Stack SNR using Astro ImageJ (AIJ) Photometry

Anything that doesn't fit into any of the other forums
Post Reply
pmohan
Posts: 9
Joined: Sat Feb 07, 2026 7:34 pm

Stack SNR using Astro ImageJ (AIJ) Photometry

#1

Post by pmohan »

I have some curious data when comparing a star’s SNR for two instances captured in SharpCap. I compare a 4-frame stack with 8.3s subframes to a single 33s frame, all else being equal. The figure shows the respective data extracted from AIJ using its photometry tools. The input images are 16-bit .fits files saved in SharpCap.

What I find curious is that, as measured by AIJ, the stack has a SNR (Source_SNR_1) that is about 1/2 that of the single 33 sec image. In AIJ, SNR is defined as (Source_1)/(Source_Err_1). Further, for the central measurement annulus on the star, the stack ADU count (Source_1) is ¼ that of the single image count. On the other hand, the sky background count (Sky/Pix_1, computed using the outer annulus) for the stack image is about ¼ the background for the single 33s frame.

I’m wondering what sort of scaling occurs in the stacking process and in the image file output that might help explain these results.
Group thoughts/feedback appreciated.
-Paul
Attachments
StackSNR.jpg
StackSNR.jpg (150.11 KiB) Viewed 950 times
archer1960
Posts: 98
Joined: Sun Feb 23, 2025 9:04 pm

Re: Stack SNR using Astro ImageJ (AIJ) Photometry

#2

Post by archer1960 »

I'm no expert, but for the same total exposure, I would expect the stack to have more noise (and therefore a lower SNR) because you have 4 exposures worth of read noise rather than one. I don't have an explanation for your other readings.
User avatar
admin
Site Admin
Posts: 17574
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Stack SNR using Astro ImageJ (AIJ) Photometry

#3

Post by admin »

Hi,

the expected results would be that the stack would have either the same noise, or higher, depending on whether the shorter exposures used in the stack were long enough to be 'well exposed' for deep sky (shot noise in the darker pixels is sufficient to significantly outweigh read noise).

However, there are more things that could be going on...

1) If you are using a colour camera and live stacking then the stack will be made up of debayered frames. The debayering process averages across pixels and will reduce noise. Unless you have done the same to the individual frame (for instance a 1 frame live stack of a single 33s frame) then the comparison will be spoilt by this.

2) You could be getting reduced noise in the stack due to the fact that the stack will be the average of the four frames (not the sum). The result in terms of S/N is theoretically the same, but in practice when you average and save as integer values, some noise will be lost as you divide by four as part of the average and throw away the remainder.

cheers,

Robin
pmohan
Posts: 9
Joined: Sat Feb 07, 2026 7:34 pm

Re: Stack SNR using Astro ImageJ (AIJ) Photometry

#4

Post by pmohan »

Thanks to you both and appreciate your feedback.
Indeed, using 4 exposures in the stack scales up the read noise by 4x, but other factors in the 'stack noise' equation [per Robin's lectures] make it such that for the two cases, the noise should be the same, as Robin has commented. Regarding the camera, it is monochrome.
I followed through a bit deeper:
*****
light pollution rate ( e/pixel/sec) [P] 25
read noise e [R] 2.55
sub exp length (sec) [s] 8.33
total stack time (sec) [T] 33.32
number subframes [n] 4
****
Stack noise (total 33sec) 29.1
single frame noise (33s) 28.8
****
So if the noise is the same in both cases, I'm trying to understand why, for the target star, AIJ gives a stack SNR [60] that is 1/2 that of the single frame [120] ? I would expect the computed SNRs to be about the same.
Per Robin's comment on the stack being an average, not a sum, we can see why AIJ shows a stack Source_1 count of (36068) while for the single frame its (145790). All I can conclude is that somehow between the scaled stack output image and how AIJ computes SNR, there is a disconnect that I need to track through.

Cheers, Paul
User avatar
admin
Site Admin
Posts: 17574
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: Stack SNR using Astro ImageJ (AIJ) Photometry

#5

Post by admin »

Hi Paul,

another thing just popped into my memory while reading your post above... Does the camera you are using have a 12 or 14 bit sensor? SharpCap typically scales up images from such cameras to the full 16 bit scale, otherwise you wouldn't be able to move here on the forums for people posting 'why are my saved images too dark/black'... However, I have heard that the scaling up does bad things for the sort of photometry measurements that you are doing, hence there is an option in the 'Saving' page of the settings to save FITS without this scaling up.

Maybe that's the issue that you are encountering?

cheers,

Robin
pmohan
Posts: 9
Joined: Sat Feb 07, 2026 7:34 pm

Re: Stack SNR using Astro ImageJ (AIJ) Photometry

#6

Post by pmohan »

Thanks Robin,
Sensor is an IMX533 14-bit monochrome, and I will try the alternate 'save' setting to get unscaled FITS and see how the comparison looks.
You know, this is somewhat an academic exercise given I will likely be using unstacked ~30sec frames for my time-series image sequences. But it's good to understand what is going on none the less.

Regards - Paul
Post Reply