Large-scale cloud-based building elevation data extraction and flood insurance estimation to support floodplain management

Abstract

Factors that often make coastal areas prone to high-cost flood damage include its low-lying topography, tropical and subtropical climates, miles of coastline exposed to hurricane and storm surge hazards, and older housing stock. In this study, we sought to develop a platform that can be used by various stakeholders to re-evaluate and reframe community risk through the use of mobile LiDAR technology to create a current 3D digital elevation model of all homes, businesses, and infrastructure in shoreline communities within the current 100-year floodplain in one county in New Jersey. An innovative strategy for extracting property elevation information from mobile LiDAR and mobile imagery was developed utilizing a web-based approach. Unlike traditional elevation extraction approaches that require field surveys, the proposed method does not require an onsite inspection from certified surveyors, which is often computationally expensive but the key method to current elevation certificate generation. Thus, the proposed approach makes it possible to obtain accurate structure elevation and realize the elevation extraction for large-scale community areas. Results indicate that overall, 96.5% of building diagrams identified with our method have the same value as the original elevation certificate documents. For the estimation of elevation certificate C2a elevation, the estimated error is between 0.082 ft (0.025 m) and 0.104 ft (0.032 m) at the 95% confidence level. Similarly, the C2b elevation had an estimated error between 0.133 ft (0.040 m) and 0.136 ft (0.041 m) at the 95% confidence level.