Background subtraction using several local off-source observations

The procedure is identical to that in section  3.1, except for the following points:

1. Several suitable off-source observations should be selected from the GINGABGD database table. The total duration of the off-source observations must not be substantially less than 24 hours and they must include the same range of SUD values as the source observation. All of the off-source observations should be within a few days of the source observations, the closer the better. If sufficient off-source data is not available within a few days of the source observation, then off-source observations taken 37 days before and/or after the source observation can also be used (Hayashida et al. 1989).

2. The off-source data cubes should all have the same number of energy channels and detector idents. Time counters describing the radioactive decays should be inserted into the source and all the off-source data cubes using the program bgdsaa (see section 3.1).

3. The program bgdmfit should then be used to simultaneously fit all the off-source observations; bgdmfit can fit up to 50 files. When the program fits M off-source observations with N specified parameters, N+M-1 coefficient files are generated; M of these files describe the constant background component associated with each of the M off-source observations, and the remaining N-1 describe the time dependent (but pointing position independent) background components.

4. The program bgdsim uses the coefficient files produced by bgdmfit to generate a model background data cube of the same shape as the source observation data cube. Program bgdsim only requires one file describing the constant background component, while bgdmfit generates one file for each off-source observation. Either a mean constant file should be created (using the program meanconst or meanpha; an example is given below), or the constant file associated with the off-source observation closest in time to the source observation should be used. Program subcb can then be used to subtract the model background cube from an observation cube to give the background subtracted cube.

   The program meanconst performs a straight or weighted mean of all spectral files satisfying an input file specification. The weighted means can either be weighted using the variance associated with each datum or by the exposure associated with each file. In the latter case the variance in the output file is set to zero. Program meanconst differs from meanpha as it allows the possibility of excluding data taken at low Galactic latitudes.

   qcl>meanconst
   
   ** MEANCONST Version 1.3 Fri Aug  8 20:07:39 1997 **
    
   FILESPEC Input file name = *.* =bgd_constant*.phad
   DATOUT Name of  output file = bgd_constant001.psum =bgd_constant.phad
   BGAL_MIN Minimum galactic latitude = 0.0 =25.0
    Number of files actually used:                 2
    Number of files before galactic latitude cut:  2
   Project from which data originates is not specified.
   Will assume files have GINGA/ASTRO-C file formats
    
         1  bgd_constant001.phad
        Shape :     1 x    48   x     2   =    96
                 TIME    PHA     IDENT   
    
         2  bgd_constant002.phad
        Shape :     1 x    48   x     2   =    96
                 TIME    PHA     IDENT   
    
   AV_OPTION straight, weighted, or exposure = STR =
    MEANCONST STRAIGHT OPTION  for Ginga
    
   0 suspect/invalid data points found.
    
   ID_MASK( 2)
    1 0101010101010101
    2 1010101010101010
        Output file:  bgd_constant.phad
   qcl>

5. The quality of the background fit can be checked in a similar way to that described in section 3.1, although the labour is greater as several background cubes need to be incestuously subtracted.