Rotation in alert postage stamps and cutouts

Science Alerts & Brokers
,
#1

I have two questions regarding the alerts and cutouts:

1. Rotation of postage stamps in alert packets

I’ve noticed that the postage stamps included in alert packets appear to vary in orientation across visits, reflecting each visit’s sky rotation. I would like to understand how DI handles this.

Specifically: before running DI,
(i) will the visit image be resampled and rotated to match the template’s orientation (so stamps would always be in template orientation),
(ii) Or is the template rotated to the visit image’s orientation (so stamps retain the visit’s native rotation)?

If it’s the former, does the orientation variation we observed in early alert stamps point to an early production issue rather than expected pipeline behavior?

2. Accessing cutouts before the 80-hour embargo

If I understand correctly, cutouts are accessible in two ways:
(i) from brokers within ~60s of readout,
(ii) via Butler/SIA, but subject to an 80-hour embargo.

However, the Data Products Definition Document (DPDD, sec 3.4.2) states: “Complete difference images will be made available for download no later than 24 hours from the end of visit acquisition.” which seems to directly conflict with my understanding based on RTN-011 (early science). Is the DPDD information stale on this point?

My main question: is there a 3rd way to access cutouts, preferably larger than the 30×30 px stamps in alert packages, before the 80-hour embargo expires?

Thank you in advance.
Best regards,
Satadru

#2

Regarding question 1: the answer is option (ii) – the template coadd images are warped, scaled, and PSF-matched to the visit images, and then subtracted from the visit image to create a difference image. Thus the on-sky orientation of the visit will be retained in the difference image as well.

Regarding image access: you are correct that the DPDD information is misleading. Images will only be accessible after the 80-hour embargo period. Thank you for pointing out the confusing text – I’ll raise this to make sure it gets addressed.

1 Like
#3

I have seen the rotation of the template. But I still don’t understand why the provided diacutout is not the science cutout - the template cutout ? Apparently the difference is much better optimised in the diacutout compared to the user-calculated (science - template) cutout. Probably the optimised template sent to the broker is not the optimized template ?

#4

Thanks for the confirmation! Although it will create a bit of issue for real-time analyses of image time-series, I understand the motivation behind this choice.