A changing snowpack
Ongoing changes in snowpack, linked to long-term climate projections, affect regional water resources and mountain ecosystems, and pose new challenges for water resource managers.
University of Washington · Civil & Environmental Engineering
Data Science and Hydrology
Our research combines field measurements, remote sensing, physics-based models, and machine learning to investigate multiscale hydrological processes across diverse regions in a changing climate. As datasets grow, we integrate modern data science into our workflows.
Why it matters
Water decisions depend on good science
Snow, rivers, forests and communities are tightly linked. When the climate shifts, the people who manage water need tools they can trust, and a workforce that can build them.
Floods, droughts & low flows
Floods put communities at risk, while summer low flows affect fish and complicate water allocation. Better characterization and monitoring of extremes supports earlier, smarter action.
Science people can use
We work with utilities, agencies and Tribal representatives, and we train students and researchers across campus in data science and AI, so that new methods turn into real decisions.
Explore
Dive into our work
Research
High-resolution snow mapping, tree mortality, extreme events, and physics-informed machine learning.
Explore research →
Applied Research
Partnerships with Seattle City Light, the UW Climate Impacts Group, and the CIVIC project.
See our partnerships →
Data Science
eScience seminars, capstone mentoring, hackweeks, and the GeoSMART curriculum.
Explore training →
Photo: USGS / UnsplashResearch
Our research themes and approaches
From meter-scale snow maps to physics-informed neural networks, our work spans observation, modeling and the connections between disciplines.
Hydrology & Ecohydrology
Identifying Snow-Covered Areas at High Spatial Resolution
Accurate mapping of snow-covered areas is critical for understanding water resources and ecological dynamics in mountainous regions. Our research develops and applies advanced remote sensing techniques to capture snow distribution at meter scale spatial resolutions. By refining these observational methods, we can bridge the gap between sparse ground measurements and broad climate models, capturing micro-topographic effects on snow accumulation and ablation.
Detecting Snowmelt Phases with SAR and Hydrological Modeling
Monitoring the physical state of the snowpack is essential for predicting spring runoff and managing water supplies. We leverage Synthetic Aperture Radar (SAR)—which can penetrate cloud cover—in combination with robust hydrological models to track liquid water content in the snowpack. This integrated approach allows us to reliably detect critical snowmelt phases, such as the onset of runoff, improving early warning systems for floods and water availability forecasts.
Detecting Tree Mortality in Forests
Tree mortality driven by climate stress, drought, and insect outbreaks significantly impacts forest health, watershed dynamics, and carbon cycling. We leverage foundation models combined with high-resolution satellite imagery to detect, map, and monitor forest tree mortality across broad spatial scales. By harnessing computer vision foundation models and fine-resolution remote sensing observations, our work enables scalable, precise identification of canopy decline and dead trees, advancing forest monitoring, wildfire risk assessment, and ecohydrological research.
Extreme Events
We leverage gridded datasets, physics-based hydrologic models, and machine learning techniques to analyze and quantify extreme environmental events across varying spatial and temporal scales. We integrate multi-source observational data with predictive algorithms to advance the characterization, risk assessment, and monitoring of severe phenomena including floods, droughts, and tornadoes.
Data-Driven Modeling
Machine Learning for Streamflow and Snowpack Modeling
Traditional hydrological models can be computationally expensive and rely heavily on precise physical parameterization. To overcome these limitations, we develop machine learning models, including recurrent and graph-based deep learning networks, that capture complex, non-linear relationships in meteorological and watershed data. These models provide fast, accurate predictions of streamflow dynamics and snowpack evolution, including snow water equivalent across the Western United States, at varying spatial and temporal scales.
Where possible, we embed physical principles, such as mass and energy conservation, through physics-informed machine learning (PIML), so predictions remain physically realistic and robust when forecasting unprecedented climate extremes or unobserved conditions.
Interdisciplinary Research
Cascading Hazards with GAIA HazLab
The most devastating disasters are coupled: atmospheric rivers trigger landslides, saturated soils liquefy, and wildfires set up debris flows. We collaborate with GAIA HazLab, which builds digital twins of the Earth by fusing open, multimodal data with AI and cloud computing to monitor and forecast geohazards in real time across the ocean, atmosphere, and solid Earth. Our contribution focuses on the hydrologic piece of these cascades, such as the December 2025 Mount Rainier floods, where seismic networks helped estimate river discharge between sparse stream gauges as atmospheric-river rainfall drove flood response in the Puyallup lahar corridor.
