== IRC imaging toolkit version 20110225 ==

++ Detail of the updates ++

(1) The main reduction script 'pipeline' has been divided into two
   scripts: 'pipeline1' & 'pipeline2'.

   In order to perform the correction for the earth-shine light,
   the reduction script 'pipeline' has been divided into 'pipeline1'
   & 'pipeline2'. How to correct the earth-shine light effect is
   described below in (6).


(2) The reduction order has been changed.
   [pipeline1 & pipeline2]
  
   Considering the principle of linearity correction functions, the
   dark subtraction should be done after the linearity correction.
   In this new version, 'wraparound', 'ircnorm', and 'linearity'
   corrections are applied to dark frames (as well as object frames),
   and the corrected dark frames are subtracted from the similarly
   corrected object frames. Generally speaking, a larger flux value
   is derived with the new version compared to the previous version,
   with the difference increasing for brighter objects. Nonetheless,
   the difference is estimated to be a few percent at most.

   Applying masks to bad/hot pixels, saturated pixels, and the slit
   area, which used to be done in separate tasks, is now performed
   in one task 'maskall' in pipeline2.


(3) New options for the dark subtraction are available.
   [pipeline1]

   In addition to super-dark and self-dark, two new options are
   available: user-dark & self-made dark. Choice of a dark type is
   controlled by the option 'darktype' in 'pipeline1'.

   A user can specify ones own dark frames by setting darktype='us'.
   In this case, dark frames should be specified in a text file
   (e.g. 'userdark.dat') and be stored in a working directory (not
   in 'rawdata/'). The option 'darkfile' needs to be the name of the
   file containing the list of dark frames. Syntax of the darkfile
   should be the same as 'lib/constants.database', and an example
   is shown in 'lib/dark/userdark.dat'.

   The other new option self-made dark is only applicable to NIR
   long exposure frames. The variation of dark currents is more
   accurately taken into account by modelling the dark current of
   a pixel to be a linear function of a typical count in the masked
   area (Tsumura et al., in prep.), and subtracting the modelled dark
   frames improves especially the data taken soon after the passage
   of the SAA region, when dark currents tend to be significantly
   higher. To activate this option, set smdark=yes. Note that this
   procedure takes a longer time than the normal dark subtraction,
   as it creates one dark frame for each object frame. If the model
   dark frames created by this process need to be kept, set delsmd=no.

   A type of the dark frames applied to an image is recorded in
   the FITS header 'DRKTYPE'.


(4) The latest flat patterns for L15 & L24 have been included.
   [pipeline1]

   The flat patterns for L15 and L24 have been significantly improved
   by removing the scattered light component due to reflections
   inside the optics (Arimatsu et al., in prep.). However, just
   dividing an image by the new flat is sometimes not enough since
   the scattered light pattern still exists in the background. Two
   new options regarding the background subtraction have thus been
   introduced.

   In order to subtract the background sky including the scattered
   light, set subsky=yes and subLscat=yes. Assuming the old flat
   frames represent the sky pattern with the scattered light,
   the script rescales the old flat to fit each object frame and
   subtracts the rescaled flat. An average of the subtracted sky flux
   is stored in the FITS header 'AVGSKY'. This value is deduced to
   overestimate an exact flux of true background sky about 5%--10%
   but still a good indicator of sky flux. Note that this process
   may not be suitable for very extended objects such as the LMC as
   it is hard to distinguish between such objects and the sky. 

   For the bands other than L15 and L24, a constant value representing
   the sky flux calculated from the mode of object frames is subtracted
   when subsky=yes. For L15 and L24, the sky flux is subtracted during
   the scattered light correction as described above. In order to
   subtract the scattered light in L15 and L24 but to keep the sky
   values positive, set subLscat=yes and subsky=no. A constant value
   from 'AVGSKY' is added back after the scattered light correction.
   A FITS header 'ADDSKY'='yes' is added in this case.

   Considering the original accuracy of flux calibration, the
   conversion factors (from ADU to Jy) derived by Tanabe et al. (2008)
   can still be applicable though these values were calculated using
   an older reduction package.


(5) Handling of masked pixels has been modified.
    [pipeline2]

   For identifying the masked/blank area, a new parameter 'blankval'
   has been introduced. During image transformation processes such as
   the aspect ratio correction and the geometry shift and rotation
   before coadding, an interpolation between a masked (i.e. bad) pixel
   and a normal pixel had been problematic. By setting a constant
   value for masked pixels and checking how the interpolation affects
   adjacent pixels, the affected pixels are properly masked in a
   transformed image. A FITS header 'BLANK' is added to identify the
   constant value for masked pixels. A large negative number such as
   -9999.9 (default) is recommended.


(6) New tasks for removing the effect of the earth-shine light and
   for coadding frames, which uses the IRAF task 'xregister', have
   been developed.

   In some of the pointing observation data sets especially taken
   around the solstices, the background flux varies during a pointing
   due to the earth-shine light. In order to remove this effect,
   a template for the light pattern needs to be created before
   subtraction. To create a pattern, use the task 'mk_el'. Inputs to
   this task should meet several criteria. See the information written
   in the top of the script. To subtract the created pattern, use
   'sub_el'. Before these processes, 'pipeline1' need to be run.
   These new tasks are included in the greenbox.

   The coadd process in pipeline2 sometimes fails when few stars are
   available for the relative shift calculation. In such cases, the
   new task named 'coadd2' stored in the bluebox would be useful.
   By using the IRAF task 'xregister', this task derives the relative
   shifts where the cross correlation of two object frames becomes
   the maximum. Note that it cannot take a rotation of frames into
   account.

   In addition, several scripts regarding the regular coadd process
   have been updated. The FITS header 'COADDTYP' indicates which of
   these coadd methods is applied (undefined means the normal coadd).


== EOF ==