Physics-Based Deep Spatiotemporal Hyperlocal Radar Nowcasting with a Multi-Variable U-Net for High-Resolution Precipitation Forecasting

📄 arXiv: 2607.16080 📥 PDF

作者: Akshay Sunil, Muhammed Rashid, Raja Sekhar Sivaraju, Sushma Nair, Subimal Ghosh

分类: cs.LG, eess.IV

发布日期: 2026-07-20


💡 一句话要点

提出基于物理的深度时空超局部雷达短期预报框架以解决降水预测问题

🎯 匹配领域: 支柱八:物理动画 (Physics-based Animation)

关键词: 降水预测 雷达数据 深度学习 U-Net 气象建模 短期预报 城市洪水管理 物理引导

📋 核心要点

  1. 现有的短期降水预测方法在高分辨率数值天气预报中存在计算延迟和模型初始化的挑战,难以满足实时需求。
  2. 本文提出了一种基于雷达数据的紧凑型短期预报框架,利用多层次反射率和速度特征,采用U-Net模型进行快速降水预测。
  3. 实验结果表明,在90分钟的预报时间内,模型在不同降水强度阈值下的成功指数显著高于传统方法,且生成预报的速度极快。

📝 摘要(中文)

降水短期预报在城市洪水管理和实时决策中至关重要。传统的高分辨率数值天气预报需要频繁的数据同化和模型初始化,导致计算延迟。本文提出了一种紧凑的雷达短期预报框架,结合多层次反射率和多种速度特征,通过编码器-解码器U-Net模型进行降水预测。该模型能够在90分钟内以7.5分钟的间隔预测未来的复合反射率场,并在实验中显示出优于传统方法的性能,尤其是在敏感性和空间相关性方面。

🔬 方法详解

问题定义:本文旨在解决传统高分辨率降水预测方法在计算延迟和实时决策中的不足,尤其是在城市洪水管理中的应用挑战。

核心思路:通过构建一个基于雷达数据的深度学习框架,直接从高频观测中学习风暴演变,快速生成降水预报,避免了传统方法的复杂性。

技术框架:该框架采用编码器-解码器U-Net结构,结合多层次反射率、Doppler径向速度和径向速度梯度特征,预测未来12个复合反射率场。

关键创新:引入高反射率注意力模块以增强对对流核心的敏感性,并通过物理引导的归因分析验证学习到的敏感性是否具有气象学意义。

关键设计:模型使用2023年5月至8月的孟买Doppler雷达观测数据进行训练,采用特定的损失函数和网络结构设计,确保在90分钟的预报时间内实现较低的均方根误差和较高的空间相关性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

在90分钟的预报时间内,模型在不同降水强度阈值下的成功指数分别为0.437、0.332和0.193,相较于持久性方法,模型在较长预报时间内表现出更低的均方根误差和更高的空间相关性,显示出显著的性能提升。

🎯 应用场景

该研究的潜在应用领域包括城市洪水管理、农业气象预报和灾害响应等。通过提供高分辨率的短期降水预报,能够帮助决策者及时采取措施,减少洪水带来的损失,提升城市的气候适应能力。

📄 摘要(原文)

Precipitation nowcasting over the immediate 10-90 min period is important for flood management and real-time decision-making in urban regions. Conventional short-range forecasting with high-resolution numerical weather prediction requires frequent data assimilation, model initialization, and spin-up, introducing computational latency. Machine learning provides an alternative by learning storm evolution directly from high-frequency observations and producing forecasts quickly after training. This is particularly relevant for Mumbai, India, where monsoon convection, land-sea interactions, and localized intense rainfall make short-term prediction difficult. Here, we develop a compact radar-only nowcasting framework that combines multi-elevation reflectivity, Doppler radial velocity, and radial-velocity-gradient proxy features within an encoder-decoder U-Net. Using the most recent radar volume scan, the model predicts 12 future composite reflectivity fields at 7.5-min intervals up to 90 min lead time. The derived velocity magnitude, divergence-like, directional-shear, and vorticity-like channels represent kinematic signatures associated with convergence and boundary interactions without requiring full wind-field retrieval. A high-reflectivity attention module improves sensitivity to convective cores, and physics-guided attribution examines whether the learned sensitivities are meteorologically meaningful. The model is trained using Mumbai Doppler radar observations from May to August 2023 and evaluated on temporally independent events. At 90 min lead time, Critical Success Index values are 0.437, 0.332, and 0.193 for $\geq$10, $\geq$20, and $\geq$30 dBZ thresholds, respectively. Compared with persistence, the model gives lower RMSE and higher spatial correlation at longer lead times. Once trained, it runs on a standard computer, generating nowcasts within seconds for real-time use.