Pixel Units in the H-alpha survey

• Transmission (Black stars on grey background)
• Intensity (White stars on black background)
• Density (White stars on black background)

Pixel values in transmission units are related to incident intensity via the established transmission (T), Density (D) and Intensity (I) relationships from the straight line portion of the log-linear photographic characteristic curve. i.e.

where gamma is the gradient of the response on the linear region of the D-Log10(I) plane. A typical tech-pan characteristic curve obtained through the standard broad-band OG590 filter is shown here as measured photographic density vs log10(I).

In the SHS all pixel units are floating point real*4. This compares with the SSS where pixel units are Integer*2 values in density space but scaled (on a field by field basis) to lie in the range 0-32767. Real*4 values are used for the SHS as they offer greater accuracy and flexibility. For example they are amenable to transformation to real flux units (e.g. to Rayleighs) via suitable calibration sources. Also application of the H-alpha flat-field will not result in data going out of range as it possibly could do for 2-byte (signed) integers.

Below we give brief explanations for each unit and partial extracts from the fits headers of extracted fits images of the same 15x15arcmin region centred on a Planetary Nebula at 08 03 12.4 -33 31 01 (J2000) from the SHS data which shows the key sections concerning the pixel units for each choice. At the bottom the equivalent section from the fitsheader of the same region extracted from the R-band of the SSS is also shown for comparison.

A) Pixel Units: TRANSMISSION
Basically a measure of transmitted light through the photographic emulsion as determined by SuperCOSMOS. Hence dark areas of the exposure (developed images) have low transmission values and lighter areas (sky background) have large values so the fits images appear as negatives with dark stars on a grey background, just like the photographic exposures themselves. It is the log of the reciprocal of the measured transmission values that is directly proportional to the photographic density.

DATATYPE= 'REAL*4  '           / Type of data                                 
DATUNITS= 'TRANSMISSION'       / Units: transmission, density or intensity    
XPIXELSZ=   9.997114974000E+00 / [microns] X pixel size                       
YPIXELSZ=   1.000000000000E+01 / [microns] Y pixel size                       
OBJCTRA = '  8  3 12.398'      / Centre Right Ascension (J2000)               
OBJCTDEC= '-33 31  0.93'       / Centre Declination (J2000)                   
OBJCTX  =   6.831369255326E+03 / [pixels] Centre X on plate                   
OBJCTY  =   6.784975815772E+03 / [pixels] Centre Y on plate                   
DATAMINI=         3.500000E+01 / [DATUNITS] Minimum pixel value               
DATAMAXI=         5.293000E+03 / [DATUNITS] Maximum pixel value               
DATAMEDI=         3.883000E+03 / [DATUNITS] Median pixel value                
DATAMODE=         3.964000E+03 / [DATUNITS] Modal pixel value                 
FLATFLD =                    F / Are the pixel data flat-fielded? 

B) Pixel Units: INTENSITY
Intended to represent the brightness of images where high values equate to high intensity i.e. bright images. Low values represent faint images and the background. Thus the fits images are positives - white stars on a grey-black background. In UKST photography the relationship between intensity (incident flux times exposure time) and measured photographic density is derived from determination of the characteristic curve. This normally comes from measures of step wedges impressed on the exposure at the same time as the photograph on the sky is obtained. Unfortunately, for the H-alpha exposures the step wedge is located in a corner of the exposure well off the clear aperture of the inteference filter. However a generic linearised fit has been found to give reasonable transformation under normal exposure conditions for converting measured SuperCOSMOS transmission values to nominal Intensity values for both plates and films. A constant gamma (slope of straight line portion of the characteristic curve) is used and it is assumed that the sky background level is always on this straight line region.

DATATYPE= 'REAL*4  '           / Type of data                                  
DATUNITS= 'INTENSITY'          / Units: transmission, density or intensity      
XPIXELSZ=   9.997114974000E+00 / [microns] X pixel size                         
YPIXELSZ=   1.000000000000E+01 / [microns] Y pixel size                         
OBJCTRA = '  8  3 12.400'      / Centre Right Ascension (J2000)                 
OBJCTDEC= '-33 31  1.00'       / Centre Declination (J2000)                     
OBJCTX  =   6.831332729995E+03 / [pixels] Centre X on plate                     
OBJCTY  =   6.784876714616E+03 / [pixels] Centre Y on plate                     
DATAMINI=         2.703063E+03 / [DATUNITS] Minimum pixel value                 
DATAMAXI=         1.569084E+04 / [DATUNITS] Maximum pixel value                 
DATAMEDI=         2.967537E+03 / [DATUNITS] Median pixel value                  
DATAMODE=         2.949120E+03 / [DATUNITS] Modal pixel value                   
FLATFLD =                    F / Are the pixel data flat-fielded? 

C) Pixel Units: DENSITY
Photographic density (usually in the range 1-4) is related to Intensity via the characteristic curve. For SuperCOSMOS scans a generic `linearised' form is adopted when converting between Density (transmission) and Intensity where a fixed `reasonable' value for the slope of the straight line portion of the curve (gamma) is used.

DATATYPE= 'REAL*4  '           / Type of data                                   
DATUNITS= 'DENSITY '           / Units: transmission, density or intensity      
XPIXELSZ=   9.997114974000E+00 / [microns] X pixel size                         
YPIXELSZ=   1.000000000000E+01 / [microns] Y pixel size                         
OBJCTRA = '  8  3 12.400'      / Centre Right Ascension (J2000)                 
OBJCTDEC= '-33 31  1.00'       / Centre Declination (J2000)                     
OBJCTX  =   6.831332729995E+03 / [pixels] Centre X on plate                     
OBJCTY  =   6.784876714616E+03 / [pixels] Centre Y on plate                     
DATAMINI=         7.931138E-01 / [DATUNITS] Minimum pixel value                 
DATAMAXI=         3.339080E+00 / [DATUNITS] Maximum pixel value                 
DATAMEDI=         9.282468E-01 / [DATUNITS] Median pixel value                  
DATAMODE=         9.192346E-01 / [DATUNITS] Modal pixel value                   
FLATFLD =                    F / Are the pixel data flat-fielded?               
END                                                               

DATATYPE= 'INTEGER*2'          / Type of data                                   
DATUNITS= 'DENSITY '           / Units: transmission, density or intensity      
XPIXELSZ=   9.997403000000E+00 / [microns] X pixel size                         
YPIXELSZ=   1.000000000000E+01 / [microns] Y pixel size                         
OBJCTRA = '  8  3 12.400'      / Centre Right Ascension (J2000)                 
OBJCTDEC= '-33 31  1.00'       / Centre Declination (J2000)                     
OBJCTX  =   1.465500341686E+04 / [pixels] Centre X on plate                     
OBJCTY  =   2.495197318687E+04 / [pixels] Centre Y on plate                     
END     

 imstat R66531.fits[0]
#               IMAGE      NPIX      MEAN    STDDEV       MIN       MAX
       R66531.fits[0]   1795600     6349.     2132.     2885.       23716.

Note also that the fits header for the H-alpha extractions also contains an entry concerning the application of a flat field. Further details concerning this option are given here. If the flat field option is chosen then a version of it appropriate for the pixel unit choice is automatically selected.