Using mtspec to obtain spectral estimates

Hi everyone,

I was wondering if anyone has experience with using mtspec (https://krischer.github.io/mtspec/tutorial.html), a Python library for multitaper spectral estimations?

My goal is to take a waveform, say 10 seconds before a P-wave and 60 seconds after the P-wave arrival, and perform multi-taper spectral estimation analysis to obtain spectral estimates for the entire seismogram.

The mtspec method (https://krischer.github.io/mtspec/multitaper_mtspec.html) reads time-series data and returns both the spectrum and the corresponding frequencies. My sample spacing would be delta = 1.0 / 40 sps, but I am curious as to what would be appropriate for the time_bandwidth, nfft, and number_of_tapers paramters? Would nfft correspond with the number of samples in the data?

Lastly, mtspec returns frequencies and the corresponding amplitudes. I was wondering if it’s possible, using this library, to display the spectral estimates in time-frequency space? Ideally, I’d like to generate spectrograms using the adaptive multitaper spectral estimates, but I’m unsure how to take the data, originally plotted in a periodogram, and represent the data in a spectrogram?

Thank you for your comments and suggestions.

Cheers,

Jason

1 Like