Diablo Dam, Skagit River · Photo: Megan Farmer, KUOWApplied Research
Partnering with stakeholders to inform water and climate decisions
Seattle City Light
Current and Future Hydrology of Seattle City Light Managed Basins
Ongoing changes in snowpack, linked to long-term climate projections, affect the region’s water resources and mountain ecosystems, posing new challenges for water resource managers. Through this collaboration, we evaluate future snowmelt trends in the South Fork Tolt River basin and develop meteorological metrics from gridded datasets for the Skagit River basin and other Washington locations. We analyze historical observations, climate projections, hydrologic model outputs, and key meteorological variables to support Seattle City Light’s climate adaptation, water resource, and energy planning.
NSF CIVICUW Climate Impacts Group
CIVIC NSF Stage 1
We develop a research-centered pilot project in partnership with the UW Climate Impacts Group to evaluate how forest management practices may affect summer low flows in Pacific Northwest watersheds, with a focus on salmon habitat, Tribal treaty rights, and climate resilience. The project will bring together scientists, Tribal representatives, resource managers, NGOs, and other partners to design applied field and modeling research on how forest age, thinning, canopy gaps, and snow storage influence streamflow.
UW Climate Impacts GroupDHSVM
Supporting Regional Stakeholders with Hydrologic Modeling, Climate Analysis and Projections of Future Hydrology
We collaborate with the UW Climate Impacts Group (CIG) to leverage the Distributed Hydrology Soil Vegetation Model (DHSVM), a high-resolution, physics-based model that translates broad Global Climate Models into precise, localized watershed projections. DHSVM simulates how future climate scenarios will impact processes like snowmelt, soil moisture, and streamflow through incorporating fine-scale data like topography, soil depth and topography. CIG uses these simulations to assess specific regional risks, including shifting flood vulnerabilities, water supply shortages, and ecosystem changes.
Data Science
Integrating machine learning, cloud computing, and AI education in geosciences
Data Science and AI Seminars and Training at the UW eScience Institute
Nicoleta is a data science fellow at the eScience Institute, where she hosts the UW Data Science & AI Tricampus seminar (Seattle, Bothell and Tacoma), an annual lecture series that hosts scholars working across applied areas of data science and AI. Students may opt to sign up for a one credit seminar (CR/NCR) listed as ENGR 591.
At the eScience Institute, Nicoleta is also contributing to training programs such as the Data Science and AI Accelerator and hackweeks.
Mentoring Data Science Research and Capstone Projects
We actively participate in team mentoring for student projects in the UW Master of Science in Data Science Capstone Program and CSE494: Guided Research in Computer Science and Engineering.
Mentorship provides guidance to undergraduate and graduate student teams working on complex data-driven challenges, with projects including applications of machine learning, interactive data visualization, and reproducible research in the geosciences.
ML and Data Science Curriculum in Geosciences
Machine learning applications in water resources and other geoscience fields are increasingly used to advance data-driven discovery. To facilitate faster adoption of these techniques, GeoSMART, an NSF-funded educational and research initiative led by PI Cristea at the University of Washington. The program supported courses at UW, such as Machine Learning in the Geosciences (ESS 469/569), as well as hands-on training events (hackweeks), end-to-end machine learning use cases, and review articles. Although the project has ended, we remain committed to maintaining and updating the GeoSMART website.
