Looking at the Data: Image Gallery
On this page, several features and artifacts of LBC images are collected and discussed.
This page is under construction. Please check back for updates.
Overscan and trim regions
Each LBC chip has a 50-column prescan and a 206-column overscan region. The science data cover 2048 x 4608 pixels and this is recorded in the headers as trimsec = [51:2098, 1:4608].
The overscan region runs from 2099:2304 and in the headers, this is recorded as biassec = [2099:2304, 1:4608]. Note, however, that there is a discontinuity and both vertical and horizontal gradients in the LBCB overscan regions. These must be dealt with in the reduction. This feature has been seen with different controllers. The LBCR overscan regions do not show similar structure and generally appear flat, although the low-level horizontal banding that is sometimes seen continues into the overscan.
For LBCB, there is a discontinuity in the overscan level that occurs 50 columns into the overscan. For this reason, it is recommended to adopt an overscan region that extends across all rows, but avoids at least the first 50 columns of the overscan section, biassec = [2149:2300,1:4608]. The last 4 columns are also rejected from the statistics on the overscan. It is always a good idea to examine the overscan of one or two images before removing it. Also, because of the horizontal banding issues described below, it is recommended to remove the overscan row-by-row rather than make a smooth fit to the combined columns as a function of row.
The plots below illustrate the discontinuities in the overscan region and also the fact that the prescan and overscan levels differ.
The LBCB overscan regions also show vertical and horizontal gradients. These are illustrated below:
The figure above shows the first 1000 rows for the 1st bias (lbcb.20210212.145619.fits) and the 18th one (lbcb.20210212.150600.fits), in a series of 25 biases. The 50 columns of prescan and the 206 columns of overscan are included in both cases. The discontinuity and the gradients are easily visible in the first bias while in the 18th, all that remains is a horizontal gradient in the data section.
In the figure above, the first 1000 rows of the overscan region of series of 3, 3 and 5 twilight sky flats are displayed. The discontinuity and gradient appear strongest in the first image of the series, and both gradually diminish throughout.
Electronic Noise
LBCB Chip 2 Noise
Between Dec 2017 and Sep 2020, we have occasionally seen noise in chip2 of LBCB. The copointing image below illustrates this noise and the histogram of counts for chips 2, 3 and 4 (below the image) shows the clustering of counts in chip 2, but not in chips 3 or 4, which is suggestive of a stuck bit. The noise shows up only in biases and low-background images, and it seems to appear only in colder conditions (less than ~0 C). In late September, 2020, the controller (#6) was swapped out and at the time of this update (Oct 2022) we have used two controllers (#0 and #2) in its place and we have not seen this LBCB chip 2 noise.
Low-level “horizontal” banding:
LBCB
Low-level (~6 ADU from peak-to-valley) “horizontal” banding was seen occasionally in October 2020, when controller #0 was being used for LBCB Science. Eventually, #0 was swapped out for #6 (and later for #2, which is the one in use at the time of this update, Oct 2022) This banding was seen only during a brief time in September/October 2020 has not recurred.
Even though the noise is low-level, the pattern causes it to stand out on biases and on low-background exposures, such as the copointing images. Below is an example of a copointing image which illustrates the noise on the raw image (left) and also the effect of overscan subtraction (right).
This noise was present on Oct 01, but did not reappear until Oct 16, and now it seems to be a constant feature. We are discussing this with the LBC team and Skytech.
LBCR
Since September 2019, we have occasionally seen low-level banding on LBCR biases and low-background exposures. This is mostly aligned with the rows and comes out pretty well, though not exactly, in the overscan subtraction. During a series of biases, the low level banding is sometimes seen, especially in the first 20 or 25; for this reason, it is a good idea to take more than 25 biases. A script to take 50 biases is now provided in the Calib_OBs directory (/lbt/lbto/Calib_OBs/BIASDARK/50Bias_Bino.ob). It takes about 35 minutes to complete.
LBCR “Chevron”
At the top of LBCR chips 2 and 4, there is a ‘chevron’-shaped artifact that does not come out through flat fielding.
The image below shows the ‘chevron’ at the top of LBCR chip 2.
The zoom below shows unrelated features which are typically seen on all chips of LBCR images: horizontal lines at multiples of 512 (perhaps a clock or timing issue) and horizontal ripples.
