3-D computer graphic of a similutated groundwater plume with numerical method being applied

The many ways water moves through and interacts with the earth’s surface gives rise to some dauntingly complex data analysis and modeling challenges. For example, characterizing groundwater and surface water interactions along river corridors requires complex modeling across space and time with data sets that are often sparse or have high uncertainty. CEE researchers have been at the forefront of developing and applying new computational approaches to optimally utilize limited available data to protect water resources.


CEE faculty and student research in advanced computational methods in water resources management addresses key needs in numerical methods, optimization, Bayesian techniques, and machine learning and deep learning algorithm development.

Highlighted Research Projects & Initiatives

Groundwater and Subsurface Characterization

Figure showing 3-D computer generated view of a groundwater plume Of the fresh water available for use worldwide, 96% of it is groundwater. Protection and careful stewarding of this valuable resource requires the ability to predict its behavior – where, when, and how it flows through the subsurface. Such predictions are achieved using advanced computational modeling; however, the difficulty and cost associated with obtaining data used to develop groundwater models results in an unavoidable uncertainty in the mathematical representation of the subsurface environment. CEE researchers study how groundwater model forecasts can represent this uncertainty and provide the decision-maker with uncertainty estimates of predicted groundwater behavior, one of the key elements in the assessment of risk. Development and application of this exceedingly important modeling capability is a focus of the groundwater research activity at UVM.



Multi-Objective Optimization to Examine Tradeoffs in River Reconnection Projects

Graphic of HEC-RAS model simulation of flooding along a roadwayFloodplain and river reconnection projects show great promise to restore water quality, improve flood resilience and enhance habitats.  However, with the exponential availability of environmental sensors and associated information, resource managers and planners need decision-support tools to help identify the more effective locations and techniques for floodplain reconnection.  Our team is developing multi-objective optimization algorithms to assess multiple stakeholder criteria that prioritize suites of restoration or conservation projects at river network and basin scales. These algorithms may be wrapped around existing hydraulic models (2D HEC-RAS, HAND) or databases (VTANR stream geomorphic data sets) to evolve solutions that optimally meet flood resiliency and water quality goals while simultaneously addressing stakeholder needs. This research is being advanced under several projects supported by the VT Agency of Transportation, VT Department of Environmental Conservation, Lake Champlain Sea Grant, Lake Champlain Basin Program, and the VT Water Resources & Lake Studies Center.


CEE Faculty: Kristen Underwood, Donna Rizzo, Mandar Dewoolkar

UVM Collaborators: Rebecca Diehl, Beverley Wemple

Collaborating Institutions & Organizations: Vermont Agency of Transportation, Vermont Agency of Natural Resources, Vermont Land Trust, The Nature Conservancy of Vermont, Gund Institute for Environment



Research Topics

  • Numerical methods
  • Optimization
  • Bayesian techniques
  • Machine learning
  • Deep learning

Graduate Study in Advanced Computational Methods in Water Resource Management

Highlighted courses are listed below:

  • CE241 – Numerical Methods for Engineering
  • CE369 – Applied Geostatistics
  • CE370 – Reliability of Engineering Systems
  • CS352 – Evolutionary Computation