← Back to Sessions

Data-Driven Performance Evaluation of Ground Motion Models Applicable for Active Crustal Region in Italy

Description: 

The influence of regional crustal wave propagation effects and complex geologic structures has been observed in strong motion records in Italy. These phenomena are critical and are captured by ground motion models (GMM) for developing regional seismic hazard maps. The aim of present work is to select and rank GMMs applicable in active crustal regions of Italy for seismic hazard analysis. In this study, we evaluated the relative performance of 15 active crustal GMMs, including regional, pan-European, globally derived default and regionally adjusted ones. We compiled a comprehensive strong motion flatfile containing over 43,000 recordings from ESM and ITACA between 1969-2022. First, we processed the strong motion records to remove duplicates and selected only records having magnitudes greater than 4.5. Then, residual analysis was performed using strong motion records from rock site stations (having V_S30 of 600-1000 m/s) and all stations with varying V_S30. We analyzed the influence of minimum magnitude, rupture distance, and fault mechanism on overall bias and between event as well as within-event residuals. Finally, data-driven scoring methods were employed, using univariate (LLH) and multi-variate (mLLH) logarithmic scores, to assess relative performance of GMMs. Regional level seismic hazard analysis was also performed using selected candidate GMMs, by assigning weight of 1.00. Results of residual analyses showed that almost all GMMs are over-predicting at the short period due to regional factors, where the degree of over-prediction is significant among earlier version of pan-European and global GMMs. On the other hand, almost all GMMs were under-predicting at the long periods (approximately after 1-2 seconds). In general, regional and pan-European GMMs showed relatively better performance (in terms of better LLH and mLLH scores), by capturing regional characteristics than global and global model with regional adjustments. Finally, rock-level hazard maps, corresponding to V_S30 of 760 m/s and 475-year return period, revealed that regional models exhibit lower hazard than pan-European model.

Session: Regional-Scale Hazard, Risk and Loss Assessments [Poster Session]

Type: Poster

Date: 5/3/2024

Presentation Time: 08:00 AM (local time)

Presenting Author: Balaji

Student Presenter: No

Invited Presentation: 

Authors