Was playing with an idea. Just a half somewhat abstract idea.
What if a time series of data is not only the signal and noise. What if you have a signal, independent noise and dependent noise.
The dependent and independent noise can perhaps be like the image below.
Some of the noise has some mean characteristic. In this case the dependent noise goes around the circle doing its thing. Where as the ?independent noise just sits still and surrounds any object.
Then I wonder. Could you take a time data series and separate it into a signal, independent and dependent noise?
I think dependent noise is noise you can lead to an object. Be it a real object or why not some calculated object. In any case you can say the noise contribution belongs to this object.