Getting too old for this game - Sum or Median/Average??

[dph1nm]

One should be clear about the physics of all this. It is the increase in total exposure time (i.e. collection of more photons) which improves the signal-to-noise. Averaging is nothing to do with it - this is just a convenient bit of maths to keep the numbers in your stacked data file in a similar range to those from a single exposure.

NigelM

[dph1nm]

also, all things being equal (!!??) you get a better SNR with a larger stack, than you'd get with a longer exposure!!

No. If you have 10 subs in the stack then the s/n increases by sqrt(10) - but if you had 10 times the exposure in a single sub the signal goes up by 10 but the noise only goes up by sqrt(10), so the ratio of the two is still sqrt(10).

NigelM

[Merlin66]

Nigel,

Thanks for your input - greatly appreciated!

I've got it now!

- reduced noise allows fainter detail to become visible....

The laws of diminishing returns seems to suggest there are some limits on "acceptable" SNR.....above SNR=200??? Or, in the case of astrophotography it's all subjective ??

[StuartJPP]

Each exposure is a measurement of the signal plus *or minus* an error....

Right...got it.

[FraserClarke]

The laws of diminishing returns seems to suggest there are some limits on "acceptable" SNR.....above SNR=200??? Or, in the case of astrophotography it's all subjective ??

Yeah, diminishing returns is very good description (every doubling of SNR needs four times more exposure).  When you start pushing towards high signal-to-noise ratios (>50, say), there are a bunch of other effects which start to come in. Unfortunately noise never goes down quite as sqrt(t) because of all sorts of correlated effects. You probably want to avoid getting into that regime though -- it can get very painful trying to understand them all

Astrophotography is no different to spectroscopy as far as the statistics goes; except that for making pretty pictures people only worry about whether things look right -- not whether they actually are right...  That allows you take a whole bunch of short-cuts and not worry about it...

[dph1nm]

limits on "acceptable" SNR.....above SNR=200???

If I ever get a DSO image with SNR that high I will let you know!

NigelM

[Merlin66]

Hmmmmm

To achieve a "shot noise" SNR of 200, you'd only need a signal of 20,000 ADU (ignoring gain etc)

With all those LOOOOONG exposures ( multiple hours) being done nowadays, I'd have thought that could be achieved????

[dph1nm]

To achieve a "shot noise" SNR of 200, you'd only need a signal of 20,000 ADU

Only if your sky is completely dark! The SNR is DSO_ADU/sqrt(sky_ADU + DSO_ADU), assuming a gain of 1 and ignoring any thermal noise etc. For faint DSOs, where sky_ADU>> DSO_ADU that is much more difficult to obtain.

NigelM