Estimating first-floor elevation for building flooding risk assessment using street view image
Topics:
Keywords: street view, image, first-floor elevation, vertical dimension, deep learning
Abstract Type: Virtual Paper Abstract
Authors:
Huan Ning, Geoinformation and Big Data Research Laboratory, Department of Geography, University of South Carolina, Columbia, USA
Zhenlong Li, Geoinformation and Big Data Research Laboratory, Department of Geography, University of South Carolina, Columbia, USA
Xinyue Ye, Department of Landscape Architecture and Urban Planning, College Station, Texas A&M University, USA
Shaohua Wang, Department of Informatics, New Jersey Institute of Technology, Newark, USA
Wenbo Wang, Department of Informatics, New Jersey Institute of Technology, Newark, USA
Xiao Huang, Department of Geosciences, University of Arkansas, Fayetteville, Arkansas, USA.
,
,
,
,
Abstract
First-floor elevation (FFE) indicates the elevation of the main entrance of a building. It can indicate the inundating risk given the water height of a flood. An FFE under the base flood elevation (the elevation of 1% chance of flood) in a flood zone might tend to be inundated, so FFE is favorable for house flood inundation early warning and damage assessment. However, FFE datasets are rarely available for hazard management because those records are not digitalized and updated timely. Our study used street view imagery (SVI) to estimate FEE. The front door bottom is adopted as an FFE proxy; then, we trained an object detector and developed algorithms to identify house doors in SVI. The known height of the door (2 meters in the U.S.) is viewed as a subtense bar, and the apex angle can be measured from SVI; hence the vertical coordinate of the door bottom can be computed based on the principle of tacheometric surveying. The results suggest that the average error of the estimated FFE is 0.218 m. The proposed method provides a novel approach for FFE estimation; it can be extended for automatic verticle measurement from SVI and is expected to benefit future SVI-related studies.
Estimating first-floor elevation for building flooding risk assessment using street view image
Category
Virtual Paper Abstract
