get_instrument_angle for OSIRIS in instruments.py has problems

[cdfassnacht]

Hi, in looking at the code for get_instrument_angle for OSIRIS in instruments.py, I think that there is a problem:

The offset between the imager and spectrograph has changed from 42.5 degrees to 43.4 degrees. I just checked with Keck, and it looks as though this change happened in May 2022. This will affect things for me since I'm trying to combine data from 2021 and 2025.

[cdfassnacht]

I see that there is a PA_IMAG header card that contains the sky PA, which I see in data as far back as 2021 (I haven't checked older files yet). It seems to know about the update in the imager offset, since for my 2021 and 2025 files the PA_SPEC is different but the PA_IMAG is the effectively the same, as intended. The "different PA" functionality mentioned in the original issue post was invoked because of differences of a few hundredths of a degree, so even though the sky PA of the two epochs was intended to be the same, the tiny differences in actual PA correctly invoke the code for different PAs. In any case, maybe have the code read the PA_IMAG header, rather than calculate it from ROTPOSN, etc. and needing to check the date of the observation to see whether to use 42.5 or 43.4

[jluastro]

Up until this past month, we were using the PA_IMAG header card as the PA. And this mostly worked great. The issue was that we found several nights of data (spread far apart) where the PA_IMAG keyword was wrong and did not correspond to the correct ROTPOSN value. We don’t know why this occurred (maybe communication issues), but we opted to go back to using ROTPOSN which is the actual motor position on the bench and re-calculating PA_IMAG ourself. You are correct that we need the exact date of any optical changes that would affect the equation between ROTPOSN and PA_IMAG.

On Sep 25, 2025, at 11:48 AM, Chris Fassnacht ***@***.***> wrote: cdfassnacht left a comment (Keck-DataReductionPipelines/KAI#48) <#48 (comment)> I see that there is a PA_IMAG header card that contains the sky PA, which I see in data as far back as 2021 (I haven't checked older files yet). It seems to know about the update in the imager offset, since for my 2021 and 2025 files the PA_SPEC is different but the PA_IMAG is the effectively the same, as intended. The "different PA" functionality mentioned in the original issue post was invoked because of differences of a few hundredths of a degree, so even though the sky PA of the two epochs was intended to be the same, the tiny differences in actual PA correctly invoke the code for different PAs. In any case, maybe have the code read the PA_IMAG header, rather than calculate it from ROTPOSN, etc. and needing to check the date of the observation to see whether to use 42.5 or 43.4 — Reply to this email directly, view it on GitHub <#48 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AALEJQCJGVW7ZIWNTH7KPRT3UQ2JVAVCNFSM6AAAAACHNRZ6PGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTGMZVGUZDGMBVGI>. You are receiving this because you are subscribed to this thread.

======================================== Jessica Lu Chair, Astronomy Department Associate Professor, Astronomy University of California, Berkeley ***@***.*** ========================================

[cdfassnacht]

Hi Jessica, This approach makes sense to me. Jim Lyke told me that the change happened on or around May 3, 2022

[nsabrams]

Hi! I've implemented a fix that defaults to the PA_IMAG header if you're in May 2022 until we find out the more exact date, but then uses 42.5 or 43.4 if you're before or after May 2022, respectively.

[cdfassnacht]

This is great. Thanks for updating the PA calculation!

[cdfassnacht]

@nsabrams - are you going to push the updated version? It looks like the current version on github is still from before your fix. Thanks again for writing the update!

[nsabrams]

It was fixed on the dev branch in this commit: a07bc9c. Are you still. having problems?

[cdfassnacht]

Sorry! That was my mistake. I was looking at the main branch and not the dev branch.