How to deal with slowly varying errors?

[odiakkoh]

Hi,

Here's a tricky sort of general problem I've been encountering quite a bit lately. It's all about errors that vary slowly.

So: the way I've always been taught to deal with errors is to lump them into two different categories: "random errors", that are independent between measurements, and "systematic errors" that are constant between measurements.

How on Earth do you deal with a type of error that fits into neither of these categories, but is somewhere in between... To be technical: the error is neither completely correlated, nor completely uncorrelated between measurements.

Does anyone have any advice, or know of any articles or books that may have some good practical advice about how to deal with this rather tricky kind of error?

Thanks in advance,
Matt

 

[Redbelly98]

It sounds like you're talking about what's known as 1/f noise (verbally, people pronounce it "one over f noise"). You might try googling "1/f noise" and see what comes up.

edit:
Or look at a book that deals with error and signal noise. There should be a few in any university physics or engineering department library.

edit #2:
This is not a trivial subject.