Research Areas
Our research investigates how human activities reshape freshwater systems and translates that knowledge into actionable solutions for water sustainability. We combine cutting-edge methods with stakeholder engagement to protect and restore water resources.
Water Quality and Nutrient Legacies
Cleaner water often takes decades longer to achieve than expected, even with extensive implementation of conservation measures. One reason for these time lags is the existence of "nutrient legacies" in intensively farmed areas. These legacies are long-term stores of nitrogen and phosphorus in soils and groundwater that keep "leaking" into rivers even after we improve management practices.
We use isotopic analysis, tracer studies, and national-scale water-quality modeling to quantify nutrient legacies. Our work shows that these legacies can delay recovery for decades, and we provide science-based recommendations to help managers set realistic timelines and design more effective strategies. Through field studies across agricultural landscapes, we have demonstrated that nutrient legacies represent one of the most significant barriers to achieving water quality goals. Our innovative approaches including isotopic tracers, groundwater age dating, and machine learning help quantify legacy stores and predict their release patterns. This research has direct implications for policy design, helping managers understand why conservation benefits take 10-30 years to fully materialize in many watersheds.
Highlights:
- Legacy nitrogen in groundwater can persist 20-30 years.
- Legacy phosphorus in soils and sediments can continue to fuel nutrient pollution for decades after upstream source reductions.
- Legacy effects explain 40-60% of current nutrient loading in many watersheds
- Policy timelines must account for legacy delays to be effective
Wetland Function, Mapping, and Restoration in Human-Impacted Landscapes
Wetlands are natural water filters and flood protectors — but more than half have been lost in the United States, with some regions losing over 90%. Our research shows how these losses degrade water quality and identifies the most effective places to restore wetlands for maximum impact.
Using remote sensing, historical records, and field studies, we map where wetlands once existed, track how their loss has altered nutrient cycles, and pinpoint where restoration would do the most good. We also measure how restored wetlands function over time, showing how quickly they regain the ability to filter pollution and buffer floods.
Highlights:
- Small wetlands, outsized benefits: Small and scattered wetlands remove nutrients at disproportionately high rates, often contributing as much nitrogen removal as larger systems.
- Targeted restoration works: Adding wetlands in nutrient "hotspots" can cut downstream nitrate loads by nearly half, a powerful tool for tackling dead zones.
- Loss is not random: Preferential drainage of small, upland wetlands has simplified wetland networks, reducing biodiversity, groundwater recharge, and nutrient retention.
- Architecture matters: The size, shape, and placement of wetlands across the landscape strongly influence their ability to filter water and provide ecosystem services.
Continental-Scale Nutrient Inventories
Our team has developed the most comprehensive nutrient budgets for the United States, tracking nitrogen and phosphorus flows from farm to watershed to coast. The TREND (Tracking Reactive Nitrogen to Environmental Systems with Data) project provides county-level estimates of nutrient inputs, outputs, and surpluses from 1930 to present, revealing how agricultural intensification has created regional hotspots of nutrient pollution that correspond with water quality problems in major river systems and coastal zones.
Highlights:
- US nitrogen inputs increased 10-fold from 1930-2000, now stabilizing
- Phosphorus use efficiency has improved 40% since 1980
- Regional nutrient imbalances drive 80% of water quality problems
- Historical data essential for understanding current environmental conditions
Next-Generation Modeling: Bridging Machine Learning & Process Understanding
Water quality challenges demand models that can capture complex realities while staying grounded in physical processes. Our lab develops hybrid modeling approaches that merge machine learning with biogeochemical knowledge, delivering predictions that are both accurate and interpretable. From random forests models to predict nutrient transport across the Great Lakes Watersheds to deep learning models predicting groundwater nitrogen legacies in the Upper Mississippi, our work shows how data-driven tools can uncover hidden patterns and improve our understanding of water quality trajectories. We are also advancing LSTM models that provide high-frequency predictions of streamflow, stream temperature, and nutrient loads in places like the Chesapeake Bay, which is critical for management decisions under changing climate and land use.
A core part of this research is interpretability: we use feature importance metrics, partial dependence plots, and SHAP values to explain how models work and to ensure predictions align with physical understanding. And through the Chesapeake Summer Water Institute and graduate training, we prepare early-career scientists to lead the next wave of innovation in water quality modeling.
Highlights:
- Machine learning improves prediction of nutrient concentrations and transport across large, complex watersheds.
- Random forests and other ensemble methods uncover key drivers of nutrient dynamics (e.g., land use, soils, climate) and highlight regional differences.
- LSTM networks enable fine-scale, high-frequency forecasts of streamflow, temperature, and nutrient loads — critical for watershed management.
- Interpretability tools (e.g., SHAP values, partial dependence plots) make models more transparent and policy-relevant.
Road Salt and Chloride Legacies
Decades of road-salt application have created a persistent and growing "chloride legacy" in groundwater and surface waters across northern regions. Our research reveals why short-term management approaches are insufficient and provides new frameworks for understanding and managing chloride pollution at regional scales.
Our research shows that chloride concentrations in many aquifers and streams are still rising decades after peak salt use, with levels in some places already exceeding thresholds harmful to aquatic life. Using mass-balance modeling approaches and extensive monitoring data across the Chicago Metro Area, we have quantified how much chloride is stored in groundwater versus exported to waterways. This work reveals why short-term management fixes are insufficient, and why new watershed-scale approaches, from reduced application to alternative deicers, are urgently needed.
Highlights:
- Chloride concentrations still rising in 60% of monitored groundwater wells
- Legacy effects delay benefits of salt reduction by 15-25 years
- Alternative deicing compounds can reduce environmental impact by 40-70%
- Targeted application strategies can maintain safety while reducing salt use by 30%
Interested in Collaboration?
We welcome opportunities to collaborate with researchers, policymakers, and practitioners working on water quality and environmental sustainability challenges.