01
About
I'm an engineer and hydrologist working toward a PhD in Civil & Environmental Engineering at Colorado State University, advised by Dr. Frances Davenport. Before returning to school in 2024, I spent a decade as a water resources engineer at Emmons & Olivier Resources, building hydrologic and hydraulic models for everything from engineering design to regional watershed planning.
My work has long sat at the intersection of hydrology, civil engineering, GIS, and data science, and some of my most rewarding projects have come from carrying an idea from one of those fields into another. The PhD is an opportunity to pair that breadth with depth, and to develop a more rigorous understanding of hydrology and its history as a scientific discipline.
Dissertation research
Advancing Streamflow Estimation in Colorado Using Mixture Distributions and Deep Learning
My dissertation examines how hydrologic models perform at the extremes of the streamflow distribution, where the practical stakes are highest and where models fit to heterogeneous records tend to fall short. The work combines mixed-population flood frequency analysis in Colorado with deep learning approaches to peak-flow and dry-year streamflow prediction.
Tools & methods
- Hydrologic & hydraulic modeling
- Deep learning (LSTMs)
- Explainable AI
- Flood frequency analysis
- Extreme value statistics
- Python / PyTorch
- R / Shiny
- GIS
- HPC / SLURM
- SWMM / PCSWMM
- DRAINMOD
- Web development
02
Projects
Current research, plus a selection of work from my decade in consulting.
Dissertation research
Mixed-population flood frequency in Colorado
Evaluating how flood quantile estimates change when snowmelt, rainfall, and rain-on-snow floods are modeled as separate populations, using peaks-over-threshold analysis at 206 USGS gages across Colorado. Site-level results are available through an interactive dashboard.
- Flood frequency
- Extreme value statistics
- R Shiny
Closing the peak-flow gap in LSTM streamflow models
Evaluating whether resampling rare flows and improving precipitation inputs reduce peak-flow underprediction in LSTM streamflow models across 494 CAMELS catchments, and at what cost to overall skill. Manuscript in preparation.
- Deep learning
- PyTorch
- Large-sample hydrology
LSTM water supply predictions in dry years
Assessing LSTM predictions of April–August runoff in Colorado during dry and normal years, and using feature attribution to identify which inputs, including high-resolution snow water equivalent, drive those predictions.
- Water supply
- Explainable AI
- Snow
Selected consulting work
Rochester Comprehensive Surface Water Management Plan
Comprehensive surface water management plan for Rochester, Minnesota.
- Watershed planning
- Flood hazard
Middle Cedar Watershed Management Plan
Watershed management plan for the Middle Cedar River watershed in Iowa.
- Watershed planning
Edmonton LID Study
Study of low impact development (LID) stormwater practices for the City of Edmonton, Alberta.
- Stormwater
- LID
Grand Marais Stormwater Management Plan
Stormwater management plan for the City of Grand Marais, Minnesota.
- Stormwater
Thunder Bay Stormwater Management Plan
Stormwater management plan for the City of Thunder Bay, Ontario.
- Stormwater
Rural Stormwater Management Model
A rural stormwater management model for managing water quality in the Lake Huron watersheds of Ontario.
- PCSWMM
- Water quality
03
Experience
Graduate Teaching Assistant
Fall 2026 – PresentColorado State University · Fort Collins, CO
- CIVE 520: Physical Hydrology
Graduate Research Assistant
2024 – PresentColorado State University · Fort Collins, CO
- Dissertation research on flood frequency analysis and deep learning (LSTM) streamflow prediction at the extremes, with a focus on Colorado
- Developing mixed-population flood frequency methods for snowmelt, rainfall, and rain-on-snow floods across 206 Colorado stream gages
- Training and evaluating LSTM model ensembles across hundreds of catchments on high-performance computing (HPC) clusters
Water Resources Engineer
2013 – 2023Emmons & Olivier Resources, Inc. · St. Paul, MN
- Built and applied hydrologic and hydraulic models for projects ranging from engineering design support to regional-scale watershed planning
- Contributed to stormwater and watershed management plans across Minnesota, Iowa, Ontario, and Alberta
- Developed web tools supporting modeling and watershed planning projects
Graduate Research Assistant
2010 – 2012University of Minnesota · St. Paul, MN
Graduate Research Contractor
2010United States Geological Survey · Minneapolis, MN
Visiting Research Assistant
2009Czech Technical University (ČVUT) · Prague, Czech Republic
Undergraduate Research Assistant
2005 – 2009University of Minnesota · St. Paul, MN
04
Publications
Also on Google Scholar and ORCID.
Peer-reviewed
- 2016
- 2015
Developing optimum subsurface drainage design procedures
Sands GR, Canelon D, Talbot M
Acta Agriculturae Scandinavica, Section B — Soil & Plant Science 65(sup1):121–127
DOI
In preparation
- 2026
Toward closing the peak-flow prediction gap in LSTM streamflow models
Talbot MT, Davenport FV
In preparation
Conference proceedings
- 2014
Evaluating and Estimating DRAINMOD's Effective Rooting Depth for Corn
Talbot MT, Sands G, Coulter J
Evapotranspiration: Challenges in Measurement and Modeling from Leaf to the Landscape Scale and Beyond. Raleigh, NC
DOI
Contributions & acknowledgements
- 2018
Agriculture — A river runs through it — The connections between agriculture and water quality
Capel PD, McCarthy KA, Coupe RH, Grey KM, Amenumey SE, Baker NT, Johnson RL
U.S. Geological Survey Circular 1433
DOI - 2013
05
Talks
Recent talks and posters.
PosterDec 2025New Orleans, LANo Free Lunch? Improving LSTM Flood Predictions with Minimal Loss in Overall Skill
Talbot MT · AGU25 Annual Meeting
TalkMay 2025Golden, COToward Enhancing Neural Networks in Predicting Streamflow Extremes
Talbot MT · 2025 AWRA & CGWA Symposium
- TalkApr 2025Fort Collins, CO
Toward Improving Machine Learning Models for Predicting Streamflow Extremes
Talbot MT · 2025 Hydrology Days
PosterDec 2024Washington, D.C.Watershed Alchemy: Transforming Data into Hydrologic Understanding
Talbot MT · AGU24 Annual Meeting
- PosterMar 2023Virtual
Evaluation of a GIS-Based Flood Hazard Assessment in Rochester, Minnesota
Talbot MT · 56th International Conference on Water Management Modeling (ICWMM)
06
Education
Ph.D., Civil & Environmental Engineering
Expected 2027Colorado State University
- Program
- Hydrologic Science & Engineering
- Dissertation
- Advancing Streamflow Estimation in Colorado Using Mixture Distributions and Deep Learning (proposed)
- Advisor
- Dr. Frances Davenport
M.S., Bioproducts & Biosystems Science, Engineering, and Management
2019University of Minnesota
- Advisor
- Dr. Gary Sands
B.B.A.E., Biosystems & Agricultural Engineering
2009University of Minnesota