物理海洋AI交叉热门领域

物理海洋学与人工智能交叉前沿研究报告:范式转移、基座模型与数字孪生(2024-2026)

1. 执行摘要与引言:物理海洋学的“第四范式”

物理海洋学,作为地球系统科学的核心分支,正处于一场深刻的认识论与方法论革命之中。长期以来,该学科依赖于基于物理定律的演绎推理(第一范式)和基于观测的归纳总结(第二范式),并在过去半个世纪中通过高性能计算模拟(第三范式)取得了长足进步。然而,随着海量观测数据(卫星遥感、Argo浮标、海底观测网)的爆发式增长以及计算算力的飞跃,人工智能(AI)驱动的第四范式正在重塑我们理解和预测海洋的方式 。[1][2]

当前,物理海洋学与人工智能的交叉已不再局限于简单的统计分析或后处理(即“软AI”),而是深入到了核心动力学过程的模拟、参数化方案的构建以及全耦合地球系统模型的开发中(即“硬AI”)。2023年至2025年的文献表明,这一交叉领域的热点主要集中在以下几个战略方向:基于Transformer和深度学习的全球海洋预报大模型(Foundation Models)科学机器学习(SciML)在次网格过程参数化中的应用“深蓝AI”在海洋遥感反演与三维重构中的突破,以及数字孪生海洋(Digital Twin Ocean, DTO)的构建

本报告旨在详尽梳理上述热点领域的最新进展、核心技术路径及未来趋势。基于广泛的文献调研,我们将深入剖析AI如何从单纯的数据拟合工具演变为具备物理一致性的动力学模拟器,并探讨这一转型过程中面临的可解释性、泛化能力及数据基础设施挑战。

2. 全球海洋预报大模型:从数值解算到智能推演

在物理海洋学与AI的所有交叉领域中,最具颠覆性且目前最“炙手可热”的方向无疑是AI驱动的全球海洋预报系统(AI-GOFS)。传统的数值海洋模式(如NEMO, HYCOM, MOM6)受限于Courant-Friedrichs-Lewy (CFL) 条件和高昂的计算成本,难以在极高分辨率下进行快速的集合预报。而新一代AI基座模型(Foundation Models)通过学习海量再分析数据中的非线性映射关系,正在打破这一瓶颈,实现了在秒级时间内完成通过传统方法需数千核·小时计算量的预报任务 。[1:1][3][4]

2.1 羲和(XiHe):涡旋分辨级数据驱动模型的突破

由中国国防科技大学团队开发的**羲和(XiHe)**模型,代表了纯数据驱动海洋环流模式在涡旋分辨尺度上的重大突破。

2.1.1 架构创新与时空建模

XiHe模型的设计核心在于解决传统卷积神经网络(CNN)在处理全球尺度海洋数据时对长距离空间依赖捕捉不足的问题。该模型引入了一种新颖的海洋专用模块,包含局部和全球空间信息提取(Spatial Information Extraction, SIE)组件。这种设计使得模型不仅能够捕捉局部的细微动力学特征(如锋面、亚中尺度涡),还能有效学习全球尺度的遥相关模式(Teleconnections)。[4:1]

为了处理海洋与陆地的复杂边界,XiHe引入了海陆掩膜机制(Ocean-Land Masking Mechanism)。这一机制在训练过程中强制模型忽略陆地网格点,专注于学习海洋网格的信息,从而消除了陆地数据对海洋动力学特征学习的干扰(即“伪影”问题),显著提升了沿岸区域的预报精度 。[4:2]

2.1.2 性能表现与计算效率

基于GLORYS12再分析数据(1993-2017年作为训练集),XiHe实现了1/12的高空间分辨率预报。在IV-TT Class 4标准化评估框架下的测试表明,XiHe对海洋流场的60天预报精度甚至优于传统数值业务系统PSY4的10天预报精度 。[4:3]

更令人瞩目的是其计算效率的提升。XiHe在单块GPU上平均仅需0.36秒即可完成全球海洋未来10天的全要素预报。这一速度比传统数值GOFS快了数千倍,为开展大规模集合预报、不确定性量化以及实时应急响应(如溢油扩散、搜救漂移预测)提供了前所未有的可能性 。[5]

2.2 伏羲海洋(FuXi-Ocean):次日尺度与多时间尺度融合

复旦大学与上海人工智能实验室推出的**伏羲海洋(FuXi-Ocean)**模型,进一步将AI预报的能力拓展到了次日(Sub-daily)尺度和多物理场耦合。

2.2.1 混合时间(Mixture-of-Time)架构

长期预报中的累积误差是自回归AI模型面临的主要挑战。FuXi-Ocean创新性地提出了混合时间(MoT)模块,该模块能够自适应地整合来自不同时间上下文的预测结果。通过学习变量在不同时间步长上的可靠性权重,MoT有效缓解了序列预报中的误差漂移问题,使得模型在长达数十天的预报中仍能保持较高的物理一致性 。[6]

2.2.2 全要素耦合与极端事件捕捉

不同于仅关注流场或温盐的模型,FuXi-Ocean对五大基本海洋变量进行联合预报:温度(T)、盐度(S)、纬向流(U)、经向流(V)以及海表面高度(SSH)。更重要的是,该模型在输入端纳入了大气变量(如10米风场),实际上构建了一个松散耦合的海气预报系统。这种跨圈层的信息输入使得FuXi-Ocean在捕捉受强大气强迫驱动的极端海洋事件方面表现卓越 。[7]

实证研究显示,在台风“艾利”(Aere)期间,FuXi-Ocean展现出了对高频海洋响应的动态追踪能力,能够准确预报台风引起的冷尾迹(Cold Wake)和流场剧烈变化,其在极端事件下的均方根误差(RMSE)显著低于气候态预测和部分传统数值模式 。[8]

2.3 风乌(FengWu)与Prithvi:无缝预报与通用基座

AI在气象领域的成功正在向海洋领域快速渗透,推动了无缝预报(Seamless Forecasting)通用基座模型的发展。

2.3.1 风乌-W2S的跨尺度能力

**风乌-W2S(FengWu-Weather to Subseasonal)模型致力于打破天气预报(0-14天)与次季节预报(3-6周)之间的界限。通过自回归的方式,它能够生成长达42天的6小时分辨率预报。该模型在预测麦厄-朱利安振荡(MJO)**等海气耦合现象上表现出了惊人的技巧,将MJO的有效预报时效延长至36天。这表明AI模型已经具备了捕捉大气与海洋之间慢变耦合过程的能力,为未来一体化的地球系统AI模型奠定了基础 。[9][10]

