Peter Ukkonen
YOU?
Author Swipe
View article: Vertically Recurrent Neural Networks for Sub‐Grid Parameterization
Vertically Recurrent Neural Networks for Sub‐Grid Parameterization Open
Machine learning has the potential to improve the physical realism and/or computational efficiency of parameterizations. A typical approach has been to feed concatenated vertical profiles to a dense neural network. However, feed‐forward ne…
View article: Distilling Machine Learning’s Added Value: Pareto Fronts in Atmospheric Applications
Distilling Machine Learning’s Added Value: Pareto Fronts in Atmospheric Applications Open
The added value of machine learning for weather and climate applications is measurable through performance metrics, but explaining it remains challenging, particularly for large deep learning models. Inspired by climate model hierarchies, …
View article: Emulation of sub-grid physics using stochastic, vertically recurrent neural networks
Emulation of sub-grid physics using stochastic, vertically recurrent neural networks Open
Machine learning (ML) has the potential to reduce systematic uncertainties in Earth System Models by replacing or complementing existing physics-based parameterizations of sub-grid processes. However, after decades of research, ensuring ge…
View article: Distilling Machine Learning's Added Value: Pareto Fronts in Atmospheric Applications
Distilling Machine Learning's Added Value: Pareto Fronts in Atmospheric Applications Open
The added value of machine learning for weather and climate applications is measurable through performance metrics, but explaining it remains challenging, particularly for large deep learning models. Inspired by climate model hierarchies, …
View article: Code and partial data used in "Vertically recurrent neural networks for sub-grid parameterization"
Code and partial data used in "Vertically recurrent neural networks for sub-grid parameterization" Open
This repository contains the RNN training and evaluation code used in the paper Vertically recurrent neural networks for sub-grid parameterization The radiative transfer emulation data can be accessed with through a Climetlab plugin (Clime…
View article: The Cycle 46 Configuration of the HARMONIE-AROME Forecast Model
The Cycle 46 Configuration of the HARMONIE-AROME Forecast Model Open
The aim of this technical note is to describe the Cycle 46 reference configuration of the HARMONIE-AROME convection-permitting numerical weather prediction model. HARMONIE-AROME is one of the canonical system configurations that is develop…
View article: Distilling Machine Learning's Added Value: Pareto Fronts in Atmospheric Applications
Distilling Machine Learning's Added Value: Pareto Fronts in Atmospheric Applications Open
The added value of machine learning for weather and climate applications is measurable through performance metrics, but explaining it remains challenging, particularly for large deep learning models. Inspired by climate model hierarchies, …
View article: The Cycle 46 Configuration of the HARMONIE-AROME Forecast Model
The Cycle 46 Configuration of the HARMONIE-AROME Forecast Model Open
The aim of this technical note is to describe the Cycle 46 reference configuration of the HARMONIE-AROME convection-permitting numerical weather prediction (NWP) model. HARMONIE-AROME is a configuration of the shared ALADIN-HIRLAM system, …
View article: First results and future plans for ecRad radiation in Météo-France models
First results and future plans for ecRad radiation in Météo-France models Open
Radiation in the atmosphere provides the energy that drives atmospheric dynamics and physics on all scales, from cloud particle growth to global weather and climate. Radiation schemes in global weather and climate models have to simplify t…
View article: Code and partial data for "Representing sub-grid processes in weather and climate models via sequence learning"
Code and partial data for "Representing sub-grid processes in weather and climate models via sequence learning" Open
This repository contains the RNN training and evaluation code used in the paper for all three parameterization problems: non-orographic gravity wave drag (NOGWD) non-local parameterization (Wang et al. 2022) moist physics (Han et al. 2023,…
View article: Twelve Times Faster yet Accurate: A New State‐Of‐The‐Art in Radiation Schemes via Performance and Spectral Optimization
Twelve Times Faster yet Accurate: A New State‐Of‐The‐Art in Radiation Schemes via Performance and Spectral Optimization Open
Radiation schemes are critical components of Earth system models that need to be both efficient and accurate. Despite the use of approximations such as 1D radiative transfer, radiation can account for a large share of the runtime of expens…
View article: Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0
Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0 Open
Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NNs), obtaining large speed-ups, bu…
View article: Fast computation of cloud 3D radiative effects in dynamical models by optimizing the ecRad scheme
Fast computation of cloud 3D radiative effects in dynamical models by optimizing the ecRad scheme Open
Radiation schemes are fundamental components of weather and climate models that need to be both efficient and accurate. In this work we refactor ecRad, a flexible radiation scheme developed at the European Centre for Medium-Range Weather F…
View article: Reply on RC1
Reply on RC1 Open
Abstract. Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large spee…
View article: Reply on RC2
Reply on RC2 Open
Abstract. Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large spee…
View article: Reply on RC1
Reply on RC1 Open
Abstract. Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large spee…
View article: Reply on AC1
Reply on AC1 Open
Abstract. Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large spee…
View article: Comment on egusphere-2022-1047
Comment on egusphere-2022-1047 Open
Abstract. Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large spee…
View article: Comment on egusphere-2022-1047
Comment on egusphere-2022-1047 Open
Abstract. Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large spee…
View article: Emulating radiative transfer in a numerical weather prediction model
Emulating radiative transfer in a numerical weather prediction model Open
Machine learning, and particularly neural networks, have been touted as a valuable accelerator for physical processes. By training on data generated from an existing algorithm a network may theoretically learn a more efficient representati…
View article: Code and partial data for "Representing sub-grid processes in weather and climate models via sequence learning"
Code and partial data for "Representing sub-grid processes in weather and climate models via sequence learning" Open
This repository contains the RNN training and evaluation code used in the paper Representing sub-grid processes in weather and climate models via sequence learning. Three parameterization problems from earlier studies are included (we have…
View article: Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0
Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0 Open
Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large speed-ups, but…
View article: Reply on AC2
Reply on AC2 Open
Abstract. In numerical weather prediction (NWP) models, physical parameterization schemes are the most computationally expensive components, despite being greatly simplified. In the past few years, an increasing number of studies have demo…
View article: Comment on egusphere-2022-866
Comment on egusphere-2022-866 Open
Abstract. In numerical weather prediction (NWP) models, physical parameterization schemes are the most computationally expensive components, despite being greatly simplified. In the past few years, an increasing number of studies have demo…
View article: Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System
Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System Open
Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large speed-ups, but…
View article: Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System
Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System Open
Radiation schemes are physically important but computationally expensive components of weather and climate models. This has spurred efforts to replace them with a cheap emulator based on neural networks (NN), obtaining large speed-ups, but…
View article: Code and extensive data for training neural networks for radiation, used in "Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0""
Code and extensive data for training neural networks for radiation, used in "Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0"" Open
Data and code used in a paper submitted to JAMES titled : Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System 1) The files ml_training_*.7z contain extensive datasets (in NetCDF format…
View article: Extensive data for training neural networks for radiation. Code and data used in: "Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System"
Extensive data for training neural networks for radiation. Code and data used in: "Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System" Open
Data and code used in a paper submitted to JAMES titled : Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System 1) The files ml_training_*.7z contain extensive datasets (in NetCDF format…
View article: Code and extensive data for training neural networks for radiation, used in "Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0""
Code and extensive data for training neural networks for radiation, used in "Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System: RRTMGP-NN 2.0"" Open
Data and code used in a paper submitted to JAMES titled : Implementation of a machine-learned gas optics parameterization in the ECMWF Integrated Forecasting System 1) The files ml_training_*.7z contain extensive datasets (in NetCDF format…