Color transforms for small subset

["Gutzwiller, Michael" <mgutzwiller@lanvision.com>]

I looked for Landolt standards in the small subset of data I've been
using to determine how well my mini TASS catalog matches the Landolt
standard magnitudes and what the color transform might look like.

I found 7 standards in my little subset.  The results are shown below:

Before transform:
           V                         I
 TASS   Landolt  Error     TASS   Landolt  Error

11.326  11.302   0.024    10.553  10.493   0.060
 9.999  10.010  -0.011     9.929   9.836   0.093
10.853  10.803   0.050    10.240  10.179   0.061
 9.587   9.574   0.013     9.036   8.963   0.073
 8.745   8.737   0.008     8.225   8.162   0.063
10.892  10.938  -0.046     9.261   9.253   0.008
11.526  11.491   0.035    10.785  10.660   0.125

Doing a least squares fit I get a color transform of:

V = Vt - 0.032 + 0.030*(Vt - It)
I = It - 0.105 + 0.052*(Vt - It)

(This isn't really right since I should transform each I filter
separately and I should weight the measurements by their errors but it
should be close.)

Using this transform I get:
           V                         I
Transf  Landolt  Error    Transf  Landolt  Error

11.318  11.302   0.016    10.488  10.493  -0.005
 9.969  10.010  -0.041     9.827   9.836  -0.009
10.840  10.803   0.037    10.166  10.179  -0.013
 9.572   9.574  -0.002     8.959   8.963  -0.004
 8.729   8.737  -0.008     8.147   8.162  -0.015
10.910  10.938  -0.028     9.240   9.253  -0.013
11.517  11.491   0.026    10.718  10.660   0.058

Standard Deviation
                 0.028                     0.026

The transformed magnitudes seem to match quite well.

All in all I think the flat compensation program's algorithm seems
sound.  With a little additional work it should be ready for general use
soon.

Mike G.