2.3.2 Prithvi WxC:NASA/IBM的开源基座

Prithvi WxC代表了另一种发展路径:构建通用的地球科学基座模型。该模型基于Vision Transformer (ViT) 架构,在MERRA-2再分析数据和Sentinel卫星数据上进行了大规模预训练。Prithvi WxC不仅用于天气预报,还被设计用于下游任务的微调(Fine-tuning),如重力波参数化极端天气事件估计以及海洋叶绿素反演。这种“预训练+微调”的范式极大地降低了针对特定海洋任务开发AI模型的门槛 。[11][12]

2.4 AI模型与数值模式的综合对比分析

为了更直观地理解AI模型在物理海洋学中的地位,我们将主流AI模型与传统数值模式进行了对比(见表1)。

表 1:2024-2025年主流AI海洋预报模型与传统数值模式对比

维度 传统数值模式 (如 NEMO, HYCOM) XiHe (国防科大) FuXi-Ocean (复旦/上海AI Lab) OceanRep (AWI)
核心机制 离散化求解Navier-Stokes方程 深度学习非线性映射 (CNN/Attention) 混合时间架构 (MoT) 基于Vision Transformer (ViT)
空间分辨率 受算力限制 (通常 1/12°) 1/12° (涡旋分辨级) 1/12° 多分辨率支持
计算耗时 极高 (数千核小时/10天预报) 极低 (0.36秒/10天预报) 低 (秒级)
物理守恒 显式守恒 (质量、能量、动量) 隐式学习 (存在漂移风险) 优化误差累积 物理一致性增强
输入数据 初始场 + 边界强迫 历史再分析数据 (GLORYS12等) 多源融合数据 (ERA5/ORAS5) 模拟数据 (FESOM2)
优势领域 长期气候态、机理研究 短期/中期快速预报、集合预报 极端事件、次日高频过程 长期动力学模拟

[4:4]
深度洞察: 这一领域的竞争正在从单纯的“物理方程求解精度”转向“数据架构的表达能力”。然而,AI模型目前仍面临分布外泛化(Out-of-Distribution, OOD)的挑战。当海洋系统因气候变化进入前所未有的状态(如极端高温、环流崩溃)时,基于历史数据训练的AI模型可能会失效,而基于物理定律的数值模式理论上更具鲁棒性。因此,未来的主流方向必然是二者的融合——即混合建模(Hybrid Modeling)[13][14]

3. 科学机器学习(SciML):解决次网格参数化难题

如果说基座模型试图替代整个动力学核心,那么**科学机器学习(Scientific Machine Learning, SciML)**则致力于解决传统物理海洋学模型中的“阿喀琉斯之踵”——次网格过程参数化(Subgrid-Scale Parameterization)

3.1 闭合问题与数据驱动方案

海洋模型无法分辨所有尺度的运动。小尺度过程(如湍流混合、内波破碎、亚中尺度涡)必须通过参数化方案来近似其对大尺度流场的影响。传统的参数化方案(如KPP方案)往往基于半经验公式,存在较大的不确定性,导致气候模型中存在系统性偏差(如热带太平洋冷舌偏差、混合层深度误差) 。[15][16]

3.1.1 从DNS/LES中学习

研究者们正在利用高分辨率的**直接数值模拟(DNS)大涡模拟(LES)生成“真值”数据,训练神经网络来预测次网格应力张量或热通量。最新的研究(2024-2025)不再局限于局部的梯度扩散假设,而是利用卷积神经网络(CNN)或Transformer捕捉非局部(Non-local)各向异性(Anisotropic)**的特征。例如,基于CNN的参数化方案能够根据周围流场的形态(如涡旋的拉伸变形)来决定能量的耗散或反向散射(Backscatter),这在传统方案中是难以实现的 。[17][18]

3.1.2 垂直混合参数化的革新

在垂直混合方面,深度学习模型被用于替代经典的K-Profile Parameterization (KPP) 方案。通过利用长期的热带太平洋水文和湍流观测数据进行训练,改进后的神经网络能够更准确地捕捉浮力频率(N2)与混合效率之间的非线性关系,显著减少了模拟的偏差 。[15:1][16:1]

3.2 物理信息神经网络(PINNs)与约束学习

为了克服纯数据驱动模型的物理不一致性,**物理信息神经网络(Physics-Informed Neural Networks, PINNs)**及其变体成为了研究热点。

3.2.1 软约束与硬约束

PINNs的核心思想是将物理方程(如浅水方程、连续性方程)作为正则化项加入到损失函数中(Loss=Lossdata+λLossphysics)。

3.2.2 方程发现(Equation Discovery)

除了训练黑盒网络,AI还被用于从数据中“发现”显式的物理方程。利用**稀疏回归(Sparse Regression)**技术(如SINDy算法),研究者可以从高分辨率海洋数据中筛选出控制流体运动的主导项,从而构建出既简洁又可解释的代数方程组。这种方法在发现未知的湍流闭合项或修正现有的动力学方程方面展现出巨大潜力 。[20][21]

4. “深蓝AI”:海洋遥感、反演与三维重构

“深蓝AI”(Deep Blue AI)是指利用AI技术将海洋观测能力从卫星可见的“表层”拓展到不可见的“内部”及“微观”尺度的前沿领域 。[22][23]

4.1 表面到内部(Surface-to-Interior)的反演

物理海洋学的一个经典难题是如何仅凭海表数据(SST, SSH, SSS)推断海洋内部的三维结构。

4.1.1 动力学模态学习

基于再分析数据或模式输出,深度学习模型(特别是CNN和Transformer)能够学习海表特征与内部温盐剖面之间的非线性映射关系。

4.2 超分辨率与数据修复(Inpainting)

卫星数据面临两大挑战:云层遮挡(导致SST缺失)和时空分辨率不足(导致亚中尺度过程模糊)。

4.2.1 基于扩散模型(Diffusion Models)的超分辨率

2024-2025年的一个显著趋势是将生成式AI(Generative AI)引入海洋数据处理。特别是扩散模型(Diffusion Models),因其强大的纹理生成能力,正逐渐取代传统的插值方法。

