How Much Temporal Information Does Earthquake Declustering Remove? An Entropy-Based Analysis of the Türkiye Catalogue
Year: 2026
Type: preprint
Abstract: Earthquake catalogues contain spatial and temporal clustering generated primarily by foreshock-aftershock sequences, which can obscure the underlying structure of background seismic activity. Declustering is therefore commonly applied before statistical analyses intended to characterize independent seismicity. However, the dependence associated with clustered events also represents an embedded feature of earthquake activity that can be quantified using appropriate statistical measures. Information-theoretic methods provide an objective framework for evaluating this temporal organization and the extent to which it remains after declustering. Within this context, this study examines the uncertainty and short-term temporal predictability of monthly earthquake activity in Türkiye using catalogues declustered by the Gardner-Knopoff, Deniz-Yücemen, and Reasenberg methods. Monthly event counts were classified into catalogue-specific activity states and analysed using Shannon entropy, conditional entropy, lagged mutual information, moving-window analysis, and permutation testing. The Gardner-Knopoff and Deniz-Yücemen catalogues exhibited weak and statistically non-significant month-to-month dependence, whereas the Reasenberg catalogue retained significant temporal organization during the early and recent periods. The findings demonstrate that the apparent predictability of seismic activity depends strongly on catalogue preparation. More broadly, information-theoretic measures provide a quantitative basis for comparing declustering algorithms according to the amount of temporal dependence they remove or retain.
Source: Research Square
Author: Hakan Karaca
Language: English
Subfield: Geophysics
Field: Earth and Planetary Sciences
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7221634893