Doctor of Philosophy in Ocean Science & Engineering

In the

School of Earth and Atmospheric Sciences

Shivaprakash Muruganandham

Will defend his dissertation

Exploring Sea Level Futures: From Antarctic Ice Loss to Coastal Flood Adaptation

16, September 2026

12:30 PM

Ford ES&T 3235 (Ocean Room)

https://gatech.zoom.us/j/94955892751

 Thesis Advisor:

Alexander Robel, Ph.D.

School of Earth and Atmospheric Sciences

Georgia Institute of Technology

Committee Members:

Winnie Chu, Ph.D.

School of Earth and Atmospheric Sciences

Georgia Institute of Technology

Joseph Montoya, Ph.D.

School of Biological Sciences

Georgia Institute of Technology

Ali Sarhadi, Ph.D.

School of Earth and Atmospheric Sciences

Georgia Institute of Technology

Kevin Haas, Ph.D.

School of Civil and Environmental Engineering

Georgia Institute of Technology

ABSTRACT:

Global mean sea level rose throughout the twentieth century and has accelerated in recent decades, driven in part by thermal expansion of the warming ocean and mass loss from glaciers and ice sheets. Continued sea-level rise is expected to increase the risk of coastal flooding, yet the future contribution of the Antarctic Ice Sheet to sea-level projections remains a significant source of uncertainty in sea-level projections. Coastal adaptation planning requires projections that connect future sea level to local flood conditions and the physical effects of potential adaptation measures over relevant policy and planning horizons. Against this background, this dissertation examines physical responses at two ends of the sea-level problem by asking: (1) How does spatiotemporal variability in ice-shelf basal melt propagate through Antarctic ice-sheet dynamics, and how much uncertainty does it contribute to sea level projections? (2) How do coastal communities adapt when sea-level rise, storm surge, tide, and rainfall interact with modifications to the coastal landscape during a tropical cyclone?

Ocean variability changes melting at the base of Antarctic ice shelves. I generate basal-melt histories by applying empirical orthogonal function decomposition and Fourier phase randomization on an MPAS-Ocean simulation that resolves circulation beneath ice shelves, then propagate them through the MPAS-Albany Land Ice model. Across the 300-year ice-sheet ensembles, the prescribed mean melt pathway controls the trajectory of the Antarctic sea-level contribution, but the relative importance of variability depends on the time horizon of interest. During the first century, this divergence among ensemble members can be substantial relative to the weak cumulative ice sheet response; however, by year 300, alternative realizations produce an ensemble spread that is two to three orders of magnitude smaller than the corresponding mean sea-level contribution. Most spread develops in regions where the background melting is strong, and grounding lines are retreating.

Along coastlines, sea-level rise raises the background water level on which storms act to cause flooding. A coupled hydrodynamic-hydrologic flood modeling framework comprising GeoClaw and LISFLOOD-FP simulates compound flooding during Hurricane Matthew in Chatham County, Georgia, under different sea-level and landscape-intervention scenarios. For the modeled event, the combined coastal and rainfall forcing produces peak depths greater than those from either driver alone. Adaptation responses vary spatially: floodwalls can reduce flooding locally while increasing it elsewhere, and surface roughness effects depend on the extent of treatment, inundation depth, and sea level. Within the sampled Hurricane Matthew design, a machine-learning surrogate (U-Net) reproduces flood fields with mean errors near 0.01 m over inundated land cells, including scenarios containing intervention types withheld from training, targeted siting choices, and two-intervention portfolios, enabling rapid exploration of sea-level and adaptation scenarios.

Across both projects, imposed climate or weather forcing sets the broad response, while geometry and physical state shape its local magnitude and spatial pattern. Emulation makes broader sampling computationally tractable by generating basal-melt histories and evaluating coastal flood scenarios, allowing these spatially uneven physical responses to be examined directly.