UC Santa Cruz

This dissertation contributes to the pressing need to improve the connections between quantitative scientific studies and policy in the context of human vulnerability to costal hazards, at multiple spatial and temporal scales. This dissertation integrates diverse tools and methods (including GIS, Machine Learning Clustering Techniques, and spatial indices), to assess risks and vulnerabilities of coastal communities to select natural hazards, at multiple geographies and scales (i.e. California, Florida, and Latin America and Caribbean). Statistically sound methods were applied to integrate data from multiple disciplines, including: natural hazards, geographical distribution of natural habitats, population, and assets, as well as socioeconomic vulnerability, providing insights to adaptation alternatives.

The first chapter of this dissertation demonstrates that flood losses in California could be mitigated through action that meets both flood risk reduction and conservation objectives. The study demonstrates that government funded buyouts, followed by restoration of targeted lands, can support social, environmental, and economic objectives: reduction of flood exposure, restoration of natural resources, and efficient use of limited governmental funds.

In the second chapter of this dissertation, a revised and improved version of the model developed in chapter 1 is applied to the state of Florida. In addition to flood exposure and natural habitats, social vulnerability was also included in the prioritization scheme. Further, inland habitats were also included, expanding the focus of the analysis beyond just the coast. Results identified lands in Florida that are eligible to receive federal funds to attain multiple benefits: (i) reduce flood risk to home owners; (ii) reduce FEMA’s financial burden (from future flood claim payments); (iii) restore/protect natural habitats; (iv) remediate social vulnerability, and (v), identify potential sources of funding for projects. There were at least 10,000 km2 of land in Florida where such objectives may be achieved simultaneously. In a targeted case-study our model identified 92 RLPs in Miami-Dade located in areas of high social vulnerability, high flood exposure, and where natural habitats coexist. Collectively, these 92 RLPs filed 207 claims against NFIP from 1978 to 2011.

In the third chapter, I employed a combination of machine learning clustering techniques (Self Organizing Maps and K-Means algorithms) and a spatial index (GIS), to assess coastal risks in Latin America and the Caribbean (LAC) on a comparative scale. The third study meets multiple objectives, including the identification of hotspots and key drivers of coastal risk, and the ability to process large-volume multidimensional and multivariate datasets - effectively reducing sixteen variables related to coastal hazards, geographic exposure, and socioeconomic vulnerability, into a single index.