Review of deep learning applications in severe convective weather prediction

doi:10.11772/j.issn.1001-9081.2025020184

Journal of Computer Applications ›› 2026, Vol. 46 ›› Issue (3): 980-992.DOI: 10.11772/j.issn.1001-9081.2025020184

• Frontier and comprehensive applications • Previous Articles Next Articles

Review of deep learning applications in severe convective weather prediction

Min CHEN¹^,²^,³, Xiaolin QIN¹^,²(), Shaohan LI³, Hao YANG³, Taohong LI³

^1.Chengdu Institute of Computer Application，Chinese Academy of Sciences，Chengdu Sichuan 610213，China
^2.University of Chinese Academy of Sciences，Beijing 100049，China
^3.School of Computer Science，Chengdu University of Information Technology，Chengdu Sichuan 610225，China

Received:2025-03-03 Revised:2025-04-18 Accepted:2025-04-28 Online:2026-03-16 Published:2026-03-10
Contact: Xiaolin QIN
About author:CHEN Min， born in 1989， Ph. D. candidate. Her research interests include spatio-temporal prediction， data mining， smart meteorology.
LI Shaohan， born in 1999， M. S. candidate. His research interests include spatio-temporal prediction， smart meteorology.
YANG Hao， born in 1981， Ph. D.， associate professor. His research interests include data mining， machine learning.
LI Taohong， born in 2000， M. S. candidate. His research interests include spatio-temporal prediction， precipitation downscaling.
Supported by:
Sichuan Science and Technology Program(2024NSFJQ0035);Sichuan Provincial Science and Technology Achievement Transfer and Transformation Demonstration Project(2024ZHCG0026)

深度学习应用于强对流天气预测的综述

陈敏¹^,²^,³, 秦小林¹^,²(), 李绍涵³, 杨昊³, 李韬弘³

^1.中国科学院成都计算机应用研究所，成都 610213
^2.中国科学院大学，北京 100049
^3.成都信息工程大学计算机学院，成都 610225

通讯作者: 秦小林
作者简介:陈敏（1989—），女，四川乐山人，博士研究生，主要研究方向：时空预测、数据挖掘、智慧气象
李绍涵（1999—），男，重庆人，硕士研究生，主要研究方向：时空预测、智慧气象
杨昊（1981—），男，四川成都人，副教授，博士，主要研究方向：数据挖掘、机器学习
李韬弘（2000—），男，广东深圳人，硕士研究生，主要研究方向：时空预测、降水降尺度。
基金资助:
四川省科技计划项目(2024NSFJQ0035);四川省科技成果转移转化示范项目(2024ZHCG0026)

Abstract

Abstract:

The groundbreaking advancements in deep learning technology establish a new paradigm for interdisciplinary research in “AI + meteorology”， and severe convective weather prediction emerges as a cutting-edge research focus due to its complex dynamical characteristics and significant socioeconomic impacts. Therefore， the theoretical advancements and methodological innovations of Deep Neural Networks （DNN） in severe convective weather prediction were reviewed systemically， and their specific applications were explored deeply. Firstly， based on the spatio-temporal sequence prediction paradigm， the mechanisms of high-frequency feature extraction by Recurrent Neural Networks （RNNs） and non-RNNs in meteorological sequence modeling were dissected. Secondly， from a generative modeling perspective， the advantages of Generative Adversarial Network （GAN） and diffusion model in probabilistic prediction of extreme weather events were demonstrated. Thirdly， the theoretical breakthroughs of meteorological large-scale models realizing multimodal data fusion and cross-scale feature learning via pre-training and fine-tuning paradigms were revealed， along with their generalization enhancement mechanisms in global numerical weather prediction. Fourthly， aiming at model evaluation systems， the limitations of traditional statistical metrics in extreme weather prediction were analyzed， and pathways for constructing novel evaluation frameworks such as physical consistency constraints were discussed. Finally， the key scientific challenges faced currently and future research directions were distilled， aiming to provide theoretical support and methodological references for constructing the next-generation intelligent system for severe convective weather prediction.

