v2.2.0

New Features

• Add velocity region support and ability to preserve WCS for single regions in spectral extraction. [#1263, #1276]

Bug Fixes

• Add appropriate axes_names to resulting GWCS when a Spectrum is created with a spectral axis
  specified and no WCS, based on the spectral axis units. [#1264]

• Fix spectra resulting from new JWST WFSS format having MaskedQuantity spectral_axis. [#1266]

Other Changes and Additions

• Dropped support for python 3.10 and bumped the minimum required version of Astropy
  to 7.0. [#1267]

v2.1.0

New Features

• Add new flux_col keyword argument to JWST x1d reader to override flux column
  auto-detected from source type column (point vs surface brightness). [#1244]

• The readers can now read IRAF MULTISPEC files with linear and log-linear wavelength
  solutions; at the same time the IRAF equispec format gains the ability to read log-linear
  wavelength solutions. [#1254]

Bug Fixes

• Fixed/updated SDSS-V mwm data loader to account for a subset of targets with bad datasums. [#1253]

Other Changes and Additions

• Updated the JWST data readers to accomodate new WFSS file formats
  with SpectrumList.read(). [#1252]

v1.20.3

Minor doc fix to make 1.20.x docs available long-term.

v2.0.0

See https://specutils.readthedocs.io/en/latest/#changes-in-version-2 for more details about major changes between version 1.x and 2.x.

New Features

• Spectral axis can now be any axis, rather than being forced to be last. See docs
  for more details. [#1033, #1226]

• specutils.Spectrum now properly handles GWCS input for wcs attribute. [#1074]

• JWST reader no longer transposes the input data cube for 3D data and retains
  full GWCS information (including spatial). [#1074]

• Spectrum1D renamed to Spectrum. [#1126]

Other Changes and Additions

• Initializing a Spectrum with only a spectral axis (not full WCS) will now
  result in a GWCS matching the dimensionality of the flux array, rather than a
  1D spectral GWCS in all cases. [#1211, #1222]

• Spectrum arithmetic now checks whether the spectral axes of the two operand Spectrum
  objects are equal, and fails if they are not. [#1211]

• Added documentation about transitioning from 1.x to 2.x. [#1226]

v1.20.2

Bug Fixes

• Fixed loaders for HST/STIS. Added an "HST/STIS multi" reader for use with
  SpectrumCollection to read multi-extension and/or multi-order echelle datasets. [#1233]

Other Changes and Additions

• Warn about use of spectral_axis_index (new Spectrum1D kwarg for 2.0) in 1.x. [#1238]

v1.20.1

Bug Fixes

• Switch to using an alias rather than a subclass for the temporary 1.x access
  to the new 2.x class name (Spectrum). [#1236]

v1.20.0

1.20.0 (2025-05-01)

Bug Fixes

• Fixed loaders for IRAF-encoded non-linear Chebychev + Legendre wavelength
  solutions that were imposing integer-valued domain limits. [#1199]
• Fixed loaders for SDSS APOGEE not correctly identifying the format [#1217]
• Fixed template comparison returning empty list of chi2 values in some cases. [#1228]

Other Changes and Additions

• Added Spectrum class wrapper around Spectrum1D to ease transition to new
  class name in 2.0. [#1229, #1230]

• Removed redshift and radial velocity setters in Spectrum1D that have been deprecated
  since 1.8. [#1229]

• Added documentation about upcoming changes in specutils 2.0. [#1231]

v1.19.0

1.19.0 (2024-11-05)

Bug Fixes

• Fixed mwmVisit SDSS-V Spectrum1D and SpectrumList default loader
  being unable to load files containing only BOSS instrument spectra. [#1185]

• Fixed automatic format detection for SDSS-V SpectrumList default loaders. [#1185]

• Fixed SplineInterpolatedResampler when input and output spectral axes are different
  physical types, e.g. wavelength and velocity. [#1190]

Other Changes and Additions

• "Multi" loaders have been removed from SDSS-V SpectrumList default loaders. [#1185]

v1.18.0

New Features

• New Spectrum1D.with_spectral_axis_and_flux_units method to convert both
  spectral axis and flux units at the same time. [#1184]

Bug Fixes

• Fixed Spectrum1D.with_flux_unit() not converting uncertainty along
  with flux unit. [#1181]

• Fixed extracting a spectral region when one of spectrum/region is in wavelength
  and the other is in frequency units. [#1187]

Other Changes and Additions

• Replaced LevMarLSQFitter with TRFLSQFitter as the former is no longer
  recommended by astropy. [#1180]

v1.17.0

v1.17.0

New Features

• Fitting and excision functions can now use inclusive upper bounds for input regions if specified. [#1171]

Bug Fixes

• Fixed specifying a single value for window in analysis.fit_lines. [#1164]

Other Changes and Additions

• Dropped support for python 3.9. [#1176]

• utils.wcs_utils.refraction_index (and thus air_to_vac and vac_to_air)
  now defaults to Morton2000 as the method instead of Griesen2006. [#1169]