Stefan F. Gary
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Stefan F. Gary has a total of 76 co-authors. Here are the top co-authors by number of shared publications:
- Stuart A. Cunningham
Scottish Association for Marine Science, Oban, UK
4 shared publications - Clare Johnson
Scottish Association for Marine Science Oban Scotland
3 shared publications - Loïc Houpert
National Oceanography Centre, Southampton, UK
3 shared publications - N. Penny Holliday
National Oceanography Centre, Southampton, UK
3 shared publications - Amy E. Goldman
Pacific Northwest National Laboratory, Richland, USA
2 shared publications - Feili Li
State Key Lab of Marine Environmental Science & College of Ocean and Earth Sciences Xiamen University Xiamen China
2 shared publications
Latest Publications
- Open Access
Harnessing artificial intelligence to automate environmental predictions
Predicting heterogeneous and non-linear processes remains a fundamental challenge in Earth sciences. Here, we present an artificial intelligence (AI)-guided framework that iteratively combines predictive modeling with targeted field sampling to rapidly improve environmental predictions. We demonstrate our workflow by predicting oxygen consumption, a key process of stream metabolism, across the contiguous United States (CONUS). Our approach consisted of 18 iterative loops of measurements and models, combining distributed participatory field sampling, lab analysis, automated machine learning (ML) predictions, and error and distinctiveness analyses to autonomously guide the next sampling at optimal site locations. Through our approach, we increased the predictive power of sediment oxygen consumption across CONUS by over fifteenfold between the first and last iteration. Relative to our last sampling iteration, our first sampling missed sites with high rates and underestimated median oxygen consumption rates by 68%. In addition to identifying areas of high oxygen consumption rates, iterations enabled refinement of laboratory and data handling methods, and engagement with a broad community of field researchers. We conclude that AI-guided iterative loops between targeted sampling and predictive modeling are a powerful and efficient approach for improving predictions of heterogeneous environmental processes.
Open AccessHarnessing artificial intelligence to automate environmental predictions
Predicting heterogeneous and non-linear processes remains a fundamental challenge in Earth sciences. Here, we present an artificial intelligence (AI)-guided framework that iteratively combines predictive modeling with targeted field sampling to rapidly improve environmental predictions. We demonstrate our workflow by predicting oxygen consumption, a key process of stream metabolism, across the contiguous United States (CONUS). Our approach consisted of 18 iterative loops of measurements and models, combining distributed participatory field sampling, lab analysis, automated machine learning (ML) predictions, and error and distinctiveness analyses to autonomously guide the next sampling at optimal site locations. Through our approach, we increased the predictive power of sediment oxygen consumption across CONUS by over fifteenfold between the first and last iteration. Relative to our last sampling iteration, our first sampling missed sites with high rates and underestimated median oxygen consumption rates by 68%. In addition to identifying areas of high oxygen consumption rates, iterations enabled refinement of laboratory and data handling methods, and engagement with a broad community of field researchers. We conclude that AI-guided iterative loops between targeted sampling and predictive modeling are a powerful and efficient approach for improving predictions of heterogeneous environmental processes.
- Open Access
- 25 March 2024
- Published by Wiley
Prediction of Distributed River Sediment Respiration Rates using Community-Generated Data and Machine Learning
River sediment microbial respiration is a key indicator of ecosystem functioning and the biogeochemical fluxes across this critical zone link surface and subsurface waters. As such, there is tremendous interest in measuring and mapping these respiration rates. Respiration observations are expensive and labor intensive; there is limited data available to the community. An open science, collaborative initiative is collecting samples for respiration rate analysis and multi-scale metadata; this evolving data set is being used for making machine learning (ML) predictions at unsampled sites to help inform continued community engagement. However, it is a challenge to find an optimum configuration for ML models to work with this feature-rich (i.e. 100+ possible input variables) data set. Here, we present results from a two-tiered approach to managing the analysis of this complex data set: 1) a stacked ensemble of models that automatically optimizes hyperparameters and manages the training of many models and 2) feature permutation importance to detect the most important features in the models. The major elements of this workflow are modular, portable, open, and cloud-based thus making this implementation a potential template for other applications. The models developed here predict that sediment organic matter chemistry is one of the most important features for predicting sediment respiration rate. Other larger-scale, important features fall into the categories of climatic, ecological, geological, and fluvial settings. Leveraging these larger-scale features to generate data-driven estimates of river sediment respiration rates reveals spatially consistent but heterogeneous patterns across the river network of the Columbia River Basin.