Key words: deep learning, severe convective weather prediction, Generative Adversarial Network (GAN), diffusion model, meteorological large-scale model

摘要：

深度学习技术的突破性进展为“AI+气象”交叉学科的研究开辟了新的范式，而强对流天气预测因为它的复杂动力学特征和重大社会经济影响成为前沿研究热点。因此，本文系统梳理深度神经网络（DNN）在强对流天气预测领域的理论进展和方法创新，并深入探讨它们的具体应用。首先，基于时空序列预测范式剖析循环神经网络（RNN）与非RNN在气象时序建模中的高频特征捕获机理；其次，从生成式建模的视角，论证生成对抗网络（GAN）和扩散模型在极端天气事件概率预测中的建模优势；再次，揭示气象大模型通过预训练-微调范式实现多模态数据融合与跨尺度特征学习的理论突破，并阐释它们在全球数值天气预报中的泛化能力提升机制；继次，针对模型评估体系，分析传统统计指标在极端天气预测中的局限性，并探讨物理一致性约束等新型评估框架的构建路径；最后，凝练出当前面临的关键科学挑战及未来研究方向，旨在为构建新一代强对流天气预测智能系统提供理论支撑与方法论参考。

关键词: 深度学习, 强对流天气预测, 生成对抗网络, 扩散模型, 气象大模型

CLC Number:

TP183

Min CHEN, Xiaolin QIN, Shaohan LI, Hao YANG, Taohong LI. Review of deep learning applications in severe convective weather prediction[J]. Journal of Computer Applications, 2026, 46(3): 980-992.

陈敏, 秦小林, 李绍涵, 杨昊, 李韬弘. 深度学习应用于强对流天气预测的综述[J]. 《计算机应用》唯一官方网站, 2026, 46(3): 980-992.