4.2.2 引导式超分(Guided Super-Resolution)

利用高分辨率的SST影像(通常易于获取)来引导低分辨率SSH数据(受限于卫星轨道间距)的降尺度。AI模型学习SST锋面与SSH梯度之间的强相关性,从而在没有直接高度计观测的区域“想象”出合理的地转流结构 。[27]

4.3 智能特征检测:从CNN到Vision Transformer

自动化检测海洋现象(如内波、涡旋、海冰)正从基于阈值的算法向语义分割模型演进。

5. 数字孪生海洋(DTO):从模拟到决策智能

数据、模型与AI的深度融合最终指向了一个宏大的目标:构建数字孪生海洋(Digital Twin Ocean, DTO)。这不仅是物理海洋学的高级应用,更是服务于政策制定和海洋治理的决策支持系统。

5.1 全球与区域倡议

5.2 关键技术:AI代理模型(Surrogate Modeling)

在DTO中,AI不仅是预报工具,更是交互引擎。传统的数值模型运行一次可能需要数小时,无法满足决策者实时交互的需求。AI代理模型通过学习物理模型的输入输出关系,能够在毫秒级时间内给出近似解,从而支撑动态的风险评估和规划 。[39][40]

6. 交叉前沿与具体应用场景

6.1 海洋二氧化碳移除(mCDR)的监测与核查

随着应对气候变化的紧迫性增加,**海洋二氧化碳移除(mCDR)**技术(如碱度增强、人工上升流)成为研究热点。然而,如何从充满自然变率的海洋背景中检测出微弱的人为碳汇信号是巨大的挑战(MRV难题)。

6.2 台风-海洋耦合强度预报

台风强度的预报(特别是快速增强过程)一直受限于对海洋热含量(OHC)估计的不足。

6.3 物理信息水深反演(Bathymetry Inversion)

利用PINNs进行浅水区地形反演是另一个新兴热点。通过观测表面的波浪传播特征或流场变化,利用浅水方程作为物理约束,PINNs可以反推出海底的地形分布。这种方法为低成本、大范围的浅海地形测绘提供了新思路,无需昂贵的船载声纳测量 。[45][46]

6.4 智慧渔业与航运优化

在商业应用端,AI正在革新渔业和航运。

7. 挑战与未来展望

尽管前景广阔,但物理海洋学与AI的深度融合仍面临严峻挑战:

8. 结论

物理海洋学与人工智能的交叉正在经历从量变到质变的过程。2024年至2026年,这一领域的核心特征将是:

对于相关领域的研究者而言,掌握**物理感知(Physics-aware)**的AI架构设计能力,不仅是技术跟进的需求,更是推动物理海洋学向第四范式转型的必由之路。


🔗 正文参考文献 (References)


🧠 思考过程 (Thoughts)

Step 1: 确立宏观研究坐标

我正在全面审视2023年至2026年物理海洋学与人工智能交叉领域的最新进展。通过梳理国际顶级学术会议的趋势,我观察到研究重点已从简单的数据驱动预测转向了更为复杂的物理-数据混合建模,这为我深入理解该领域的演进脉络奠定了基础。

Step 2: 破解物理机制与数据融合的挑战

我正在重点审视传统动力模型在处理精细物理过程(如次格点参数化和湍流模拟)时的局限性。通过调研物理驱动与数据融合的最新方法,我旨在探寻如何利用物理信息神经网络等前沿技术,解决模型预测中的不确定性并填补动力学机制理解上的空白,从而在物理一致性与计算效率之间找到平衡点。

Step 3: 深入挖掘特定技术与未来路径

接下来,我准备对气候预测大模型、海洋遥感影像的超分辨率处理以及新兴的海洋数字孪生技术展开针对性研究。我将通过分析一系列核心应用场景,识别出最具潜力的技术突破方向,并思考如何将这些离散的技术点整合为一套系统性的研究框架。

Step 4: [54][55][56][57][58][59][60][61][62][63][64][65][66][67][68][69][70][71][72][73][74][75][76][77]

Step 5: 海洋大模型与数字孪生平台的崛起

我观察到“海洋大模型”(Ocean Foundation Models)正成为跨学科研究的新高地。通过分析最新的技术趋势,我发现研究者们正借鉴大气科学的成功经验,利用视觉 Transformer 等先进架构开发如 OceanRep 这样的大规模海洋动力学模型。同时,数字孪生海洋(Digital Twin Ocean)项目已进入实质性应用阶段,通过整合多源观测数据与 AI 模拟工具,为全球气候适应和生态保护提供精细化的虚拟仿真支持。

Step 6: 物理一致性与数据驱动的深度融合

在解决模型偏差方面,我发现“物理信息神经网络”(PINNs)的应用正从简单的曲线拟合转向复杂的动力学重构。例如,通过在地学规律约束下利用漂流瓶数据重构海表面流场,或者将经典物理方案作为损失函数,成功解决了纯数据模型在模拟海洋垂直混合时缺乏物理一致性的难题。这种通过物理机制修正 AI 预测误差的方法,正在有效填补传统动力学机制与观测数据之间的空白。

Step 7: 专项攻克亚格点过程与极端气候预测

接下来的研究中,我将重点探究亚格点参数化的前沿进展,特别是人工智能在海洋湍流闭合方案中的具体应用。同时,我计划深入调研专门针对厄尔尼诺(ENSO)及台风、风暴潮等极端事件设计的 AI 预测模型,旨在揭示 AI 技术在提升预测时效、空间分辨率以及捕捉非线性气候信号方面的最新突破。

Step 8: [78][79][80][81][82][83][84][85][86][87][88][89][90][91][92][93][94][95][96][97][98][99][100][101][102][103][104][105][106][107][108][109][110][111][112][113][114][115][116][117][118][119][120][121]

Step 9: 跨时空尺度的预报模型突破

我在梳理中发现,一系列高分辨率海洋大模型正在重新定义预报精度。例如,“伏羲海洋”通过创新的时间混合模块实现了亚日级别的预报,而“天海”模型则能处理多达107个物理变量并达到涡旋分辨率。这种向无缝隙天气-气候预测跨越的技术路径,不仅提升了模拟的精细度,更在耦合大气、陆地与海洋的复杂相互作用中展示了巨大潜力。

