Add effects required for GEOSNAP

[hugobuddel]

This is more or less a placeholder for now to collect the effects we need for GEOSNAP because we got a RIX about it.

RIQ https://jira.eso.org/browse/MET-2163 "Risk management - GeoSnap does not reach performance: scopesim capabilities?"

ID072 lists 2 risks associated with the GeoSnap detector performance: multiplexer glow and 1/f noise. I understand you are waiting for the results of the evaluation by ESO.

If these risks get concrete, in order to maintain the instrument performance, some modifications might be needed at the level of ETC, operations, calibration, and pipeline design. For this, it would help if the effects can be properly taken into account in the simulator.

Can the simulator (scopesim) be modified to take into account these effects, if they are confirmed? Have you already considered some mitigation strategy?

Answer

-) the 1/f noise has been characterised by Derek based on measurements at TIS, and has been characterised very extensively in Michigan (ROIC 1 version older than ours, but the 1/f noise appears to be the same). We can include it as a SNR factor >~0.9. The good news is that the 1/f noise scales with operating frequency (DIT) in such a way that if we operate in a regime where we fill the detector well with background to a fixed level (say 25 or 33 percent), then the 1/f noise will always have the same relative contribution in the error budget, regardless of the DIT. In other words, we can adapt the DIT according to the flux levels of a given application (*), without taking a 1/f penalty if we have long DITs.

(*) with METIS we have a large dynamic range from longslit spectroscopy at R ~ 400 to broad-band imaging at R ~ 6.

-) multiplexer glow: this we can include as an extra background. We have to see whether we need an image or whether a single value applied to all pixels is sufficient - given the physical origin it is not guaranteed to be as-near-as-we-ever-care homogeneous of the array. But in either case, it can be easily included, either as a single value for all pixels (the amplitude will depend on the DIT, shorter DITs = higher power dissipation = higher glow; the glow_emin_per_pix_per_sec(DIT) function will be determined in the characterisation campaign in Michigan), or as an image (or more precisely, a series of images to account for the dependence on the frame rate / DIT)

-) Bad (hot) pixels: these can be implemented in ScopeSim easily. We can include the effects on 2 levels:
a) include the bad pixel map as is, and include the chop-nod sequence with dither offsets between the individual chop cycles, to make a series of images that comprise an observation and that the PIP can take as input to generate a final product. This is the way to do it for PIP development / testing.
b) as a SNR vs. position map as already experimented with in E-ANR-MPIA-MET-1169, where I did what is describe in option a) but not using ScopeSim. With this I can assess the impact on SNR in the final data product; it should be the same as the end product of a time series as described in option a) after running through the PIP. Alain, if you like I can send you the current version of E-ANR-MPIA-MET-1169, I think an earlier version was shared with ESO (I discussed it extensively with Joel Vernet and Michele). Next year we will optimise the dither scheme, but the optimal solution depends strongly on which individual GeoSnap we will put into METIS in the end.

I've not really looked into what this practically means for ScopeSim.