- 25 March 2024
- Published by Wiley
Open AccessPrediction of Distributed River Sediment Respiration Rates using Community-Generated Data and Machine Learning
River sediment microbial respiration is a key indicator of ecosystem functioning and the biogeochemical fluxes across this critical zone link surface and subsurface waters. As such, there is tremendous interest in measuring and mapping these respiration rates. Respiration observations are expensive and labor intensive; there is limited data available to the community. An open science, collaborative initiative is collecting samples for respiration rate analysis and multi-scale metadata; this evolving data set is being used for making machine learning (ML) predictions at unsampled sites to help inform continued community engagement. However, it is a challenge to find an optimum configuration for ML models to work with this feature-rich (i.e. 100+ possible input variables) data set. Here, we present results from a two-tiered approach to managing the analysis of this complex data set: 1) a stacked ensemble of models that automatically optimizes hyperparameters and manages the training of many models and 2) feature permutation importance to detect the most important features in the models. The major elements of this workflow are modular, portable, open, and cloud-based thus making this implementation a potential template for other applications. The models developed here predict that sediment organic matter chemistry is one of the most important features for predicting sediment respiration rate. Other larger-scale, important features fall into the categories of climatic, ecological, geological, and fluvial settings. Leveraging these larger-scale features to generate data-driven estimates of river sediment respiration rates reveals spatially consistent but heterogeneous patterns across the river network of the Columbia River Basin.
- Top Cited PapersOpen Access
- 1 February 2019
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 363 (6426) , 516-521
A sea change in our view of overturning in the subpolar North Atlantic
An array of overturning data: The Atlantic meridional overturning circulation (AMOC) has a strong influence on climate, so it is important to understand how global warming may affect it. Lozier et al. report initial results from the Overturning in the Subpolar North Atlantic Program (OSNAP) (see the Perspective by Rhein). OSNAP has been measuring the flux of water transported by overturning in the high latitudes in the North Atlantic. The measurements reveal the strong variability of transport in the region and show that deep water formation in the Labrador Sea may not, as previously believed, be the major determinant of AMOC variability. Science , this issue p. 516 ; see also p. 456
- 1 February 2019
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 363 (6426) , 516-521
Top Cited PapersOpen AccessA sea change in our view of overturning in the subpolar North Atlantic
An array of overturning data: The Atlantic meridional overturning circulation (AMOC) has a strong influence on climate, so it is important to understand how global warming may affect it. Lozier et al. report initial results from the Overturning in the Subpolar North Atlantic Program (OSNAP) (see the Perspective by Rhein). OSNAP has been measuring the flux of water transported by overturning in the high latitudes in the North Atlantic. The measurements reveal the strong variability of transport in the region and show that deep water formation in the Labrador Sea may not, as previously believed, be the major determinant of AMOC variability. Science , this issue p. 516 ; see also p. 456
- Open Access
- 29 August 2018
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Oceans
- Vol. 123 (8) , 6019-6038
Structure and Transport of the North Atlantic Current in the Eastern Subpolar Gyre From Sustained Glider Observations
Repeat glider sections obtained during 2014–2016, as part of the Overturning in the Subpolar North Atlantic Program, are used to quantify the circulation and transport of North Atlantic Current (NAC) branches over the Rockall Plateau. Using 16 glider sections collected along 58°N and between 21°W and 15°W, absolute geostrophic velocities are calculated, and subsequently the horizontal and vertical structure of the transport are characterized. The annual mean northward transport (± standard deviation) is 5.1 ± 3.2 Sv over the Rockall Plateau. During summer (May to October), the mean northward transport is stronger and reaches 6.7 ± 2.6 Sv. This accounts for 43% of the total NAC transport of upper‐ocean waters (σO3) estimated by Sarafanov et al. (2012, https://doi.org/10.1029/2011JC007572) along 59.5°N, between the Reykjanes Ridge and Scotland. Two quasi‐permanent northward flowing branches of the NAC are identified: (i) the Hatton Bank Jet (6.3 ± 2.1 Sv) over the eastern flank of the Iceland Basin (20.5°W to 18.5°W) and (ii) the Rockall Bank Jet (1.5 ± 0.7 Sv) over the eastern flank of the Hatton‐Rockall Basin (16°W to 15°W). Transport associated with the Rockall Bank Jet is mostly depth independent during summer, while 30% of the Hatton Bank Jet transport is due to vertical geostrophic shear. Uncertainties are estimated for each individual glider section using a Monte Carlo approach, and mean uncertainties of the absolute transport are less than 0.5 Sv. Although comparisons with altimetry‐based estimates indicate similar large‐scale circulation patterns, altimetry data do not resolve small mesoscale current bands in the Hatton‐Rockall Basin which are strongly needed for the right transport estimates.
