Comparison of sampling methods for post-hurricane damage survey

In the United States, residents and businesses in the Atlantic and Gulf Coastal regions are confronted with an increasing risk of hurricane-induced damages. It is often too costly and labor intensive to survey every structure within a large area affected by a major storm. Therefore, this paper is aimed at proposing and evaluating three sampling methods specifically designed for posthurricane damage survey: Simple Random Sampling, Equal Spatial Sampling, and Route Based Sampling. After describing their general formation, a case study is presented in which these three sampling methods are applied to 1,020 residential houses affected by Hurricane Katrina in 2005, allowing for systematic and objective selection of samples for ground inspection. Then the distributions of damage conditions of samples are compared with that of population. The result shows that the sample sets selected by the Equal Spatial Sampling and Route Based Sampling methods reasonably represent the full data set with a few exceptions in high damage categories while the performance of the Simple Random Sampling is far less consistent. This work could serve as a guidance for designing a balanced sampling scheme with the consideration of expected performance and available resources in future post-disaster damage surveys.