Adding low to preexisting high signal to noise ratio imaging data

That depends a lot on your usual location. A last- or first-quarter moon is about 11 times less bright than the full moon which illuminates the sky to about 18 mag/arc-sec^2), so very approx. 20.5 mag/arc-sec^2. This means in Bortle 5 and above, there will be no significant effect - this is also in keeping with what I see at my home location. This is highly directional, of course and does not account for glare and reflections, so don't try and shoot your target 10° from the moon. If you have darker skies, it's not quite that easy, because the narrowband filters now need to be accounted for.

But you also put in 28h of exposure time already. To further increase your total SNR by 1.4 times, you will need an additional 28 hours even with the same quality subs. I find I see rather little real benefit unless I increase SNR by a factor of 2 - which means 4 times the total exposure time. I'ld add narrower bandpass filters will have the same result - but you're already using 3nm filters.

Can adding low quality data ever worsen a stack?  (Assuming standard WBPP settings etc.)  I've never seen a straight answer to this question...

Yes and no. It depends on how you add to the existing stack and how you define "low" quality.

Sorry, I meant "low SNR" - for example data captured under much brighter skies as per the OP's question. Adding to the stack with default WBPP settings, assuming the frames are just above the quality cut-off so that they are included in the stack.

Preamble: I don't use WBPP so I have no way to know what the "default" settings would be.

If SNR is your criterion then, if the choice of the weights are reasonable, then yes, it would still add meaningful data (assuming they are are normalized to each other, say with LinearFit). If the criterion is resolution then maybe no (if the high SNR is also high resolution) or yes (if the low SNR is also at much higher resolution of the high SNR stack). Or could be a combination of these and other factors which must be given proper weight in the combination of the two. Note, also, that if you combine two stack together you would need to use PM rather than ImageIntegration to carry out the task so the weights need to be pre-calculated for this to work. Alternatively you would need to create a dummy copy of each to be able to use ImageIntegration using pre-assigned weights.

OTOH, if the criterion is the smoothness of the backbroud (i.e., its noise) chances are that adding a much noiser background to a definitely more pristine one won't help the aesthetics of the final image.

With PixInsight's weighting system you lose no SNR by adding low SNR.
https://en.wikipedia.org/wiki/Inverse-variance_weighting
https://pixinsight.com/doc/docs/ImageWeighting/ImageWeighting.html#__PSF_Flux_Weighting_Algorithms_:_The_SNR_Estimator__

Alex Woronow:
Thanks for the reference...do you have one that verifies that PI uses this method, and when and where it uses it?

Also, image noise variance is not the only factor to consider. Resolution, for instance, is also important in determining the properties/quality of a composite image.

Alex

This is from the Pix link that I cited:
SNR allows us to implement a scaled version of the inverse variance weighting method.[8] 

I agree that other factors are important.
The questions was whether you lose anything by adding noisy subs (low SNR subs) to high quality subs (high SNR subs).
With inverse variance weighting, you lose no SNR, you only gain SNR.