obs_nodding and obs_nodding_irregular do not seem to behave well with multiple exposures per nod

[Hoeijmakers]

obs_nodding and obs_nodding_irregular do not seem to behave well with multiple exposures per nod

[tomasstolker]

Hi Jens,

Thanks for opening this issue! Currently it is indeed not the desired behavior of those routines. I had some discussion about this recently with Rico, but we hadn't found the time yet to implement this and deprecate obs_nodding_irregular.

The obs_nodding recipe from EsoRex can only do a pair-wise subtraction if I am not mistaken. Since we want indeed to maintain the individual integrations, we were thinking that we should first collapse several integrations to build up the background image that will be subtracted, in order to limit the noise in the individual nod-subtracted images? So e.g. collapse 6 to 10 and subtract this mean image from 1 to 5.

[Hoeijmakers]

I agree that merging frames is probably the best, most generic way to deal with this. That may be a bit easier said than done, if all those frames need to be individually pipeline reduced before being co-added. The most logical place where this should ultimately be addressed is probably the cr2res_obs_nodding pipeline recipe itself.

For now I have created a PR with a temporary fix based on obs_nodding_irregular, adding an optional keyword to pair A's and B's one to one in sequence.

[tomasstolker]

That's true, although util_calib can be used with calib_type='nodding' for calibrating the science frames. The output could then be collapsed by subsets and used for the pair-wise subtraction with obs_nodding. But this may not be something that can be quickly implemented indeed.

Thanks a lot for creating the PR with a temporary fix! That seem indeed a good solution 👍.