Position Descriptions

Climate change and aquatic invasive species are two of the most important stressors driving long-term change in the Great Lakes ecosystems, but the potential synergies of these two stressors are not well understood. The Great Lakes Aquatic Nonindigenous Species Information System (GLANSIS) contains full literature reviews, assessments and summaries for 192 established species, 26 range expansion and cryptogenic species and 88 watchlist species which includes information about environmental variables linked to climate change (e.g., thermal tolerances). This information is currently distributed throughout the separate documents; compiling it all into a cross-species synthesis will facilitate investigating synergies between these two drivers.

Research Questions:

• 1. What are the potential impacts of climate change on nonindigenous species in the Great Lakes?
  2. Which species will pose a greater threat under climate change scenarios, which will pose a lesser threat, and what is the net change in risk?

Project Activities:

The core task of this fellow will be to compile, organize and synthesize existing information within the GLANSIS database into a supporting publication (likely a NOAA Technical Memorandum). Activities include:

• • For each of the watchlist species, review the species risk assessments and species profiles to extract/organize the information relating to the impact of climate/climate change on the potential for each species to become established.
  • Analyze the extracted data to uncover key patterns such as taxonomic subsets, particular species, or habitats that will be of greatest concern.
  • Create appropriate graphics to present these key patterns and write a synthesis of this information for inclusion in a NOAA Technical Memorandum.
  • If time allows, examine the historic and current distribution patterns and environmental tolerances of established invasive species to determine which (if any) species would be expected to expand, accelerate expansion, move farther northward, or increase in impact under a 2050 climate change scenario.

Required Skills:

• • Excellent organization and technical writing skills.
  • Familiarity with ecological concepts and invasion biology.
  • Biostatistics coursework or other demonstration of familiarity with basic biological statistics needed to determine significance of apparent trends would be useful.

Location: NOAA Great Lakes Environmental Research Laboratory in Ann Arbor, MI, remote, or hybrid. 

The Great Lakes, as the planet’s largest freshwater reservoir, not only provide drinking water for over 38 million people, but also serve as a vital resource for irrigation, shipping, hydroelectric power generation, and recreation. Because of their importance, observing and monitoring the Great Lakes is a critical activity. Due to the limited resources available for this much-needed observing and monitoring, it is crucial to make the most efficient use of available observing platforms (e.g., buoys, research vessels). For this fellowship project, we aim to deploy a machine learning method that leverages convolutional Gaussian neural processes to optimize the design of the Great Lakes observing system. Our overall objective is to develop a quantitative framework for strategic placement of the next generation of observing stations, in order to best capture the variability of the dynamic Great Lakes. 

Research Questions:

We aim to build on previous work done on observing surface temperature variability. GLOS is acquiring a buoy and seeks input on its optimal placement. In this context, our research question is simply “where should the next GLOS buoy be placed to most efficiently improve our quantitative understanding of Great Lakes variability?” The data-driven recommendations from this work will be considered in the context of local community support and socioeconomic factors.

Project Activities: 

The summer fellow will use DeepSensor, an open source Python package for probabilistically modeling environmental data with neural processes, to characterize Great Lakes surface variability and to make informed suggestions for future sensor locations. Specifically, the student will:

• • Use existing observational and model data, which the mentoring team will prepare prior to the start of the fellowship, to train DeepSensor on Great Lakes surface temperature;
  • Create a list of target observing sites that would most efficiently reduce uncertainties in our quantitative representation of Great Lakes variability; and
  • Prepare a brief report to visualize, characterize, and document the results.

Required Skills:

Some experience or relevant coursework involving programming in Python and working with large datasets would be useful. Additional helpful skills include machine learning, high-performance computing, and data visualization. The mentor team will offer training on these skills as needed.

Location: NOAA Great Lakes Environmental Research Laboratory, Ann Arbor, MI, remote, or hybrid.

Ice cover in large lakes significantly influences shipping and navigation, coastal erosion, regional weather patterns, climate, and aquatic ecosystems. Previous studies have indicated substantial links between ice cover variability and multidecadal climate teleconnections. However, the observational records are often too brief to definitively establish these connections. This project aims to extend the available 2D ice charts, spanning from 1973 back to 1897, using deep learning techniques applied to land-based air temperature observations. While physics-based models are commonly used for ice cover predictions, the spatial distribution and the nature of land-based observation variables pose challenges for such models. The student researcher will develop and refine a deep-learning model to predict the ice cover from 1897 to 2023. This model will be trained, validated, and tested with the ice chart data available from 1979 to 2023, and the complete historical dataset will be used to explore the influence of multidecadal climate patterns on ice cover variability. 

