A hybrid deep neural operator/finite element method for ice-sheet modeling
One of the most challenging and consequential problems in climate modeling is to provide probabilistic projections of sea level rise. A large part of the uncertainty of sea level projections is due to uncertainty in ice sheet dynamics. At the moment, accurate quantification of the uncertainty is hin...
Published in: | Journal of Computational Physics |
---|---|
Main Authors: | , , , , |
Language: | unknown |
Published: |
2024
|
Subjects: | |
Online Access: | http://www.osti.gov/servlets/purl/1998151 https://www.osti.gov/biblio/1998151 https://doi.org/10.1016/j.jcp.2023.112428 |
id |
ftosti:oai:osti.gov:1998151 |
---|---|
record_format |
openpolar |
spelling |
ftosti:oai:osti.gov:1998151 2024-09-15T18:07:45+00:00 A hybrid deep neural operator/finite element method for ice-sheet modeling He, QiZhi Perego, Mauro Howard, Amanda A. Karniadakis, George Em Stinis, Panagiotis 2024-08-19 application/pdf http://www.osti.gov/servlets/purl/1998151 https://www.osti.gov/biblio/1998151 https://doi.org/10.1016/j.jcp.2023.112428 unknown http://www.osti.gov/servlets/purl/1998151 https://www.osti.gov/biblio/1998151 https://doi.org/10.1016/j.jcp.2023.112428 doi:10.1016/j.jcp.2023.112428 71 CLASSICAL AND QUANTUM MECHANICS GENERAL PHYSICS 2024 ftosti https://doi.org/10.1016/j.jcp.2023.112428 2024-08-20T23:40:42Z One of the most challenging and consequential problems in climate modeling is to provide probabilistic projections of sea level rise. A large part of the uncertainty of sea level projections is due to uncertainty in ice sheet dynamics. At the moment, accurate quantification of the uncertainty is hindered by the cost of ice sheet computational models. In this work we develop a hybrid approach to approximate existing ice sheet models at a fraction of their cost. Our approach consists of replacing the finite element model for the momentum equations for the ice velocity, the most expensive part of an ice sheet model, with a Deep Operator Network, while we retain a classic finite element discretization for the evolution of the ice thickness. We show that the resulting hybrid model is very accurate and it is an order of magnitude faster than the traditional finite element model. Further, a distinctive feature of the proposed model, compared to other neural network approaches, is that it can handle high-dimensional parameter spaces (parameter fields) such as the basal friction at the bed of the glacier and can therefore be used for generating samples for uncertainty quantification. Further, we study the impact of hyper-parameters, number of unknowns and correlation length of the parameter distribution on the training and accuracy of the Deep Operator Network on a synthetic ice sheet model. We then target the evolution of the Humboldt glacier in Greenland and show that our hybrid model can provide accurate statistics of the glacier mass loss and can be effectively used to accelerate the quantification of uncertainty. Other/Unknown Material glacier Greenland Humboldt Glacier Ice Sheet SciTec Connect (Office of Scientific and Technical Information - OSTI, U.S. Department of Energy) Journal of Computational Physics 492 112428 |
institution |
Open Polar |
collection |
SciTec Connect (Office of Scientific and Technical Information - OSTI, U.S. Department of Energy) |
op_collection_id |
ftosti |
language |
unknown |
topic |
71 CLASSICAL AND QUANTUM MECHANICS GENERAL PHYSICS |
spellingShingle |
71 CLASSICAL AND QUANTUM MECHANICS GENERAL PHYSICS He, QiZhi Perego, Mauro Howard, Amanda A. Karniadakis, George Em Stinis, Panagiotis A hybrid deep neural operator/finite element method for ice-sheet modeling |
topic_facet |
71 CLASSICAL AND QUANTUM MECHANICS GENERAL PHYSICS |
description |
One of the most challenging and consequential problems in climate modeling is to provide probabilistic projections of sea level rise. A large part of the uncertainty of sea level projections is due to uncertainty in ice sheet dynamics. At the moment, accurate quantification of the uncertainty is hindered by the cost of ice sheet computational models. In this work we develop a hybrid approach to approximate existing ice sheet models at a fraction of their cost. Our approach consists of replacing the finite element model for the momentum equations for the ice velocity, the most expensive part of an ice sheet model, with a Deep Operator Network, while we retain a classic finite element discretization for the evolution of the ice thickness. We show that the resulting hybrid model is very accurate and it is an order of magnitude faster than the traditional finite element model. Further, a distinctive feature of the proposed model, compared to other neural network approaches, is that it can handle high-dimensional parameter spaces (parameter fields) such as the basal friction at the bed of the glacier and can therefore be used for generating samples for uncertainty quantification. Further, we study the impact of hyper-parameters, number of unknowns and correlation length of the parameter distribution on the training and accuracy of the Deep Operator Network on a synthetic ice sheet model. We then target the evolution of the Humboldt glacier in Greenland and show that our hybrid model can provide accurate statistics of the glacier mass loss and can be effectively used to accelerate the quantification of uncertainty. |
author |
He, QiZhi Perego, Mauro Howard, Amanda A. Karniadakis, George Em Stinis, Panagiotis |
author_facet |
He, QiZhi Perego, Mauro Howard, Amanda A. Karniadakis, George Em Stinis, Panagiotis |
author_sort |
He, QiZhi |
title |
A hybrid deep neural operator/finite element method for ice-sheet modeling |
title_short |
A hybrid deep neural operator/finite element method for ice-sheet modeling |
title_full |
A hybrid deep neural operator/finite element method for ice-sheet modeling |
title_fullStr |
A hybrid deep neural operator/finite element method for ice-sheet modeling |
title_full_unstemmed |
A hybrid deep neural operator/finite element method for ice-sheet modeling |
title_sort |
hybrid deep neural operator/finite element method for ice-sheet modeling |
publishDate |
2024 |
url |
http://www.osti.gov/servlets/purl/1998151 https://www.osti.gov/biblio/1998151 https://doi.org/10.1016/j.jcp.2023.112428 |
genre |
glacier Greenland Humboldt Glacier Ice Sheet |
genre_facet |
glacier Greenland Humboldt Glacier Ice Sheet |
op_relation |
http://www.osti.gov/servlets/purl/1998151 https://www.osti.gov/biblio/1998151 https://doi.org/10.1016/j.jcp.2023.112428 doi:10.1016/j.jcp.2023.112428 |
op_doi |
https://doi.org/10.1016/j.jcp.2023.112428 |
container_title |
Journal of Computational Physics |
container_volume |
492 |
container_start_page |
112428 |
_version_ |
1810445131093901312 |