Machine Learning Global Simulation of Nonlocal Gravity Wave Propagation
0
Sign in to get full access
Overview
- This paper presents a machine learning approach to simulate the global propagation of atmospheric gravity waves, which are an important but challenging aspect of climate modeling.
- The researchers developed a neural network model that can emulate the effects of gravity waves on the large-scale atmospheric circulation, without having to explicitly resolve the waves themselves.
- This allows for more computationally efficient global climate simulations while still capturing the important influence of gravity waves.
- The paper builds on previous work on using machine learning for sub-grid scale parameterizations and hybrid physics-based and data-driven modeling in Earth system models.
Plain English Explanation
Atmospheric gravity waves are ripples or disturbances in the air that play an important role in Earth's weather and climate. These waves are caused by things like airflow over mountains or convection in thunderstorms. However, they are difficult to model accurately in global climate simulations because the waves are much smaller in scale than the model grid.
The researchers in this paper found a way to use machine learning to approximate the effects of these small-scale gravity waves on the large-scale atmospheric circulation. They trained a neural network model to learn the relationship between the properties of the gravity waves and their impact on factors like wind and temperature in the atmosphere.
By using this machine learning model instead of explicitly resolving the gravity waves, the researchers were able to run global climate simulations much more efficiently, without losing the important influence of the waves. This is a promising approach for improving the representation of sub-grid scale processes in Earth system models.
Technical Explanation
The key technical innovation in this paper is the development of a neural network model that can emulate the effects of atmospheric gravity waves on the large-scale atmospheric circulation. The model takes as input variables like wind, temperature, and humidity profiles, and outputs the tendencies (rates of change) for these variables due to the unresolved gravity wave effects.
The neural network architecture consists of multiple fully-connected layers that map the input variables to the output tendencies. The model is trained on data from high-resolution simulations that explicitly resolve the gravity waves, allowing the neural network to learn the nonlinear relationship between the wave properties and their impacts on the resolved-scale flow.
In experiments, the researchers coupled this neural network gravity wave parameterization into a global atmospheric model and found that it could reproduce the large-scale circulation patterns and variability as well as a model that explicitly resolved the gravity waves, but at a fraction of the computational cost.
Critical Analysis
One key limitation mentioned in the paper is the reliance on high-resolution simulation data for training the neural network model. In practice, such data may not always be available, especially for certain regions or time periods. The authors note that further research is needed to explore techniques for building emulators from sparse or heterogeneous data.
Additionally, the neural network approach assumes that the relationship between the gravity wave properties and their impacts can be accurately captured by the model architecture and training data. There may be complex, nonlinear interactions or feedback loops that are difficult to fully represent in this type of emulator. Exploring hybrid approaches that combine machine learning and physics-based modeling could help address this limitation.
Finally, the paper focuses on the ability of the neural network model to reproduce the large-scale circulation patterns, but does not extensively evaluate its performance on other relevant metrics like precipitation, extreme events, or regional climate features. Further research would be needed to fully assess the capabilities and limitations of this approach for comprehensive Earth system modeling.
Conclusion
This paper presents an innovative machine learning approach to simulating the global propagation of atmospheric gravity waves, a key component of Earth's climate system. By using a neural network to emulate the effects of these small-scale waves, the researchers were able to develop a computationally efficient global climate model that still captures the important influence of gravity waves.
This work builds on previous efforts to embed machine-learned sub-grid scale variability and explore hybrid physics-based and data-driven modeling in Earth system models. The authors have demonstrated the potential of this approach to improve the representation of sub-grid scale processes and enhance the computational efficiency of global climate simulations.
Further research is needed to address the limitations of this approach, such as the reliance on high-resolution training data and the challenges of fully capturing complex nonlinear interactions. But overall, this work represents an exciting step forward in the use of machine learning for more accurate and efficient Earth system modeling.
This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!
Related Papers
0
Machine Learning Global Simulation of Nonlocal Gravity Wave Propagation
Aman Gupta, Aditi Sheshadri, Sujit Roy, Vishal Gaur, Manil Maskey, Rahul Ramachandran
Global climate models typically operate at a grid resolution of hundreds of kilometers and fail to resolve atmospheric mesoscale processes, e.g., clouds, precipitation, and gravity waves (GWs). Model representation of these processes and their sources is essential to the global circulation and planetary energy budget, but subgrid scale contributions from these processes are often only approximately represented in models using parameterizations. These parameterizations are subject to approximations and idealizations, which limit their capability and accuracy. The most drastic of these approximations is the single-column approximation which completely neglects the horizontal evolution of these processes, resulting in key biases in current climate models. With a focus on atmospheric GWs, we present the first-ever global simulation of atmospheric GW fluxes using machine learning (ML) models trained on the WINDSET dataset to emulate global GW emulation in the atmosphere, as an alternative to traditional single-column parameterizations. Using an Attention U-Net-based architecture trained on globally resolved GW momentum fluxes, we illustrate the importance and effectiveness of global nonlocality, when simulating GWs using data-driven schemes.