Research Questions: 

1. 1. Can the deep learning model reliably predict the spatial ice cover distribution in the Great Lakes directly from land-based temperature observations?
   2. What is the impact of multidecadal climate oscillations on the spatial and temporal variability of ice cover in the Great Lakes?

Project Activities:

Under the guidance of the project mentors, the fellow will independently develop a deep learning model for each of the Great Lakes (Superior, Michigan, Huron, Erie, Ontario). Initially, the fellow will build upon pre-existing models and use the reconstructed daily air temperature data for each of the Great Lakes back to 1897 to expand the currently available ice cover charts for the Great Lakes. Responsibilities include coding in Python, analyzing data, writing a technical memo to describe results, and participating in weekly group meetings.

Required Skills:

Some experience or relevant coursework involving programming and data analysis in Python would be useful. Previous exposure to machine learning and oceanography/limnology would also be advantageous. The mentor team will offer training on these skills as needed. The fellow should be eager to enhance their programming skills and learn new machine-learning techniques.

Location: NOAA Great Lakes Environmental Research Laboratory in Ann Arbor, MI, remote, or hybrid.

Global Climate Models (GCMs) provide future climate projections and are often used to force regional models in the Great Lakes region (GLR). However, it is important to assess whether or not a specific GCM is appropriate for use in the GLR before climate projections are used. GLISA, the NOAA Climate Adaptation Partnership (CAP) team for the Great Lakes, provides climate information, resources, and decision support tools for users looking for regional climate projections, including technical summaries of specific GCMs. This project will expand GLISA’s existing portfolio of climate model evaluations by providing a ‘report card’ of GFDL-CM4, a member of the CMIP6 ensemble.

Research Questions: 

How well does GFDL-CM4, a global climate model and member of the CMIP6 ensemble, replicate the weather and climate patterns unique and important to the Great Lakes region?

Project Activities:

This project involves evaluating the performance of GFDL-CM4 in the Great Lakes region by comparing historical model output to observations, and producing a report card for use by those who need specific climate model data, but do not have the time or expertise needed to use it effectively. Specific activities include:

• • Download variables from the historical run of GFDL-CM4 and use data analysis to evaluate the performance, including the biases and limitations, of the global climate model within the Great Lakes region by comparing model output to popular reanalyses and meteorological station observations.
  • Summarize GFDL-CM4 performance and technical specifications (e.g., lake treatment, model parameterizations) in a Climate Model Report Card.

Required Skills:  

The ideal candidate will be interested in atmospheric science and have or be working toward a degree in a related discipline. Some experience or relevant coursework involving programming (Python, Matlab, R, or similar program) and familiarity with gridded and observational atmospheric datasets (e.g., netCDF) would be useful. The mentor team will offer training on these skills as needed.

Location: NOAA Great Lakes Environmental Research Laboratory in Ann Arbor, MI, remote, or hybrid.

Hydrologic modeling has become an essential tool for managing water resources, especially in regions such as the Great Lakes. Recently, research teams at CIGLR/GLERL have been focused on developing applications of deep learning approaches (e.g., Long Short-Term Memory, LSTM) that build on existing regional scale process-based models (e.g., Large Basin Runoff Model, LRBM) to predict hydrologic dynamics within the Great Lakes basins. However, the inherent complexity of these basins—characterized by a substantial surface water area and a significant proportion of ungauged regions—has led to limitations in the accuracy of current models in predicting total water supplies to the lakes. This shortcoming impedes more effective water management in the Great Lakes basin. By leveraging the reliable estimates provided by the Large Lake Statistical Water Balance Model (L2SWBM), this study proposes to integrate total water supply as a physical constraint to enhance existing hydrologic models. The objective of this research project is to refine the hydrologic modeling framework using the Great Lakes basins as a testbed, with potential benefits that may extend to regional and global water resource management.

Research Questions: 

• 1. How can we enhance the accuracy and reliability of hydrologic models for predicting total water supply in large and ungauged basins?
  2. What are the benefits of incorporating the physical constraints of total water supply from other data sources into regional hydrologic models?

Project Activities: 

The primary responsibility of a summer fellow will be to integrate total water supply estimates from L2SWBM into the existing hydrologic models developed at CIGLR/GLERL. Project tasks may include:

• • Learning to manipulate L2SWBM datasets and hydrologic models developed by CIGLR and GLERL.
  • Incorporating total water supply as a constrained variable into the existing modeling framework.
  • Training the models and evaluating performance.