- 29 August 2018
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Oceans
- Vol. 123 (8) , 6019-6038
Open AccessStructure and Transport of the North Atlantic Current in the Eastern Subpolar Gyre From Sustained Glider Observations
Repeat glider sections obtained during 2014–2016, as part of the Overturning in the Subpolar North Atlantic Program, are used to quantify the circulation and transport of North Atlantic Current (NAC) branches over the Rockall Plateau. Using 16 glider sections collected along 58°N and between 21°W and 15°W, absolute geostrophic velocities are calculated, and subsequently the horizontal and vertical structure of the transport are characterized. The annual mean northward transport (± standard deviation) is 5.1 ± 3.2 Sv over the Rockall Plateau. During summer (May to October), the mean northward transport is stronger and reaches 6.7 ± 2.6 Sv. This accounts for 43% of the total NAC transport of upper‐ocean waters (σO3) estimated by Sarafanov et al. (2012, https://doi.org/10.1029/2011JC007572) along 59.5°N, between the Reykjanes Ridge and Scotland. Two quasi‐permanent northward flowing branches of the NAC are identified: (i) the Hatton Bank Jet (6.3 ± 2.1 Sv) over the eastern flank of the Iceland Basin (20.5°W to 18.5°W) and (ii) the Rockall Bank Jet (1.5 ± 0.7 Sv) over the eastern flank of the Hatton‐Rockall Basin (16°W to 15°W). Transport associated with the Rockall Bank Jet is mostly depth independent during summer, while 30% of the Hatton Bank Jet transport is due to vertical geostrophic shear. Uncertainties are estimated for each individual glider section using a Monte Carlo approach, and mean uncertainties of the absolute transport are less than 0.5 Sv. Although comparisons with altimetry‐based estimates indicate similar large‐scale circulation patterns, altimetry data do not resolve small mesoscale current bands in the Hatton‐Rockall Basin which are strongly needed for the right transport estimates.
- Open Access
- 2 July 2018
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Oceans
- Vol. 123 (7) , 4538-4559
Subpolar North Atlantic Overturning and Gyre‐Scale Circulation in the Summers of 2014 and 2016
The Atlantic Meridional Overturning Circulation (AMOC) is a key component of the global climate system through its transport of heat and freshwater. The subpolar North Atlantic (SPNA) is a region where the AMOC is actively developed and shaped though mixing and water mass transformation and where large amounts of heat are released to the atmosphere. Two hydrographic transbasin sections in the summers of 2014 and 2016 provide highly spatially resolved views of the SPNA velocity and property fields on a line from Canada to Greenland to Scotland. Estimates of the AMOC, isopycnal (gyre‐scale) transport, and heat and freshwater transport are derived from the observations. The overturning circulation, the maximum in northward transport integrated from the surface to seafloor and computed in density space, has a high range, with 20.6 ± 4.7 Sv in June–July 2014 and 10.6 ± 4.3 Sv in May–August 2016. In contrast, the isopycnal (gyre‐scale) circulation was lowest in summer 2014: 41.3 ± 8.2 Sv compared to 58.6 ± 7.4 Sv in 2016. The heat transport (0.39 ± 0.08 PW in summer 2014, positive is northward) was highest for the section with the highest AMOC, and the freshwater transport was largest in summer 2016 when the isopycnal circulation was high (−0.25 ± 0.08 Sv). Up to 65% of the heat and freshwater transport was carried by the isopycnal circulation, with isopycnal property transport highest in the western Labrador Sea and the eastern basins (Iceland Basin to Scotland).