Step 10: 解决数据驱动模型中的物理失真

我正在深入分析如何通过物理约束来解决纯数据模型中的伪影问题。最新的进展显示,通过在模型中引入陆地掩膜信息和地转平衡约束,海表面高度预报的误差显著降低。我意识到,仅仅依靠数据堆叠无法完全克服物理一致性的挑战,必须将经典的流体动力学规律深度嵌入网络架构中,才能真正填补动力学反演中的信息鸿沟并提高模型的解释力。

Step 11: 探索特征识别与反演问题的新路径

接下来,我准备将研究重心转向计算机视觉与反演算法的深度结合。我计划探讨扩散模型和 Transformer 架构在海洋遥感超分辨率处理、中尺度涡旋及内波自动识别中的前沿表现。同时,我将深入研究如何利用人工智能解决海洋反演中的复杂难题,如通过稀疏回归发现潜在的动力学方程,以及对海底地形和声学层析成像的精确反演,以构建更具深度的海洋数字化认知。

Step 12: [122][123][124][125][126][127][128][129][130][131][132][133][134][135][136][137][138][139][140][141][142][143][144][145][146][147][148][149][150][151][152][153][154][155][156][157][158][159][160][161][162][163][164][165][166][167][168][169][170]

Step 13: 计算机视觉架构的代际更迭

我注意到海洋遥感领域正经历从传统卷积神经网络向 Transformer 和扩散模型(Diffusion Models)的范式演进。通过分析最新的识别算法,我发现 Swin-Unet 和 SegFormer 等架构在处理中尺度涡旋和内波边缘特征时表现出更强的鲁棒性;而扩散模型的引入,则显著提升了海表多要素信息的超分辨率重建能力,甚至能通过生成式算法有效修补云层遮挡带来的观测缺失。这种生成式技术正在为观测数据的时空精细化提供全新的解决方案。

Step 14: 动力学方程的自动化发现与反演

我正在深入思考如何通过人工智能自动化地提取未知的海洋物理规律。目前,我发现利用演化算法和稀疏回归技术,研究者已能从海量的稀疏观测数据中直接推导出潜在的动力学偏微分方程。此外,物理信息神经网络在解决复杂的反演难题上取得了实质性突破,例如通过集成地理掩膜和地转平衡约束,成功解决了海表面高度预测中的伪影问题,并实现了对海底地形参数的精确反演。这不仅缓解了传统数值模拟中参数不确定的难题,更在机制层面加深了我对非线性海洋动力系统的理解。

Step 15: 迈向全方位数字孪生与蓝海应用

我已梳理了如 EDITO 和 DITTO 等全球性数字孪生海洋工程的最新进展,它们正通过集成多模态数据构建海洋生物多样性与动力系统的虚拟映射。接下来,我将把研究视野扩展至“蓝色经济”与产业应用,重点调研人工智能在绿色航行路径优化、智能渔业精准预测以及海洋碳汇监测中的落地表现。同时,我计划深入剖析跨学科基础大模型在海洋碳循环模拟中的潜力,力求在完成技术趋势研判的基础上,为未来海洋治理提供更具前瞻性的洞察。

Step 16: [171][172][173][174][175][176][177][178][179][180][181][182][183][184][185][186][187][188][189][190][191][192][193][194][195][196][197][198][199][200][201][202][203][204][205]

Step 17: 海洋基础大模型的代际跨越

我正在深入分析物理海洋学向“大模型时代”迈进的标志性突破。通过对 Prithvi WxC 和“羲和”(XiHe)等模型的审视,我发现基础大模型已不再局限于单一变量预测,而是展现出处理全球叶绿素分布、重力波模拟以及涡旋分辨率预报的多任务处理能力。这种从专用模型向通用地理空间人工智能(Geo-AI)的演进,正显著增强我们捕捉海洋非线性特征和跨尺度动力演变的能力。

Step 18: 数字孪生驱动的生态与碳汇监测

我观察到数字孪生海洋项目(如 EDITO 和 DITTO)正利用 AI 实现从物理场到生物地球化学循环的深度映射。特别是在海洋碳循环领域,3D-ABC 等模型正在为海洋二氧化碳移除(mCDR)的监测与验证提供精准工具。同时,通过分析 SeaSwipe 等视觉识别技术在深海影像中的应用,我意识到 AI 正将零散的观测数据整合为系统性的生态指标,这为解决全球气候变化背景下的海洋治理难题提供了关键的技术支撑。

Step 19: 赋能蓝色经济并构建全景研究综述

我进一步梳理了 AI 在绿色航运路径优化、智能渔业精准预报等“蓝色经济”场景中的落地表现。虽然海洋声学层析成像等特定领域的 AI 应用仍处于探索期,但在捕捉极端天气信号和优化海洋资源利用方面,AI 已展现出强大的实用价值。目前,我已系统性地完成了从底层算法革新到高层产业应用的调研,下一步我将对这些碎片化的前沿成果进行深度合成,构建出一份关于物理海洋与人工智能交叉领域的全景式研究报告。

🔗 思考过程参考文献


  1. Crafting the Future: Machine learning for ocean forecasting - Reports ↩︎ ↩︎ ↩︎

  2. Crafting the Future: Machine learning for ocean forecasting - Reports ↩︎

  3. NOAA deploys new generation of AI-driven global weather models ↩︎

  4. XiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting - arXiv ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎

  5. [2402.02995] XiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting ↩︎

  6. NeurIPS Poster FuXi-Ocean: A Global Ocean Forecasting System with Sub-Daily Resolution ↩︎

  7. FuXi-ENS: A machine learning model for efficient and accurate ensemble weather prediction ↩︎

  8. A data-driven global ocean forecasting model with sub-daily and eddy-resolving resolution - arXiv ↩︎ ↩︎ ↩︎

  9. FengWu-W2S: A deep learning model for seamless weather-to-subseasonal forecast of global atmosphere - Semantic Scholar ↩︎

  10. FengWu-W2S: A deep learning model for seamless weather-to-subseasonal forecast of global atmosphere | Request PDF - ResearchGate ↩︎