Read more6/24/2024
0
On the importance of learning non-local dynamics for stable data-driven climate modeling: A 1D gravity wave-QBO testbed
Hamid A. Pahlavan, Pedram Hassanzadeh, M. Joan Alexander
Machine learning (ML) techniques, especially neural networks (NNs), have shown promise in learning subgrid-scale parameterizations for climate models. However, a major problem with data-driven parameterizations, particularly those learned with supervised algorithms, is model instability. Current remedies are often ad-hoc and lack a theoretical foundation. Here, we combine ML theory and climate physics to address a source of instability in NN-based parameterization. We demonstrate the importance of learning spatially $textit{non-local}$ dynamics using a 1D model of the quasi-biennial oscillation (QBO) with gravity wave (GW) parameterization as a testbed. While common offline metrics fail to identify shortcomings in learning non-local dynamics, we show that the concept of receptive field (RF) can identify instability a-priori. We find that NN-based parameterizations that seem to accurately predict GW forcings from wind profiles ($mathbf{R^2 approx 0.99}$) cause unstable simulations when RF is too small to capture the non-local dynamics, while NNs of the same size but large-enough RF are stable. We examine three broad classes of architectures, namely convolutional NNs, Fourier neural operators, and fully-connected NNs; the latter two have inherently large RFs. We also demonstrate that learning non-local dynamics is crucial for the stability and accuracy of a data-driven spatiotemporal emulator of the zonal wind field. Given the ubiquity of non-local dynamics in the climate system, we expect the use of effective RF, which can be computed for any NN architecture, to be important for many applications. This work highlights the necessity of integrating ML theory with physics to design and analyze data-driven algorithms for weather and climate modeling.
Read more7/17/2024
0
Probabilistic Emulation of a Global Climate Model with Spherical DYffusion
Salva Ruhling Cachay, Brian Henn, Oliver Watt-Meyer, Christopher S. Bretherton, Rose Yu
Data-driven deep learning models are on the verge of transforming global weather forecasting. It is an open question if this success can extend to climate modeling, where long inference rollouts and data complexity pose significant challenges. Here, we present the first conditional generative model able to produce global climate ensemble simulations that are accurate and physically consistent. Our model runs at 6-hourly time steps and is shown to be stable for 10-year-long simulations. Our approach beats relevant baselines and nearly reaches a gold standard for successful climate model emulation. We discuss the key design choices behind our dynamics-informed diffusion model-based approach which enables this significant step towards efficient, data-driven climate simulations that can help us better understand the Earth and adapt to a changing climate.
Read more6/24/2024
0
Exploring the Potential of Hybrid Machine-Learning/Physics-Based Modeling for Atmospheric/Oceanic Prediction Beyond the Medium Range
Dhruvit Patel, Troy Arcomano, Brian Hunt, Istvan Szunyogh, Edward Ott
This paper explores the potential of a hybrid modeling approach that combines machine learning (ML) with conventional physics-based modeling for weather prediction beyond the medium range. It extends the work of Arcomano et al. (2022), which tested the approach for short- and medium-range weather prediction, and the work of Arcomano et al. (2023), which investigated its potential for climate modeling. The hybrid model used for the forecast experiments of the paper is based on the low-resolution, simplified parameterization atmospheric general circulation model (AGCM) SPEEDY. In addition to the hybridized prognostic variables of SPEEDY, the current version of the model has three purely ML-based prognostic variables. One of these is 6~h cumulative precipitation, another is the sea surface temperature, while the third is the heat content of the top 300 m deep layer of the ocean. The model has skill in predicting the El Ni~no cycle and its global teleconnections with precipitation for 3-7 months depending on the season. The model captures equatorial variability of the precipitation associated with Kelvin and Rossby waves and MJO. Predictions of the precipitation in the equatorial region have skill for 15 days in the East Pacific and 11.5 days in the West Pacific. Though the model has low spatial resolution, for these tasks it has prediction skill comparable to what has been published for high-resolution, purely physics-based, conventional operational forecast models.
Read more5/31/2024