2dfdr "Rockhopper" (6.00) update released

Dear Users of 2dfdr,

The AAO is proud to announce a new public release of our flagship fibre-data-reduction software, 2dfdr. Considerable effort has gone into improving the scientific quality of the data reduction, as well as improvements to the package's usability.

The new version, code named "Rockhopper", is available from the AAO's website:


Changes and improvements to the software

  • Tramline tracking accuracy is improved to ~0.1pixel across all currently supported instruments
  • Optimal extraction routines have been extensively revisited:
    • Routines are more stable (resulting in fewer pixels incorrectly marked bad)
    • Variance estimation is much more accurate
  • 2dfdr now supports HERMES, KOALA and SAMI data
  • Tramline mapping in extremely sparse fields has been improved
  • A b-spline scattered light option has been added to optimal extraction
  • Reduction messages are now shown in the user interface, and can be saved to file.
  • All the settings of the current session can be saved to ".idx" configuration files for future use.
  • Plotting widget is more user-friendly, although most old keyboard shortcuts still work
  • Plotting multiple items simultaneously is now possible
  • Additional diagnostic plots and output have been added.
  • Reduction directory structure has been cleaned up, with some files now appearing in corresponding sub-directories.
  • Wavelength model prediction for AAOmega rewritten.
  • Our bug zapper has been used extensively, most notably to remove a crippling memory leak.
  • Source code has been made available.
  • Removal of dependence on external libraries

What does this mean for my data?

The tramline mapping and optimal extraction improvements above affect data quality at a fundamental level. Therefore, we have seen many improvements in testing in "higher level" data reduction quantifiers.

  • Better splicing for AAOmega data.
  • Better sky subtraction.
  • More stable wavelength solutions, especially for high-dispersion SPIRAL data
  • Fewer negative spectra
  • Less "ringing" around bright sky emission lines
  • Scientifically useable variance estimates from optimal extraction
  • Fewer bad/invalid pixels in the output data

Future Improvements

The working group will continue to improve 2dfdr. The following improvements are anticipated for future releases.

  • Better stability for the wavelength solution in HERMES data.
  • Improved wavelength solution accuracy for HERMES data, greater than 0.1 pixel.
  • Implementation of a variable width PSF for the fibre traces across the detector.

We recommend that anyone looking to improve the quality of their existing data consider re-reducing archival data with this new release. 

The improvements to the software are in thanks to the hard work of the 2dfdr Working Group, which combines astronomers from the community with the research and technical staff at the AAO. Particular thanks to Mike Birchall, who is the primary 2dfdr developer. Thanks also to many GALAH, SAMI, and OzDES team members who have beta tested the many incarnations on their data. 

Anyone with an interest in improving their science by contributing to the future development of 2dfdr is encouraged to contact myself.

Bug reports are always appreciated, and should be directed to your support astronomer, or to myself at andrew.green@aao.gov.au


Andy Green
2dfdr Working Group