Figures/Tables 9

References 81

[1]	World Meteorological Organization. State of the global climate 2024［M］. Geneva： World Meteorological Organization， 2025： 23-45.
[2]	Intergovernmental Panel on Climate Change. Climate change 2023： the physical science basis ［M］. Cambridge： Cambridge University Press， 2023.
[3]	中国气象局. 中国气象年鉴2022［M］. 北京：气象出版社， 2022： 95-98.
	China Meteorological Administration. China meteorological yearbook 2022 ［M］. Beijing： China Meteorological Press， 2022： 95-98.
[4]	PALMER T. A vision for numerical weather prediction in 2030 ［EB/OL］. ［2024-11-12］..
[5]	RIHAN F A， COLLIER C G， ROULSTONE I. Four-dimensional variational data assimilation for Doppler radar wind data ［J］. Journal of Computational and Applied Mathematics， 2005， 176（1）： 15-34.
[6]	YAN Q， JI F， MIAO K， et al. Convolutional residual-attention： a deep learning approach for precipitation nowcasting ［J］. Advances in Meteorology， 2020， 2020： No.6484812.
[7]	曾明剑，王桂臣，吴海英，等. 基于中尺度数值模式的分类强对流天气预报方法研究［J］. 气象学报， 2015， 73（5）： 868-882.
	ZENG M J， WANG G C， WU H Y， et al. Study of the forecasting method for the classified severe convection weather based on a meso-scale numerical model ［J］. Acta Meteorologica Sinica， 2015， 73（5）： 868-882.
[8]	RONNEBERGER O， FISCHER P， BROX T. U-net： convolutional networks for biomedical image segmentation ［C］// Proceedings of the 2015 International Conference on Medical Image Computing and Computer-assisted Intervention， LNCS 9351. Cham： Springer， 2015： 234-241.
[9]	SHI X， CHEN Z， WANG H， et al. Convolutional LSTM network： a machine learning approach for precipitation nowcasting ［C］// Proceedings of the 29th International Conference on Neural Information Processing Systems. Cambridge： MIT Press， 2015： 802-810.
[10]	LAM R， SANCHEZ-GONZALEZ A， WILLSON M， et al. Learning skillful medium-range global weather forecasting ［J］. Science， 2023， 382（6677）： 1416-1421.
[11]	中国气象局. 中国气象局印发《人工智能气象应用工作方案（2023—2030年）》［EB/OL］. ［2024-11-12］..
	China Meteorological Administration. China meteorological administration issues the Artificial Intelligence Meteorological Application Work Plan （2023—2030）［EB/OL］. ［2024-11-12］..
[12]	AGAPIOU A. Remote sensing heritage in a petabyte-scale： satellite data and heritage Earth Engine© applications ［J］. International Journal of Digital Earth， 2017， 10（1）： 85-102.
[13]	高雨佳，王鹏飞，刘亮，等. 基于注意力增强元学习网络的个性化联邦学习方法［J］. 计算机研究与发展， 2024， 61（1）： 196-208.
	GAO Y J， WANG P F， LIU L， et al. Personalized federated learning method based on attention-enhanced meta-learning network ［J］. Journal of Computer Research and Development， 2024， 61（1）： 196-208.
[14]	焦鹏飞，陈舒欣，郭翾，等. 图神经常微分方程综述［J］. 计算机研究与发展， 2024， 61（8）： 2045-2066.
	JIAO P F， CHEN S X， GUO X， et al. Survey on graph neural ordinary differential equations ［J］. Journal of Computer Research and Development， 2024， 61（8）： 2045-2066.
[15]	CHEN T， GUESTRIN C. XGBoost： a scalable tree boosting system ［C］// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York： ACM， 2016： 785-794.
[16]	LEGLER S， JANJIĆ T. Combining data assimilation and machine learning to estimate parameters of a convective-scale model ［J］. Journal of Advances in Modeling Earth Systems， 2021， 13（12）： e2021MS002534.
[17]	NERINI D， FORESTI L， LEUENBERGER D， et al. A reduced-space ensemble Kalman filter approach for flow-dependent integration of radar extrapolation nowcasts and NWP precipitation ensembles ［J］. Monthly Weather Review， 2019， 147（3）： 987-1006.
[18]	DOSWELL C A， Ⅲ， BROOKS H E， MADDOX R A， et al. Flash flood forecasting： an ingredients-based methodology ［J］. Weather and Forecasting， 1996， 11（4）： 560-581.
[19]	PRUDDEN R， ADAMS S， KANGIN D， et al. A review of radar-based nowcasting of precipitation and applicable machine learning［EB/OL］. ［2024-03-02］..
