4.6 Article

A Baseline for Global Weather and Climate Simulations at 1 km Resolution

期刊

出版社

AMER GEOPHYSICAL UNION
DOI: 10.1029/2020MS002192

关键词

explicitly simulated convection; atmosphere; winter season; high performance computing; stratosphere; MJO

资金

  1. DOE Office of Science User Facility [DE-AC05-00OR22725]
  2. European Union [800897, 801101, 754337, 823988]
  3. [CLI900]

向作者/读者索取更多资源

In an attempt to advance the understanding of the Earth's weather and climate by representing deep convection explicitly, we present a global, four-month simulation (November 2018 to February 2019) with ECMWF's hydrostatic Integrated Forecasting System (IFS) at an average grid spacing of 1.4 km. The impact of explicitly simulating deep convection on the atmospheric circulation and its variability is assessed by comparing the 1.4 km simulation to the equivalent well-tested and calibrated global simulations at 9 km grid spacing with and without parametrized deep convection. The explicit simulation of deep convection at 1.4 km results in a realistic large-scale circulation, better representation of convective storm activity, and stronger convective gravity wave activity when compared to the 9 km simulation with parametrized deep convection. Comparison of the 1.4 km simulation to the 9 km simulation without parametrized deep convection shows that switching off deep convection parametrization at a too coarse resolution (i.e., 9 km) generates too strong convective gravity waves. Based on the limited statistics available, improvements to the Madden-Julian Oscillation or tropical precipitation are not observed at 1.4 km, suggesting that other Earth system model components and/or their interaction are important for an accurate representation of these processes and may well need adjusting at deep convection resolving resolutions. Overall, the good agreement of the 1.4 km simulation with the 9 km simulation with parametrized deep convection is remarkable, despite one of the most fundamental parametrizations being turned off at 1.4 km resolution and despite no adjustments being made to the remaining parametrizations.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.6
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

Editorial Material Multidisciplinary Sciences

Opportunities and challenges for machine learning in weather and climate modelling: hard, medium and soft AI

Matthew Chantry, Hannah Christensen, Peter Dueben, Tim Palmer

Summary: The workshop in September 2019 highlighted the growing area of applying machine learning to improve weather and climate prediction, outlining motivations, opportunities, and challenges ahead. This article is part of the theme issue on 'Machine learning for weather and climate modelling'.

PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY A-MATHEMATICAL PHYSICAL AND ENGINEERING SCIENCES (2021)

Article Multidisciplinary Sciences

Deep learning for post-processing ensemble weather forecasts

Peter Gronquist, Chengyuan Yao, Tal Ben-Nun, Nikoli Dryden, Peter Dueben, Shigang Li, Torsten Hoefler

Summary: The study introduces a mixed model that utilizes a subset of original weather trajectories combined with deep neural networks for post-processing to improve ensemble forecast skills. The model shows a relative improvement of over 14% in ensemble forecast skill (CRPS), with larger improvements seen for extreme weather events on select case studies. By using fewer trajectories, computational costs can be reduced while maintaining comparable results to the full ensemble.

PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY A-MATHEMATICAL PHYSICAL AND ENGINEERING SCIENCES (2021)

Review Environmental Sciences

Bridging observations, theory and numerical simulation of the ocean using machine learning

Maike Sonnewald, Redouane Lguensat, Daniel C. Jones, Peter D. Dueben, Julien Brajard, V Balaji

Summary: Progress in physical oceanography has been enhanced by the use of machine learning techniques, offering new possibilities for study and discovery. Challenges unique to ocean study, such as sparse spatial data and limited time series, can be addressed with ML. The use of ML in observations, theory, and numerical modeling shows promising opportunities for advancing oceanographic exploration.

ENVIRONMENTAL RESEARCH LETTERS (2021)

Article Meteorology & Atmospheric Sciences

A Comparison of Data-Driven Approaches to Build Low-Dimensional Ocean Models

Niraj Agarwal, D. Kondrashov, P. Dueben, E. Ryzhov, P. Berloff

Summary: This study conducted a comprehensive comparison of LR, stochastic, and deep-learning approaches for ocean circulation emulation, with the multi-level linear stochastic method performing the best for both short- and long-timescale forecasts. LR with white noise extension showed better performance on long timescales, while pure deep learning implementations performed worse than LR and its white noise augmentation. Overall, multi-level LR stochastic models and hybrid models augmented by stochastic terms learned via deep learning were found to be more practical, accurate, and cost-effective for ocean emulation.

JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS (2021)

Article Meteorology & Atmospheric Sciences

The role of the timing of sudden stratospheric warmings for precipitation and temperature anomalies in Europe

Erika Monnin, Marlene Kretschmer, Inna Polichtchouk

Summary: The study found that in Europe, sudden stratospheric warmings (SSWs) occurring in early winter have a more pronounced surface impact compared to SSWs occurring later in the winter.

INTERNATIONAL JOURNAL OF CLIMATOLOGY (2022)

Article Meteorology & Atmospheric Sciences

Resolved gravity waves in the tropical stratosphere: Impact of horizontal resolution and deep convection parametrization

Inna Polichtchouk, Nils Wedi, Young-Ha Kim

Summary: Parametrized deep convection inhibits CGGWs, while explicitly resolved deep convection is almost unaffected by horizontal resolutions. As resolution increases, the contribution of small-scale CGGWs increases and that of long and mesoscale waves decreases.

QUARTERLY JOURNAL OF THE ROYAL METEOROLOGICAL SOCIETY (2022)

Article Meteorology & Atmospheric Sciences

Machine Learning Emulation of Urban Land Surface Processes

David Meyer, Sue Grimmond, Peter Dueben, Robin Hogan, Maarten van Reeuwijk

Summary: The study developed an urban neural network (UNN) to train on the mean predicted fluxes from multiple urban land surface models (ULSMs), showing improved stability and accuracy. By combining the strengths of multiple ULSMs, the UNN successfully improved the modeling of surface fluxes.

JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS (2022)

Article Meteorology & Atmospheric Sciences

Machine Learning Emulation of 3D Cloud Radiative Effects

David Meyer, Robin J. Hogan, Peter D. Dueben, Shannon L. Mason

Summary: In this study, we propose to use computationally cheap neural networks to correct the treatment of cloud structure in numerical weather and climate models. By learning the difference between the 1D radiation scheme and the 3D SPARTACUS solver, which includes 3D cloud effects but is computationally expensive, neural networks can improve the accuracy of the models without significantly increasing the runtime.

JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS (2022)

Article Meteorology & Atmospheric Sciences

Deep Learning to Estimate Model Biases in an Operational NWP Assimilation System

Patrick Laloyaux, Thorsten Kurth, Peter Dominik Dueben, David Hall

Summary: Model bias is a major obstacle to accuracy and reliability in numerical weather prediction, and the weak-constraint 4D-Var method has been developed to correct bias accumulated along model trajectories. Deep learning is used to correct model bias using temperature retrievals from radio occultation measurements.

JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS (2022)

Article Meteorology & Atmospheric Sciences

Gravity-Wave-Driven Seasonal Variability of Temperature Differences Between ECMWF IFS and Rayleigh Lidar Measurements in the Lee of the Southern Andes

Sonja Gisinger, Inna Polichtchouk, Andreas Doernbrack, Robert Reichert, Bernd Kaifler, Natalie Kaifler, Markus Rapp, Irina Sandu

Summary: This study evaluates temperature and gravity wave activity in ECMWF Integrated Forecasting System (IFS) over Rio Grande using long-term high-resolution temperature data from Compact Rayleigh Autonomous Lidar (CORAL). The results show that seasonal and altitudinal variations play an important role in explaining the temperature differences between IFS and lidar, and the gravity wave activity in the middle atmosphere is a major driver of seasonal variability.

JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES (2022)

Article Meteorology & Atmospheric Sciences

Resolved Gravity Waves in the Extratropical Stratosphere: Effect of Horizontal Resolution Increase from O(10) to O(1) km

Inna Polichtchouk, Annelize Van Niekerk, Nils Wedi

Summary: Global ECMWF IFS simulations are used to assess gravity wave forcing in the extratropical stratosphere at different horizontal resolutions. The results indicate a doubling in zonal-mean resolved gravity wave forcing with higher horizontal resolution. The parameterization of gravity wave forcing is still required at 9 and 4 km resolutions, and a rebalancing of the parameterized nonorographic and orographic gravity wave forcing is suggested.

JOURNAL OF THE ATMOSPHERIC SCIENCES (2023)

Article Meteorology & Atmospheric Sciences

Mixed-Precision for Linear Solvers in Global Geophysical Flows

Jan Ackmann, Peter D. Dueben, Tim Palmer, Piotr K. Smolarkiewicz

Summary: This paper investigates the potential computational savings of using mixed precision arithmetic in elliptic solvers for atmosphere and ocean models. The study shows that for certain key components, using half precision can achieve a speed-up of 4 times compared to double precision.

JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS (2022)

Article Meteorology & Atmospheric Sciences

A Generative Deep Learning Approach to Stochastic Downscaling of Precipitation Forecasts

Lucy Harris, Andrew T. T. McRae, Matthew Chantry, Peter D. Dueben, Tim N. Palmer

Summary: In this study, the authors utilize generative adversarial networks (GANs) to improve the accuracy and resolution of low-resolution weather forecasting model outputs, using high-resolution radar measurements as ground truth. By learning to add resolution and structure, GANs can generate high-resolution and spatially coherent precipitation maps, matching the statistical properties of state-of-the-art downscaling methods.

JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS (2022)

Article Geosciences, Multidisciplinary

Model intercomparison of COSMO 5.0 and IFS 45r1 at kilometer-scale grid spacing

Christian Zeman, Nils P. Wedi, Peter D. Dueben, Nikolina Ban, Christoph Schar

Summary: The study compares the performance of the COSMO model and ECMWF's IFS model in simulating strong convective days over Europe, finding that for deep convective processes, time step and horizontal diffusion are important factors affecting model resolution.

GEOSCIENTIFIC MODEL DEVELOPMENT (2021)

Article Meteorology & Atmospheric Sciences

Tropical Cyclones in Global Storm-Resolving Models

Falko Judt, Daniel Klocke, Rosimar Rios-Berrios, Benoit Vanniere, Florian Ziemen, Ludovic Auger, Joachim Biercamp, Christopher Bretherton, Xi Chen, Peter Dueben, Cathy Hohenegger, Marat Khairoutdinov, Chihiro Kodama, Luis Kornblueh, Shian-Jiann Lin, Masuo Nakano, Philipp Neumann, William Putman, Niklas Roeber, Malcolm Roberts, Masaki Satoh, Ryosuke Shibuya, Bjorn Stevens, Pier Luigi Vidale, Nils Wedi, Linjiong Zhou

Summary: Recent advancements in computing and model development have ushered in the era of global storm-resolving modeling, showing great potential to revolutionize weather and climate prediction. This study evaluates the performance of nine global-storm resolving models in simulating tropical cyclones, with results showing that while they produce realistic TCs overall, each model has unique biases in terms of TC number, intensity, size, and structure. Some models perform better than others in simulating TCs, indicating the need for improvement to unleash the full potential of global storm-resolving models in TC prediction.

JOURNAL OF THE METEOROLOGICAL SOCIETY OF JAPAN (2021)

暂无数据