Hi,
The cospectra values can be positive and negative.
How can I plot the cospectra vs frequency on log-log scale?
Is that correct: absolute (cospectra) vs frequency on log-log scale?
Thank you.
.
there is nothing wrong in plotting the absolute value of the cospectrum in loglog ; you should just indicate also the sign of the cospectrum on the relevant frequency bands and be quite careful when drawing straight line fits ... they should no be analyzed across the sign change boundaries !
in fact there are two different sources of sign changes :
- "true" sign changes : depending on the physical process studied, you may or may not expect strong sign reversals in cospectrum
- "noise" sign changes : often you would expect from theoretical considerations that the cospectrum goes to zero with increasing frequency ; therefore, at the high frequency end of the cospectrum, the signal becomes very small and noise contributions may move it across the sign change boundary ; this is an good indication of the end of reliable measurements !
.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics