Tonight we've submitted our updated high-z paper on SMACS to arxiv and the journal w\ et al
There are 2 main changes 1) adds a comparison to other studies and 2) updates the imaging to include post-flight zero points, a big issue as I'll explain 1/6
Conversation
First, we compare to Atek2022 and Yan2022, both whom focussed on SMACS. We find the agreement between our studies to be poor, with out photoz's agreeing with only 6/15 Atek objects and 1/16 of the brightest Yan objects. 2/6
2
2
11
The reasons are a combination of 1) different teams finding 2 sigma detections in F090W or none at all and 2) different teams preferring a lower-z (2-5) solution to the higher-z. All dependent on image processing, SED templates, code choice etc. 3/6
3
1
8
On the subject of image processing. Zero Points.
There was a big update to the pipeline CRDSv0942 on July 30-31st. These include post-flight callibrations for NIRCam which are significant. The reason is the red-side is over performing by 20+% (Rigby2022). 4/6
2
6
21
However, this was previously not considered in the pipeline, meaning sources in the red images were up to 0.3 mags brighter than they should be. A significant issue for SED fits conducted on early versions of the images. I show here the magnitude of the changes that happened. 5/6
3
15
25
We thus urge caution to those using early JWST data products, especially those made in the first 3 weeks. As there will be systematic zero point issues present which can influence redshift, SFR and stellar mass estimations. 6/6
4
14
34
Heading to bed now. But want to emphasise we don't want to say A is right or B wrong when talking conflicting photo-z solutions, nor say bin everything published in 1st few weeks. Is all a learning experience & we need to understand the instrument/contaminating population more.
1
1
12
Also a massive thanks to all those working at and elsewhere on the pipeline and calibration of the instrumentation. We're all appreciative of the work you're all doing 🥰
11
Show replies
I think most people do, but the issue lies in people using different models for z=2-5 galaxies or z>9 galaxies. There's also the image processing. Different people are doing it different ways, when we're dealing with small faint things, how we handle background&noise is important
6