seaman at noao.edu
Tue Jun 16 15:01:33 PDT 2009
On Jun 16, 2009, at 2:47 PM, Arnold Rots wrote:
> RELATIVE and NORMALIZED fluxes are dimensionless quantities: they are
> an ABSOLUTE flux divided by some reference flux value. Therefore the
> unit should be and empty string.
Question: Would it make sense to reserve a token "UNITLESS" to make
such cases explicit?
> An UNCALIBRATED flux could be expressed in some form of instrumental
> units, e.g., count/s.
Isn't the problem as described that the data are calibrated, but the
calibration differs by a multiplicative scaling factor from some
standard unit? Presumably there may be cases with additive offsets as
well: Celsius vs Kelvin spring to mind as an example. (We should
hesitate before going beyond linear :-)
Tack BSCALE and BZERO onto every value?
More information about the dm