This page archives Slack comments from the splinter session on image alignment of the Improving JWST Data Products Workshop (IJDPW).
Macarena Garcia Marin
Thanks very much for taking part of the image alignment exercise! We will kick this off on Friday at 9am, in the Boardroom. For those not familiar with the Muller building I expect we will meet first at the Auditorium. Below Im adding a few notes that I shared by email with those Im aware will be participating. Sharing it here in case I missed anyone.
Splinter objectives:
- Compare the pipeline output of a Carina Nebula mosaic (or a few tiles) with the products in MAST.
- PID2731. Find program details here
- If you check the APT file, Observation 1 is NIRCam (3 filters, 2x2 tiles), Observation 2 is MIRI (4 filters, 1x4 tiles)
- In the attached figure, kindly provided by Mattia, the red circles mark Gaia DR3 targets. You will see that the MAST products, to the left, are badly misaligned. The right is Mattia’s manual alignment to the Gaia DR3.
- Discuss pipeline shortcomings and understand if any of your tools and strategies could be successfully implemented to improve it.
- Considerations:
- Time performance
- Algorithm flexibility, i.e. does it only run for specific datasets/instruments?
- How much better does you code perform the alignment?
- How much “human intervention” is needed?
if imaging is not your area of "alignment expertise" let me know and will find a good alternative
Thomas Williams
top is the PHANGS approach, bottom is the MAST products
Macarena Garcia Marin
that looks really neat. I know you cant join tomorrow. Could you give an estimate of how well aligned yours are? e.g. by comparing with Gaia?
Thomas Williams
I haven’t done any alignment to Gaia here, so no idea
Varun Bajaj
Here's a link to my PSF fitting package (it has some docs, but the example notebook is a good starting point) https://github.com/Vb2341/One-Pass-Fitting
Thanks to those who joined the image alignment splinter. It was a very lively discussion that resulted in a list of recommendations
- Use PSF fitting to generate the catalogue to be used in the alignment (Varun’s package https://github.com/Vb2341/One-Pass-Fitting)
- Make the matching more robust (Armin)
- Make tweakreg more modular, ability to stop, check the results and continue (Armin working on that for JHAT)
- It would be very useful if the pipeline could generate alignment plots a-la-JHAT
- Have tweakreg with the option to align everything to the reference catalogue
- Guider-related:
- We should have the necessary infrastructure for the guider team to evaluate if a certain visit locked into the wrong guide star. This leads to large errors and bad alignment. Users should be warned about such situations, and notebooks on how to fix the issue.
- Improve the reconstruction of the observatory V2V3 and roll angle.
- Link the analysis of the science images to the guider, to improve the alignment outcome
- Guider uses triads, how about asterisms? FSW guider challenging
- JDox page on alignment strategies, and provide notebook for observers to deal with aligning issues.
- Coronagraphy:
- On the operations side have a more robust and repeatable TA (new SIAF, pixel offset)
- JHAT working on alignment improvements, these could be ported to the pipeline. Coronagrpahic users generally don’t use distortion map. Demonstrate that using it will yield better results and share that with community.
- Have products that informs on the alignment quality.
- Mark the position of the star on the end product.
In addition, while running the Carina exercise Varun run into a TweakReg bug (in the explan_Ref catalog, see above). He will open a github issue after some additional testing
Julien Girard
Mattia’s paper on MIRI imaging, they get an error on the pixel scale measurement < 0.4µas , well done!
Of course this is with ePSF fitting. High-precision astrometry and photometry with the JWST/MIRI imager