Multi-Hazard Bayesian Hierarchical Model for Damage Prediction

TL;DR

This paper introduces a probabilistic multi-hazard damage prediction model that captures hazard interactions and outperforms traditional deterministic models in accuracy, demonstrated through cyclone damage data from the Philippines.

Contribution

The work develops the first fully probabilistic hierarchical model for multi-hazard damage prediction, addressing limitations of fixed parameters and hazard independence assumptions in existing models.

Findings

Reduced damage prediction error by 61% compared to single-hazard models

Achieved 80% error reduction over deterministic benchmarks

Improved damage estimation accuracy by USD 0.8 billion to USD 2 billion

Abstract

A fundamental theoretical limitation undermines current disaster risk models: existing approaches suffer from two critical constraints. First, conventional damage prediction models remain predominantly deterministic, relying on fixed parameters established through expert judgment rather than learned from data. Second, probabilistic frameworks are fundamentally restricted by their underlying assumption of hazard independence, which directly contradicts the observed reality of cascading and compound disasters. By relying on fixed expert parameters and treating hazards as independent phenomena, these models dangerously misrepresent the true risk landscape. This work addresses this challenge by developing the Multi-Hazard Bayesian Hierarchical Model (MH-BHM), which reconceptualizes the classical risk equation beyond its deterministic origins. The model's core theoretical contribution lies in reformulating a classical risk formula as a fully probabilistic model that naturally accommodates hazard interactions through its hierarchical structure while preserving the traditional hazard-exposure-vulnerability framework. Using tropical cyclone damage data (1952-2020) from the Philippines as a test case, with out-of-sample validation on recent events (2020-2022), the model demonstrates significant empirical advantages. Key findings include a reduction in damage prediction error by 61% compared to a single-hazard model, and 80% compared to a benchmark deterministic model. This corresponds to an improvement in damage estimation accuracy of USD 0.8 billion and USD 2 billion, respectively. The improved accuracy enables more effective disaster risk management across multiple domains, from optimized insurance pricing and national resource allocation to local adaptation strategies, fundamentally improving society's capacity to prepare for and respond to disasters.