Variational inference of ice shelf rheology with physics-informed machine learning
Floating ice shelves that fringe the coast of Antarctica resist the flow of grounded ice into the ocean. One of the key factors governing the amount of flow resistance an ice shelf provides is the rigidity (related to viscosity) of the ice that constitutes it. Ice rigidity is highly heterogeneous an...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Cambridge University Press
2023-10-01
|
| Series: | Journal of Glaciology |
| Subjects: | |
| Online Access: | https://www.cambridge.org/core/product/identifier/S0022143023000084/type/journal_article |
| _version_ | 1797652278754148352 |
|---|---|
| author | Bryan Riel Brent Minchew |
| author_facet | Bryan Riel Brent Minchew |
| author_sort | Bryan Riel |
| collection | DOAJ |
| description | Floating ice shelves that fringe the coast of Antarctica resist the flow of grounded ice into the ocean. One of the key factors governing the amount of flow resistance an ice shelf provides is the rigidity (related to viscosity) of the ice that constitutes it. Ice rigidity is highly heterogeneous and must be calibrated from spatially continuous surface observations assimilated into an ice-flow model. Realistic uncertainties in calibrated rigidity values are needed to quantify uncertainties in ice sheet and sea-level forecasts. Here, we present a physics-informed machine learning framework for inferring the full probability distribution of rigidity values for a given ice shelf, conditioned on ice surface velocity and thickness fields derived from remote-sensing data. We employ variational inference to jointly train neural networks and a variational Gaussian Process to reconstruct surface observations, rigidity values and uncertainties. Applying the framework to synthetic and large ice shelves in Antarctica demonstrates that rigidity is well-constrained where ice deformation is measurable within the noise level of the observations. Further reduction in uncertainties can be achieved by complementing variational inference with conventional inversion methods. Our results demonstrate a path forward for continuously updated calibrations of ice flow parameters from remote-sensing observations. |
| first_indexed | 2024-03-11T16:27:34Z |
| format | Article |
| id | doaj.art-d333fbb6e44f400cb87bd99e9b422a56 |
| institution | Directory Open Access Journal |
| issn | 0022-1430 1727-5652 |
| language | English |
| last_indexed | 2024-03-11T16:27:34Z |
| publishDate | 2023-10-01 |
| publisher | Cambridge University Press |
| record_format | Article |
| series | Journal of Glaciology |
| spelling | doaj.art-d333fbb6e44f400cb87bd99e9b422a562023-10-24T09:48:09ZengCambridge University PressJournal of Glaciology0022-14301727-56522023-10-01691167118610.1017/jog.2023.8Variational inference of ice shelf rheology with physics-informed machine learningBryan Riel0https://orcid.org/0000-0003-1940-3910Brent Minchew1https://orcid.org/0000-0002-5991-3926School of Earth Sciences, Zhejiang University, 310027 Hangzhou, China Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA, USADepartment of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA, USAFloating ice shelves that fringe the coast of Antarctica resist the flow of grounded ice into the ocean. One of the key factors governing the amount of flow resistance an ice shelf provides is the rigidity (related to viscosity) of the ice that constitutes it. Ice rigidity is highly heterogeneous and must be calibrated from spatially continuous surface observations assimilated into an ice-flow model. Realistic uncertainties in calibrated rigidity values are needed to quantify uncertainties in ice sheet and sea-level forecasts. Here, we present a physics-informed machine learning framework for inferring the full probability distribution of rigidity values for a given ice shelf, conditioned on ice surface velocity and thickness fields derived from remote-sensing data. We employ variational inference to jointly train neural networks and a variational Gaussian Process to reconstruct surface observations, rigidity values and uncertainties. Applying the framework to synthetic and large ice shelves in Antarctica demonstrates that rigidity is well-constrained where ice deformation is measurable within the noise level of the observations. Further reduction in uncertainties can be achieved by complementing variational inference with conventional inversion methods. Our results demonstrate a path forward for continuously updated calibrations of ice flow parameters from remote-sensing observations.https://www.cambridge.org/core/product/identifier/S0022143023000084/type/journal_articleGlacial rheologyglacier modelingice dynamicsice rheologyice shelves |
| spellingShingle | Bryan Riel Brent Minchew Variational inference of ice shelf rheology with physics-informed machine learning Journal of Glaciology Glacial rheology glacier modeling ice dynamics ice rheology ice shelves |
| title | Variational inference of ice shelf rheology with physics-informed machine learning |
| title_full | Variational inference of ice shelf rheology with physics-informed machine learning |
| title_fullStr | Variational inference of ice shelf rheology with physics-informed machine learning |
| title_full_unstemmed | Variational inference of ice shelf rheology with physics-informed machine learning |
| title_short | Variational inference of ice shelf rheology with physics-informed machine learning |
| title_sort | variational inference of ice shelf rheology with physics informed machine learning |
| topic | Glacial rheology glacier modeling ice dynamics ice rheology ice shelves |
| url | https://www.cambridge.org/core/product/identifier/S0022143023000084/type/journal_article |
| work_keys_str_mv | AT bryanriel variationalinferenceoficeshelfrheologywithphysicsinformedmachinelearning AT brentminchew variationalinferenceoficeshelfrheologywithphysicsinformedmachinelearning |