Mary Ruth Keller
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
Research Subjects
Research Interests
Mary Ruth Keller has a total of 157 co-authors. Here are the top co-authors by number of shared publications:
B.L. Gotwols
Johns Hopkins University
5 shared publicationsChristine Piatko
3 shared publicationsChristopher D. Hash
Applied Coherent Technology, Herndon, USA
3 shared publications- Noam. R. Izenberg
The Johns Hopkins University Applied Physics Laboratory Laurel MD USA
3 shared publications - Scott L. Murchie
Johns Hopkins Applied Physics Laboratory Laurel MD USA
3 shared publications W.C. Keller
3 shared publications
Latest Publications
- Open Access
- 10 December 2025
- Published by Wiley
NATURE CLIMATE CHANGE
Extended Data Fig. 5 | Snowmelt increases during winter months are widespread. Geographic distribution of the monthly melt trends shown in Extended Data Fig. 4 for stations with data records ≥ 30 years (small black markers; n = 634) that have statistically significant (p ≤ 0.05) long-term increases (red markers) and decreases (blue markers) in monthly melt as a fraction of total annual melt for the months of (a) October, (b) November, (c) December, (d) January, (e) February, and (f) March. 30-year dataset from SNOTEL sites documents increased winter-time snow-melt as temperatures rise, thus decreasing the amount of water available in the spring/summer and increasing drought.
- 10 December 2025
- Published by Wiley
Open AccessNATURE CLIMATE CHANGE
Extended Data Fig. 5 | Snowmelt increases during winter months are widespread. Geographic distribution of the monthly melt trends shown in Extended Data Fig. 4 for stations with data records ≥ 30 years (small black markers; n = 634) that have statistically significant (p ≤ 0.05) long-term increases (red markers) and decreases (blue markers) in monthly melt as a fraction of total annual melt for the months of (a) October, (b) November, (c) December, (d) January, (e) February, and (f) March. 30-year dataset from SNOTEL sites documents increased winter-time snow-melt as temperatures rise, thus decreasing the amount of water available in the spring/summer and increasing drought.
- Open Access
- 5 February 2025
- Published by Wiley
Physics-informed Machine Learning for Characterization of Arctic Sea Ice
Melting Arctic sea ice fractures into floes, adversely affecting safe navigation. Short-term forecasts of sea ice motion are required, but remain challenging due to multi-scale spatiotemporal atmosphere-ice-ocean couplings. Limitations of current methods include: optical satellite data are high-resolution but may be blocked by clouds or unable to observe in polar night; passive microwave data are low-resolution; SAR data have high resolution, but exact overpasses can be days apart; and climate models have insufficient spatial resolution. Machine learning (ML) has the potential to solve the prediction problem but can require large amounts of training data with no guarantee that these models obey the laws of physics. Physics-informed neural networks (PINNs) reduce the amount of data required for learning by constraining the model to generate solutions that satisfy a given partial differential equation. PINNs are fully differentiable with respect to all input coordinates and free parameters. We present a PINN that characterizes sea ice motion and predicts sea ice deformation.
- 5 February 2025
- Published by Wiley
Open AccessPhysics-informed Machine Learning for Characterization of Arctic Sea Ice
Melting Arctic sea ice fractures into floes, adversely affecting safe navigation. Short-term forecasts of sea ice motion are required, but remain challenging due to multi-scale spatiotemporal atmosphere-ice-ocean couplings. Limitations of current methods include: optical satellite data are high-resolution but may be blocked by clouds or unable to observe in polar night; passive microwave data are low-resolution; SAR data have high resolution, but exact overpasses can be days apart; and climate models have insufficient spatial resolution. Machine learning (ML) has the potential to solve the prediction problem but can require large amounts of training data with no guarantee that these models obey the laws of physics. Physics-informed neural networks (PINNs) reduce the amount of data required for learning by constraining the model to generate solutions that satisfy a given partial differential equation. PINNs are fully differentiable with respect to all input coordinates and free parameters. We present a PINN that characterizes sea ice motion and predicts sea ice deformation.
