PS1 Image Artifacts


"Burns" are the result of bright sources landing on the detector in the image under consideration, or a recent image, causing trails along the detector columns. It seems that lots of excess charge upsets the pixels. When this charge is clocked out, some of it remains, causing a burn trail that goes upward (away from the amplifier) in the image with a bright source. Then in subsequent exposures, the angry pixels slowly leak charge, causing burn trails that go downward (towards the amplifier).

'burntool' has been developed by the Camera group (specifically, John Tonry) to attempt to remedy these artifacts in software. Since it requires operation on images in the order in which they were observed, it must be run on data in the IPP after registration but before the chip stage is executed (output is named *.burn.fits, set rawImfile.user_1 = 1.0, and then turn on USE.DEBURNED.IMAGE in the PPIMAGE recipe). Efforts are underway in the Camera group to fix these artifacts in hardware.

Three different artifact patterns can arise from burntool. The first has a bright streak that burntool should have removed but didn't. The second has the burn from a star on a previous image oversubtracted either due to a bad fit or the images being processed in the wrong order. The third shows an oversubtraction of the burn.

Star with positive burntool feature

Oversubtraction from burntool

Ghost image

The origin of these is unknown. Are they true ghosts (due to multiple partial reflections), or are they electronic (crosstalk or burns)?

Ghost image

Star glint

A bright star just beyond the edge of the populated focal plane can cause a glint. See discussion here: [PS1_GPC1_GlintMasks].

Edge glint

Moon glints may also be seen, due to light from the moon avoiding the baffles and hitting the detector.


Ghosts are due to reflection of a bright star off the focal plane, and re-reflection and focusing back on the focal plane by L1 or L2. This results in an out-of-focus image (image of the pupil) on the detector. We attempt to mask these using the astrometric model and a star catalogue, but the model of the position, size and exact shape is not particularly precise. This results in a partially masked ghost.


Jagged noise

These jagged noise lines are not stable in position or time, moving and changing brightness between exposures. Their cause is unknown, and no measures are in place to address them.

jagged noise pattern

More examples of jagged noise, note that these are from from a diff:

impressive jagged noise example (from a diff)

Pattern noise

RFI or other conditions in the readout electronics cause a variable signal in the readout of each row. We initially attempted to address this in software using a program written by the Camera group, tiltystreaks. This produced more artifacts and was not particularly stable. We now use a polynomial subtraction of each row (with appropriate outlier rejection) in ppImage.


Burn tool?

Note: substamps show template, image, difference image

Burn tool?

Note: substamps show template, image, difference image

This looks on first glance like a real object. There are many like that in the images.

The same artifact at the same position in different images, different days!!!

Note: substamps show template, image, difference image

Poor Cells

Noted in MD01.refstack.20120803, bad cell causing poor arc in primarily skycell.036,033,024,025,026,047 y-band from ota14, cell 31 and not in other filters. Triggered inexposures with higher background?

Noted in reference stack MD10.refstack.20120705, poor/bad cell causing poor arc in primarily skycell.054 g-band (and less so in other filters) -- ota43, cell 54?