Description

Fits a quadratic to the detector response using 3 exposures and dark frames. At present this is primarily a test program, and is not completed to perform actual production runs. The program reads 4 sets of 3 exposures plus dark frames from 9/08/07 and calculates {a, b, c} for y = ax^2 + bx + c fitting the 3 exposures. At present, only one quadratic is calculated fitting the mean of each exposure stack, rather then pixel-by-pixel.

Usage

```quadflats.py raw-data-folder
```

The program takes a single argument: the location of the raw data for 9/08/07 (or no arguments if it happens to be in '09807/raw').

Examples

```> quadflats.py /Users/sky/Projects/Python/VenusData/090807/raw

dark: mean1 =        7.227067, mean2 =       18.515120, mean3 =       42.905338
422: mean1 =      623.193564, mean2 =     2564.759774, mean3 =     9963.936523
422: min   =     -142.901587, max   =     9921.031185
942: mean1 =     2895.637321, mean2 =    10818.978527, mean3 =    35971.496094
942: min   =    -1922.302551, max   =    35928.590755
1462: mean1 =     2615.845815, mean2 =    10145.250667, mean3 =    34929.968750
1462: min   =    -1668.133294, max   =    34887.063412
2030: mean1 =     2471.854942, mean2 =     9323.951035, mean3 =    32721.347656
2030: min   =    -1145.716724, max   =    32678.442318
``` Fits are performed for sky frames starting with #422, 942, 1462, and 2030. Dark frames starting at #452 are subtracted from each exposure average before fitting. What is interesting to note about the results is that, while all curves have the same shape, the exposures and fit for images 422+ are less than 1/3 the values of the other 3, even though the exposures and coadds are the same.

I don't know if the quadratic fits will improve the flat-fielding over affine fits. I may come back to this question later this month.