[20]	BROWNING K A. Review lecture： local weather forecasting ［J］. Proceedings of the Royal Society of London. Series A， Mathematical and Physical Sciences， 1980， 371（1745）： 179-211.
[21]	田济扬. 天气雷达多源数据同化支持下的陆气耦合水文预报［D］. 北京：中国水利水电科学研究院， 2017.
	TIAN J Y. Coupled atmospheric-hydrological modeling for hydrological forecast based on multi-source data assimilation containing radar data ［D］. Beijing： China Institute of Water Resources and Hydropower Research， 2017.
[22]	曹伟华，陈明轩，高峰，等. 雷暴区域追踪矢量与雷暴单体追踪矢量融合临近预报研究［J］. 气象学报， 2019， 77（6）： 1015-1027.
	CAO W H， CHEN M X， GAO F， et al. A vector blending study based on object-based tracking vectors and cross correlation tracking vectors ［J］. Acta Meteorologica Sinica， 2019， 77（6）： 1015-1027.
[23]	CHEN L， DAI J， TAO L. A modified cross-correlation method （COTREC） and its application in precipitation nowcasting ［J］. Journal of Tropical Meteorology， 2009， 25（1）： 117-122.
[24]	张亚萍，程明虎，夏文梅，等. 天气雷达回波运动场估测及在降水临近预报中的应用［J］. 气象学报， 2006， 64（5）： 631-646.
	ZHANG Y P， CHEN M H， XIA W M， et al. Estimation of weather radar echo motion field and its application to precipitation nowcasting ［J］. Acta Meteorological Sinica， 2006， 64（5）： 631-646.
[25]	MUELLER C， SAXEN T， ROBERTS R， et al. NCAR auto-nowcast system ［J］. Weather and Forecasting， 2003， 18（4）： 545-561.
[26]	曹春燕，陈元昭，刘东华，等. 光流法及其在临近预报中的应用［J］. 气象学报， 2015， 73（3）： 471-480.
	CAO C Y， CHEN Y Z， LIU D H， et al. The optical flow method and its application to nowcasting ［J］. Acta Meteorologica Sinica， 2015， 73（3）： 471-480.
[27]	XU H， ZHANG J， CAI J， et al. GMFlow： learning optical flow via global matching ［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 8111-8120.
[28]	MOHAMMED A D， MORRIS T. Optical flow estimation using local features ［EB/OL］. ［2024-03-02］. .
[29]	HAMDAN A， IBEKWE K I， ETUKUDOH E A， et al. AI and machine learning in climate change research： a review of predictive models and environmental impact ［J］. World Journal of Advanced Research and Reviews， 2024， 21（1）： 1999-2008.
[30]	胡家晖，卢楚翰，姜有山，等. 深度学习模型TAGAN在强对流回波临近预报中的应用［J］. 大气科学， 2022， 46（4）： 805-818.
	HU J H， LU C H， JIANG Y S， et al. Application of deep learning model TAGAN in nowcasting of strong convective echo［J］. Chinese Journal of Atmospheric Sciences， 2022， 46（4）： 805-818.
[31]	SIT M， SEO B C， DEMIR I. IowaRain： a statewide rain event dataset based on weather radars and quantitative precipitation estimation ［EB/OL］. ［2024-03-02］..
[32]	AYZEL G， SCHEFFER T， HEISTERMANN M. RainNet v1.0： a convolutional neural network for radar-based precipitation nowcasting ［J］. Geoscientific Model Development， 2020， 13： 2631-2644.
[33]	NOAA National Centers for Environmental Prediction. Multi-Radar/Multi-Sensor （MRMS） precipitation data ［DS/OL］. ［2024-01-15］. .
[34]	VEILLETTE M S， SAMSI S， MATTIOLI C J. SEVIR： a storm event imagery dataset for deep learning applications in radar and satellite meteorology ［C］// Proceedings of the 34th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2020： 22009-22019.
[35]	HERRUZO P， GRUCA A， LLISO L， et al. High-resolution multi-channel weather forecasting： first insights on transfer learning from the Weather4Cast Competitions 2021 ［C］// Proceedings of the 2021 IEEE International Conference on Big Data. Piscataway： IEEE， 2021： 5750-5757.
[36]	Météo-France. MeteoNet： the open weather dataset ［DS/OL］. ［2024-01-15］..
[37]	European Centre for Medium-range Weather Forecasts （ECMWF）. ERA5 fifth generation reanalysis data ［DS/OL］. ［2024-01-17］..
[38]	中国气象局国家气象信息中心. CMA全球大气再分析数据集（CMA-RA V1.5）［DS/OL］. ［2024-01-22］..
	China Meteorological Administration， National Meteorological Information Center. CMA-RA V1.5 global atmospheric reanalysis dataset ［DS/OL］. ［2024-01-22］..
[39]	HINTON G E， SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks ［J］. Science， 2006， 313（5786）： 504-507.
