Detecting Periodic Components in a White Gaussian Time Series
Abstract
A family of tests for periodic components in a white Gaussian series is proposed. The test is based on a statistic which is proportional to the ratio of the maximum periodogram to the trimmed mean of the periodograms. The asymptotic distribution of the statistic is obtained. It is shown that the test proposed and Fisher's test have the same asymptotic powers at the alternative hypotheses that the series contains a single periodic component at a non-zero Fourier frequency. The tests are applied to detect the eigenfrequencies of the Earth. The proposed test detects some peaks that Fisher's test fails to detect.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1986
- Accession Number
- ADA453917
Entities
People
- Shean-tsong Chiu
Organizations
- Rice University