Claim Missing Document
Check
Articles

Found 1 Documents
Search

Entropy-Based Model Dominance Analysis in PSHA Logic Trees for Seismic Hazard Assessment: Timor-Leste Case Study Florindo Martins; Yoyong Arfiadi
International Journal Of Economics Social And Technology Vol. 5 No. 3 (2026): September 2026
Publisher : Lembaga Riset Ilmiah

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59086/ijest.v5i3.2401

Abstract

Probabilistic Seismic Hazard Analysis (PSHA) is widely used for seismic hazard assessment by considering uncertainties associated with earthquake sources and ground motion prediction models. However, conventional logic tree approaches commonly apply static weighting schemes that may not represent variations in model contributions across different seismic intensity levels. This study evaluates the application of Shannon entropy as a quantitative indicator to identify model dominance and characterize epistemic uncertainty within the PSHA logic tree framework. A case study was conducted in Timor-Leste using the United States Geological Survey (USGS) earthquake catalog from 1960–2024. The earthquake data were processed through declustering using the Gardner-Knopoff (1974) method and classified into three seismic source types based on hypocenter depth: crustal, subduction interface, and intraslab sources. Gutenberg–Richter recurrence parameters were estimated using the Maximum Likelihood Estimation (MLE) method. A logic tree consisting of three seismic sources and three Ground Motion Prediction Equation (GMPE) models was developed, resulting in nine model combinations. For each PGA level, the contribution proportion of each branch relative to the total annual exceedance rate was computed, from which Shannon entropy and normalized entropy were derived to quantify the distribution of hazard contributions. The results show that subduction sources provide the dominant contribution to seismic hazard in Timor-Leste across all PGA levels. Entropy analysis reveals that model dominance varies with increasing PGA: at lower PGA levels, contributions are distributed across multiple models (higher normalized entropy), whereas at higher PGA levels, the subduction source increasingly dominates (lower normalized entropy). Entropy-based weighting assigns substantially different weights (subduction: 0.7830, crustal: 0.2158, intraslab: 0.0012) compared to the equal-weighting assumption (0.333 each), resulting in higher hazard estimates at moderate-to-high PGA levels. This study demonstrates that normalized entropy provides a quantitative, intensity-dependent indicator for evaluating epistemic uncertainty distribution within PSHA logic trees, with the specific empirical finding that model dominance in Timor-Leste is intensity-dependent and predominantly controlled by the subduction source.