Photometric Calibration

Elixir automatically provides measurements of the CCD zeropoints. This is done with modest accuracy in realtime during the observing run through the Ptolemy I analysis, and to high accuracy at the end of the run with the Ptolemy II analysis. As discussed elsewhere, the Ptolemy analysis is applied to every image of the sky (IMAGETYP = OBJECT), regardless of the content of the images. The analysis includes object detection, photometric measurement, and astrometric calibration. The resulting data are incorporated in a database of stellar photometry. Since the same photometric database can be made to include the measurements of all standard stars, and since the association between stars on the sky is performed on the basis of the celestial coordinates, observations of the standard stars are automatically associated with the standard star measurements themselves. The determination of the zeropoints is at this point a simple matter of extracting all and only the measurements of the standard stars and comparing measurements in the appropriate filters. The rest of the process are mere details.

Of course, the details can be fairly complex. The first point to mention is the mechanism used to speed up the extraction of the correct stars. When we talk about standard stars in this context, we assume that we are talking about the Landolt standard stars. It is possible, of course, to use any standard stars, and in most cases the same arguments would apply. Landolt's standard stars are not uniformly distributed on the sky, nor is there a large number of such stars. Instead, they are found in a small number of special locations, providing a way to speed up their selection. The organization of the photometry database makes use of the geometry of the sky. The individual stellar measurements are stored in roughly 8000 files ('region files' or 'cpt files'), each representing a small patch on the sky, bounded by lines of constant RA and DEC. These files are in turn stored in 12 separate directories representing 7.5° bands of declination. The standard stars are located in a small subset of the 8000 region files, greatly speeding up the search for the standard star measurements.

Another point to discuss is the use of photcodes to distinguish the standard stars from other stars. One type of photcode is the REFERENCE type, which refers to photometry not determined by Elixir (external photometry). These REF type photcodes can be used for photometric standard measurements as well. For example, the Landolt photometry standards are included in the photometry database with REF photcodes names B_L92, V_L92, etc, where L92 refers to the Landolt (1992) reference.

We now have all the elements necessary to determine the photometric calibration for a given set of images, whether from a specific night, or 12k run. If we want to determine the current zero points for the V filter images, we simply extract the measurements from the specific region files for the Landolt fields, then we look for all stars with a measurement in the V_L92 band, and then check if it has also been measured in any of the dependent filters associated with V.

Having identified all of these V measurements, there are a variety of operations we might want to perform. These may include selecting a subset on the basis of some criteria (ie, time range, airmass range, CCD number, object quality flag, stellar magnitude, or stellar color). We may want to plot the zero point, or the difference from a nominal zeropoint, as a function of any of these various parameters. We provide mechanisms to perform each of these functions.

The selection process can be viewed in two ways. In the first version, we apply various cuts of interest and plot the data in one shot. The other method is to extract the photometry and all interesting parameters to a set of vectors which can then be manipulated by the user, or by a program.

Photometric Zeropoint database

End users of Elixir need a method to access the specific photometric zero points which are valid for a time or image of interest. The Zeropoint database stores all determined zeropoints along with necessary anciliary information, and provides the mechanism to associate a specific image and star measurement with the appropriate calibration values. In a method similar to the Detrend database, tools are provided to add to the database and to search for specific data in the database.

The zeropoint database follows the format of the other databases used in Elixir: a file written with binary data defined by a C header structure, along with a FITS-like header giving summary information. File locking is used to maintain the database integrity.

 /* Zeropoint Database structure */
 typedef struct {
   short int photcode;
   unsigned long int tstart;
   unsigned long int tstop;
   unsigned long int treg;
   float zp-meas, zp-pred;
   float sigma1, sigma2;
   float Ko, Ao, Kp, Ap;
   short int c1, c2;
   float mean-color;
   short int quality
 } PhotReg;    /* 272 bytes */

Above we show the C structure used to define the zeropoint database data. Zeropoints are defined for specific primary / secondary photcodes. The calibration provides the information necessary to apply a transformation of the form:

M = mdb + cΛ + KΛ(sec z - 1) + AΛ(M1 - M2) - ΔM

where M is the calibrated magnitude, mdb is the magnitude as stored in the database, cΛ is the zeropoint, KΛ is the airmass coefficient for this filter, sec z is the airmass, AΛ is the color coefficient for this filter, M1 - M2 is the color of this star from two specific filters, and ΔM is the a reference offset that allows some system flexibility on the stored database values.

The zeropoint values are not generally determined in isolation for a single moment. Rather, a collection of data spanning a time period are used to fit for the various coeffecients. Typically, the standard star data from a night are used to determine, in addition to the values of KΛ and AΛ, the stability of the sky transparency. As discussed above, during the Queue processing, we will obtain standard star data in all filters 3 times per night and in a single filter an additional 5 - 7 times per night. The stability of the zero point solution for the reference filter can be better used to assess the transparency of the night sky. We have therefore maintained seveal pieces of information relevant to these issues. First, we store the scatter of the main filter. Second, we store the scatter of the reference filter. Finally, we also store a quality related to the measured transparency of the sky. If the sky is not photometric, one would expect not only a significant scatter for the zeropoint, but also that the zeropoint be smaller than expected. The deviation of the zeropoint from the nominal value is used to determine the 'quality'. TBD.

It is necessary for a given set of zeropoint measurements to be associated with a range of times for which that zeropoint applies. Typically, this range would represent the night or nights used to determine the zeropoint. We provide the values tstart and tstop to represent the time endpoints. Normally, with Elixir, the zeropoint calculation is made for a specific night, so these values are filled with the sunset and sunrise times for that night, which ensures any images taken that night are covered by the range.

Finally, the calibration includes a color transformation term. It is necessary to define which is the color used for the transformation, and what is a typical color in case the star in question does not have a measured color.