  11. Prithvi Model: Geospatial Foundation - Emergent Mind ↩︎

  12. Prithvi WxC: Foundation Model for Weather and Climate - arXiv ↩︎ ↩︎

  13. A Framework for Hybrid Physics-AI Coupled Ocean Models - arXiv ↩︎ ↩︎

  14. Data-Driven Dynamic Modal Bias Analysis and Correction for Earth System Models ↩︎

  15. Physically Constrained Adaptive Deep Learning for Ocean Vertical-Mixing Parameterization ↩︎ ↩︎

  16. Physics-informed deep-learning parameterization of ocean vertical mixing improves climate simulations | National Science Review | Oxford Academic ↩︎ ↩︎ ↩︎

  17. The Critical Role of Small-Scale Dissipation in Deriving Subgrid Forcing Within an Ocean Quasi-Geostrophic Model - MDPI ↩︎

  18. An Analysis of Deep Learning Parameterizations for Ocean Subgrid Eddy Forcing - arXiv ↩︎

  19. Investigation of Physics-Informed Methods for Improving Sea Surface Height Prediction Based on Neural Networks in the South China Sea - MDPI ↩︎

  20. Potential for Equation Discovery with AI in the Climate Sciences - ESD ↩︎

  21. Ensemble-SINDy: Robust sparse model discovery in the low-data, high-noise limit, with active learning and control | Proceedings A ↩︎

  22. Deep blue artificial intelligence for knowledge discovery of the intermediate ocean - Frontiers ↩︎

  23. Deep blue artificial intelligence for knowledge discovery of the intermediate ocean ↩︎

  24. Physics-informed neural networks to reconstruct surface velocity field from drifter data ↩︎

  25. ReconMOST: Multi-Layer Sea Temperature Reconstruction with Observations-Guided Diffusion - arXiv ↩︎

  26. Spatial Downscaling of Sea Surface Temperature Using Diffusion Model - MDPI ↩︎

  27. Sea surface height super-resolution using high-resolution sea surface temperature with a subpixel convolutional residual network | Environmental Data Science | Cambridge Core ↩︎

  28. Stripe segmentation of oceanic internal waves in SAR images based on SegFormer ↩︎

  29. SAR Image Semantic Segmentation of Typical Oceanic and Atmospheric Phenomena ↩︎

  30. Semantic Segmentation of Typical Oceanic and Atmospheric Phenomena in SAR Images Based on Modified Segformer - MDPI ↩︎

  31. Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps - IEEE Xplore ↩︎

  32. Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps - IEEE Xplore ↩︎

  33. The European Digital Twin Ocean: Integrating Data, Models, and AI for Sustainable Ocean Management. ↩︎

  34. The European Digital Twin Ocean Launched at the 2024 Digital Ocean Forum | CMEMS ↩︎ ↩︎

  35. International Digital Twins of the Ocean Summit 2023 ↩︎

  36. About - DITTO - Digital Twins of the Ocean ↩︎

  37. DITTO - Digital Twins of the Ocean ↩︎

  38. 2025 Intervention Area: Digital Twins of the Ocean - Sustainable Blue Economy Partnership ↩︎

  39. Abstract EGU25-15061 - CO Meeting Organizer ↩︎

  40. Data-driven Surrogate Model Improves Storm Surge Prediction - SIAM.org ↩︎

  41. New AI Foundation Models for Science - Forschungszentrum Jülich ↩︎

  42. The ocean-climate-biodiversity nexus and marine carbon dioxide removal (mCDR) - Search for Funding - EuroAccess - Fördermittelsuche der EuroVienna EU-consulting & -management GmbH ↩︎

  43. FEATURE ARTICLE • Perspectives on Marine Carbon Dioxide Removal from the Global Ocean Acidification Observing Network - The Oceanography Society ↩︎

  44. Real-Time Operational Trial of Atmosphere–Ocean–Wave Coupled Model for Selected Tropical Cyclones in 2024 ↩︎

  45. Bathymetry Inversion Using a Deep-Learning-Based Surrogate for Shallow Water Equations Solvers - Penn State Research Database ↩︎

  46. (PDF) Bathymetry Inversion Using a Deep‐Learning‐Based Surrogate for Shallow Water Equations Solvers - ResearchGate ↩︎

  47. Top 50 Companies in AI in Sustainable Fisheries and Aquaculture Market 2025 ↩︎

  48. 'Here to stay and evolving fast': How GreenFish's AI-powered fish-forecasting tech is modernizing commercial fisheries - Responsible Seafood Advocate ↩︎

  49. AI Route Optimization: Enhancing Delivery Efficiency in 2025 - Descartes Systems Group ↩︎

  50. What is AI Route Optimization? - Windward.AI ↩︎

  51. XAI Helps in Storm Surge Forecasts: A Case Study for the Southeastern Chinese Coasts ↩︎

  52. Prediction of storm surge in the Pearl River Estuary based on data-driven model - Frontiers ↩︎

  53. Developing a Digital Twin of the Ocean - Tue, 05/13/2025 - 12:29 | Department of Economic and Social Affairs ↩︎

  54. sp.copernicus.orgCrafting the Future: Machine learning for ocean forecasting - Reports ↩︎

  55. researchgate.netApplications of deep learning in physical oceanography: a comprehensive review ↩︎

  56. frontiersin.orgApplications of deep learning in physical oceanography: a comprehensive review - Frontiers ↩︎

  57. meetingorganizer.copernicus.orgAbstract EGU25-15061 - CO Meeting Organizer ↩︎

  58. destine.ecmwf.intRapid progress on machine learning Earth system modelling showcased at EGU25 ↩︎

  59. agu.orgAI - AGU ↩︎

  60. agu.confex.comArtificial Intelligence in Oceanography IV Oral ↩︎

  61. community.climatechange.aiMachine Learning for Ocean Science at EGU2025 - Climate Change AI ↩︎

  62. sp.copernicus.orgCrafting the Future: Machine learning for ocean forecasting - Reports ↩︎

  63. mdpi.comA Review of Machine Learning Applications in Ocean Color Remote Sensing - MDPI ↩︎

  64. sp.copernicus.orgCrafting the Future: Machine Learning for Ocean Forecasting - Reports ↩︎

  65. iapjournals.ac.cnPhysically Constrained Adaptive Deep Learning for Ocean Vertical-Mixing Parameterization ↩︎

  66. frontiersin.orgRemote Sensing Applications in Oceanography with Deep Learning - Frontiers ↩︎

