Environmental Engineer, Environmental Statistician
Dr. Jack Elsey is an environmental engineer and statistician with deep expertise in computational modeling, machine learning, and applied statistics for environmental and water resources projects. He is skilled at applying modern statistical approaches to navigate complex engineering challenges under tight deadlines, and explaining everything along the way in easy-to-understand terms. Since joining CDM Smith in September 2023, Jack has served as a lead statistician on high-profile projects for the EPA, FEMA, and NPS.
Jack specializes in developing Bayesian statistical models that are designed to directly answer questions most relevant to the project’s objectives. Unlike traditional statistical approaches, Bayesian models are highly flexible; this allows for a much closer approximation of reality, and more power to detect signals in noisy environmental datasets. With an ability to closely tailor the model’s structure and output, the emphasis is on “practical significance” rather than abstract discussions of “statistical significance”.
Jack implements his models and analyses in Python, R, and the statistical programming language Stan. On projects where he is reviewing work in an advisory role, he is able to recommend statistical approaches that use standard spreadsheets and GIS software.
Environmental Engineer and Environmental Statistician, September 2023–Present
Jack currently works at CDM Smith, where he develops and reviews statistical analyses for environmental remediation, flood risk, water resources, and ecological restoration projects. Selected project experience from his current role is organized below by practice area.
Black Butte Mine Superfund Site, EPA Region 10, Cottage Grove, OR, 2025-present. Using a dataset of soil sample mercury concentrations that reflected a heterogenous mixture of background atmospheric deposition and local mining activities, but no clear distinction as to which source the mercury in each sample originated from, Jack developed a Bayesian statistical model that is capable of probabilistically identifying which parts of the site are contaminated above background.
Hanlin-Allied-Olin Superfund Site, EPA Region 3, Moundsville, WV, 2025-present. Jack developed a Bayesian model that was not only capable of estimating the mercury concentration present in periphyton and mussel tissue at different locations, but could also estimate ratios comparing concentrations within the site to background, as well as ratios comparing concentrations at different depths.
PFAS Analysis Research Project, ESTCP, 2024-present. Jack is the statistical subject matter expert in a project comparing the accuracy and precision of different laboratory analyses for quantifying the total PFAS present in soil and groundwater samples. Using a dataset comprised of triplicate analyses performed for a dozen PFAS-impacted sites, he is developing innovative Bayesian statistical models for evaluating how aspects of the sample (e.g. pH, organic carbon content) can degrade the ability of practitioners to measure the total PFAS concentration.
Martin County Utilities, FL, 2026-present. Using a sparse dataset, Jack developed a Bayesian statistical model for predicting the range of PFOS and PFOA concentrations that could potentially be observed in network of water supply wells, as well as estimating the risk that a blend of water from these wells would exceed regulatory limits.
Northwest Pipe & Casing Superfund Site, EPA Region 10, Clackamas, OR, 2024-present. To inform the timing of sampling following groundwater remediation, Jack developed a Bayesian model that could quantify an annual pattern of seasonality in chlorinated solvent concentrations. Jack also developed a statistical testing approach that was powerful enough to estimate the efficacy of in-situ remediation using a severely limited dataset, but could be practicably implemented by the remediation contractor in a spreadsheet.
Portland Harbor Superfund Site, EPA Region 10, Portland, OR, 2023-present. Jack is CDM Smith’s primary on-staff statistician at this $1.05B sediment remediation project, where he has been responsible for reviewing geostatistical analyses of contamination delineation, developing long-term monitoring plans for estimating temporal recovery trends, and performing Bayesian uncertainty analysis to estimate the probability of reactive cap failure.
Olin Corp. (McIntosh Plant) Superfund Site, EPA Region 4, McIntosh, AL, 2023-present. Approximately $500k was spent at this site to deploy ultrasonic flow meters for groundwater flux measurements, but the dataset was marred by measurement bias attributable to the meters settling in soft sediment. Jack developed a statistically rigorous correction technique for this bias, obviating the need to estimate groundwater flux via less accurate indirect methods.
CDM Smith Internal R&D Project, 2024-present. Jack developed an innovative statistical model for accurately predicting heavy metal concentrations in surface soil using hyperspectral imaging data collected by unmanned aerial vehicles. Initial results have achieved R-squared values well over 80%.
Silver Bow Creek/Butte Area Superfund Site, Butte Soils Operable Unit (BPSOU), EPA Region 8, Butte, MT, 2024-present. Jack has served as a statistical subject matter expert for this large site in the western US that has been severely impacted by mine waste. He has primarily been consulted on questions related to replacing expensive laboratory analyses with field analyses, and how remediation contractors can rigorously demonstrate the validity of this replacement with spreadsheet-based methods.
Future of Flood Risk Data Initiative, FEMA, 2025-present. Jack is the lead statistical subject matter expert on a research project for reducing the computational cost of Monte Carlo flood risk analysis. Instead of running a hydraulic flooding model tens of thousands of times and incurring a large cloud computing bill, the methods Jack is implementing will approximate the flood risk with one to two orders of magnitude fewer model runs.
