[bug #60644] Brightness error over-estimated in scaled noise (STD less than one)

[Mohammad Akhlaghi]

URL:
  <https://savannah.gnu.org/bugs/?60644>

                 Summary: Brightness error over-estimated in scaled noise (STD
less than one)
                 Project: GNU Astronomy Utilities
            Submitted by: makhlaghi
            Submitted on: Thu 20 May 2021 02:03:57 AM BST
                Category: MakeCatalog
                Severity: 3 - Normal
              Item Group: Output not reasonable
                  Status: In Progress
                 Privacy: Public
             Assigned to: makhlaghi
             Open/Closed: Open
         Discussion Lock: Any

    _______________________________________________________

Details:

As explained Section 3.3 (S/N Equation Modifications) of the NoiseChisel
paper <https://arxiv.org/abs/1505.01664>, when the image noise has a value
that is less than 1, it means that the data are not in units of counts (they
are often scaled to other units, like electrons/sec). However, even though it
is scaled (multiplied by a factor) from the original counts, the original
noise was Poisson noise, it just needs to be scaled back. 

In MakeCatalog, we were applying this scaling when calculating the
signal-to-noise ratio (and thus the magnitude error). However, it was missing
in the raw Brightness error columns. So when the data are in very small units
(like Janskies!) the brightness errors become larger than the brightness
itself, even for very high S/N objects!




    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?60644>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/