  67. essd.copernicus.orgA machine-learning reconstruction of sea surface pCO2 in the North American Atlantic Coastal Ocean Margin from 1993 to 2021 - ESSD Copernicus ↩︎

  68. frontiersin.orgForecasting sea level rise using enhanced deep learning models - Frontiers ↩︎

  69. academic.oup.comDeep-learning-based information mining from ocean remote-sensing imagery | National Science Review | Oxford Academic ↩︎

  70. mdpi.comInvestigation of Physics-Informed Methods for Improving Sea Surface Height Prediction Based on Neural Networks in the South China Sea - MDPI ↩︎

  71. academic.oup.comPhysics-informed deep-learning parameterization of ocean vertical mixing improves climate simulations | National Science Review | Oxford Academic ↩︎

  72. frontiersin.orgPhysics-informed neural networks to reconstruct surface velocity field from drifter data ↩︎

  73. agu.confex.comThe European Digital Twin Ocean: Integrating Data, Models, and AI for Sustainable Ocean Management. ↩︎

  74. research-and-innovation.ec.europa.euEuropean Digital Twin of the Ocean (European DTO) - Research and innovation ↩︎

  75. digitaltwinocean.mercator-ocean.euDigital Twin Ocean ↩︎

  76. oceanx.orgOceanX's AI Revolution ↩︎

  77. marine.copernicus.euThe European Digital Twin Ocean Launched at the 2024 Digital Ocean Forum | CMEMS ↩︎

  78. mdpi.comThe Critical Role of Small-Scale Dissipation in Deriving Subgrid Forcing Within an Ocean Quasi-Geostrophic Model - MDPI ↩︎

  79. arxiv.orgAn of Deep Learning Parameterizations for Ocean Subgrid Eddy Forcing - arXiv ↩︎

  80. usclivar.orgOcean Model Development, Data-driven Parameterizations, and Machine Learning in Ocean Models of the Earth System Workshop Agenda | US CLIVAR ↩︎

  81. agu.confex.comNumerical and Computational Challenges in the Development of Ocean Models and Their Coupling to Earth System Components II Oral ↩︎

  82. sp.copernicus.orgCrafting the Future: Machine learning for ocean forecasting - Reports ↩︎

  83. arxiv.orgEvaluation of Analytical Turbulence Closures for Quasi-Geostrophic Ocean Flows with Coastal Boundaries - arXiv ↩︎

  84. www7320.nrlssc.navy.milAn Empirical Evaluation of Turbulence Closure Models in the Coastal Ocean ↩︎

  85. ojs.ukscip.comHybrid Physics-Data Driven Ocean-Atmosphere Coupling Model for Tropical Cyclone Track Prediction - UK Scientific Publishing ↩︎

  86. openreview.netOcean-E2E: Hybrid Physics-Based and Data-Driven Global Forecasting of Marine Heatwaves with End-to-End Neural Assimilation | OpenReview ↩︎

  87. academic.oup.comPhysics-informed deep-learning parameterization of ocean vertical mixing improves climate simulations | National Science Review | Oxford Academic ↩︎

  88. mdpi.comA Hybrid Physics-Informed and Data-Driven Approach for Predicting the Fatigue Life of Concrete Using an Energy-Based Fatigue Model and Machine Learning - MDPI ↩︎

  89. arxiv.orgA Framework for Hybrid Physics-AI Coupled Ocean Models - arXiv ↩︎

  90. research-repository.uwa.edu.auData-Driven Dynamic Modal Bias and Correction for Earth System Models ↩︎

  91. sp.copernicus.orgCrafting the Future: Machine learning for ocean forecasting - Reports ↩︎

  92. journals.ametsoc.orgBias and Trend Correction of Precipitation Datasets to Force Ocean Models - AMS Journals ↩︎

  93. frontiersin.orgData-Driven Modeling of Dissolved Iron in the Global Ocean - Frontiers ↩︎

  94. researchgate.net(PDF) Application of Artificial Intelligence in ENSO Prediction Research - ResearchGate ↩︎

  95. climate.govMarch 2025 ENSO update: neutral conditions expected soon | NOAA Climate.gov ↩︎

  96. journals.ametsoc.orgAI Weather and Climate Prediction and Applications in - AMS Journals - American Meteorological Society ↩︎

  97. cpc.ncep.noaa.govENSO: Recent Evolution, Current Status and Predictions ↩︎

  98. iri.columbia.eduIRI – International Research Institute for Climate and Society | December 2025 Quick Look ↩︎

  99. mdpi.comXAI Helps in Storm Surge Forecasts: A Case Study for the Southeastern Chinese Coasts ↩︎

  100. arxiv.orgRGB-Encoded Physics-Aware Deep Learning for Storm Surge Forecasting - arXiv ↩︎

  101. siam.orgData-driven Surrogate Model Improves Storm Surge Prediction - SIAM.org ↩︎

  102. frontiersin.orgPrediction of storm surge in the Pearl River Estuary based on data-driven model - Frontiers ↩︎

  103. metoffice.gov.ukAI model could 'revolutionise' marine operational decision-making - Met Office ↩︎

  104. noaa.govNOAA deploys new generation of AI-driven global weather models ↩︎

  105. eodyn.comForecasting the Ocean: eOdyn Unveils AI Model for High-Fidelity Current Prediction ↩︎

  106. ecmwf.intRepresenting ocean wind waves in ECMWF's AIFS ↩︎

  107. alphaxiv.orgFuXi-Ocean: A Global Ocean Forecasting System with Sub-Daily ↩︎

  108. neurips.ccNeurIPS Poster FuXi-Ocean: A Global Ocean Forecasting System with Sub-Daily Resolution ↩︎

  109. pmc.ncbi.nlm.nih.govFuXi-ENS: A machine learning model for efficient and accurate ensemble weather prediction ↩︎

  110. arxiv.orgA data-driven global ocean forecasting model with sub-daily and eddy-resolving resolution - arXiv ↩︎

  111. researchgate.netFuXi-Ocean: A Global Ocean Forecasting System with Sub-Daily Resolution - ResearchGate ↩︎

  112. semanticscholar.orgFengWu-W2S: A deep learning model for seamless weather-to-subseasonal forecast of global atmosphere - Semantic Scholar ↩︎

