Paul Martini, 01/08/2013 06:46 AM
This page contains analysis for [[Cal-R5]] and Cal-G6.
h2. TAMU Presentation on Fringing
Here are the main results from my presentation at the TAMU meeting:
* Fringing amplitude is 0.1-0.2% of the sky level in z, 0.3-0.5% in Y, the spatial scale is 1" to 60"
* Fringe amplitude appears strongest at the center of the FOV
* Fringe frames exhibit A/B offsets and these offsets appear more prominent in CCDs that exhibit more significant nonlinearities (see also the [[AB_ratio_variation]] wiki page)
* Fringe correction appears stable, and even quite similar between z and Y.
The slides are available as "docdb#6830":http://des-docdb.fnal.gov:8080/cgi-bin/ShowDocument?docid=6830
-- Paul Martini, 8 Jan 2013
h2. z-band fringing and the SN data
The typical peak-to-valley amplitude of the z-band fringing is about 0.1% and the largest variations are about 0.2%. These correspond to about 5-10 DN variations in a sky level of about 5000 DN in the 330s z-band images obtained for the SN deep fields. The fringe pattern is not readily apparent in the individual z-band images without smoothing, as the rms = 30 DN in these data. The fringe pattern is more obvious once the 11 330s z-band exposures have been combined, as the combined images should have a noise amplitude comparable to the amplitude of the fringe pattern.
I created a fringe frame from 10 regular survey (90s) z-band observations from 20121124 for CCD35, scaled it to match the fringe amplitude in the longer (330s) SN observations (of C3 obtained 20121202), and created stacked versions of the data for this CCD with and without the fringe correction. The uncorrected (left) and corrected (middle) images are shown in the two panels below. The third panel contains a smoothed version of the fringe correction frame.
Because the fringe pattern is still comparable to the noise level, I also smoothed both stacked images with a 3-pixel Gaussian to make the fringe pattern more obvious. This is shown below.
Note that the fringe-corrected data shown in the middle panel exhibits a large-scale gradient. This may be an artifact of using fringe
correction files generated on a different night. In any case, the feature is on the scale of the entire array and should be removed with background subtraction.
Note also that the fringe correction frame is not perfect. There are several residual features (e.g. a saturated star 'footprint' in the upper left corner) that should disappear with either better masking or a larger input sample for construction of the fringe frame.
-- Paul Martini, 5 December 2012
h2. Fringe pattern on CCD N4 in Y band
The typical peak-to-valley amplitude of the Y-band fringing is about 0.3% and the largest variations are about 0.5%. These correspond to about 3-5 DN variations in a sky level of about 900 DN in 50s Y-band images.
A number of Y-band images were obtained under OBSTAC control during the bright time at the end of the SV period. I created fringe correction frames from observations on two separate nights to evaluate if the fringe pattern remains constant from night to night. The two datasets I evaluated are from 20121122 (9 observations of MS1a starting with 154495) and 20121124 (9 observations of RXT 2248.7-4431 starting with 155251).
I started with the data processed through firstcut by DESDM for CCD N4. I created object masks and then combined the frames to create fringe frames for each night. The image below shows these two fringe frames, along with the difference of the two. The difference image shows no residual fringe pattern and implies the fringe pattern is quite stable, at least under photometric conditions. The grayscale stretch is +/- 2x the rms in the difference image.
These results also suggest that 9 Y-band images are sufficient to produce an adequate fringe correction frame in most circumstances.
-- Paul Martini, 2 December 2012