Publications
Recent publications
2026
Agroseismology and the impact of farming practices on soil hydrodynamics
Science, 392(6795), 306–310.Read more →
Sensitivity of spatial snowmelt simulations to radiative forcing and model top layer thickness
Water Resources Research, 62(8), e2025WR042968.Read more →
Comparing 3 m resolution snow cover downscaled from MODIS, VIIRS, and HLS using commercial satellite imagery and terrain information
EGUsphere [preprint], 1–32.Read more →
2025
Could climate change decrease landslide hazard in snow-dominated mountainous regions? Insights from a distributed hydrology-shallow landslide model of the North Cascades, USA
Water Resources Research, 61(12), e2025WR040071.Read more →
Open-source models for development of data and metadata standards
Patterns, 6(7), 101316.Read more →
Projected Changes in Peak Flows for the Snohomish River Basin
Climate Impacts Group, University of Washington, Seattle, prepared for the King County Flood Control District.Read more →
2024
Using commercial satellite imagery to reconstruct 3 m and daily spring snow water equivalent
Water Resources Research, 60(11), e2024WR037983.Read more →
Substantial cold bias during wintertime cold extremes in the southern Cascadia region in historical CMIP6 simulations
Journal of Geophysical Research: Atmospheres, 129(19), e2024JD041483.Read more →
Towards practical artificial intelligence in Earth sciences
Computational Geosciences, 28(6), 1305–1329.Read more →
Using photographs and deep neural networks to understand flowering phenology and diversity in mountain meadows
Remote Sensing in Ecology and Conservation, 10(4), 480–499.Read more →
Earlier
High-resolution CubeSat imagery and machine learning for detailed snow-covered area
Remote Sensing of Environment, 258, 112399.
Crowd-sourced data reveal social-ecological mismatches in phenology driven by climate
Frontiers in Ecology and the Environment, 18(2), 76–82.
Separating snow and forest temperatures with thermal infrared remote sensing
Remote Sensing of Environment, 209, 764–779.
An evaluation of terrain-based downscaling of fractional snow covered area data sets based on LiDAR-derived snow data and orthoimagery
Water Resources Research, 53(8), 6802–6820.
Team
Who we are and why we do it
Current team
Nicoleta Cristea
Principal Investigator
Nicoleta is a research assistant professor in the Department of Civil and Environmental Engineering and a senior data science fellow at the eScience Institute at the University of Washington. Her research focuses on hydrology, ecohydrology, and the impacts of climate change on water resources, using remote sensing combined with data-driven and physics-based modeling. Her work integrates geographic information systems, satellite imagery, spatial statistics, spatiotemporal visualization, and machine learning. Nicoleta holds a PhD degree from the University of Washington.
Hernán Querbes Duhart
Research Scientist
Hernán Querbes Duhart is a Research Scientist working with Dr. Nicoleta Cristea in the Department of Civil and Environmental Engineering. He earned his M.S. in Civil Engineering (Hydrology) from the University of Washington in June 2026, advised by Dr. Bart Nijssen. His research focuses on advancing hydrological understanding through machine learning and process-based modeling. His current project, conducted in collaboration with Seattle City Light, centers on developing hydrological models for the Skagit River watershed using the Distributed Hydrology Soil Vegetation Model (DHSVM), with applications to water resource management and climate adaptation in the Pacific Northwest.
LinkedIn →
Ryan Richards
Research Scientist
Ryan Richards is a Research Scientist in the Data Science in Hydrology group working on detecting insect-infested trees using high-resolution imagery and geospatial foundation models. His work supports scalable forest health monitoring and advances our understanding of climate- and pest-driven tree mortality.
Ross Mower
PhD Student
He is a PhD student in the Data Science in Hydrology and Mountain Hydrology groups, working with PIs Cristea and Lundquist on analyzing snowmelt using Synthetic Aperture Radar (SAR) and physics-based models. Ross is also a part-time research scientist at the National Center for Atmospheric Research, working on improving snow models through high-performance computing and machine learning. In his free time, Ross can likely be found hiking, biking, or skiing in the mountains.
Email →Zhuoshi Li
PhD Student
Zhuoshi Li is a PhD student working with PIs Cristea and Reed on analyzing and modeling tornado patterns and their effects on buildings using machine learning models.
Niteesh Kumar
Data Science Graduate Student Assistant
Niteesh Kumar is a Data Science Graduate Student Assistant in the Data Science in Hydrology group working on analyzing floods and hydrological risk in the Skagit River basin. His work combines data science methods, hydrologic modeling, and spatial analysis to evaluate flood hazards and support water management and climate adaptation in the Pacific Northwest.
LinkedIn →
Simran Dhankar
Data Science Graduate Student Assistant
Simran Dhankar is a Data Science Graduate Student in the Data Science in Hydrology group working on using Long Short-Term Memory (LSTM) models to simulate snowpack dynamics across the Western United States. Her research applies deep learning and hydroclimatic data modeling to improve snowpack estimation and water resource forecasting.