[40]	尹文君，张大伟，严京海，等. 基于深度学习的大数据空气污染预报［J］. 中国环境管理， 2015， 7（6）： 46-52.
	YIN W J， ZHANG D W， YAN J H， et al. Deep learning based air pollutant forecasting with big data ［J］. Chinese Journal of Environmental Management， 2015， 7（6）： 46-52.
[41]	SHI J， SHIRALI A， JIN B， et al. Deep learning and foundation models for weather prediction： a survey ［EB/OL］. ［2025-03-01］..
[42]	WANG Y， LONG M， WANG J， et al. PredRNN： recurrent neural networks for predictive learning using spatiotemporal LSTMs ［C］// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2017： 879-888.
[43]	WANG Y， JIANG L， YANG M H， et al. Eidetic 3D LSTM： a model for video prediction and beyond ［EB/OL］. ［2024-11-17］..
[44]	WANG Y， ZHANG J， ZHU H， et al. Memory in memory： a predictive neural network for learning higher-order non-stationarity from spatiotemporal dynamics ［C］// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2019： 9146-9154.
[45]	LE GUEN V， THOME N. Disentangling physical dynamics from unknown factors for unsupervised video prediction ［C］// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2020： 11471-11481.
[46]	WU H， YAO Z， WANG J， et al. MotionRNN： a flexible model for video prediction with spacetime-varying motions ［C］// Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2021： 15430-15439.
[47]	BAI L， YAO L， LI C， et al. Adaptive graph convolutional recurrent network for traffic forecasting ［C］// Proceedings of the 34th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2020： 17804-17815.
[48]	LIN H， GAO Z， XU Y， et al. Conditional local convolution for spatio-temporal meteorological forecasting ［C］// Proceedings of the 36th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2022： 7470-7478.
[49]	VASWANI A， SHAZEER N， PARMAR N， et al. Attention is all you need ［C］// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2017： 6000-6010.
[50]	LIN Z， LI M， ZHENG Z， et al. Self-attention ConvLSTM for spatiotemporal prediction ［C］// Proceedings of the 34th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2020： 11531-11538.
[51]	TANG S， LI C， ZHANG P， et al. SwinLSTM： improving spatiotemporal prediction accuracy using Swin Transformer and LSTM ［C］// Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2023： 13424-13433.
[52]	CHANG Z， ZHANG X， WANG S， et al. MAU： a motion-aware unit for video prediction and beyond ［C］// Proceedings of the 35th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2021： 26950-26962.
[53]	TANG Y， DONG P， TANG Z， et al. VMRNN： integrating Vision Mamba and LSTM for efficient and accurate spatiotemporal forecasting ［C］// Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. Piscataway： IEEE， 2024： 5663-5673.
[54]	GAO Z， TAN C， WU L， et al. SimVP： simple yet better video prediction ［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 3160-3170.
[55]	CASTRO R， SOUTO Y M， OGASAWARA E， et al. STConvS2S： Spatiotemporal convolutional sequence-to-sequence network for weather forecasting ［J］. Neurocomputing， 2021， 426： 285-298.
[56]	NING S， LAN M， LI Y， et al. MIMO is all you need： a strong multi-in-multi-out baseline for video prediction ［C］// Proceedings of the 37th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2023： 1975-1983.
[57]	TAN C， LI S， GAO Z， et al. OpenSTL： a comprehensive benchmark of spatio-temporal predictive learning ［J］. Advances in Neural Information Processing Systems， 2023： 69819-69831.
[58]	DOSOVITSKIY A， BEYER L， KOLESNIKOV A， et al. An image is worth 16x16 words： Transformers for image recognition at scale［EB/OL］. ［2024-02-08］..
[59]	LIU Z， LIN Y， CAO Y， et al. Swin Transformer： hierarchical Vision Transformer using shifted windows ［C］// Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2021： 9992-10002.
[60]	RAO Y， ZHAO W， TANG Y， et al. HorNet： efficient high-order spatial interactions with recursive gated convolutions ［C］// Proceedings of the 36th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2022： 10353-10366.
