Date of Award

12-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

School of Mathematical and Statistical Sciences

Committee Chair/Advisor

Whitney Huang

Committee Member

Andrew Brown

Committee Member

Brook Russell

Committee Member

Qiong Zhang

Abstract

Uncertainty is an inherent feature of scientific modeling and decision-making, necessitating rigorous quantification to ensure reliable inference and robust applications. This dissertation addresses uncertainty quantification (UQ) across three distinct yet interconnected projects, each illustrating different facets and challenges of UQ. The first project focuses on estimating r-year return levels for storm surges caused by hurricanes, where UQ is critical to propagate uncertainty from hurricane characteristics through complex input-output relationships and extreme value statistical modeling. The second project develops dynamic surrogate models to improve sequential decision-making in autonomous vehicle control, capturing temporal dependencies using methodologies commonly applied in multifidelity contexts to enhance online optimization performance under uncertainty. The third project investigates multivariate bias correction for climate model outputs, aiming to correct future projections by leveraging past model and observational data while preserving essential temporal and inter-variable structures. Beyond traditional statistics, this work emphasizes comprehensive modeling of uncertainty sources, including model discrepancy and distributional shifts, and employs Gaussian Processes and conditional modeling to robustly propagate uncertainty. Together, these projects demonstrate innovative methodologies for UQ in environmental risk assessment, autonomous systems, and climate science, providing generalizable frameworks that enhance decision-making reliability under uncertainty.

Author ORCID Identifier

0000-0001-8380-911X

Available for download on Thursday, December 31, 2026

Share

COinS