Flood Mitigation Assistance Program, FEMA, 2024-2025. Jack was brought on to perform linear regressions, but he demonstrated to the client that their annual grant review task was a manifestation of a classical optimization problem, and application of optimization algorithms would make their $800M budget go 15% farther.
Board of Water Supply, Honululu, HI, 2026-present. Jack developed a Bayesian model for forecasting the rate of water main breaks a year, given different rates of proactive replacement of pipes with a high estimated probability of failure.
Philadelphia Water Department, 2024-present. Jack developed an innovative technique for estimating the probability distribution of total combined sewer overflow released into the river each year. The technique involves the use of copulas to simulate rainfall events, and surrogate models to approximate the predictions of computationally expensive storm water management models.
Breakwater Installation Project, NPS, 2024-present. Jack used statistical models to quantify the benefits of a breakwater installation on submerged aquatic vegetation recovery and reduced wave activity.
Freshwater Wetland Canal Backfilling Project, NPS, 2024-present. Jack used Bayesian statistics to quantify how backfilling canals affected saltwater infiltration into a freshwater wetland.
The following headings summarize Dr. Elsey’s contributions to research projects he participated in during his MS and PhD degrees at Tufts University.
On the Reliable Estimation of Sequential Monod Kinetic Parameters, 2020-2023. Dr. Elsey used Hamiltonian Monte Carlo to randomly select sets of microbial degradation kinetic parameters that result in realistic synthetic datasets. He performed Monte Carlo experiments with more than a million individual model fittings to measure accuracy and precision of parameter estimates derived from different types of laboratory treatability test designs. He also implemented a likelihood-ratio test to show that a commonly used treatability test design will usually result in failure to uniquely estimate subsets of the kinetic parameters. Dr. Elsey demonstrated that the ill-posed nature of inverse problem can be largely attributed to extreme collinearity of model-fitting results. He compared nonlinear regression analysis and Markov chain Monte Carlo to determine which method results in more reliable interval estimates of kinetic parameters.
Microbial Reductive Dechlorination by a Commercially Available Dechlorinating Consortium Is Not Inhibited by Perfluoroalkyl Acids at Field-Relevant Concentrations, 2020-2023. Dr. Elsey fit a microbial model to laboratory experiments designed to measure how the presence of PFAS can slow the degradation of other subsurface contaminants. He used a superposition of an analytical solution to the diffusion equation to refine predictions of experimental mass loss.
Quantifying Impacts of Microcosm Mass Loss on Kinetic Constant Estimation, 2018-2021. Dr. Elsey summarized 20+ studies of microbially mediated contaminant degradation to illustrate how incidental mass loss of contaminants from laboratory equipment was common but also routinely excluded from the numerical models used to estimate parameters. He conducted a Monte Carlo experiment to demonstrate that excluding mass loss from the numerical model frequently resulted in biased kinetic parameter estimates.
Exploration of Processes Governing Microbial Reductive Dechlorination in a Heterogeneous Aquifer Flow Cell, 2018-2020. Dr. Elsey wrote a numerical model describing the microbially mediated degradation of contaminants in laboratory treatability tests. He estimated kinetic parameters of the degradation process by fitting a model to experimental data.
Subsurface Source Zone Characterization and Uncertainty Quantification Using Discriminative Random Fields, 2016-2018. Dr. Elsey assisted paper authors by debugging code and running subsurface multiphase contaminant transport models to generate synthetic training data for a machine learning model that used soil boring data for predictions of where denser-than-water contaminants had pooled in the subsurface.
Elsey, Jack L., Eric L. Miller, John A. Christ, and Linda M. Abriola. “On the Reliable Estimation of Sequential Monod Kinetic Parameters.” Journal of Contaminant Hydrology 262 (2024): 104323. https://doi.org/10.1016/j.jconhyd.2024.104323.
Hnatko, Jason P., Chen Liu, Jack L. Elsey, Sheng Dong, John D. Fortner, Kurt D. Pennell, Linda M. Abriola, and Natalie L. Cápiro. “Microbial Reductive Dechlorination by a Commercially Available Dechlorinating Consortium Is Not Inhibited by Perfluoroalkyl Acids (PFAAs) at Field-Relevant Concentrations.” Environmental Science & Technology 57, no. 22 (June 6, 2023): 8301–12. https://doi.org/10.1021/acs.est.2c04815.
Elsey, Jack L., John A. Christ, and Linda M. Abriola. “Quantifying Impacts of Microcosm Mass Loss on Kinetic Constant Estimation.” Environmental Science & Technology 55, no. 20 (October 19, 2021): 13822–33. https://doi.org/10.1021/acs.est.1c03452.
Yang, Lurong, Jason P. Hnatko, Jack L. Elsey, John A. Christ, Kurt D. Pennell, Natalie L. Cápiro, and Linda M. Abriola. “Exploration of Processes Governing Microbial Reductive Dechlorination in a Heterogeneous Aquifer Flow Cell.” Water Research 193 (April 2021): 116842. https://doi.org/10.1016/j.watres.2021.116842.