[61]	LI S， WANG Z， LIU Z， et al. MogaNet： multi-order gated aggregation network ［EB/OL］. ［2024-11-17］..
[62]	TAN C， GAO Z， WU L， et al. Temporal attention unit： towards efficient spatiotemporal predictive learning ［C］// Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2023： 18770-18782.
[63]	CHEN M， YANG H， LI S， et al. STAA： spatiotemporal alignment attention for short-term precipitation forecasting ［J］. IEEE Geoscience and Remote Sensing Letters， 2024， 21： No.1003105.
[64]	WU H， XU F， CHEN C， et al. PastNet： introducing physical inductive biases for spatio-temporal video prediction ［C］// Proceedings of the 32nd ACM International Conference on Multimedia. New York： ACM， 2024： 2917-2926.
[65]	NIE X， YAN Y， LI S， et al. Wavelet-driven spatiotemporal predictive learning： bridging frequency and time variations ［C］// Proceedings of the 38th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2024： 4334-4342.
[66]	TULYAKOV S， LIU M Y， YANG X， et al. MoCoGAN： decomposing motion and content for video generation ［C］// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2018： 1526-1535.
[67]	ZHANG C， YANG X， TANG Y， et al. Learning to generate radar image sequences using two-stage generative adversarial networks［J］. IEEE Geoscience and Remote Sensing Letters， 2020， 17（3）： 401-405.
[68]	LUO C， LI X， YE Y， et al. Experimental study on generative adversarial network for precipitation nowcasting ［J］. IEEE Transactions on Geoscience & Remote Sensing， 2022， 60（1）： 1-20.
[69]	CHANG Z， ZHANG X， WANG S， et al. STRPM： a spatiotemporal residual predictive model for high-resolution video prediction ［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 13926-13935.
[70]	YIN J， MEO C， ROY A， et al. Precipitation nowcasting using physics-informed discriminator generative models ［C］// Proceedings of the 32nd European Signal Processing Conference. Piscataway： IEEE， 2024： 967-971.
[71]	AKAN A K， ERDEM E， ERDEM A， et al. SLAMP： stochastic latent appearance and motion prediction ［C］// Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2021： 14708-14717.
[72]	GAO Z， SHI X， HAN B， et al. PreDiff： precipitation nowcasting with latent diffusion models ［C］// Proceedings of the 37th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2024： 78621-78656.
[73]	YU D， LI X， YE Y， et al. DiffCast： a unified framework via residual diffusion for precipitation nowcasting ［C］// Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2024： 27758-27767.
[74]	ZHANG Z， HU J， CHENG W， et al. ExtDM： distribution extrapolation diffusion model for video prediction［C］// Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2024： 19310-19320.
[75]	PATHAK J， SUBRAMANIAN S， HARRINGTON P， et al. FourCastNet： a global data-driven high-resolution weather model using adaptive Fourier neural operators ［EB/OL］. ［2024-03-11］..
[76]	ZHANG Y， LONG M， CHEN K， et al. Skilful nowcasting of extreme precipitation with NowcastNet ［J］. Nature， 2023， 619（7970）： 526-532.
[77]	BI K， XIE L， ZHANG H， et al. Accurate medium-range global weather forecasting with 3D neural networks ［J］. Nature， 2023， 619（7970）： 533-538.
[78]	CHEN K， HAN T， GONG J， et al. FengWu： pushing the skillful global medium-range weather forecast beyond 10 days lead ［EB/OL］. ［2024-03-24］..
[79]	ANDRYCHOWICZ M， ESPEHOLT L， LI D， et al. Deep learning for day forecasts from sparse observations ［EB/OL］. ［2024-11-15］..
[80]	CHEN L， ZHONG X， LI H， et al. A machine learning model that outperforms conventional global subseasonal forecast models ［J］. Nature Communications， 2024， 15： No.6425.
[81]	PRICE I， SANCHEZ-GONZALEZ A， ALET F， et al. GenCast： diffusion-based ensemble forecasting for medium-range weather［EB/OL］. ［2024-11-16］..

