Research

We use and develop physically-based hydrologic models to investigate water-related issues at the macro scale in the context of coupled human and natural systems.

Our modeling work spans process-based simulation and hybrid, data-driven approaches — from edge-of-field and watershed runoff prediction to nutrient loading and coastal flooding. We combine these models with machine learning to improve fine-scale prediction while preserving physical interpretability.

From individual behavior to emergent phenomena at large scale, agent-based modeling (ABM) is used to bridge the gap, studying the interactions between human and water systems.

We derive agents' behavioral rules by combining human and machine intelligence, and couple ABMs with hydrologic and groundwater models to study decisions such as crop choice and irrigation — and the system-scale consequences they produce.

Cloud-based algorithms and infrastructure address the computational issues arising from the coupled agent-based models and water models.

We build web-based applications and high-performance, cloud-native workflows — including work on distributed frameworks such as Hadoop — that make large-scale socio-hydrological modeling tractable, reproducible, and shareable.

Coupled agent-based models and water models are used to assist water policy making, addressing threats to water security — such as groundwater depletion and pollution, seawater intrusion, and urban flooding.

By linking human decision-making with hydrologic response, our work helps evaluate how policies and interventions play out across coupled human–water systems under a changing environment.

Coastal communities increasingly face compound flooding, where storm surge, tides, intense rainfall, river discharge, and groundwater interact to overwhelm drainage and defenses in ways no single driver would predict. We study these coupled mechanisms to make coastal flood risk easier to anticipate and act on.

Our work spans street-to-pipe diagnosis of combined sewer systems, high-resolution "sunny-day" and nuisance-flood digital twins, coupled 1D/2D hydrodynamic modeling, wave and skew-surge analysis, and tidal coastal-groundwater modeling. We pair physics-based simulation (WRF-Hydro, SCHISM, FUNWAVE-TVD, EPA-SWMM) with machine-learning surrogates to deliver fast, interpretable predictions that support resilient coastal infrastructure.