Date of Award

Spring 1-1-2018

Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Waleed Abdalati

Second Advisor

Barbara Buttenfield

Third Advisor

Weiqing Han

Fourth Advisor

Stefan Leyk

Fifth Advisor

Seth Spielman


Digital elevation models (DEMs) are critical components of coastal flood models. Both present-day storm surge models and future flood risk models require these representations of the Earth’s elevation surface to delineate potentially flooded areas. The National Oceanic and Atmospheric Administration (NOAA) National Centers for Environmental Information (NCEI) develops DEMs for United States’ coastal communities by seamlessly integrating bathymetric and topographic data sets of disparate age, quality, and measurement density. A current limitation of the NOAA NCEI DEMs is the accompanying non-spatial metadata, which only provide estimates of the measurement uncertainty of each data set utilized in the development of the DEM.

Vertical errors in coastal DEMs are deviations in elevation values from the actual seabed or land surface, and originate from numerous sources, including the elevation measurements, as well as the datum transformation that converts measurements to a common vertical reference system, spatial resolution of the DEM, and interpolative gridding technique that estimates elevations in areas unconstrained by measurements. The magnitude and spatial distribution of vertical errors are typically unknown, and estimations of DEM uncertainty are a statistical assessment of the likely magnitude of these errors. Estimating DEM uncertainty is important because the uncertainty decreases the reliability of coastal flood models utilized in risk assessments.

I develop methods to estimate the DEM cell-level uncertainty that originates from these numerous sources, most notably, the DEM spatial resolution, to advance the current practice of non-spatial metadata with NOAA NCEI DEMs. I then incorporate the estimated DEM cell-level uncertainty, as well as the uncertainty of storm surge models and future sea-level rise projections, in a future flood risk assessment for the Tottenville neighborhood of New York City to demonstrate the importance of considering DEM uncertainty in coastal flood models. I generate statistical products from a 500-member Monte Carlo ensemble that incorporates these main sources of uncertainty to more reliably assess the future flood risk. The future flood risk assessment can, in turn, aid mitigation efforts to reduce the vulnerability of coastal populations, property, and infrastructure to future coastal flooding.