Proceedings of the 12th International INQUA meeting on paleoseismology, active tectonic and archaeoseismology

far, CPSR has been determined through empirical models (e.g., Wells & Coppersmith, 1993, Youngs et al., 2003, Moss & Ross, 2011, Pizza et al., 2023) using earthquake databases via logistic regression. However, these curves are estimatedwith both global and regional catalogues, sometimes distinguished by fault style or considering all fault styles. There are no selection criteria based on fault angle or seismogenic crust thickness, so different seismotectonic environments with quite different fault parameters are mixed. The use of historical earthquake catalogues is also subject to data availability and accuracy. The objective of this study is to overcome the lack of information related to epistemic uncertainty, using an analytical approach distinguished by fault styles, seismogenic crust thickness, fault geometry and rupture size (i.e., earthquake magnitude). Our approach lies in assessing the CPSR, specifically on the principal fault. This is well-suited, given that nearly all instances of surface fault rupture involve displacement on the principal fault, contrasting with cases where distributed faulting may not accompany every surface rupture event (Nurminen et al., 2022, Sarmiento et al., 2022). Themethoddescribedhereafterwasappliedto8different seismotectonic zones distributed globally and distinguished by seismotectonic homogeneous context (e.g., normal, reverse, and strike-slip faulting regimes). However, for the sake of clarity, here we will only discuss the reverse fault data processed for the Japan case study. D ATA A N D M E T H O D S We utilized the stress map from Uchide et al. (2022) to identify homogeneous seismotectonic areas, assigning the northern region of the Japan island as characterized by predominantly reverse faulting. The analytical curves of CPSR were then computed by combining specific probability distributions and input parameters such as rupture width (W), average fault dip and seismogenic depth as described below. To compute all possible W and the relative uncertainty, magnitude scaling relations were used (e.g., Leonard, 2014, Thingbaijam et al, 2017). These are based on the linear relationship between the fault rupture area and the seismic moment (e.g., Kanamori and Anderson, 1975, Abe, 1975, Geller, 1976, Kanamori, 1977, Kanamori and Brodsky, 2004) providing an estimate of rupture width as the magnitude changes and the related uncertainty. Specific distributions were realized to place the W values at different locations within the seismogenic crust: the Hypocentral Depth Distribution (HDD) and the Hypocentral Depth Ratio (HDR). HDD is used to assign a different weight to all possible hypocenters located within the seismogenic crust. The earthquake catalogue used for analyzing hypocentral depths is JUICE (Yano et al., 2017), with data selected specifically within the seismotectonic area for reverse faults. The quality of the hypocentral depths was set selecting data with a vertical error of less than 5 km and a horizontal error of less than 3 km. We cut earthquake depths using Moho values obtained from numerical data of the most updated available literature (Farag et al., 2022) to exclude deep-crustal or subduction events. We selected only events with a magnitude greater than or equal to 2.5 to estimate the base of the seismogenic crust, which we assumed to be 90% seismicity cut-off depth (D90). Thus, for each seismotectonic zone, we calculated D90 as the 90th percentile based on employing a grid strategy. The grid structure is based on the method used by Latorre et al. (2023) and consists of a 20x20 km square centered in a 5x5 km grid moving into the study area. We then calculated HDD for iterations with magnitudes > 3.5 normalizing the depth to D90, ensuring applicability of the same method across various seismotectonic contexts. HDR distribution is used to assign different weight to all possible positions of the nucleation with respect to the rupture width. This concept was first developed by Chiou and Youngs (2008) and is defined as the ratio of the portion of the rupture located above the nucleation to the total width of the rupture. In our computation model HDR is an output parameter calculated as a uniform probability distribution and recomputed whenever rupture tends to violate the boundary layer condition (e.g., the ground surface and the base of the seismogenic crust).