Improving Earthquake Disaster Models with Post-Event Data: Insights from the 2015 Gorkha, Nepal Earthquake

Immense amounts of data are collected following earthquake disasters. Yet, it remains unclear how researchers’ might take full advantage of diverse post-disaster datasets. Using data from the 2015 Gorkha Nepal earthquake, this dissertation explores three ways in which post- disaster survey and assessment datasets can be used to inform models of seismic risk, vulnerability, and recovery processes. The first article presents an empirical analysis of scale issues in disaster vulnerability indices using a novel dataset of 750,000 households. This study finds that using aggregated household data to create social vulnerability indices can produce results that are meaningfully different from equivalent indices produced directly with household-level data. These results inform future development of vulnerability indices. The second article develops a Bayesian item-response theory modeling framework for estimating household-level reconstruction behavior from reconstruction progress surveys. This study provides a new way to quantitatively assess earthquake recovery, with results showing large differences in reconstruction probabilities among different levels of aid receipt, household willingness to commit additional resources, and geographic location. The final article uses engineering damage assessment data to develop a model for spatially interpolating geolocated clusters of rapid damage assessments onto a high-resolution grid. Incorporating ground truthed data significantly improves existing rapid estimates for completely damaged buildings and is feasible with the current scope of rapid damage assessment collection. Together, these contributions cast a vision for an improved disaster modeling ecosystem that more effectively integrates novel post-disaster data streams