- Open Access
- 15 December 2024
- Published by Wiley
Evaluating the interactions of winds, waves, and thin ice types in the coastal margins of the Baltic Sea with ICEYE data IN31B-2009
- 15 December 2024
- Published by Wiley
Open AccessEvaluating the interactions of winds, waves, and thin ice types in the coastal margins of the Baltic Sea with ICEYE data IN31B-2009
- Open Access
- 1 October 2023
- Published by American Meteorological Society in Artificial Intelligence for the Earth Systems
- Vol. 2 (4) , 1-36
Short-Term (7 Day) Beaufort Sea Ice Extent Forecasting with Deep Learning
Ships inside the Arctic basin require high-resolution (1–5 km), near-term (days to semimonthly) forecasts for guidance on scales of interest to their operations where forecast model predictions are insufficient due to their coarse spatial and temporal resolutions. Deep learning techniques offer the capability of rapid assimilation and analysis of multiple sources of information for improved forecasting. Data from the National Oceanographic and Atmospheric Administration’s Global Forecast System, Multi-scale Ultra-high Resolution Sea Surface Temperature (MEaSUREs), and the National Snow and Ice Data Center’s Multisensor Analyzed Sea ice Extent (MASIE) were used to develop the sea ice extent deep learning forecast model, over the freeze-up periods of 2016, 2018, 2019, and 2020 in the Beaufort Sea. Sea ice extent forecasts were produced for 1–7 days in the future. The approach was novel for sea ice extent forecasting in using forecast data as model input to aid in the prediction of sea ice extent. Model accuracy was assessed against a persistence model. While the average accuracy of the persistence model dropped from 97% to 90% for forecast days 1–7, the deep learning model accuracy dropped only to 93%. A k-fold (fourfold) cross-validation study found that on all except the first day, the deep learning model, which includes a U-Net architecture with an 18-layer residual neural network (Resnet-18) backbone, does better than the persistence model. Skill scores improve the farther out in time to 0.27. The model demonstrated success in predicting changes in ice extent of significance for navigation in the Amundsen Gulf. Extensions to other Arctic seas, seasons, and sea ice parameters are under development. Significance Statement: Ships traversing the Arctic require timely, accurate sea ice location information to successfully complete their transits. After testing several potential candidates, we have developed a short-term (7 day) forecast process using existing observations of ice extent and sea surface temperature, with operational forecasts of weather and oceanographic variables, and appropriate machine learning models. The process included using forecasts of atmospheric and oceanographic conditions, as a human forecaster/analyst would. The models were trained for the Beaufort Sea north of Alaska using data and forecasts from 2016 combined with 2018–20. The results showed improvement in short-term forecasts of ice locations over current methods and also demonstrated correctly predicted changes in the sea ice that are important for navigation.
- 1 October 2023
- Published by American Meteorological Society in Artificial Intelligence for the Earth Systems
- Vol. 2 (4) , 1-36
Open AccessShort-Term (7 Day) Beaufort Sea Ice Extent Forecasting with Deep Learning
Ships inside the Arctic basin require high-resolution (1–5 km), near-term (days to semimonthly) forecasts for guidance on scales of interest to their operations where forecast model predictions are insufficient due to their coarse spatial and temporal resolutions. Deep learning techniques offer the capability of rapid assimilation and analysis of multiple sources of information for improved forecasting. Data from the National Oceanographic and Atmospheric Administration’s Global Forecast System, Multi-scale Ultra-high Resolution Sea Surface Temperature (MEaSUREs), and the National Snow and Ice Data Center’s Multisensor Analyzed Sea ice Extent (MASIE) were used to develop the sea ice extent deep learning forecast model, over the freeze-up periods of 2016, 2018, 2019, and 2020 in the Beaufort Sea. Sea ice extent forecasts were produced for 1–7 days in the future. The approach was novel for sea ice extent forecasting in using forecast data as model input to aid in the prediction of sea ice extent. Model accuracy was assessed against a persistence model. While the average accuracy of the persistence model dropped from 97% to 90% for forecast days 1–7, the deep learning model accuracy dropped only to 93%. A k-fold (fourfold) cross-validation study found that on all except the first day, the deep learning model, which includes a U-Net architecture with an 18-layer residual neural network (Resnet-18) backbone, does better than the persistence model. Skill scores improve the farther out in time to 0.27. The model demonstrated success in predicting changes in ice extent of significance for navigation in the Amundsen Gulf. Extensions to other Arctic seas, seasons, and sea ice parameters are under development. Significance Statement: Ships traversing the Arctic require timely, accurate sea ice location information to successfully complete their transits. After testing several potential candidates, we have developed a short-term (7 day) forecast process using existing observations of ice extent and sea surface temperature, with operational forecasts of weather and oceanographic variables, and appropriate machine learning models. The process included using forecasts of atmospheric and oceanographic conditions, as a human forecaster/analyst would. The models were trained for the Beaufort Sea north of Alaska using data and forecasts from 2016 combined with 2018–20. The results showed improvement in short-term forecasts of ice locations over current methods and also demonstrated correctly predicted changes in the sea ice that are important for navigation.