类别	指标
全局精度	MAE， MSE，RMSE， PCC
二值精度	CSI， Accuracy， FSS， F1-score， POD， Precision， HSS， Recall，FAR
降尺度精度	CSI-neighborhood， FSS， Pooled CRPS
清晰度精度	GDL， LPIPS， PSNR， SSIM

类别	指标
全局精度	MAE， MSE，RMSE， PCC
二值精度	CSI， Accuracy， FSS， F1-score， POD， Precision， HSS， Recall，FAR
降尺度精度	CSI-neighborhood， FSS， Pooled CRPS
清晰度精度	GDL， LPIPS， PSNR， SSIM

方法类别	代表模型	优势	劣势
循环方法	ConvLSTM	结合CNN和LSTM结构，捕捉空间和时间依赖关系，适用于短时天气预测。ConvLSTM在气象数据集（如 Radar Echo、ERA5、GFS、ClimateNet）上能有效捕捉降水、温度、风场等时空依赖关系，适用于短期天气预测	计算成本高，对长期趋势（如ERA5的长期气候模式）捕捉较弱，易受累积误差影响，在GFS这样的数值天气预报数据上可能不及传统物理模型精确
	MotionRNN	采用时间-空间变换机制，提升对运动信息的建模能力，适用于复杂天气场景	计算复杂度较高，对长时间序列预测仍有误差累积问题
	SwinLSTM	结合Swin Transformer进行局部和全局信息交互，提高时序预测精度	结构复杂，对计算资源需求较高，训练难度比较大
非循环方法	SimVP	采用ETD架构，实现端到端多步预测，计算效率高	主要依赖CNN，难以捕捉长时间依赖关系
	MIMO-VP	采用Transformer实现MIMO，提高长时序预测能力	计算量大，训练难度较大
	TAU	通过静态注意力和动态注意力优化特征提取，提升计算效率	对数据量依赖较高，对不同天气类型的适应性有待提升
生成模型	MoCoGAN	采用运动-内容解耦，增强降水等天气场景的时空一致性	生成结果受随机性影响，存在模式崩溃风险
	PreDiff	结合扩散模型进行天气预测，生成效果稳定，适用于不确定性建模	计算量大，训练和推理开销较高
	DiffCast	结合全局UNet，增强降水预测的细节保留能力	对高分辨率天气场景的泛化能力有待提升
大模型	GraphCast	采用GNN进行全球天气预测，具备良好的泛化能力	对超长时间尺度的预测仍存一定误差
	FourCastNet	结合Transformer提高天气预测效率，计算速度快	需要高质量数据支持，缺乏物理约束
	Pangu-Weather	端到端建模全球天气，精度超过部分NWP方法	训练数据和计算资源需求极大

方法类别	代表模型	优势	劣势
循环方法	ConvLSTM	结合CNN和LSTM结构，捕捉空间和时间依赖关系，适用于短时天气预测。ConvLSTM在气象数据集（如 Radar Echo、ERA5、GFS、ClimateNet）上能有效捕捉降水、温度、风场等时空依赖关系，适用于短期天气预测	计算成本高，对长期趋势（如ERA5的长期气候模式）捕捉较弱，易受累积误差影响，在GFS这样的数值天气预报数据上可能不及传统物理模型精确
	MotionRNN	采用时间-空间变换机制，提升对运动信息的建模能力，适用于复杂天气场景	计算复杂度较高，对长时间序列预测仍有误差累积问题
	SwinLSTM	结合Swin Transformer进行局部和全局信息交互，提高时序预测精度	结构复杂，对计算资源需求较高，训练难度比较大
非循环方法	SimVP	采用ETD架构，实现端到端多步预测，计算效率高	主要依赖CNN，难以捕捉长时间依赖关系
	MIMO-VP	采用Transformer实现MIMO，提高长时序预测能力	计算量大，训练难度较大
	TAU	通过静态注意力和动态注意力优化特征提取，提升计算效率	对数据量依赖较高，对不同天气类型的适应性有待提升
生成模型	MoCoGAN	采用运动-内容解耦，增强降水等天气场景的时空一致性	生成结果受随机性影响，存在模式崩溃风险
	PreDiff	结合扩散模型进行天气预测，生成效果稳定，适用于不确定性建模	计算量大，训练和推理开销较高
	DiffCast	结合全局UNet，增强降水预测的细节保留能力	对高分辨率天气场景的泛化能力有待提升
大模型	GraphCast	采用GNN进行全球天气预测，具备良好的泛化能力	对超长时间尺度的预测仍存一定误差
	FourCastNet	结合Transformer提高天气预测效率，计算速度快	需要高质量数据支持，缺乏物理约束
	Pangu-Weather	端到端建模全球天气，精度超过部分NWP方法	训练数据和计算资源需求极大

模型	MSE	MAE	RMSE
ConvLSTM	1.521 4	0.794 9	1.233 5
PredRNN	1.331 1	0.724 6	1.153 7
MAU	1.251 4	0.703 6	1.118 7
Swin Transformer	1.142 9	0.673 5	1.069 1
HorNet	1.201 4	0.690 6	1.096 1
TAU	1.161 9	0.670 7	1.077 9

Review of deep learning applications in severe convective weather prediction

深度学习应用于强对流天气预测的综述

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 9

References 81

Related Articles 15

Recommended Articles

Metrics

模型	MSE	MAE	RMSE
ConvLSTM	1.897 6	0.921 5	1.377 5
PredRNN	1.881 0	0.906 8	1.371 5
MAU	1.900 1	0.919 4	1.378 4
Swin Transformer	1.499 6	0.814 5	1.224 6
HorNet	1.553 9	0.825 4	1.246 6
TAU	1.592 5	0.842 6	1.261 9

