Get started
Work

A Probabilistic Catastrophe Risk Model for Site-Specific Tropical Cyclone Loss Quantification

Year: 2026
Type: preprint
Abstract: Quantifying the financial risk posed by landfalling tropical cyclones requires statistical methods that extend beyond the limits of a short and non-stationary historical observational record. This paper presents the mathematical formulation of an open, modular probabilistic catastrophe risk model for site-specific tropical cyclone loss estimation, structured around the four components used throughout the catastrophe modeling industry: hazard, exposure, vulnerability, and financial loss. Storm intensities drawn from NOAA's HURDAT2 database over the satellite era (1970--2025) are fit to a continuous Weibull distribution via maximum likelihood estimation, from which a 10,000-year synthetic event catalog is generated and propagated to individual properties using a Holland-type wind-field decay model. The hazard catalog is evaluated against a synthetic Industry Exposure Database of 10,000 properties (total insured value of \$39.01 billion) distributed across six high-hazard U.S. Gulf and Atlantic coast counties, allocated using NOAA Office for Coastal Management county-level economic indices. Physical damage is estimated via construction-class-specific log-normal vulnerability functions, and resulting ground-up losses are passed through standard insurance contract terms (deductibles and policy limits) to obtain net insured liabilities. The resulting annualized loss distribution yields insured loss thresholds of \$124,968, \$138.3 million, \$381.7 million, and \$1.87 billion at the 1-in-10, 1-in-50, 1-in-100, and 1-in-500 year return periods, respectively. We discuss the model's relationship to existing open-source and commercial catastrophe modeling platforms, the limitations introduced by its stationarity assumptions and synthetic exposure base, and the implications of a non-stationary, warming climate for long-tail insured loss estimation.
DOI: 10.31223/x5dv48
Language: English
Open Access status: green
OpenAlex ID: W7214149087
Cites:
Cited by:
Related to: 10