WCS adds third axis even though WCSAXES = 2

I’ve been trying to use astropy.wcs.WCS to preserve WCS information when constructing some data cubes, and getting an extra axis when I try to use the WCS information. I only have a defined WCS for two of the three axes (because of #5462), and want to just use the same WCS for all the planes in the cube.

When I initialize the WCS, though:

>>> mywcs = wcs.WCS(hdulist['BLENDED'].header)
>>> mywcs

WCS Keywords

Number of WCS axes: 3
CTYPE : 'RA---TAN'  'DEC--TAN'  ''  
CRVAL : 150.11632130000001  2.2009730969999999  0.0  
CRPIX : -265.5  -234.5  0.0  
PC1_1 PC1_2 PC1_3  : -6.6666679999999995e-05  0.0  0.0  
PC2_1 PC2_2 PC2_3  : 0.0  6.6666679999999995e-05  0.0  
PC3_1 PC3_2 PC3_3  : 0.0  0.0  1.0  
CDELT : 1.0  1.0  1.0  
NAXIS    : 83 94

>>> mywcs.all_pix2world([[1, 2]], 0)
ValueError: When providing two arguments, the array must be of shape (N, 3)

The process, roughly speaking, is to cut the same WxH pixel region out of N 2D images taken in different filters and resampled to the same pixel grid, then construct a (N, H, W) (or (W, H, N) in FITS axis ordering) data cube where each plane is a different filter.

Here’s the header of one extension in the file I created:

XTENSION= 'IMAGE   '           / Image extension                                
BITPIX  =                  -64 / array data type                                
NAXIS   =                    3 / number of array dimensions                     
NAXIS1  =                   83                                                  
NAXIS2  =                   94                                                  
NAXIS3  =                    6                                                  
PCOUNT  =                    0 / number of parameters                           
GCOUNT  =                    1 / number of groups                               
WCSAXES =                    2 / Number of coordinate axes                      
CRPIX1  =               -265.5 / Pixel coordinate of reference point            
CRPIX2  =               -234.5 / Pixel coordinate of reference point            
PC1_1   =        -6.666668E-05 / Coordinate transformation matrix element       
PC2_2   =         6.666668E-05 / Coordinate transformation matrix element       
CDELT1  =                  1.0 / [deg] Coordinate increment at reference point  
CDELT2  =                  1.0 / [deg] Coordinate increment at reference point  
CUNIT1  = 'deg'                / Units of coordinate increment and value        
CUNIT2  = 'deg'                / Units of coordinate increment and value        
CTYPE1  = 'RA---TAN'           / Right ascension, gnomonic projection           
CTYPE2  = 'DEC--TAN'           / Declination, gnomonic projection               
CRVAL1  =          150.1163213 / [deg] Coordinate value at reference point      
CRVAL2  =          2.200973097 / [deg] Coordinate value at reference point      
LONPOLE =                180.0 / [deg] Native longitude of celestial pole       
LATPOLE =          2.200973097 / [deg] Native latitude of celestial pole        
RADESYS = 'FK5'                / Equatorial coordinate system                   
EQUINOX =               2000.0 / [yr] Equinox of equatorial coordinates         
MJD-OBS =              55924.0 / [d] MJD of observation matching DATE-OBS       
DATE-OBS= '2011-12-29'         / ISO-8601 observation date matching MJD-OBS     
EXTNAME = 'BLENDED '                                                            
FWHMAS  =                  0.7 / Kernel FWHM in arcseconds   

Here’s the wcslint output:

$ wcslint COSMOS_150.06694880+2.2103795_at_0.70_arcsec.fits
HDU 1 (BLENDED):
  WCS key ' ':
    No issues.

HDU 2 (TRUTH):
  WCS key ' ':
    No issues.

HDU 3 (BANDS):
  WCS key ' ':
    No issues.

HDU 4 (CATALOG):
  WCS key ' ':
    No issues.

HDU 5 (SEGMAP):
  WCS key ' ':
    No issues.

[josePhoenix]

Thanks! I also found a workaround by supplying naxis=2 when initializing the WCS from the header (i.e. wcs.WCS(hdulist['BLENDED'].header, naxis=2)).

[josePhoenix]

Sure. Works for me.