Title: From the Cosmos to the Canopy: Quantifying Uncertainty Across Cosmological and Environmental Systems

Program: Computing PhD

Committee Chair: Nancy F. Glenn

Committee: Nancy F. Glenn, Alejandro Flores, Jodi Mead

Abstract: In this dissertation, we present three data-driven projects that quantify uncertainty from the cosmological to the watershed scale. First, we use cosmological simulations and mock optical cluster catalogs (MiniUchuu and Cardinal) to show that projection-induced selection effects can bias weak-lensing mass calibration of galaxy clusters high by roughly 20–50%, and we outline how this systematic must be modeled and mitigated in future optical cluster cosmology. Second, we integrate NASA's GEDI and ICESat-2 spaceborne lidar with airborne lidar and ensemble machine-learning models to estimate vegetation height in dryland, complex terrain at Reynolds Creek Experimental Watershed, and quantify detection uncertainties. This presents an opportunity to assess implications for carbon monitoring and upcoming missions such as EDGE. Third, we develop a physically based, two-stage U-Net downscaling framework that fuses GOES-R, MODIS, and Sentinel-2 to generate 20 m-resolution, high-spatiotemporal blue-sky albedo maps over the Upper Colorado River Basin, improving our ability to resolve mountain snowmelt processes and water-resource variability. Together, these projects demonstrate how combining simulations, remote sensing, and modern machine learning can reduce key sources of uncertainty in both cosmology and environmental predictions.