Revisiting Machine Learning Approaches for Short‐ and Longwave Radiation Inference in Weather and Climate Models

Guardado en:
Bibliografiske detaljer
Udgivet i:Journal of Advances in Modeling Earth Systems vol. 17, no. 9 (Sep 1, 2025)
Hovedforfatter: Bertoli, Guillaume
Andre forfattere: Mohebi, Salman, Ozdemir, Firat, Jucker, Jonas, Rüdisühli, Stefan, Perez‐Cruz, Fernando, Salzmann, Mathieu, Schemm, Sebastian
Udgivet:
John Wiley & Sons, Inc.
Fag:
Online adgang:Citation/Abstract
Full Text
Full Text - PDF
Tags: Tilføj Tag
Ingen Tags, Vær først til at tagge denne postø!

MARC

LEADER 00000nab a2200000uu 4500
001 3254545794
003 UK-CbPIL
022 |a 1942-2466 
024 7 |a 10.1029/2025MS004956  |2 doi 
035 |a 3254545794 
045 0 |b d20250901 
084 |a 151828  |2 nlm 
100 1 |a Bertoli, Guillaume  |u Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland 
245 1 |a Revisiting Machine Learning Approaches for Short‐ and Longwave Radiation Inference in Weather and Climate Models 
260 |b John Wiley & Sons, Inc.  |c Sep 1, 2025 
513 |a Journal Article 
520 3 |a This paper explores Machine Learning (ML) parameterizations for radiative transfer in the ICOsahedral Nonhydrostatic weather and climate model (ICON) and investigates the achieved ML model speed‐up with ICON running on graphics processing units (GPUs). Five ML models, with varying complexity and size, are coupled to ICON; more specifically, a multilayer perceptron (MLP), a Unet model, a bidirectional recurrent neural network with long short‐term memory (BiLSTM), a vision transformer (ViT), and a random forest (RF) as a baseline. The ML parameterizations are coupled to the ICON code that includes OpenACC compiler directives to enable GPU support. The coupling is done with the PyTorch‐Fortran coupler developed at NVIDIA. The most accurate model is the BiLSTM with a physics‐informed normalization strategy, a penalty for the heating rates during training, a Gaussian smoothing as postprocessing and a simplified computation of the fluxes at the upper levels to ensure stability of the ICON model top. The presented setup enables stable aquaplanet simulations with ICON for several weeks at a resolution of about 80 km and compares well with the physics‐based default radiative transfer parameterization, ecRad. Our results indicate that the compute requirements of the ML models that can ensure the stability of ICON are comparable to GPU optimized classical physics parameterizations in terms of memory consumption and computational speed. 
653 |a Long wave radiation 
653 |a Machine learning 
653 |a Simulation 
653 |a Physics 
653 |a Datasets 
653 |a Graphics 
653 |a Solar energy 
653 |a Radiative transfer 
653 |a Climate and weather 
653 |a Climate models 
653 |a Neural networks 
653 |a Aerosols 
653 |a Variables 
653 |a Architecture 
653 |a Climate 
653 |a Parameterization 
653 |a Radiation 
653 |a Systematic review 
653 |a Environmental 
700 1 |a Mohebi, Salman  |u Swiss Data Science Center, ETH Zurich and EPFL, Zurich, Switzerland 
700 1 |a Ozdemir, Firat  |u Swiss Data Science Center, ETH Zurich and EPFL, Zurich, Switzerland 
700 1 |a Jucker, Jonas  |u Center for Climate Systems Modeling, ETH Zurich, Zurich, Switzerland 
700 1 |a Rüdisühli, Stefan  |u Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland 
700 1 |a Perez‐Cruz, Fernando  |u Swiss Data Science Center, ETH Zurich and EPFL, Zurich, Switzerland 
700 1 |a Salzmann, Mathieu  |u Swiss Data Science Center, ETH Zurich and EPFL, Zurich, Switzerland 
700 1 |a Schemm, Sebastian  |u Department of Applied Mathematics and Theoretical Physics, Cambridge University, Cambridge, UK 
773 0 |t Journal of Advances in Modeling Earth Systems  |g vol. 17, no. 9 (Sep 1, 2025) 
786 0 |d ProQuest  |t Publicly Available Content Database 
856 4 1 |3 Citation/Abstract  |u https://www.proquest.com/docview/3254545794/abstract/embedded/L8HZQI7Z43R0LA5T?source=fedsrch 
856 4 0 |3 Full Text  |u https://www.proquest.com/docview/3254545794/fulltext/embedded/L8HZQI7Z43R0LA5T?source=fedsrch 
856 4 0 |3 Full Text - PDF  |u https://www.proquest.com/docview/3254545794/fulltextPDF/embedded/L8HZQI7Z43R0LA5T?source=fedsrch