1) Source detection tables, containing the lists of sources to be merged. There should be one table per wavelength, and they must be in IPAC table format (see Table 2.1).
2) Parameter (namelist) files, used to control the input parameters to Bandmerge. There are two types of namelist file – a primary namelist, which controls the overall input, and a number of secondary namelists, which are used for each successive iteration of the source matching algorithm (see §2.2).
3) Bandpair uncertainty file giving the first crude guess of the offsets of the source coordinates between the different bands (see §2.3 and Table 2.2).
4) Either a Fiducial Image Frame table, defining the common grid onto which all of the source tables can be projected, or a reference FITS image containing all relevant WCS header keywords, that can be used to create this table. The FIF.tbl file is standard output from MOPEX (see §2.4)
2.1 Source Detection Tables
The source detection tables are the ASCII tables containing the lists of sources to be merged. There should be one source table per wavelength band. These tables must be in IPAC table format, which is the default output format of APEX, the Spitzer Science Center’s Astronomical Point-source Extraction software (distributed as part of MOPEX). The table must contain the columns and header keywords as shown in Table 2.1.
The Bandmerge software will check for (and if absent will add) two additional header keywords during processing – INSTRUME and CHNLNUM, which describe the instrument used to take the data, and the instrument channel number respectively (e.g. INTRUMEN = MIPS, CHNLNUM = 2 identifies the data as being from the MIPS 70 micron detector). If they are missing from your table headers, the values for these keywords must be included in the namelist file (see §4 for more information on how to do this, and the allowed values for these parameters). The data from the input bands will internally be assigned the following indices (and they will be included in the bandmerged output file in this order):
1 = IRAC-3.6
2 = IRAC-4.5
3 = IRAC-5.8
4 = IRAC-8.0
5 = MIPS-24
6 = MIPS-70
7 = MIPS-160
The indices are used throughout the Bandmerge module to sort, process and identify the input bands. For example, if a processing run has three input bands, given in the order MIPS-70, IRAC-3.6 and IRAC-8.0, these will be assigned index values 6, 1 and 4, respectively, and will be re-ordered as 1, 4 and 6 in the output.
Bandmerge will also check and update a number of other header keywords. Specifically, it will count the actual number of sources listed in the file, and update TOTAL_PS_NUMBER to match. Similarly, keywords CDELT and NAXIS are compared to the values in either FIF.tbl or the reference FITS image, and are updated to be consistent.
The source detection tables can be kept in any directory, as long as they are specified in the namelist with their full or relative paths.
The input parameters for Bandmerge are specified in the primary namelist file (e.g. bandmerge.nl). It must be located in a subdirectory cdf/ of the directory in which Bandmerge is going to be run, i.e. if you plan to run Bandmerge from /Users/joe/data then the namelist must be in /Users/joe/data/cdf/. The namelist file is a simple ASCII text file, with two main sections:
1) Global Parameters section, which controls input/output files, header keywords etc.
2) Module Parameter Block, which controls the values of the parameters to be used in the bandmerging process.
A quick-look annotated example is given in Appendix A, and assumes that you are trying to merge data from IRAC channels 1 and 3, and MIPS-24. The input parameters are described in detail in Chapter 3.
2.2.2 Secondary namelists
Bandmerge runs as an iterative process. It starts with a crude estimate of the positional offset between different bands (given in the Bandpair uncertainty file – see §2.3), runs the bandmerging module, followed by the positional offset calculation, and then uses the calculated offsets to refine the source positions and re-run the bandmerging module. This whole process of bandmerge - positional refinement - bandmerge is defined as a single iteration. In the primary namelist, you can set the number of iterations you wish to run (must be 1 or more), and for each iteration you must provide a secondary namelist file to control the bandmerging parameters. For example, Bandmerge runs the first iteration using the parameters in the file bmg1.nl, and the second by using bmg2.nl, and so on. These secondary namelists are a stripped-down version of the primary namelist file, and only contain the information for the Module Parameter Block. An example of a secondary namelist is given in Appendix A, and can be found in the cdf/ directory of your Bandmerge installation.
2.3 Bandpair Uncertainty File
Bandmerge requires an initial crude estimate of the offsets of the source positions between bands to start the iterative matching process. This file is called bpru.tbl, and a default version is provided in the cdf/ subdirectory of your Bandmerge installation. Most users will find this default file is fine for their needs. Columns 1 and 2 in the file specify the pair of bands that are being compared. Columns 3, 4 and 5 are the positional shifts between the pair of bands in X, Y and a cross term shift (XY). This file can include initial values for all 7 bands, even though you may only merge a subset of them. The values in columns 3, 4 and 5 will be updated by Bandmerge during the iterative source-matching process.
In order to merge data taken with different detectors, the source positions in each band must be converted to a common reference frame. The source lists from Spitzer give the source positions in RA and Dec, which must be converted to Cartesian (X,Y) coordinates. This is done using a Fiducial Image Frame (FIF) – a table defining the independent coordinate system and the spatial boundaries onto which the sources from each waveband can be projected. The FIF is a standard output from MOPEX (FIF.tbl, found in the output directory). If the user does not have an FIF table then Bandmerge must create one from one of the data frames that was used to extract the source list. In the example below we set the IRAC Channel 1 mosaic image to define the FIF. The choice of reference frame must be made carefully, and should usually be the waveband with the largest spatial coverage. Bandmerge will ignore any sources that fall outside the spatial coverage of the reference frame, regardless of the waveband.
The user must set one of the following in the primary namelist:
