Advancing building data models for the automation of high-fidelity regional loss estimations using open data

Abstract

Hurricanes cause significant building damages, whose losses are aggregated by existing hurricane regional loss estimation frameworks, compromising model granularity and fidelity. The need to reduce mounting losses through targeted mitigation investments instead demands tools that enable site-specific, building-specific, and component-level loss estimations. Delivering such granularity in high-fidelity loss modeling creates new challenges in efficiently assembling and managing spatial and geometric data of thousands of constructed buildings. In response, this paper offers two interrelated contributions: (1) a conceptual data model that leverages open data for efficient integration and querying of component- to site-level information across the loss estimation workflow and (2) the implementation of this data model in Python to tractably generate building models within existing open-source loss modeling workflows. The accompanying case study and scenario-based verifications demonstrate how the data model meets stated design requirements in its generation of building models.