Benchmarking of Machine Learning Ocean Subgrid Parameterizations in an Idealized Model
Benchmarking of Machine Learning Ocean Subgrid Parameterizations in an Idealized Model
复制标题
DOI：
10.1029/2022ms003258
复制
发表时间：
2022-12
影响因子：
6.8
通讯作者：
Andrew Ross;Ziwei Li;P. Perezhogin;C. Fernandez‐Granda;L. Zanna
中科院分区：
文献类型：
--
作者：
Andrew Ross;Ziwei Li;P. Perezhogin;C. Fernandez‐Granda;L. Zanna
最近，越来越多的研究使用机器学习（ML）模型来参数化海洋模型中计算密集的亚网格尺度过程。这些研究通常使用过滤和粗粒度的高分辨率数据来训练ML模型，并离线评估其预测性能，然后在粗分辨率模型中实现它们并评估其在线性能。在这项工作中，我们系统地基准测试了这些模型的在线性能，它们对训练过程中未遇到的领域的泛化，以及它们对数据集设计选择的敏感性。我们应用这个框架来比较大量基于物理和神经网络（NN）的参数化。我们发现过滤和粗粒化算子的选择特别关键，这种选择应该由应用程序指导。我们还表明，我们所有的物理约束NN都是稳定的，并且在在线实现时表现良好，但对新的制度推广效果不佳。为了提高泛化能力和可解释性，我们提出了一种新的方程发现方法，将线性回归和遗传编程与空间导数相结合。我们发现这种方法在训练域上的表现与神经网络相当，但在训练域之外的推广效果更好。我们发布代码和数据来重现我们的结果，并为研究社区提供易于使用的资源来开发和评估其他参数化。
Recently, a growing number of studies have used machine learning (ML) models to parameterize computationally intensive subgrid‐scale processes in ocean models. Such studies typically train ML models with filtered and coarse‐grained high‐resolution data and evaluate their predictive performance offline, before implementing them in a coarse resolution model and assessing their online performance. In this work, we systematically benchmark the online performance of such models, their generalization to domains not encountered during training, and their sensitivity to data set design choices. We apply this proposed framework to compare a large number of physical and neural network (NN)‐based parameterizations. We find that the choice of filtering and coarse‐graining operator is particularly critical and this choice should be guided by the application. We also show that all of our physics‐constrained NNs are stable and perform well when implemented online, but generalize poorly to new regimes. To improve generalization and also interpretability, we propose a novel equation‐discovery approach combining linear regression and genetic programming with spatial derivatives. We find this approach performs on par with neural networks on the training domain but generalizes better beyond it. We release code and data to reproduce our results and provide the research community with easy‐to‐use resources to develop and evaluate additional parameterizations.