  113. researchgate.netFengWu-W2S: A deep learning model for seamless weather-to-subseasonal forecast of global atmosphere | Request PDF - ResearchGate ↩︎

  114. imweather.comFengWu - Forecast Model Details - Im Weather ↩︎

  115. arxiv.org[2411.10191] FengWu-W2S: A deep learning model for seamless weather-to-subseasonal forecast of global atmosphere - arXiv ↩︎

  116. graphics.cs.uni-magdeburg.deFengWu – Skillful Global Weather Forecasting Beyond 10 Days Lead Lei Bai Shanghai AI Laboratory ↩︎

  117. journals.ametsoc.orgStatistical Prediction of Tropical Cyclone Rapid Intensification with Explainable AI in - AMS Journals ↩︎

  118. journals.ametsoc.orgAn Operations-Based Evaluation of Tropical Cyclone Track and Intensity Forecasts from Artificial Intelligence Weather Prediction Models in - AMS Journals ↩︎

  119. orca.atmos.washington.eduReal-Time Operational Trial of Atmosphere–Ocean–Wave Coupled Model for Selected Tropical Cyclones in 2024 ↩︎

  120. preventionweb.netNew study unveils ways to predict typhoon intensity with 'eye of deep learning ↩︎

  121. frontiersin.orgForecasting storm tides during strong typhoons using artificial intelligence and a physical model - Frontiers ↩︎

  122. ieeexplore.ieee.orgTransformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps - IEEE Xplore ↩︎

  123. ieeexplore.ieee.orgTransformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps - IEEE Xplore ↩︎

  124. researchgate.net(PDF) Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps - ResearchGate ↩︎

  125. mdpi.comA New Transformer Network for Short-Term Global Sea Surface Temperature Forecasting: Importance of Eddies - MDPI ↩︎

  126. tandfonline.comFull article: Towards submesoscale eddy detection in SDGSAT-1 data through deep learning ↩︎

  127. arxiv.orgReconMOST: Multi-Layer Sea Temperature Reconstruction with Observations-Guided Diffusion - arXiv ↩︎

  128. mdpi.comSpatial Downscaling of Sea Surface Temperature Using Diffusion Model - MDPI ↩︎

  129. os.copernicus.orgDeep learning for the super resolution of Mediterranean sea surface temperature fields - OS ↩︎

  130. frontiersin.orgSuper-resolution reconstruction of sea surface pollutant diffusion images based on deep learning models: a case study of thermal discharge from a coastal power plant - Frontiers ↩︎

  131. cambridge.orgSea surface height super-resolution using high-resolution sea surface temperature with a subpixel convolutional residual network | Environmental Data Science | Cambridge Core ↩︎

  132. journals.ametsoc.orgThe Internal Waves Service Workshop: Observing Internal Waves Globally with Deep Learning and Synthetic Aperture Radar in - AMS Journals ↩︎

  133. frontiersin.orgStripe segmentation of oceanic internal waves in SAR images based on SegFormer ↩︎

  134. mdpi.comResearch on the Detection of Ocean Internal Waves Based on the Improved Faster R-CNN in SAR Images - MDPI ↩︎

  135. ieeexplore.ieee.orgDetecting Internal Waves From Altimeter Data Using Deep Learning Method - IEEE Xplore ↩︎

  136. archimer.ifremer.frobserving internal waves globally with Deep Learning and Synthetic Aperture Radar - Archimer ↩︎

  137. essd.copernicus.orgSAR Image Semantic Segmentation of Typical Oceanic and Atmospheric Phenomena ↩︎

  138. arxiv.org[2405.11500] Interpreting a Semantic Segmentation Model for Coastline Detection - arXiv ↩︎

  139. mdpi.comSemantic Segmentation of Typical Oceanic and Atmospheric Phenomena in SAR Images Based on Modified Segformer - MDPI ↩︎

  140. frontiersin.orgEstimating daily semantic segmentation maps of classified ocean eddies using sea level anomaly data from along-track altimetry - Frontiers ↩︎

  141. essd.copernicus.orgSAR Image Semantic Segmentation of Typical Oceanic and Atmospheric Phenomena - ESSD Copernicus ↩︎

  142. mdpi.comRestoration of Missing Patterns on Satellite Infrared Sea Surface Temperature Images Due to Cloud Coverage Using Deep Generative Inpainting Network - MDPI ↩︎

  143. mdpi.comClassical vs. Machine Learning-Based Inpainting for Enhanced Classification of Remote Sensing Image - MDPI ↩︎

  144. github.comsatellite-image-deep-learning/techniques - GitHub ↩︎

  145. egusphere.copernicus.orgConvolutional Neural Networks for Sea Surface Data Assimilation in Operational Ocean Models: Test Case in the Gulf of Mexico - EGUsphere ↩︎

  146. pure.psu.eduBathymetry Inversion Using a Deep-Learning-Based Surrogate for Shallow Water Equations Solvers - Penn State Research Database ↩︎

  147. researchgate.net(PDF) Bathymetry Inversion Using a Deep‐Learning‐Based Surrogate for Shallow Water Equations Solvers - ResearchGate ↩︎

  148. arxiv.orgPhysics-Informed Neural Networks for the Korteweg-de Vries Equation for Internal Solitary Wave Problem: Forward Simulation and Inverse Parameter Estimation - arXiv ↩︎

  149. mdpi.comAn Improved PINN Algorithm for Shallow Water Equations Driven by Deep Learning - MDPI ↩︎

  150. tandfonline.comFull article: NDAWL-PINN: a new non-dimensionalization and multi-task learning approach for efficient training of physics-informed neural networks to solve the shallow water equations - Taylor & Francis Online ↩︎

  151. mdpi.comData Assimilation and Parameter Identification for Water Waves Using the Nonlinear Schrödinger Equation and Physics-Informed Neural Networks - MDPI ↩︎

  152. arxiv.org[2104.00615] Physics-informed neural networks for the shallow-water equations on the sphere - arXiv ↩︎

  153. esd.copernicus.orgPotential for Equation Discovery with AI in the Climate Sciences - ESD ↩︎

  154. d-nb.infoTowards data-driven discovery of governing equations in geosciences ↩︎

  155. researchgate.net(PDF) Discovery of the data-driven differential equation-based models of continuous metocean process - ResearchGate ↩︎

