Update: This scope is hackable, and it is surprisingly simple: just use Siglent keygen you can find on the forum (although I don't know which keygen exactly)
SDS804X ---> SDS824X, 50Mpts ---> 100Mpts mem-depth, 70M ---> 200M bandwidth.
The mem-depth is real, you can export a nearly 200MBytes binary wave file, which, you know 100Mpts*12bits/8 = 150MBytes
Still you use RMS for measure noise. (even if input coupling is AC but your measurement is True RMS from data after ADC.
For base noise measurement you need use Stdev (aka AC RMS and this is now AC RMS from data after ADC)
I have not time to demonstrate now all how it can affect in different signal situations.
In this particular case there is only but least internal DC offset residue.
Right way is
Stdev. Here example: (not current SDS800X HD or current SDS1000X HD)
Now DC coupled input ... no big effect. But clear difference still between RMS and ACRMS
Then note, this is 200MHz BW.. for compare different oscilloscopes with different BW we can use 20MHz BW what is available in every scope and then compare these (if scopes full BW's are different...as we know noise rise (part of total noise least) when BW rise...independent of oscilloscope brand etc)
Ch1 full BW: RMS, ACRMS (stdev), Ch4: 20M BW, RMS, ACRMS (stdev)
then just other sample
Note also here small but clearly visible DC offset in Ch4 (what is included in RMS of course but removed from Stdev (AC RMS)