8 INPUT DATA FILES

 8.1 Reduction Database
 8.2 Data Frame Format
 8.3 Saving Disk Space
 8.4 Arc Line Lists
 8.5 Structure Definition File
 8.6 Multiple Arc Frames
 8.7 Multiple Object Frames
 8.8 Flux Calibration Frames

echomop performs most of its data access via Starlink subroutine libraries. All data frames should be held in files recognised by the CONVERT utility. For efficiency, store your data in the Starlink Extensible N-dimesional Data Format (NDF).

A few ASCII format files are also read by echomop:

Structure definition files (.DEF files, stored in the ECHOMOP_EXEC directory) are used by the data-access subroutines to create and access files used to store data.

A list of identified features and their properties (ECHARC.LIS) is used for data exchange with the ECHARC program.

8.1 Reduction Database

The reduction database is a file used to contain various information produced during the reduction; it is a database of reference information for the reduction. For example, it might contain a list of order locations, trace polynomials, sky and object channels, wavelength polynomial fits, etc.

The file is created at the start of a reduction cycle with a supplied name, and grows in size until the reduction is complete. At this point it contains a fairly complete record of any intermediate results, thus allowing particular sections of a reduction to be tuned and repeated as required.

In general you will not need to access the file except via echomop. The package provides facilities for examining and plotting data from the file. You can, however, use hdstrace to look at any data in the file you wish. See SUN/102 for more details.

8.2 Data Frame Format

The input data frames should be in NDF format. The data themselves being stored in a 2-dimensional array accessible via the path:

  /filename/.DATA.DATA_ARRARY

This capability may be useful when a foreign data format has been translated into an NDF, but with non-standard object names.

8.3 Saving Disk Space

echomop makes substantial use of disk space throughout the reduction process. This is due to the intermediate storage of information in the reduction file. Much of the information stored is on a per-order basis. For data frames with many orders the disk usage may be substantially reduced by the following:

  % cp $ECHOMOP_SOURCE/ECH_COMPACT.DEF ECH_RDCTN.DEF

and:

  % source $ECHOMOP_EXEC/echomop.csh      (UNIX)
  
  ICL> LOAD ECHOMOP_EXEC:ECHOMOP           (ICL)

This must be done before any reductions are embarked upon as it will not save any space retroactively on reductions already in progress.

When this procedure is used, it is essential to ensure that the per-order mode of operation is used and that the orders are fully processed one-by-one. i.e., flat-field, model sky, extract, and save result for order 1, then for order 2 etc. This is because the compact version only stores reduction arrays for the last processed order. Thus flat-field modelling order 3 and then order 4 would cause the calculated model for order 3 to be overwritten by the one for order 4.

8.4 Arc Line Lists

Line list files are stored in ASCII format for ease of maintenance. echomop reads the line list(s) and creates its database file. These feature databases are created by echomop itself using the ASCII format files as input. Line list filenames are expected to have a .ARC extension and will be looked for in the directory pointed to by ARCDIRS.

The format of arc line lists is as follows:

   !  comment lines starting with a !
      nnnn.mmm   iii.jj    text
      nnnn.mmm   iii.jj    text
      nnnn.mmm   iii.jj    text

where:

nnnn.mmm is a wavelength in Angstroms.

iii.jj is an (optional) intensity estimate.

text is a description of the line e.g., Th II.

In addition a special comment line is recognised which allows the units of wavelength to be specified:

  !WAVELENGTH_UNITS=some-unit-specifier

The default units are Angstroms. The units string is only used for output messages. Future echomop releases may support greater use of the wavelength units in constraining fits etc.

User-created lists should be placed in the directory pointed to by the ADAM_USER environment-variable.

8.5 Structure Definition File

Definition files are used by echomop to specify the structure of the files for storing objects produced during the reduction. Each of these objects is referred to internally by the package using a reference name. The definition file contains translations of these reference names into actual object paths within datafiles.

For instance; the object referred to as OBS_LINES internally is actually stored in the reduction file in the following object:

  MORE.ECHELLE.WAVE_FIT_1D.OBS_LINES[tune_mxrfln,num_orders]

All such objects can be renamed in the reduction file by altering their specified paths.

The default definitions are likely to be adequate for most circumstances. For more information see the on-line HELP sub-topic Redefining Object paths.

8.6 Multiple Arc Frames

Multiple arc lamp images are usually taken to monitor any possible shift in the object spectrum which may have occurred during the exposure.

When no shift is evident then the simplest course is to add the two arc frames together and provide the result to echomop whenever it prompts for the arc frame name.

When a shift can be discerned, you must make a judgement as to whether meaningful wavelength calibration can still be applied. If so then there are two courses of action available:

8.7 Multiple Object Frames

echomop is primarily orientated to processing single frame observations. Facilities exist in fiGARO for the combining of multiple spectra obtained from the reduction of a set of data frames. echomop does however provide assistance for the situations where the data frames are well registered, removing the necessity for steps such as Order location, tracing, and profiling, to be performed for each frame. The ‘clone’ option allows the selective copying of sections of a previous reduction file, which then becomes the baseline for the reduction of another similar frame. Optionally the definition file itself can be altered to permit the extension of the reduction database to include multiple spectra from a single observed source. For more information see the on-line HELP sub-topic Redefining Object paths.

8.8 Flux Calibration Frames

If flux calibration is to be done then the calibration frame must be presented in a Starlink datafile (NDF) in the same form as the data frame. echomop presently insists that all frames are of the same X- and Y-dimensions (a pretty safe bet). Take care to ensure the correct extraction of the calibration frame. For example, a different CCD bias voltage may have been used to that used for the object data frame etc.

Flux calibration is not usually performed on the very high resolution spectra produced by UCLES. This is due in the main to the lack of suitable reference tables. As more UCLES data becomes available this situation should improve.

A detailed description of the steps required is available in the fiGARO documentation.