SDFITS

SDFITS is still a draft standard written up by Liszt (1995), with some more documentation by Garwood (2000?). Many observatories have implemented it, but this resulted in a few dialects, notably in the TDIM that is associated with the DATA:

  • LMT : simple 1-dimensional spectra (our plan)

  • GBT : simple 1-dimensional spectra

  • FAST : (4,64k): pol goes before chan, and NCHAN is a column in the table, so it can vary!

  • Parkes: (nchan,2,1,1): chan goes first, then next pol (not sure if 4 is also available)

  • Arecibo: CIMAFITS

  • 30m: CLASS exports a SDFITS lookalike, with the MATRIX extension name

Why use SDFITS for LMT?

The idea for LMT is to take a RAW (netCDF) data from the different Spectral Line instruments, and allow a first pass (TSYS and ON/OFF) calibration, such that the SDFITS file is a collection of calibrated spectra, where inspection, baseline subtraction, binning are all done, and optionally gridding. The conversion from RAW to SDFITS is done with current software (lmtslr, dreampy)

Using SDFITS also gives us a number of advantages:

  • Data format matches that of a spreadsheet, closely resembling the mental image users will have (and sadly might result in some new NIH data reduction code)

  • Use existing tools in specutils/astropy

  • Use GBTIDL and/or its python successor

  • A variety of 3rd party gridders (but: dialects)

CLASS

The path from CLASS to SDFITS (and vice versa?) is not well described, or at least I don’t know it, and should be described here.

CLASS/30m uses the MATRIX method, which a the SPECTRUM column is used instead of the DATA column. Why oh why is there an SDFITS standard.

HERA is multi-beam.

vector\fits outfile.fits from infile.lmv