Previous abstract Next abstract

Session 2 - Everything Else.
Display session, Friday, June 27
Ballroom C, Chair: Richard Canfield

[2.15] Multitaper Spectral Analysis and Wavelet Denoising Applied toHelioseismic Data

R. Komm, Y. Gu (NSO), P. Stark (UC Berkeley), F. Hill (NSO)

Our goal is to improve the estimates of mode frequencies, amplitudes, and widths derived from helioseismic observations. To this end, we apply Multitaper Spectral Analysis (MTSA) to the observed time series to derive power spectrum estimates, and then we apply wavelet denoising to the spectra to improve the signal-to-noise ratio of the modes. The rationale behind this approach is that MTSA leads to a more accurate and robust power spectrum estimate than the conventional periodogram and that since the log multitaper spectrum is close to Gaussian distributed wavelet denoising is the optimum method to reduce the noise level in the calculated spectra. We have put together a `pipeline' to calculate a multitaper spectral estimate from a given time series, to apply wavelet denoising to the log spectra and then to derive mode parameters using the GONG peak-fitting algorithm. This pipeline was applied to a set of simple artificial data in order to check for systematic errors and consistency. The wavelet denoising method was already applied to m-averaged South Pole spectra and to some GONG spectra of different L values reducing the noise level considerably and improving the fit. At the moment, we apply the pipeline to GONG and SOHO-SOI/MDI time series. We intend to present a comparison of two multitaper estimates using Slepian and Sinusoidal tapers with a conventional periodogram and a comparison of each of the three spectra with the corresponding wavelet denoised spectrum. This comparison will allow us to discuss the benefits of adding these methods to existing helioseismic data analysis packages.

Program listing for Friday