying quality of PanSTARRS astrometry before Gaia matching:

Image Modified

The black histogram shows the match distance for mean object (typically Gaia-adjusted) positions; it peaks at 5 mas, with a significant tail beyond 20 mas.  The red histogram uses the mean position of detections; it peaks at 20 mas, with a significant tail beyond 50 mas.  More analysis is in progress to better understand how proper motions might impact the relative astrometry for both matched and unmatched PanSTARRS sources.  In particular, we plan to use detection positions to obtain proper motion estimates or bounds for sources that do not have Gaia proper motions.  Note that separations based on mean detection positions have not been adjusted for the (known) proper motion of the corresponding Gaia source; this is ongoing.

...