- Open Access
- 31 July 2023
- Published by American Astronomical Society
GNSS (and other) Observations from Autonomous Seagoing Platforms
Whitepaper #310 in the Decadal Survey for Solar and Space Physics (Heliophysics) 2024-2033. Main topics: space weather applications; basic research. Additional topics: planetary ionospheres / upper atmospheres; space weather research/applications/operations pipeline; […]
- 31 July 2023
- Published by American Astronomical Society
Open AccessGNSS (and other) Observations from Autonomous Seagoing Platforms
Whitepaper #310 in the Decadal Survey for Solar and Space Physics (Heliophysics) 2024-2033. Main topics: space weather applications; basic research. Additional topics: planetary ionospheres / upper atmospheres; space weather research/applications/operations pipeline; […]
- Open Access
- 12 October 2021
- Published by Wiley in ESS Open Archive
Short-Term Sea Ice Extent Forecasting with Deep Learning
Current predictions for the changing Arctic focus on the evolution of sea ice from perennial to seasonal cover, with sea ice thinning as the Arctic warms. Significant efforts in developing climate forecasts for the Arctic on the decadal time-frame are underway. But, ships in transit inside the Arctic basin require high-resolution, near-term forecasts to provide guidance on scales of interest to their operations: 1-5 km. Climate predictions are insufficient for such short-term forecasts, as they are computationally intensive and at lower resolution than needed for single vessel operations. Deep learning techniques can offer the capability to enable rapid assimilation and analysis of multiple sources of information for these critical sea ice forecasts. In this study, data from NOAA’s Global Forecast System (GFS), Multi-scale Ultra-high Resolution (MUR) Sea Surface Temperature (SST), and Multisensor Analyzed Sea Ice Extent (MASIE) data from the National Snow and Ice Data Center (NSIDC) were used to develop the sea ice extent forecast model. All layers were sampled to the same 1 km Equal-Area Scalable Earth (EASE) grid. Sea ice prediction was achieved for subsequent weeks from seven days prior plus seven days of historical forecast data. The novelty in this approach was using forecast data as data to aid in the prediction of sea ice extent. The models were trained using data from the freeze-up periods of 2016, 2018, 2019, and 2020. A four-fold cross-validation was performed, with each year held out for validation. Our deep learning models are trained using overlapping 256x256 pixel tiles over the entire region of interest, the Beaufort Sea. Sigmoid outputs are averaged together in overlap regions, weighted by the distance from the edge to ensure smooth transitions from one subregion to the next. These are converted to a binary ice mask by thresholding at 0.5, then computing a per-pixel accuracy, which is averaged over the ocean/ice regions. Accuracy for a baseline persistence model was also computed over the seven-day forecast period. While the average accuracy of the persistence model dropped from 97% to 90% for forecast days one to seven, the deep learning model accuracy dropped only to 93%. With these initial positive results, we plan to extend the model outputs from ice extent to ice concentration and thickness in future work.
