Loading...
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 58743 (09)
- University: Sharif University of Technology
- Department: Civil Engineering
- Advisor(s): Toufigh, Vahab; Mofid, Masood; Haj Kazem Kashani, Hamed
- Abstract:
- Reliable estimation of earthquake fatalities is essential for effective disaster preparedness and risk mitigation. This study introduces a novel probabilistic framework that integrates Bayesian linear regression with reliability analysis to quantify earthquake fatality risk. Using a global dataset of 83 destructive earthquakes (1980–2018), fatalities are modeled as a normalized fatality rate relative to a peak ground acceleration (PGA)-weighted exposed population, with three predictor categories: (i) hazard intensity measures such as PGA and focal depth; (ii) socioeconomic resilience indicators including GDP per capita and income inequality; and (iii) community preparedness factors such as time of occurrence and prior seismic experience. The Bayesian formulation rigorously characterizes epistemic uncertainty, provides credible-interval-based sensitivity assessments, and enables continuous model updating as new data become available. Monte Carlo simulation produces fatality fragility functions and curves showing the probability of exceeding specified fatality thresholds across a range of earthquake intensities. Results indicate that PGA is a more reliable predictor of fatalities than magnitude, and that interactions among economic capacity, inequality, and historical experience strongly shape community vulnerability. The framework offers a generalizable, transparent basis for risk quantification aligned with probabilistic safety analysis, and supports classification of countries into policy-relevant risk groups to strengthen evidence-based mitigation planning
- Keywords:
- Bayesian Regression Modeling ; Monte Carlo Sampling ; Probabilistic Methods ; Earthquake Fatality Estimation ; Socioeconomic Resilience Indicators ; Risk Analysis ; Reliability
-
محتواي کتاب
- view
