{"title":"Ice-flow model emulator based on physics-informed deep learning","authors":"Guillaume Jouvet, Guillaume Cordonnier","doi":"10.1017/jog.2023.73","DOIUrl":null,"url":null,"abstract":"Abstract Convolutional neural networks (CNN) trained from high-order ice-flow model realisations have proven to be outstanding emulators in terms of fidelity and computational performance. However, the dependence on an ensemble of realisations of an instructor model renders this strategy difficult to generalise to a variety of ice-flow regimes found in the nature. To overcome this issue, we adopt the approach of physics-informed deep learning, which fuses traditional numerical solutions by finite differences/elements and deep-learning approaches. Here, we train a CNN to minimise the energy associated with high-order ice-flow equations within the time iterations of a glacier evolution model. As a result, our emulator is a promising alternative to traditional solvers thanks to its high computational efficiency (especially on GPU), its high fidelity to the original model, its simplified training (without requiring any data), its capability to handle a variety of ice-flow regimes and memorise previous solutions, and its relatively simple implementation. Embedded into the ‘Instructed Glacier Model’ (IGM) framework, the potential of the emulator is illustrated with three applications including a large-scale high-resolution (2400x4000) forward glacier evolution model, an inverse modelling case for data assimilation, and an ice shelf.","PeriodicalId":15981,"journal":{"name":"Journal of Glaciology","volume":"88 1","pages":"0"},"PeriodicalIF":2.8000,"publicationDate":"2023-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Glaciology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1017/jog.2023.73","RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"GEOGRAPHY, PHYSICAL","Score":null,"Total":0}
引用次数: 1
Abstract
Abstract Convolutional neural networks (CNN) trained from high-order ice-flow model realisations have proven to be outstanding emulators in terms of fidelity and computational performance. However, the dependence on an ensemble of realisations of an instructor model renders this strategy difficult to generalise to a variety of ice-flow regimes found in the nature. To overcome this issue, we adopt the approach of physics-informed deep learning, which fuses traditional numerical solutions by finite differences/elements and deep-learning approaches. Here, we train a CNN to minimise the energy associated with high-order ice-flow equations within the time iterations of a glacier evolution model. As a result, our emulator is a promising alternative to traditional solvers thanks to its high computational efficiency (especially on GPU), its high fidelity to the original model, its simplified training (without requiring any data), its capability to handle a variety of ice-flow regimes and memorise previous solutions, and its relatively simple implementation. Embedded into the ‘Instructed Glacier Model’ (IGM) framework, the potential of the emulator is illustrated with three applications including a large-scale high-resolution (2400x4000) forward glacier evolution model, an inverse modelling case for data assimilation, and an ice shelf.
期刊介绍:
Journal of Glaciology publishes original scientific articles and letters in any aspect of glaciology- the study of ice. Studies of natural, artificial, and extraterrestrial ice and snow, as well as interactions between ice, snow and the atmospheric, oceanic and subglacial environment are all eligible. They may be based on field work, remote sensing, laboratory investigations, theoretical analysis or numerical modelling, or may report on newly developed glaciological instruments. Subjects covered recently in the Journal have included palaeoclimatology and the chemistry of the atmosphere as revealed in ice cores; theoretical and applied physics and chemistry of ice; the dynamics of glaciers and ice sheets, and changes in their extent and mass under climatic forcing; glacier energy balances at all scales; glacial landforms, and glaciers as geomorphic agents; snow science in all its aspects; ice as a host for surface and subglacial ecosystems; sea ice, icebergs and lake ice; and avalanche dynamics and other glacial hazards to human activity. Studies of permafrost and of ice in the Earth’s atmosphere are also within the domain of the Journal, as are interdisciplinary applications to engineering, biological, and social sciences, and studies in the history of glaciology.