Required skills: 

The ideal candidate will be interested in hydrology, data science, and statistics and have or be working toward a degree in an appropriate discipline (such as engineering, hydrology, data science, etc.). Some experience or relevant coursework involving data analysis and computer science would be useful. Additional helpful skills include programming in Python and high-performance computing. The mentor team will offer training on these skills as needed.

Location: NOAA Great Lakes Environmental Research Laboratory in Ann Arbor, MI, remote, or hybrid.

The oil spill threats posed by the submerged pipelines at the Straits of Mackinac have raised significant concerns. Numerical tools are crucial for predicting oil spill movements and identifying affected water areas, providing effective guidance for response strategies. However, running a high-fidelity 3D hydrodynamic-ice model and transferring its results to drive an oil spill model can be complex and time-consuming, causing potential threats to a timely and effective response. This project employs a process-based model system to simulate hypothetical oil spills at the Straits of Mackinac, after which the simulation results will be used to feed two machine-learning surrogate models (Random Forest and Artificial Neuron Network), facilitating rapid prediction of oil spill impacts and emergency response for management departments.

Research Questions:

1. 1. If an oil spill occurs at the Strait, how extensive would the oil-affected water areas be after specific time frames (0.5 and 1 day), particularly in months with ice cover?
   2. Can the surrogate models accurately predict oil-affected water areas?

Project Activities: 

• • Review of literature provided by the mentors. Search for additional literature and review them as needed.
  • Run the GNOME oil spill model using the hydrodynamic-ice forcings provided by the mentors.
  • Calculate and analyze the oil-affected areas from the model results using the code provided by the mentors.
  • Use the input features including currents, wind, ice, etc, and the output oil-affected water areas to build surrogate machine-learning models (provided by the mentors) and test model performance. Data will be randomly divided into training (80%), validation (10%), and test (10%).
  • Identify the input feature importance for the established surrogate machine-learning models.

Required Skills: 

• • Background knowledge in oceanography, water resource engineering, machine learning, or artificial intelligence.
  • Interest in hydrodynamic models, oil spill models, or data analysis using Python, R, or similar programming languages.

Location: NOAA Great Lakes Environmental Research Laboratory in Ann Arbor, MI, remote, or hybrid.

Lake Erie’s harmful algal blooms (HABs), which have been a significant issue since the late 1990s, threaten water quality, aquatic ecosystems, and public health. Meteorological parameters such as wind speed and wave height play critical roles in nutrient resuspension and mixing, directly influencing bloom dynamics. Despite extensive monitoring and modeling efforts in Lake Erie, a research gap exists in analyzing historical seasonal and annual trends and shifts (1990 – 2023, 33 years) in wind data and wave height and their direct correlations with bloom severity indices. This research aims to address these knowledge gaps by assessing historical trends and shifts in wind speed, wave height, and nutrient data using statistical non-parametric tests, correlation with bloom severity, and additionally to develop a statistical resuspension model using three years of hindcast data from FVCOM-WAVEWATCH III (WWIII) to forecast resuspension events that trigger HABs frequency and severity. The outcomes of this research will provide valuable insights into the physical drivers of HABs and aid in the development of predictive tools for improved management and mitigation strategies.

Research Questions:

1. 1. How are wind speed, wave height, and nutrient concentrations correlated with the bloom severity index in Lake Erie?
   2. How can wind data and hydrodynamic outputs be leveraged to develop a statistical model for predicting resuspension or mixing events that impact the frequency and severity of harmful algal blooms (HABs)?

Project Activities:

This research involves compiling historical data on wind speed, wave height, nutrient concentrations, and bloom severity index from HAB monitoring initiatives. Statistical non-parametric tests, such as the Mann-Kendall and Pettitt tests, will be conducted to analyze historical seasonal and annual trends and detect shifts in wind speed and wave height. Relationships between wind speed, wave height, nutrient concentrations, and the bloom severity index will be evaluated using Spearman’s rank correlation. Based on these results, an additional objective is to develop a statistical resuspension model using hindcast data from FVCOM-WWIII model outputs to forecast resuspension events that trigger bloom occurrences.

Required Skills:

Minimum qualifications include some familiarity with data analysis, statistical tests (non-parametric tests, correlation), and programming skills in R, Python, or a similar program. The mentors will provide some training for these skills if necessary.

Location: NOAA Great Lakes Environmental Research Laboratory in Ann Arbor, MI, remote, or hybrid.