- 12 October 2021
- Published by Wiley in ESS Open Archive
Open AccessShort-Term Sea Ice Extent Forecasting with Deep Learning
Current predictions for the changing Arctic focus on the evolution of sea ice from perennial to seasonal cover, with sea ice thinning as the Arctic warms. Significant efforts in developing climate forecasts for the Arctic on the decadal time-frame are underway. But, ships in transit inside the Arctic basin require high-resolution, near-term forecasts to provide guidance on scales of interest to their operations: 1-5 km. Climate predictions are insufficient for such short-term forecasts, as they are computationally intensive and at lower resolution than needed for single vessel operations. Deep learning techniques can offer the capability to enable rapid assimilation and analysis of multiple sources of information for these critical sea ice forecasts. In this study, data from NOAA’s Global Forecast System (GFS), Multi-scale Ultra-high Resolution (MUR) Sea Surface Temperature (SST), and Multisensor Analyzed Sea Ice Extent (MASIE) data from the National Snow and Ice Data Center (NSIDC) were used to develop the sea ice extent forecast model. All layers were sampled to the same 1 km Equal-Area Scalable Earth (EASE) grid. Sea ice prediction was achieved for subsequent weeks from seven days prior plus seven days of historical forecast data. The novelty in this approach was using forecast data as data to aid in the prediction of sea ice extent. The models were trained using data from the freeze-up periods of 2016, 2018, 2019, and 2020. A four-fold cross-validation was performed, with each year held out for validation. Our deep learning models are trained using overlapping 256x256 pixel tiles over the entire region of interest, the Beaufort Sea. Sigmoid outputs are averaged together in overlap regions, weighted by the distance from the edge to ensure smooth transitions from one subregion to the next. These are converted to a binary ice mask by thresholding at 0.5, then computing a per-pixel accuracy, which is averaged over the ocean/ice regions. Accuracy for a baseline persistence model was also computed over the seven-day forecast period. While the average accuracy of the persistence model dropped from 97% to 90% for forecast days one to seven, the deep learning model accuracy dropped only to 93%. With these initial positive results, we plan to extend the model outputs from ice extent to ice concentration and thickness in future work.
- Open Access
- 26 February 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computing in Science & Engineering
- Vol. 23 (1) , 108
Call for 2021 Major Awards Nominations
Presents information on CS award guidelines and nominations.
- 26 February 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Computing in Science & Engineering
- Vol. 23 (1) , 108
Open AccessCall for 2021 Major Awards Nominations
Presents information on CS award guidelines and nominations.
- 13 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 59 (7) , 5434-5448
Active/Passive Multiple Polarization Sea Ice Detection During Initial Freeze-Up
An algorithm for ice/water detection of ice less than 30 cm in thickness (thin ice) using dual-polarized synthetic aperture radar (SAR) and passive microwave data has been developed based on insights from previous work. The frequency chosen from those available was in the C-band. For the radiometer layer ice concentrations, land contaminated pixels were removed. For the SAR layers, a novel approach to their use in ice detection was developed using the coefficient of variation (COV), defined as the ratio of the standard deviation of the normalized radar cross section, σ₀, to the mean σ₀. The COV reduced the complicated signatures of breaking waves and thin ice to bright ice pixels on a dark water background. The co-pol and cross-pol COV layers and the radiometer layer were merged according to physics-based rules, except in regions adjacent to land or the ice edge where only the SAR layers were valid. The detection algorithm was applied to 233 Sentinel 1a and 1b dual-polarization images of the freeze-up from 2015, 2016, 2018, and 2019 in the Beaufort and Chukchi Seas. The synergistic ice retrievals showed improved agreement with manual operational analyses in the marginal ice zone over other automated operational products.
- 13 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 59 (7) , 5434-5448
Active/Passive Multiple Polarization Sea Ice Detection During Initial Freeze-Up
An algorithm for ice/water detection of ice less than 30 cm in thickness (thin ice) using dual-polarized synthetic aperture radar (SAR) and passive microwave data has been developed based on insights from previous work. The frequency chosen from those available was in the C-band. For the radiometer layer ice concentrations, land contaminated pixels were removed. For the SAR layers, a novel approach to their use in ice detection was developed using the coefficient of variation (COV), defined as the ratio of the standard deviation of the normalized radar cross section, σ₀, to the mean σ₀. The COV reduced the complicated signatures of breaking waves and thin ice to bright ice pixels on a dark water background. The co-pol and cross-pol COV layers and the radiometer layer were merged according to physics-based rules, except in regions adjacent to land or the ice edge where only the SAR layers were valid. The detection algorithm was applied to 233 Sentinel 1a and 1b dual-polarization images of the freeze-up from 2015, 2016, 2018, and 2019 in the Beaufort and Chukchi Seas. The synergistic ice retrievals showed improved agreement with manual operational analyses in the marginal ice zone over other automated operational products.
- Open Access
- 1 July 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Intelligent Systems
- Vol. 35 (3) , 115
HONORING WOMEN IN COMPUTING Call for Nominations
Presents a call for nominations for the CS Honoring Women in Computing award.
- 1 July 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Intelligent Systems
- Vol. 35 (3) , 115
Open AccessHONORING WOMEN IN COMPUTING Call for Nominations
Presents a call for nominations for the CS Honoring Women in Computing award.