Charles K. Huyck

Improving Catastrophe Modeling for Business Interruption Insurance Needs.

While catastrophe (CAT) modeling of property damage is well developed, modeling of business interruption (BI) lags far behind. One reason is the crude nature of functional relationships in CAT models that translate property damage into BI. Another is that estimating BI losses is more complicated because it depends greatly on public and private decisions during recovery with respect to resilience tactics that dampen losses by using remaining resources more efficiently to maintain business function and to recover more quickly. This article proposes a framework for improving hazard loss estimation for BI insurance needs. Improved data collection that allows for analysis at the level of individual facilities within a company can improve matching the facilities with the effectiveness of individual forms of resilience, such as accessing inventories, relocating operations, and accelerating repair, and can therefore improve estimation accuracy. We then illustrate the difference this can make in a case study example of losses from a hurricane.

A Comprehensive Analysis of Building Damage in the 2010 Haiti Earthquake Using High-Resolution Imagery and Crowdsourcing

This chapter provides a detailed account of how technology, inspiration and collaboration were used to rapidly assess damage caused by the devastating January 12, 2010 Haiti earthquake. This was one of the first events where remote sensing technology (especially high spatial resolution imagery) was embraced in a truly operational sense to support post-disaster recovery planning. Sub-meter satellite imagery was available the day following the earthquake, and provided the first glimpse of the destruction caused by the earthquake. Days later, finer spatial resolution aerial imagery became available and provided even more detail on building damage. Together, these datasets allowed over 600 remote sensing experts and engineers to generate one of the most comprehensive assessments of earthquake building damage in the last decade. Furthermore, this information was shared with Haitian government in the form of a Building Damage Assessment Report in support of the Post-Disaster Needs Assessment (PDNA) and Recovery Framework.

Utilizing New Technologies in Managing Hazards and Disasters

This chapter introduces new and emerging technologies that have proven effective in disaster management or show promise in future deployments. These technologies are discussed in the context of the four major phases of disaster management: preparedness, response, recovery and mitigation. Examples of some technologies discussed in detail include real-time hazard warning or monitoring systems; advanced loss estimation methodologies and tools; remote sensing for response and recovery; and field data collection and visualization systems, especially those that are GIS and/or GPS-based. The chapter concludes with a brief discussion of research or implementation issues, focusing specifically on the above technologies, and including issues related to real-time event monitoring; privacy protection; and information sharing and trust management.

Sample Treatment of Uncertainties in Earthquake Portfolio Risk Analysis

Previous papers and presentations by the authors have proposed that robust simulation should be used to define uncertainties in catastrophe risk analyses for portfolios and/or systems. Robust simulation begins with a “preferred” comprehensive model that turns out to be comprised of non-unique solutions for many technical issues. Uncertainties in this “preferred” model can be estimated either endogenously, that is, through alternative distributions used in the simulation process, or exogenously, through alternative comprehensive model simulations. This paper elucidates how these uncertainties are estimated through the examination on the one hand of available earthquake hazard models and (e.g., GMPE, kinematic) and selected uncertainties (e.g., directivity, focal depth), and on the other hand of available building vulnerability models (statistical, opinion-based, engineering) and selected uncertainties (e.g., structural period, strength, ductility). The goal of this paper is thus to define more clearly what count as “alternative credible models” and how they may be used to estimate uncertainties in the resulting portfolio loss distributions. BACKGROUND In recent years, the overarching question “How does one account for uncertainties in catastrophe risk analysis?” has become more prominent. Within traditions of statistical and probability theory, the narrow tradition of using the distinction between “epistemic” and “aleatory” uncertainty has been demonstrated to yield considerable incoherence. Endogenous or nominal uncertainties account for uncertainties given the models used, but not those uncertainties resulting from the use of alternative models (parameters, data, or assumptions).