Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 28: Data Analysis Methods and Modelling of Geophysical Systems
DY 28.8: Vortrag
Donnerstag, 29. März 2012, 16:45–17:00, MA 144
Similarity measures for irregularly sampled time series — •Kira Rehfeld1,2, Norbert Marwan1, Jobst Heitzig1, and Jürgen Kurths1,2 — 1Potsdam Institute for Climate Impact Research, Potsdam, Germanyesearch, Potsdam, Germany — 2Department of Physics, Humboldt University Berlin, Berlin, Germany
Automated and joint analysis and inter-comparison of palaeo-climatological time series from proxy archives, e.g. stalagmites, ice and sediment cores, is of much interest in the study of past climate and its changes. Due to heterogeneous archive properties, reconstructed observation times are not spaced at regular intervals. This introduces an additional, substantial, error source when applying standard linear and nonlinear measures, as necessary interpolation introduces bias, especially for high-frequency signal components. Using kernel-based approaches, we circumvent the need for interpolation and use the information contained in the time series at the different time scales directly. In benchmark tests we compare results for kernel-based Pearson correlation and mutual information to estimates obtained from standard interpolation-based methods. We illustrate robustness, reliability and superiority of the new methods using synthetic time series of known inter-sampling time distributions similar to those found in reality and show that the results we obtain from palaeo records show the same characteristics. To illustrate the capability of our approach we construct, analyze and interpret small complex networks from palaeo records of Asian Monsoon variability.