Abstract The problem of augmenting low-resolution ocean circulation models with the information extracted from the data relevant to the unresolved subgrid processes is addressed. A highly nonlinear model of eddy-resolving oceanic circulation – quasigeostrophic wind-driven double gyres – is considered. The model solutions are characterized by a vigorous dynamic coupling between the resolved large-scale and small-scale (eddy) flow features. This solution provides the data for augmenting the low-resolution model with the same configuration. The eddy forcing field, which contains the essential information about coupling between the large and eddy scales, is obtained, modified, coarse-grained and added to augment the low-resolution model. The implemented modification involves novel data-adaptive harmonic decomposition analysis and dynamical constraining based on the low-resolution nonlinear advection operator. The resulting augmentation of the low-resolution model significantly improves the solution, including its time-mean circulation and low-frequency variability. This result also paves the way for a systematic data-driven emulation of unresolved and under-resolved scales of motion.

中文翻译：

---

低分辨率海洋模型动力学的数据驱动增强

摘要 解决了从与未解析子网格过程相关的数据中提取的信息来增强低分辨率海洋环流模型的问题。考虑了涡旋解决海洋环流的高度非线性模型——准地转风驱动的双环流。模型解决方案的特点是解析的大尺度和小尺度（涡流）流动特征之间存在强烈的动态耦合。该解决方案提供了用于增强具有相同配置的低分辨率模型的数据。涡强迫场包含关于大尺度和涡尺度之间耦合的基本信息，被获取、修改、粗粒度并添加以增强低分辨率模型。实施的修改涉及基于低分辨率非线性对流算子的新型数据自适应谐波分解分析和动态约束。由此产生的低分辨率模型增强显着改善了解决方案，包括其时间平均循环和低频可变性。这一结果也为对未解析和未解析的运动尺度进行系统的数据驱动模拟铺平了道路。