As we know the wavelet power spectra (WPS) is just the norm of the wavelet coefficients to square. So the wavelet amplitude spectra is just to take the square root of the WPS.
When I smooth the WPS in time, I must to multiply by the variance of the signal and we get the global wavelet power spectrum. (gwps)
So a natural extension to the global amplitude wavelet spectra is take the square root of (gwps) but i have doubts about the last procedure.
Do you have any information about the procedure or am I in the correct approach?
Please, check the following Matlab program:
http://paos.colorado.edu/research/wavelets/wave_matlab/wavetest.m
there was a litle mistake. torrence multiply the global wavelet spectra by the variance because they normalized the signal before apply the wavelet transform in the code. In case you don't do that you must to apply the relation (22) that appear in the paper .
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series