Deriving First Floor Elevations within Residential Communities Located in Galveston Using UAS Based Data

Abstract

Flood damages occur when just one inch of water enters a residential household and models of flood damage estimation are sensitive to first-floor elevation (FFE). The current sources for FFEs consist of costly survey-based elevation certificates (ECs) or assumptions based on year built, foundation type, and flood zone. We sought to address these limitations by establishing the role of an Unmanned Aerial System (UAS) to efficiently derive accurate FFEs. Four residential communities within Galveston Island, Texas were selected to assess efficient flight parameters required for UAS photogrammetry within the built environment. A real-time kinematic positioning enabled (RTK) UAS was then used to gather georeferenced aerial imagery and create detailed 3D photogrammetric models with ±0.02 m horizontal and ±0.05 m vertical accuracies. From these residential models, FFEs and other structural measurements present in traditional ECs were obtained. Comparative statistical analyses were performed using the UAS-based measurements and traditional EC measurements. UAS based FFE measurements achieved 0.16 m mean absolute error (MAE) across all comparative observations and were not statistically different from traditional EC measures. We conclude the RTK enabled UAS approach is an efficient, cost-effective method in establishing accurate FFEs and other flood-sensitive measures in residential communities.

Exploring flood mitigation governance by estimating first-floor elevation via deep learning and google street view in coastal Texas

“…These neighborhoods have FEMA’s E.C. data and reliable FFE estimates (Diaz et al, 2022) for validation. The samples showed that our workflow is well applicable to Galveston, so we applied the method to all buildings recorded in Galveston Island’s building footprint data.…”

“…Different from traditional field surveys that rely on manual measurements, our measurement from SVI based on computer vision technology is capable of processing enormous amounts of data. Contributing to the generalizability of our research, the initial data for the proposed workflow can be derived either from the building footprints or a single list of addresses of the target area; for instance, the dataset Galveston_FFEData2020 (Diaz et al, 2022), which is used for accuracy validation in this research, identifies unique houses using only street addresses.…”

“…In this study, we compare the FFE detected via SVIs with the FFE values from two sources: a set of City of Galveston approved E.C.s and the Galveston_FFEData2020 dataset (Diaz et al, 2022). For a detailed description of these comparison datasets, please refer to Section F of the Supplementary Materials.…”

3D visualization of hurricane storm surge impact on urban infrastructure

“…The coordinates are the basis to download the street view images (SVIs) from Google Maps as well as the depth maps corresponding with each SVI. Considering the generalizability of our research, the initial data of the proposed workflow can either start from the building footprints or a single list of addresses of the target area (Diaz et al, 2022). For instance, the dataset of Galveston used for accuracy validation only contains addresses but not the coordinate position as the major label for data.…”

Automated first floor height estimation for flood vulnerability analysis using deep learning and Google Street View

“…Also, the developed RF model was based on the information derived from EC that is unavailable for most flood‐prone areas worldwide, and its preparation is time‐consuming and a tedious task. Diaz et al (2022) used a positioning‐enabled unmanned aerial system (UAS) and created a detailed 3D photogrammetric model to estimate the FFE of the buildings in Galveston Island, Texas. Although the reported mean absolute error was promising (0.16 m), the method is costly and does not apply to large areas.…”