Regensburg 2004 – scientific programme

Parts | Days | Selection | Search | Downloads | Help

DY: Dynamik und Statistische Physik

DY 46: Poster

DY 46.33: Poster

Thursday, March 11, 2004, 16:00–18:00, Poster D

The effect of long-term correlations on clustering of extreme events — •Jan F. Eichner¹, Armin Bunde¹, Jan W. Kantelhardt¹, and Shlomo Havlin² — ¹Institut für Theoretische Physik III, Justus-Liebig-Universität, Giessen, Germany — ²Dept. of Physics and Minerva Center, Bar-Ilan University, Ramat-Gan, Israel

Long-term correlations, indicated by a power-law decay of the auto-correlation function C(s) ∼ s^−γ, appear in many natural records (e.g. temperatures, river flows, and heartbeat intervals). In uncorrelated data the return intervals r between events above a certain threshold q are uncorrelated and follow the Poissonian statistics. We show that in the presence of long-term correlations, the return intervals are also correlated with the same correlation exponent γ, and follow a stretched exponential distribution. As a consequence the extreme events show a tendency of clustering. We give many examples of observational and reconstructed data where this behavior is observed. The long-term correlations may present a natural mechanism for the observed clustering of hazardous floods between 1300 and 2000.