LinkedIn →Alumni
Balaji Boopal
Data Science Graduate Student Assistant
Balaji Boopal was a Graduate Student Assistant in the Data Science in Hydrology group while pursuing his Master’s in Data Science at the University of Washington. His background spans artificial intelligence, machine learning, deep learning, and scalable data engineering pipelines, applying data-driven modeling and AI solutions to hydroclimatic and environmental datasets. He is currently a Data Scientist at Amazon.
LinkedIn →
Jesse Akes
Graduate Student Assistant
Jesse Akes was a Student Assistant in the Data Science in Hydrology group while completing his M.S. in Applied Mathematics at the University of Washington. He completed his undergraduate degree at the University of North Carolina at Chapel Hill with a B.S. in Computer Science and a minor in Mathematics, and worked as a software engineer in Seattle focusing on backend and search technologies. His research analyzed the impact of climate change on the Pacific Northwest’s water resources, notably in the Skagit River Basin. In his free time, he enjoyed exploring the outdoors, mountain running, and skiing. He is now a Senior Software Engineer at Headway.
LinkedIn →
Kehan Yang
Postdoctoral Fellow
Kehan Yang was a Data Science Postdoctoral Fellow whose work focused on advancing our understanding of spatiotemporal variability of seasonal snowpack using multi-platform remote sensing observations and statistical learning models. Within the Data Science and Hydrology group, her project developed machine learning models to generate snow-covered area maps from commercial smallsat imagery provided by Planet. She is now a Research Scientist at SSAI and NASA Goddard Space Flight Center.
LinkedIn →
Steven Pestana
Research Scientist
Steven Pestana was a Research Scientist in the Data Science in Hydrology group. His contributions focused on working with distributed hydrologic model outputs and meteorological datasets in the Skagit River basin, as well as supporting, participating in, and organizing GeoSMART hackweeks and machine learning use case studies. He is now a scientist at TealWaters.
LinkedIn →
Emma Boudreau
Graduate Student
Emma Boudreau was a UW CEE student working with PIs Dr. Nicoleta Cristea and Dr. Jessica Lundquist. Her research investigated how the fine-scale distribution of snow influences late-season streamflow, with downstream effects on groundwater levels and ecosystem health, leveraging high-resolution satellite imagery and machine learning to map snow cover in complex, forested terrain. She is now with CDM Smith’s water resources engineering team.
LinkedIn →
Ian Chiu
Data Science Undergraduate Student Assistant
Ian Chiu was a Data Science Undergraduate Student Assistant in the Data Science in Hydrology group. His work focused on developing a Python library for a random forest machine learning model trained to detect snow using satellite imagery. He is now at Amazon.
LinkedIn →
Wenyu Jiao
Graduate Student Assistant
Wenyu Jiao was a Graduate Student Assistant in the Data Science in Hydrology group working on analyzing snowmelt patterns in the Tolt River basin and evaluating changes in snowpack using WRF-CMIP5 climate projections. She is now pursuing a Ph.D. degree at Cornell University.
LinkedIn →
Aji John
Postdoctoral Fellow
Aji John was a postdoctoral researcher in the Data Science and Hydrology Group and an eScience Institute postdoctoral fellow at the University of Washington. His research focused on using fine-scale satellite imagery to study montane flowering patterns and snow distribution, a critical driver of ecological productivity. Currently, Aji is a research scientist/software engineer at the University of Washington.
LinkedIn →
Stefan Todoran
Data Science Graduate Student Assistant
Stefan Todoran was a University of Washington Computer Science student working primarily with Dr. Nicoleta Cristea and Dr. Marine Denolle as part of the GeoSMART team. Stefan built infrastructure for the advancement of ML tools within the geosciences, ranging from the GeoSMART and DSHydro websites and blogs to CI/CD automation for GeoSMART curricula. In addition, Stefan was part of multiple research projects, including the use of computer vision in the reconstruction of misaligned image datasets, as well as the use of aerial photography for cold water refuge mapping. In his free time he enjoyed arts & crafts, skiing, dancing and traveling! He is now with UiPath.
LinkedIn →