  156. royalsocietypublishing.orgEnsemble-SINDy: Robust sparse model discovery in the low-data, high-noise limit, with active learning and control | Proceedings A ↩︎

  157. pnas.orgLearning dynamical systems from data: An introduction to physics-guided deep learning | PNAS ↩︎

  158. bluepartnership.eu2025 Intervention Area: Digital Twins of the Ocean - Sustainable Blue Economy Partnership ↩︎

  159. mercator-ocean.euInternational Digital Twins of the Ocean Summit 2023 ↩︎

  160. sdgs.un.orgDeveloping a Digital Twin of the Ocean - Tue, 05/13/2025 - 12:29 | Department of Economic and Social Affairs ↩︎

  161. ditto-oceandecade.orgAbout - DITTO - Digital Twins of the Ocean ↩︎

  162. nesdis.noaa.govIndustry Study: AI-Driven Digital Twin Technology Could Improve NOAA's Ability To Integrate Diverse Environmental Observations ↩︎

  163. nesdis.noaa.govNOAA AI-BASED 3D EARTH AND SPACE OBSERVING DIGITAL TWIN (EO-DT) ↩︎

  164. noaa.govNOAA Center for Artificial Intelligence (NCAI) | National Oceanic and Atmospheric Administration ↩︎

  165. nesdis.noaa.govJoint Venture Digital Twin Report | NESDIS ↩︎

  166. noaa.govDigital Twins to Support Ocean Science and Technology - NOAA ↩︎

  167. globalocean.noaa.govVirtual Registration Open: International Digital Twins of the Ocean Summit (May 4-5) ↩︎

  168. ditto-oceandecade.orgDITTO - Digital Twins of the Ocean ↩︎

  169. ditto-oceandecade.orgDITTO at OOSC in Nice, 4 June 2025 - DITTO - Digital Twins of the Ocean ↩︎

  170. oceandecade.orgDigital Twins of the Ocean (DITTO) ↩︎

  171. emergentmind.comPrithvi Model: Geospatial Foundation - Emergent Mind ↩︎

  172. earthdata.nasa.govPrithvi-weather-climate: Advancing Our Understanding of the Atmosphere | NASA Earthdata ↩︎

  173. arxiv.orgPrithvi WxC: Foundation Model for Weather and Climate - arXiv ↩︎

  174. arxiv.org[2409.13598] Prithvi WxC: Foundation Model for Weather and Climate - arXiv ↩︎

  175. earthdata.nasa.govNASA and IBM Research Apply AI to Weather and Climate ↩︎

  176. fz-juelich.deNew AI Foundation Models for Science - Forschungszentrum Jülich ↩︎

  177. helmholtz.deAn AI model of the global carbon cycle - Helmholtz ↩︎

  178. gfz.dePioneering AI project on global carbon budget - GFZ Helmholtz-Zentrum für Geoforschung ↩︎

  179. ics.hawaii.eduUH Researchers Publish AI Foundation Model for Ocean Remote Sensing ↩︎

  180. mbari.orgNew AI approach sharpens picture of carbon export in the Southern Ocean - MBARI ↩︎

  181. descartes.comAI Route Optimization: Enhancing Delivery Efficiency in 2025 - Descartes Systems Group ↩︎

  182. windward.aiWhat is AI Route Optimization? - Windward.AI ↩︎

  183. rtslabs.comAI Route Optimization: Everything You Need to Know (2025) - RTS Labs ↩︎

  184. nuvizz.comAI Route Optimization Future of Smart Deliveries in 2025 - Nuvizz ↩︎

  185. lumenalta.comA guide to freight optimization in 2025 - Lumenalta ↩︎

  186. sphericalinsights.comTop 50 Companies in AI in Sustainable Fisheries and Aquaculture Market 2025 ↩︎

  187. globalseafood.org'Here to stay and evolving fast': How GreenFish's AI-powered fish-forecasting tech is modernizing commercial fisheries - Responsible Seafood Advocate ↩︎

  188. thebusinessresearchcompany.comAi In Sustainable Fisheries And Aquaculture Market Size 2025 - The Business Research Company ↩︎

  189. forbes.comData Is The New Soil: Why Global Agriculture Needs A Sovereign AI Factory - Forbes ↩︎

  190. mdpi.comRedefining Aquaculture Safety with Artificial Intelligence: Design Innovations, Trends, and Future Perspectives - MDPI ↩︎

  191. euro-access.euThe ocean-climate-biodiversity nexus and marine carbon dioxide removal (mCDR) - Search for Funding - EuroAccess - Fördermittelsuche der EuroVienna EU-consulting & -management GmbH ↩︎

  192. bidenwhitehouse.archives.govNATIONAL MARINE CARBON DIOXIDE REMOVAL RESEARCH STRATEGY | Biden White House Archives ↩︎

  193. marineboard.euMonitoring, Reporting and Verification for marine Carbon Dioxide Removal ↩︎

  194. tos.orgFEATURE ARTICLE • Perspectives on Marine Carbon Dioxide Removal from the Global Ocean Acidification Observing Network - The Oceanography Society ↩︎

  195. oceanvisions.orgA Comprehensive Program to Prove or Disprove Marine Carbon Dioxide Removal Technologies by 2030 | Ocean Visions ↩︎

  196. sp.copernicus.orgCrafting the Future: Machine learning for ocean forecasting - Reports ↩︎

  197. ieeexplore.ieee.orgDeepBlue: Advanced convolutional neural network applications for ocean remote sensing ↩︎

  198. frontiersin.orgDeep blue artificial intelligence for knowledge discovery of the intermediate ocean - Frontiers ↩︎

  199. github.comXiongWeiTHU/Awesome-AI-for-Atmosphere-and-Ocean - GitHub ↩︎

  200. researchgate.netDeep blue artificial intelligence for knowledge discovery of the intermediate ocean ↩︎

  201. arxiv.orgXiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting - arXiv ↩︎

  202. pmc.ncbi.nlm.nih.govData-driven global ocean modeling for seasonal to decadal prediction - PubMed Central ↩︎

  203. github.comOcean-Intelligent-Forecasting/XiHe-GlobalOceanForecasting - GitHub ↩︎

  204. arxiv.org[2402.02995] XiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting ↩︎

  205. researchgate.netXiHe: A Data-Driven Model for Global Ocean Eddy-Resolving Forecasting - ResearchGate ↩︎