QSpace Community: Queen's UniversityInformation
http://hdl.handle.net/1974/6
Queen's UniversityInformation2014-11-29T03:43:41ZApplications of Multitaper Spectral Analysis to Nonstationary Data
http://hdl.handle.net/1974/12584
Title: Applications of Multitaper Spectral Analysis to Nonstationary Data
Authors: Rahim, KARIM
Abstract: This thesis is concerned with changes in the spectrum over time observed in Holocene climate data as recorded in the Burgundy grape harvest date series. These changes represent nonstationarities, and while spectral estimation techniques are relatively robust in the presence of nonstationarity--that is, they are able to detect significant contributions to power at a given frequency in cases where the contribution to power at that given frequency is not constant over time--estimation and prediction can be improved by considering nonstationarity. We propose improving spectral estimation by considering such changes. Specifically, we propose estimating the level of change in frequency over time, detecting change-point(s) and sectioning the time series into stationary segments. We focus on locating a change in frequency domain in time, and propose a graphical technique to detect spectral changes over time. We test the estimation technique in simulation, and then apply it to the Burgundy grape harvest date series. The Burgundy grape harvest date series was selected to demonstrate the introduced estimator and methodology because the time series is equally spaced, has few missing values, and a multitaper spectral analysis, which the methodology proposed in this thesis is based on, of the grape harvest date series was recently published. In addition, we propose a method using a test for goodness-of-fit of autoregressive estimators to aid in assessment of change in spectral properties over time.
This thesis has four components: (1) introduction and study of a level-of-change estimator for use in the frequency domain change-point detection, (2) spectral analysis of the Burgundy grape harvest date series, (3) goodness-of-fit estimates for autoregressive processes, and (4) introduction of a statistical software package for multitaper spectral analysis. We present four results. (1) We introduce and demonstrate the feasibility of a level-of-change estimator. (2) We present a spectral analysis and coherence study of the Burgundy grape harvest date series that includes locating a change-point. (3) We present a study showing an advantage using multitaper spectral estimates when calculating autocorrelation coefficients. And (4) we introduce an R software package, available on the CRAN, to perform multitaper spectral estimation.
Description: Thesis (Ph.D, Mathematics & Statistics) -- Queen's University, 2014-10-15 00:52:05.8422014-10-15T04:00:00ZMr.
http://hdl.handle.net/1974/12558
Title: Mr.
Authors: Haoyu, Sun2014-10-03T04:00:00ZBivariate Models for Mortality and Morbidity
http://hdl.handle.net/1974/12557
Title: Bivariate Models for Mortality and Morbidity
Authors: Gao, Yu
Abstract: In this report, to assess the adverse effects of air pollution on health outcomes, gen- eralized additive models (GAMs) were fitted to mortality and hospital admissions data for Toronto, Montreal, Ottawa and Vancouver. The model used in the National Morbidity, Mortality, and Air Pollution Study (NMMAPS) [1] is discussed and it is referred as the standard model in this report. Previous studies have established that the health risks are associated with short-term exposure to air pollutants within one to several days. To focus on the short-term health risks by removing long-term risks, I modify the smoother on time from natural cubic splines ns( ), which was used in the standard model , to Slepian sp( ) due to Dr. Wesley Burr’s based on spectral ideas [2] in the GAM model. In the mean time, a second modification is that only the high-pass pollutant is included in the model since it is assumed that the low-frequency pollutant is orthogonal to health outcomes.
From the output of model fitting, we can get the estimate of log relative risk and the standard error . The covariance matrix between log relative risk of mortality and morbidity can be obtained based on generalized estimating functions (GEE) [3].
Bayesian hierarchical models are used to analyze time series data from multiple locations [4]. Hierarchical bivariate time series models are used to combine and pool information, estimating the overall log relative rates of mortality and morbidity, and
ii
the heterogeneity across multiple locations.
This report aims at fitting the standard model and the modified model to Cana-
dian data, and try to compare the results obtained.2014-10-03T04:00:00ZNonparametric and parametric methods for solar oscillation spectra
http://hdl.handle.net/1974/12502
Title: Nonparametric and parametric methods for solar oscillation spectra
Authors: Haley, Charlotte
Abstract: The study of the systematic oscillations of the Sun has led to better understanding
of the Sun’s inner structure and dynamics, and may help to resolve inconsistencies
between observations and the standard solar model. Recent studies have concluded
that solar modal structure remains coherent past turbulence in the convection zone
and imprints its signatures on the solar wind and the interplanetary magnetic field
fluctuations, and these structures are coherent with atmospheric pressure variations,
terrestrial seismic oscillations, and data from communications systems. Time series
containing modal structure can be expected to contain several thousands of resolved
and unresolved line components in very short bands in frequency, and the measure-
ment of these modes pushes spectrum estimation methods for time series to its limit.
This thesis presents two theoretical contributions for modeling solar oscillations in
power spectra (i) expressions for the expected number and shape of significant spuri-
ous peaks in spectrum estimates are given, in the absence of modal structure, and a
permutation test for the identification of spectra containing pathological numbers of
modal components. (ii) A model for maximum likelihood estimation of the solar os-
cillation parameters in composite spectra is given. The scientific contributions of this
thesis are (a) identification of highly significant modal artifacts in solar wind mea-
surements as seen by the Advanced Composition Explorer (ACE) on the 2 − 3mHz
band and (b) quantification of the presence of modal structure in secondary cosmic
rays (specifically neutrons) on Earth.
Description: Thesis (Ph.D, Mathematics & Statistics) -- Queen's University, 2014-09-25 19:20:15.2252014-09-27T04:00:00Z