Project    People    Science    Observations    Data Reduction  
 Image Gallery    WFI links    OAC Home   Internal page   BACK  

New astrometry and image stacking

Mario Radovich

A new procedure for the astrometry and coaddition of the OACDF

The usage of the MSCRED package for astrometry and coaddition of the images has two limitations:

  1. It is not possible to combine in one image the two OACDF2 and OACDF4 fields; it is therefore necessary to extract two separate catalogs.
  2. It is not possible to run SExtractor in dual mode using e.g. the R-band image as a detection image. To build the multi-band catalog it is therefore necessary to make a cross-correlation between the catalogs in the individual bands: this introduces the risk that either wrong associations are found or objects are lost.

During my stay at the Institut d'Astrophysique de Paris (IAP) I developed, in the framework of a cooperation between OAC and the TERAPIX team, a new procedure (ASTROMETRIX) to compute the astrometric solution in wide-field images. The procedure is based on the "global astrometry" technique: the astrometric solutions is computed by simultaneously minimizing the differences in the positions of the sources in a given images and [i] those from a reference astrometric catalog (the USNO A-2) and [ii] those of the same sources in other overlapping images. This approach allows to compute the astrometric solution in fields that are wider than a single mosaic.

Before the coaddition is done, it is necessary to compute the scaling factors to be applied to each frame to take into account the airmass and the changes in the atmospheric transparency (note that the images may come from different nights and observing runs). To this aim I developed a procedure (PHOTOMETRIX) that uses the same approach as for astrometry: scaling factors are computed by minimizing the differenc in magnitude of sources in overlapping images.

Finally, the resampling of the images with the astrometric solution and their coaddition are done using the   SWarp  software developed by E. Bertin. SWarp allows to resample images on the top of a refeence image and then combine the coadded images from the individual bands into one chi-squared image: this can be then used as a detection image. This approach is the optimal one in the case of multi-band images since the detection image will contain the sources detected in one or more individual bands.

As a test I compared the magnitudes obtained using Iraf+MSCRED and Astrometrix+SWarp.
The following table shows the difference in magnitude (average and RMS) obtained using the sources in the given range of magnitudes (OACDF2+OACDF4):

Filter Mag. range Mag(As+Sw)-Mag(iraf) Average Mag(As+Sw)-Mag(iraf) RMS
B 17 - 23.5 0.01 0.04
V 17 - 23.5 0.05 0.04
R 17 - 23.5 0.06 0.08
I (shallow) 17 - 21.0 0.02 0.10
753 17 - 22.5 0.01 0.09


The plots show the difference in magnitudes vs. the position (RA and DEC) and magnitude.