Correcting DATE-OBS and adding MJDREF/MJDREFI keywords in spectra

[Parrazyte]

The pha spectra obtained when running batspectrum_analysis (and probably the event files before) lack some temporal information:

-the file header only has a TSTART and a TSTOP

-the DATE-OBS is a constant set to '2022-01-26T12:16:00'

example:

f=fits.open('4U_1630-47_survey_point_20212731133.pha')
f[1].header
Out[4]: 
XTENSION= 'BINTABLE'           / binary table extension                         
BITPIX  =                    8 / array data type                                
NAXIS   =                    2 / number of array dimensions                     
NAXIS1  =                   20 / length of dimension 1                          
NAXIS2  =                    8 / length of dimension 2                          
PCOUNT  =                    0 / number of group parameters                     
GCOUNT  =                    1 / number of groups                               
TFIELDS =                    4 / number of table fields                         
TTYPE1  = 'CHANNEL '                                                            
TFORM1  = 'J       '                                                            
TTYPE2  = 'RATE    '                                                            
TFORM2  = 'E       '                                                            
TUNIT2  = 'count/s '                                                            
TTYPE3  = 'STAT_ERR'                                                            
TFORM3  = 'E       '                                                            
TUNIT3  = 'count/s '                                                            
TTYPE4  = 'SYS_ERR '                                                            
TFORM4  = 'D       '                                                            
EXTNAME = 'SPECTRUM'           / extension name                                 
TELESCOP= 'SWIFT   '           / Telescope (mission) name                       
INSTRUME= 'BAT     '           / Instrument name                                
HDUCLASS= 'OGIP    '           / Conforms to OGIP/GSFC standards                
HDUCLAS1= 'SPECTRUM'           / Contains spectrum                              
GAINAPP = 'T       '           / Gain correction has been applied               
GAINMETH= 'FIXEDDAC'           / Cubic ground gain/offset correction using DAC-b
TELAPSE =                443.0 / [s] Total elapsed time from start to stop      
ONTIME  =                443.0 / [s] Accumulated on-time                        
LIVETIME=                443.0 / [s] ONTIME multiplied by DEADC                 
EXPOSURE=                443.0 / [s] Accumulated exposure                       
BAT_RA  =    248.5019983058676 / [deg] Right ascension of source                
BAT_DEC =   -47.39400153565722 / [deg] Declination of source                    
TSTART  =          654694420.0 / Start time                                     
TSTOP   =          654694863.0 / Stop time                                      
AREASCAL=                  1.0 / Nominal effective area                         
BACKSCAL=                  1.0 / Background scale factor                        
CORRSCAL=                  0.0 / Correction scale factor                        
BACKFILE= 'none    '           / Background FITS file                           
CORRFILE= 'none    '           / Correction FITS file                           
RESPFILE= '4U_1630-47_survey_point_20212731133.rsp' / Redistribution Matrix file
ANCRFILE= 'none    '           / Effective Area file (ARF)                      
QUALITY =                    0 / Data quality flag                              
GROUPING=                    0 / Spectra are not grouped                        
POISSERR= 'F       '           / Poisson errors do not apply                    
SYS_ERR =                  0.0 / Systematic error value                         
DETCHANS=                    8 / Total number of detector channels available    
CHANTYPE= 'PI      '           / Pulse height channel type                      
HDUCLAS2= 'NET     '           / Spectrum is background subtracted              
HDUCLAS3= 'RATE    '           / Spectrum is count/s                            
PHAVERSN= '1992a   '           / Version of spectrum format                     
HDUVERS = '1.2.0   '           / Version of spectrum header                     
FLUXMETH= 'WEIGHTED'           / Flux extraction method                         
BAT_XOBJ=  -0.2964718762975234 / [cm] Position of source in BAT_X               
BAT_YOBJ=   0.2575742108354418 / [cm] Position of source in BAT_Y               
BAT_ZOBJ=   0.9196520822991298 / [cm] Position of source in BAT_Z               
COORTYPE= 'sky     '           / Type of coordinates specified for weighting    
FFAPP   =                    1 / Projection correction applied?                 
NFAPP   = 'F       '           / Near-field correction applied? ~(COS+RSQ)      
PCODEAPP=                    1 / Partial coding correction applied?             
PCODEFR =            0.8671875 / Partial coding fraction of target              
NGPIXAPP=                    1 / Normalized by number of detectors?             
NGOODPIX=                13813 / Number of enabled detectors                    
DATE-OBS= '2022-01-26T12:16:00' / fake date-obs on UTC                          
MSKWTSQF=    0.997248493515814 / Half-variance of mask weight map               

The date-obs is signaled as fake, but it would be helpful to add MJDREF/MJDREFI keywords (from i.e. https://swift.gsfc.nasa.gov/analysis/suppl_uguide/time_guide.html) as well as leapseconds if they are necessary, and correct the DATE-OBS according to TSTART.

[parsotat]

This is a good suggestion, as I know that astropy highlights this as a warning. With the way that the current implementation of the survey data analysis is, this change is non-trivial.

In the future, when we improve the survey analysis capability we will be sure to do this so I will keep this open for now.

[Parrazyte]

The fact that you tell me this is non-trivial makes me wonder if I haven't made a mistake.

In order to retrieve the Time of the observations, what I do is manually add a fixed MJDREF+MJDREFI according to the page I gave you, then "simply" add the Tstart in seconds.

Is this good enough ?

Of course, I'm missing the leapseconds, but for my purposes (daily integrations or at least hour to hour comparisons) few seconds don't change much.

[parsotat]

Hi @Parrazyte,

Apologies for my confusion, I thought that this was a mosaic spectrum. The change is relatively straightforward to do and I think that you have already done it? Your value for DATE-OBS is different from what I have coded up in the bat_survey.py file. In any case, you can also get the real UTC date for a pointing observation from the BatSurvey object. There is some additional information here: https://github.com/parsotat/BatAnalysis/blob/main/notebooks/BatAnalysis_objects_intro.ipynb