Next: Warning Up: XRTSUB Prev: The HRI

Variance

XRTSUB will produce a variance array in the output file. The formula is:

  VAR(IX,IY,IT,IP) = SVAR(IX,IY,IT,IP) + BVAR(IX,IY,IT,IP) *
                                     (CFACTOR**2 + PFACTOR(IP)**2)

Where VAR is the output variance, SVAR is the source variance, BVAR is the background variance, CFACTOR is the area normalisation factor and PFACTOR is the positional correction factor. IX,IY,IT and IP denote the pixel position in the X,Y,Time and corrected PH axes respectively.

Typically the source and background files won't contain variance arrays by this stage. In which case poissonian statistics are assumed and the source and background data values are used in place of their variances to calculate the output variance array. If a source pixel has a value of zero then by default its variance will be assumed to be zero, however, it is possible to override this by setting SDEFVAR=1.0 (for instance) on the command line. Similarly, empty background pixels may have their variance set by specifying BDEFVAR=0.8 (for instance) on the command line. BDEFVAR is the default variance for the background which is actually going to be subtracted from the source data, e.g. if the background file is a spectral image, it will be binned into a spectrum before being normalised to the source box position and subtraction. In this case BDEFVAR will only be used if a bin in the background spectrum is zero.

Next: Warning Up: XRTSUB Prev: The HRI