eprintid: 1725 rev_number: 10 eprint_status: archive userid: 6 dir: disk0/00/00/17/25 datestamp: 2016-09-15 11:28:08 lastmod: 2019-03-07 13:43:13 status_changed: 2016-09-15 11:28:08 type: article metadata_visibility: show creators_name: Lasocki, Stanislaw creators_name: Papadimitriou, Eleftheria E. creators_id: lasocki@igf.edu.pl creators_id: ritsa@geo.auth.gr corp_creators: Institute of Geophysics, Polish Academy of Sciences, Warszawa, Poland corp_creators: Geophysics Department, School of Geology, Aristotle University of Thessaloniki, Thessaloniki, Greece title: Magnitude distribution complexity revealed in seismicity from Greece subjects: MP2_2 subjects: RGR full_text_status: none abstract: The structure of the magnitude distribution of earthquakes from three different seismotectonically homogeneous areas of Greece has been investigated by means of statistical inference methods. Unlike in previous studies, a nonparametric approach, namely, the smoothed bootstrap test for multimodality applied in this work makes it possible to test the complexity of the distribution without specifying any particular probabilistic model. Two null hypotheses, the number of modes in magnitude density equals 1 and the number of bumps in magnitude density equals 1, have been considered. Their alternatives mean that the magnitude population has a multicomponent structure. In two out of three studied cases the significance of the null hypotheses is less than 10%, which indicates that the magnitude distribution follows neither the log linear nor any smoothly non–log linear law but is more complex. Consequently, when the log linear model is applied to represent the magnitude distribution of each of the cases studied, estimates of mean return periods dramatically disagree with earthquake recurrence observations. The most significant differences are for large magnitude range; that is, such return period estimates are the most erroneous for large earthquakes. It has also been shown that the hazard estimation can be improved considerably by using the model-free approach with the kernel estimator of magnitude density. This approach ensures a satisfactory agreement between the mean return period estimates and actual observations, and in most of the cases, the agreement is very good. date: 2006-11 date_type: published publication: Journal of Geophysical Research volume: 111 number: B11 publisher: American Geophysical Union pagerange: n/a-n/a id_number: doi:10.1029/2005JB003794 issn: 0148-0227 official_url: http://doi.org/10.1029/2005JB003794 access_IS-EPOS: limited software_references: AD_Test_for_Magnitude_Distribution software_references: Test_for_Multimodal_Magnitude_Distrybution owner: Publisher citation: Lasocki, Stanislaw and Papadimitriou, Eleftheria E. (2006) Magnitude distribution complexity revealed in seismicity from Greece. Journal of Geophysical Research, 111 (B11). n/a-n/a. DOI: https://doi.org/10.1029/2005JB003794