- 2 July 2018
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Oceans
- Vol. 123 (7) , 4538-4559
Open AccessSubpolar North Atlantic Overturning and Gyre‐Scale Circulation in the Summers of 2014 and 2016
The Atlantic Meridional Overturning Circulation (AMOC) is a key component of the global climate system through its transport of heat and freshwater. The subpolar North Atlantic (SPNA) is a region where the AMOC is actively developed and shaped though mixing and water mass transformation and where large amounts of heat are released to the atmosphere. Two hydrographic transbasin sections in the summers of 2014 and 2016 provide highly spatially resolved views of the SPNA velocity and property fields on a line from Canada to Greenland to Scotland. Estimates of the AMOC, isopycnal (gyre‐scale) transport, and heat and freshwater transport are derived from the observations. The overturning circulation, the maximum in northward transport integrated from the surface to seafloor and computed in density space, has a high range, with 20.6 ± 4.7 Sv in June–July 2014 and 10.6 ± 4.3 Sv in May–August 2016. In contrast, the isopycnal (gyre‐scale) circulation was lowest in summer 2014: 41.3 ± 8.2 Sv compared to 58.6 ± 7.4 Sv in 2016. The heat transport (0.39 ± 0.08 PW in summer 2014, positive is northward) was highest for the section with the highest AMOC, and the freshwater transport was largest in summer 2016 when the isopycnal circulation was high (−0.25 ± 0.08 Sv). Up to 65% of the heat and freshwater transport was carried by the isopycnal circulation, with isopycnal property transport highest in the western Labrador Sea and the eastern basins (Iceland Basin to Scotland).
- 27 February 2018
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Oceans
- Vol. 123 (2) , 1471-1484
Seasonal Cycles of Oceanic Transports in the Eastern Subpolar North Atlantic
The variability of the Atlantic Meridional Overturning Circulation (AMOC) may play a role in sea surface temperature predictions on seasonal to decadal time scales. Therefore, AMOC seasonal cycles are a potential baseline for interpreting predictions. Here we present estimates for the seasonal cycle of transports of volume, temperature, and freshwater associated with the upper limb of the AMOC in the eastern subpolar North Atlantic on the Extended Ellett Line hydrographic section between Scotland and Iceland. Due to weather, ship‐based observations are primarily in summer. Recent glider observations during other seasons present an opportunity to investigate the seasonal variability in the upper layer of the AMOC. First, we document a new method to quality control and merge ship, float, and glider hydrographic observations. This method accounts for the different spatial sampling rates of the three platforms. The merged observations are used to compute seasonal cycles of volume, temperature, and freshwater transports in the Rockall Trough. These estimates are similar to the seasonal cycles in two eddy‐resolving ocean models. Volume transport appears to be the primary factor modulating other Rockall Trough transports. Finally, we show that the weakest transports occur in summer, consistent with seasonal changes in the regional‐scale wind stress curl. Although the seasonal cycle is weak compared to other variability in this region, the amplitude of the seasonal cycle in the Rockall Trough, roughly 0.5–1 Sv about a mean of 3.4 Sv, may account for up to 7–14% of the heat flux between Scotland and Greenland.
- 27 February 2018
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Oceans
- Vol. 123 (2) , 1471-1484
Seasonal Cycles of Oceanic Transports in the Eastern Subpolar North Atlantic
The variability of the Atlantic Meridional Overturning Circulation (AMOC) may play a role in sea surface temperature predictions on seasonal to decadal time scales. Therefore, AMOC seasonal cycles are a potential baseline for interpreting predictions. Here we present estimates for the seasonal cycle of transports of volume, temperature, and freshwater associated with the upper limb of the AMOC in the eastern subpolar North Atlantic on the Extended Ellett Line hydrographic section between Scotland and Iceland. Due to weather, ship‐based observations are primarily in summer. Recent glider observations during other seasons present an opportunity to investigate the seasonal variability in the upper layer of the AMOC. First, we document a new method to quality control and merge ship, float, and glider hydrographic observations. This method accounts for the different spatial sampling rates of the three platforms. The merged observations are used to compute seasonal cycles of volume, temperature, and freshwater transports in the Rockall Trough. These estimates are similar to the seasonal cycles in two eddy‐resolving ocean models. Volume transport appears to be the primary factor modulating other Rockall Trough transports. Finally, we show that the weakest transports occur in summer, consistent with seasonal changes in the regional‐scale wind stress curl. Although the seasonal cycle is weak compared to other variability in this region, the amplitude of the seasonal cycle in the Rockall Trough, roughly 0.5–1 Sv about a mean of 3.4 Sv, may account for up to 7–14% of the heat flux between Scotland and Greenland.
- 1 January 2017
- Published by Elsevier
- p. 117
Microplastics in the Final Ocean Frontier
- 1 September 2011
- Published by Elsevier in Deep Sea Research Part II: Topical Studies in Oceanography
- Vol. 58 (17-18) , 1798-1818
Export of Labrador Sea Water from the subpolar North Atlantic: A Lagrangian perspective
- 1 September 2011
- Published by Elsevier in Deep Sea Research Part II: Topical Studies in Oceanography
- Vol. 58 (17-18) , 1798-1818
Export of Labrador Sea Water from the subpolar North Atlantic: A Lagrangian perspective