Beam FITS header & description

[Joshuaalbert]

@JSKenyon Could you post the header of a valid beam model fits? Also, any other constraints it needs to satisfy would be great.

[bennahugo]

Think the most concise header formulation is here: https://github.com/ratt-ru/eidos/blob/a47fa43d78a814a8d0ea31535f0ebd6433812f3d/eidos/util.py#L78-L107

…

On Tue, Nov 14, 2023 at 1:02 PM Joshua George Albert < ***@***.***> wrote: @JSKenyon <https://github.com/JSKenyon> Could you post the header of a valid beam model fits? Also, any other constraints it needs to satisfy would be great. — Reply to this email directly, view it on GitHub <#314>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB4RE6U3SZ2VKRMU5TDPMUDYENFTHAVCNFSM6AAAAAA7KTMFR2VHI2DSMVQWIX3LMV43ASLTON2WKOZRHE4TENJRGM4TENI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- -- Benjamin Hugo

[JSKenyon]

You can also look at the very minimal beams used in the tests.

[Joshuaalbert]

Thanks for the replies. I notice some wierd things with the test beams:

1. The headers have a mismatch of NAXIS=3 and the actual number of axes defined (4).

2. The axis types for direction are X and Y which I'm not sure about. Shouldn't it be RA--SIN, and DEC--SIN? Are X and Y equal to arccos(l) and arccos(m)? Is this already assuming pointing direction is the phase-tracking centre, so the beam is always assumed to point along the bore-sight of the dish (for dish antennas)? I suppose this wouldn't work for phased arrays.

3. If the axes [RA--SIN, DEC--SIN, FREQ] are used will it work too? Or must I convert to [X, Y, FREQ]?

SIMPLE  =                    T / conforms to FITS standard                      
BITPIX  =                  -64 / array data type                                
NAXIS   =                    3 / number of array dimensions                     
NAXIS1  =                  513                                                  
NAXIS2  =                  513                                                  
NAXIS3  =                    3                                                  
EXTEND  =                    T                                                  
DATE    = '2015-05-20 12:40:12.507624'                                          
DATE-OBS= '2015-05-20 12:40:12.507624'                                          
ORIGIN  = 'GFOSTER '                                                            
TELESCOP= 'VLA     '                                                            
OBJECT  = 'beam    '                                                            
EQUINOX =               2000.0                                                  
CTYPE1  = 'X       '           / points right on the sky                        
CUNIT1  = 'DEG     '                                                            
CDELT1  =             0.011082 / degrees                                        
CRPIX1  =                  257 / reference pixel (one relative)                 
CRVAL1  =      0.0110828777007                                                  
CTYPE2  = 'Y       '           / points up on the sky                           
CUNIT2  = 'DEG     '                                                            
CDELT2  =             0.011082 / degrees                                        
CRPIX2  =                  257 / reference pixel (one relative)                 
CRVAL2  =   -2.14349358381E-07                                                  
CTYPE3  = 'FREQ    '                                                            
CDELT3  =     30303030.3030303 / frequency step in Hz                           
CRPIX3  =                    1 / reference frequency postion                    
CRVAL3  =         1280000000.0 / reference frequency                            
CTYPE4  = 'STOKES  '                                                            
CDELT4  =                    1                                                  
CRPIX4  =                    1                                                  
CRVAL4  =                   -5                                                  
GFREQ1  =         1280000000.0                                                  
GFREQ2  =         1306120000.0                                                  
GFREQ3  =         1333330000.0                                                  
END     

[JSKenyon]

I must confess that I am no expert when it comes to the beams - QuartiCal itself doesn't really use them bar feeding them into the underlying predict (which is implemented in codex-africanus). @sjperkins may have a better idea of precisely what will and won't work. The documentation for the beam term is here. Note that the application of the beam in the predict is known to be a bit of a bottleneck at present. Tagging @landmanbester in case he has better intuition about what we should be doing.

[sjperkins]

The axis types for direction are X and Y which I'm not sure about. Shouldn't it be RA--SIN, and DEC--SIN? Are X and Y equal to arccos(l) and arccos(m)? Is this already assuming pointing direction is the phase-tracking centre, so the beam is always assumed to point along the bore-sight of the dish (for dish antennas)? I suppose this wouldn't work for phased arrays.

IIRC it's in DEG because it's sampled in the ALT-AZ coordinate system, rather the LM coordinate system. @bennahugo @o-smirnov can you confirm? I recall we had a long discussion about this a year or two back.

NAXIS = 3, so I think the CTYPE4 etc. are ignored. Separate FITS cubes are maintained per correlation/stokes parameter.

The FREQ axis can be irregularly spaced and the GFREQn coordinates establish the centre point frequency of the associated voxel. This is a custom modification to the FITS schema introduced by @o-smirnov for use in MeqTrees I believe.

[Joshuaalbert]

Ah ok, makes sense then!