Activity Number:
|
628
- Statistical Applications in the Physical Sciences
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
Abstract #322984
|
|
Title:
|
Spatio-Temporal Multitaper Spectrum Estimation
|
Author(s):
|
Christopher Geoga* and Charlotte Haley and Mihai Anitescu
|
Companies:
|
Argonne National Laboratory and Argonne National Laboratory and Argonne National Laboratory
|
Keywords:
|
Time Series Analysis ;
Spatio-Temporal Analysis ;
Spectrum Estimation ;
Multitaper
|
Abstract:
|
Multitaper spectrum analysis is the average of multiple windowed spectrum estimates using the time-limited and optimally band-limited discrete prolate spheroidal sequences (DPSSs). The multitaper spectrum estimate is a powerful tool for explicitly controlling the bias and variance of power spectrum point estimates. Since the fundamental tool of spectrum estimation is the FFT, multidimensional spectrum estimation can provide a computationally efficient method for estimating covariance structure and linear transfer functions in spatio-temporal data, with the multitaper estimator providing additional hypothesis testing tools. Currently, higher-dimensional multitaper spectrum estimation is largely done using approximations of the multidimensional DPSSs with limited accuracy that meaningfully reduce the power of the estimator and can introduce bias into spectrum estimates. In this presentation, we will discuss recent developments in the computation of multidimensional DPSSs by F.J. Simons and D.V. Wang in 2011 and then demonstrate the implementation of these methods for higher-dimension multitaper spectrum estimation on a few simple examples from physical data.
|
Authors who are presenting talks have a * after their name.