Библиографическое описание:TESTING OF SPECTRUM ANALYSIS RESULTS IN DENDROCHRONOLOGY: POSSIBILITIES TO FIT AND FORECAST LONG-TERM TREE-RING CHRONOLOGIES : доклад, тезисы доклада / V. Shishov [et al.]. - [S. l. : s. n.], 2015. - Текст : непосредственный // 15th International Multidisciplinary Scientific GeoConference SGEM 2015 / 15th International Multidisciplinary Scientific GeoConference SGEM 2015 (2015 ; 18.06 - 24.06 ; Albena, Bulgaria). - Sofia : Общество с ограниченной ответственностью СТЕФ92 Технолоджи, 2015. - 2. - P. 537-544. - ISBN 978-619-7105-37-7, DOI 10.5593/SGEM2015/B32/S14.072.
Аннотация:Tree-ring chronologies (dendrochronological time series) are an important proxy source for oblique high-resolution information about climate and environmental changes in the past and present. Often the time series signals are associated with direct external periodic forcing (e.g., annual irradiance, seasonal moisture regimes, etc.), or with the internal oscillations within biological systems themselves (e.g., age-dependent trends, components of competition, etc). In most cases, the observed signal is interpreted as superposition of different internal and external influences. In most cases due to unstable frequency, amplitude and phase of analyzed signals the significance of power spectrum peaks may be tested by the “red-noise” null hypothesis, with a number of additional assumptions concerning possible causes for the observed instability. The goal of this paper is to verify information losses in the case of testing a power spectrum by the “white-noise” null hypothesis in order to detect significant cycles in dendrochronological time series. The new approach described herein allows us to (1) obtain an adequate spectral decomposition of different tree-ring chronologies; (2) analyze spatial comparisons of different time series, specifying possible causes for disagreement; and (3) build new long-term reconstructions of different climatic series by different cyclical components. Moreover, the approach helps to extend super long-term tree-ring chronologies by low-frequency components, to verify temporal periods in the past for which there are no good statistical estimations, which will enable extension of existing climatic reconstructions.
