Comprehensive damage assessment and analysis of damage mechanisms from Hurricane Harvey
Description
TitleComprehensive damage assessment and analysis of damage mechanisms from Hurricane Harvey
Date Created2019
Other Date2019-05 (degree)
Extent1 online resource (xiv, 127 pages) : illustrations
DescriptionA combination of mobile data collection and new damage assessment methods with spatial analysis and machine learning algorithms were used to correlate structural characteristics with damage and iterate upon damage assessment protocols for further development. More specifically, data was collected using a mobile scanning vehicle, reducing volunteer exposure to the harsh post-disaster environment, collecting high volumes of panoramic and LiDAR imagery in a relatively short period of time. This new data collection method was deployed in Texas during Hurricane Harvey. Among many datasets collected by this method, the dataset used in this study consisted of almost purely wind-caused damage from Hurricane Harvey to 553 homes in southeast Texas. A damage assessment methodology was created, combining lessons learned and protocols from previous studies, to increase efficiency and include more external public sources of data for better damage analysis. Statistical analysis was combined with spatial analysis revealing structural components which can be expected to reduce or increase damage from single-hazard wind damage. Spatial analysis indicated that damage rating was related to peak wind speed and explanatory regression revealed that the most significant variables to classification were: Age, Latitude, Metal Roofs, Distance to Coast, Total Area, Asphalt Roofs, Wood Siding, Stucco Siding, Two Story Buildings, and Building Value. Machine Learning classifiers were used improve the efficiency of damage assessments by indicating the multicollinearity and the feature importance of each variable. The variables with the highest feature importance include: Distance to Coast, Longitude, Single-Family, Age, Total Area, Wind Speed, and Single Story. These variables should be prioritized in future studies, while variables with low feature importance, such as Grade Level Entry, Intersecting or Overlapping Roofs, 10/12 Roof Pitch, Commercial uses, and Vinyl Siding, should be reconsidered in future damage assessments.
NoteM.S.
NoteIncludes bibliographical references
Genretheses, ETD graduate
LanguageEnglish
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Statistical Profile