Last modified: 11/04/2009, F. Masci

Gain (and possibly simultaneous Read-Noise) Validation/Derivation
-----------------------------------------------------------------

The plan is to fly with the ground gain estimates (one number per band) derived
by SDL. There is also a plan to validate/rederive the gain using on-orbit data. 
This can be done using independent exposures with varying background 
intensity and examining the dependence of pixel noise noise variance on
pixel flux. Some options are below. The read-noise (RN) can either be estimated 
simultaneously in the fitting method, or, assumed as a prior using 
ground RN or cover-on RN maps (for w1,w2) from the dark derivation exercise 
(see ../darks_w1w2).

Options:

- Try estimating gain using cover-on image differencing for w1,w2?

- Use cover cool down as a proxy for flux level variation? 

- Use pole-to-ecliptic background variation for w3,w4 gain derivation
  and use fact that read-noise << photon noise.

The program ValidateNoiseMod.r computes the expected sigma in observed 
DEB slope versus DEB slope value given predetermined values of gains 
and read-noises. It also does the reverse: given an input run of 
sigma(slope) vs. slope, it performs a linear fit to estimate the gain 
and read-noise. This is on some coarse level (not the pixel scale).

This program is written in the "R" language, and the interpreter can be
downloaded from: http://cran.stat.ucla.edu/ The installation is 
straightforward. The program can then be executed at the command-prompt 
using:
/usr/bin/R -q --no-restore --no-save < ValidateNoiseMod.r

The program has been tested and the (SDL) ground values of gain/read-noise
validated independently by me using MIC2 (flat) test data in:
caustic:/wise-ops/01/wise/fmasci/NedSim_May09/RefPixels/stacking