Background correction

[HolgerPollyNet]

3. Why is adding the BG signal after time-averaging correct? If, e.g., the period is one hour, then the BG signal can drastically change (so one loses the temporal information). Why not take the raw signal and extract the signal noise from it?

totSig = signal + bg; totSig(totSig < 0) = 0; noise = sqrt(totSig);

https://github.com/PollyNET/Pollynet_Processing_Chain/blob/b4f0cb38613f1059cf0e9cb2b60211e62e59bdb4/lib/retrievals/pollyFernald.m#L58C1-L60C22

[HolgerPollyNet]

@ZPYin : This is because of the signal-to-noise ratio. If we use the raw resolution (30 s and 7.5 m), the results will be very noisy. And the reference height cannot be guaranteed. So temporal information must be sacrifised if we want to have less noisy results.

[HolgerPollyNet]

@ZPYin . Also here I have a question. If one substract the background on each raw profile and the doe the time averaging it should be more correct, because as adi (github still missing) pointed out, the background could change during time. If it is really implemented this way, we would ned to change it or to implement a new option for doing differently, then we should move this thread and open a new issue.