Strange CTI behaviors

[drsamadi]

Hi,
I'm trying to build some simulations with CTI.

My simulated image (50 rows, 6 col.) is located near the top of the CCD. I assume mean trap densities twice larger than the level expected at the "EOL" (4.5 years).
The simulation contains stars taken from a realistic catalogue. The background includes the zodiacal light (BackgroundValue=-1).

Plotting a slice along the vertical direction (Y axis), gives the following figure:

The legend is the following:

• blue squares : CTI turn-off
• black circles: CTI on, the first exposure
• red circles: CTI on, the second exposure

First, I don't understand why the first exposure is very different from the second one (note that the subsequent exposures are identical to the second exposure).
Second, I don't understand the shape of the signal in the first ~ 5-7 rows.

This looks like as an edge effect. It seams that a non negligible fraction of the charges present in the first rows are taken to fill the traps (hence the decrease of the signal in the first rows), however the traps should be already partially filled by the sky background which is present everywhere in the field of view.

Note that such feature is not seen with PIS, because the latter operates the CTI by taking into account all the rows located below the simulated imagette. Also, the CTI in PIS is stationary.

PlatoSim 3.6.0-393-g31c21d7f

CTI.zip

[Dries-Seynaeve]

Hi,

Thank you for pointing this out. It think that the we indeed don't take the sky background into effect into at the first exposure, so when we start the exposure we do have more traps. I'll take a look through to code to see what is going on.

For now, can you share your python code that you used to create your plot. I seem to get something similar, but I can't seem to recreate the image entirely.

Kind regards
Dries

[drsamadi]

Hi,
I did not recorded my code. Actually, I just did a plot along the vertical direction. I have considered the first column. The zip file contains the two platosim simulations (with and without CTI).

[drsamadi]

I think I have also subtracted the offset (1000 ADU) and multiplied by the gain (25) to get values in electrons

[JorisDeRidder]

I changed the way the initial occupation of the traps is computed, which solves the problem that the first exposure is different from the next one(s). For example here are the first 3 exposures (EOL config).

and a cross-section along the rows:

As you can see, the bottom row(s) (i.e. closest to the readout register) have lower flux values (the color scheme I used greatly exaggerates this). The reason, I think, is that the trap occupation equilibrium is not yet reached. The Short et al. model was developed for an entire image, not for an imagette. I'm not sure how to fix that, and/or whether it's a limitation of the Short et al. model.