模型	MSE	MAE	RMSE
ConvLSTM	0.049 4	0.154 2	0.222 3
PredRNN	0.055 0	0.158 8	0.234 6
MAU	0.049 5	0.151 6	0.222 6
Swin Transformer	0.046 4	0.147 3	0.215 4
HorNet	0.046 9	0.147 5	0.216 6
TAU	0.047 2	0.146 0	0.217 3

[1]	Songsen YU, Huang HE, Guopeng XUE, Hengtuo CUI. Quantitation and grading method for ceramic tile chromatic aberration based on improved fractal encoding network [J]. Journal of Computer Applications, 2026, 46(3): 959-968.
[2]	Peirong SHAO, Suzhen LIN, Yanbo WANG. Human-centric detail-enhanced virtual try-on method [J]. Journal of Computer Applications, 2026, 46(3): 915-923.
[3]	Yan HU, Peng LI, Shuyan CHENG. Adversarial purification method based on directly guided diffusion model [J]. Journal of Computer Applications, 2026, 46(3): 821-829.
[4]	Yongwei JIANG, Xiaoqing CHEN, Linjie FU. Elastic medical image registration model with high-frequency preservation based on spectrum decomposition [J]. Journal of Computer Applications, 2026, 46(3): 924-932.
[5]	Jian ZHANG, Jianbo YU, Jian TANG. Municipal solid waste incineration state recognition method based on multilayer preprocessing [J]. Journal of Computer Applications, 2026, 46(3): 940-949.
[6]	Xiaolei CHEN, Zhiwei ZHENG, Xue HUANG, Zhenbin QU. Panoramic video super-resolution network combining spherical alignment and adaptive geometric correction [J]. Journal of Computer Applications, 2026, 46(2): 528-535.
[7]	Quanjie LIU, Zhaoyi GU, Chunyuan WANG. Unsafe driving behavior detection under complex lighting conditions [J]. Journal of Computer Applications, 2026, 46(2): 613-619.
[8]	Jinjiao LIN, Canshun ZHANG, Shuya CHEN, Tianxin WANG, Jian LIAN, Yonghui XU. Vehicle insurance fraud detection method based on improved graph attention network [J]. Journal of Computer Applications, 2026, 46(2): 437-444.
[9]	Ming LI, Mengqi WANG, Aili ZHANG, Hua REN, Yuqiang DOU. Image steganography method based on conditional generative adversarial networks and hybrid attention mechanism [J]. Journal of Computer Applications, 2026, 46(2): 475-484.
[10]	Jie CAO, Lingfeng XIE, Bingjin WANG, Changhe ZHANG, Zidong YU, Chao DENG. Adolescent scoliosis screening method considering class imbalance and background diversity [J]. Journal of Computer Applications, 2026, 46(2): 630-639.
[11]	Zeyi GUO, Fenglian LI, Lichun XU. Double decision mechanism-based deep symbolic regression algorithm [J]. Journal of Computer Applications, 2026, 46(2): 406-415.
[12]	Kejian CUI, Zhiming WANG, Zhaowen QIU. Method for retinal vessel segmentation and coronary artery disease prediction using optical coherence tomography angiography [J]. Journal of Computer Applications, 2026, 46(2): 640-651.
[13]	Haoqian JIANG, Dong ZHANG, Guanyu LI, Heng CHEN. SetaCRS： Conversational recommender system with structure-enhanced hierarchical task-oriented prompting strategy [J]. Journal of Computer Applications, 2026, 46(2): 368-377.
[14]	Jincheng FU, Shiyou YANG. Short-term wind power prediction using hybrid model based on Bayesian optimization and feature fusion [J]. Journal of Computer Applications, 2026, 46(2): 652-658.
[15]	Feng HAN, Yongfeng BU, Haoxiang LIANG, Shuwen HUANG, Zhaoyang ZHANG, Shijie SUN. Vehicle trajectory anomaly detection based on multi-level spatio-temporal interaction dependency [J]. Journal of Computer Applications, 2026, 46(2): 604-612.