Investigating Wetland Ecosystem Dynamics Through Modeling Across Scales
Abstract
Wetlands are complex ecosystems that provide a wide range of ecological services, including water quality improvement, flood control, erosion reduction, carbon sequestration, and habitat provision. Although they cover a small fraction of the Earth’s land surface, wetlands serve as effective nature-based solutions that offer substantial benefits to both human and environmental systems. This dissertation advances the understanding of wetland processes through two complementary approaches: the development of process-based numerical models for simulating hydrology and water quality in constructed wetlands (CWs), and the application of data-driven methods to investigate carbon dynamics in natural wetlands across the United States. The first part focuses on modeling hydrological and biogeochemical processes in CWs by introducing a novel cell-averaged subsurface flow model capable of capturing wetland-scale hydrodynamics, which is then extended to simulate water quality through modules for suspended sediment transport, nutrient cycling, and primary productivity. The second part explores the spatial and temporal drivers of carbon fluxes in natural wetlands using a national-scale database of greenhouse gas measurements and environmental predictors. Machine learning models are developed and evaluated to predict gross primary production (GPP), ecosystem respiration (RECO), net ecosystem exchange (NEE), and methane (CH₄) emissions, and are integrated into an ensemble framework to spatially extrapolate carbon dynamics across the Southeastern United States. Collectively, this work contributes to the advancement of both mechanistic and statistical modeling approaches for wetlands, providing valuable tools for ecosystem management, climate change mitigation, and the design of sustainable water infrastructure.