Utility of Graph Neural Networks in Short-to Medium-Range Weather Forecasting
Source abstract
Abstract
Weather forecasting is crucial for agriculture, transportation, and industry. Deep Learning (DL) has greatly improved the prediction accuracy. Among them, Graph Neural Networks (GNNs) excel at processing weather data by establishing connections between regions. This allows them to understand complex patterns that traditional methods might miss. As a result, achieving more accurate predictions becomes possible. The paper reviews the role of GNNs in short-to medium-range weather forecasting. The methods are classified into three categories based on dataset differences. The paper also further identifies five promising research frontiers. These areas aim to boost forecasting precision and enhance computational efficiency. They offer valuable insights for future weather forecasting systems.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Energy Load and Power Forecasting · Hydrological Forecasting Using AI · Computational Physics and Python Applications
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (100+)
Show source-provided references
1. Machine learning-based extreme event attribution10.1126/sciadv.adl3242 ↗
2. Predicting uncertainty in forecasts of weather and climate10.1088/0034-4885/63/2/201 ↗
3. Hybrid physics-AI outperforms numerical weather prediction for extreme precipitation nowcasting10.1038/s41612-024-00834-8 ↗
4. A novel deep learning approach for typhoon-induced storm surge modeling through efficient emulation of wind and pressure fields10.1038/s41598-023-35093-9 ↗
5. Deep learning for twelve hour precipitation forecasts10.1038/s41467-022-32483-x ↗
6. Big data analytics in weather forecasting: a systematic review10.1007/s11831-021-09616-4 ↗
7. The quiet revolution of numerical weather prediction10.1038/nature14956 ↗
8. PredRNN: a recurrent neural network for spatiotemporal predictive learning10.1109/TPAMI.2022.3165153 ↗
9. Artificial intelligence revolutionises weather forecast, climate monitoring and decadal prediction10.3390/rs13163209 ↗
10. Scaling transformer neural networks for skillful and reliable medium-range weather forecasting
11. Fourcastnet: accelerating global high-resolution weather forecasting using adaptive fourier neural operators
12. Spherical fourier neural operators: learning stable dynamics on the sphere
13. Gao · Proceedings of the 36th Annual Conference on Neural Information Processing Systems (NeurIPS ‘24) · 2024 · 36
14. Vision GNN: an image is worth graph of nodes
15. Learning skillful medium-range global weather forecasting10.1126/science.adi2336 ↗
16. WeatherGNN: exploiting meteo-and spatial-dependencies for local numerical weather prediction bias-correction
17. PN-HGNN: precipitation nowcasting network via hypergraph neural networks10.1109/TGRS.2024.3513294 ↗
18. Conditional local convolution for spatio-temporal meteorological forecasting
19. ClimaX: a 25 foundation model for weather and climate
20. Global-local feature learning via dynamic spatial-temporal graph neural network in meteorological prediction10.1109/TKDE.2024.3397840 ↗
21. Heterogeneous graph neural network
22. Forecasting precision: the role of graph neural networks and dynamic GNNs in weather prediction
23. Yin · IEEE Trans Inf Forensics Secur · 2022 · 18 · 20610.1109/TIFS.2022.3219342 ↗
24. Yasunaga · North American Chapter of the Association for Computational Linguistics (NAACL) · 2021
25. Graph neural networks for intelligent transportation systems: a survey10.1109/TITS.2023.3257759 ↗
26. The graph neural network model10.1109/TNN.2008.2005605 ↗
27. Graph convolutional networks: a comprehensive review10.1186/s40649-019-0069-y ↗
28. Prediction of solar radiation based on spatial and temporal embeddings for solar generation forecast10.24251/HICSS.2020.363 ↗
29. node2vec: scalable feature learning for networks
30. Sparse graph attention networks10.1109/TKDE.2021.3072345 ↗
31. European Centre for Medium-Range Weather Forecasts (ECMWF) [Internet]. [cited 2025 Apr 28]. Available from: https://www.ecmwf.int/.
32. Global Modeling and Assimilation Office (GMAO). NASA. Modern-era retrospective analysis for research and applications, Version 2 (MERRA-2) [Internet]. [cited 2025 Apr 28]. Available from: https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/.
33. Japan Meteorological Agency (JMA). JRA-55: Japanese 55-year reanalysis [Internet]. [cited 2025 Apr 28]. Available from: https://jra.kishou.go.jp/JRA-55/index_en.html.
34. National Centers for Environmental Information (NCEI). Integrated Surface Database (ISD) [Internet]. [cited 2025 Apr 28]. Available from: https://www.ncei.noaa.gov/products/land-based-station/integrated-surface-database.
35. Deep learning for precipitation nowcasting: a benchmark and a new model
36. Sevir: a storm event imagery dataset for deep learning applications in radar and satellite meteorology
37. National Centers for Environmental Information (NCEI). Next Generation Weather Radar (NEXRAD) [Internet]. [cited 2025 Apr 28]. Available from: https://www.ncei.noaa.gov/products/radar/next-generation-weather-radar.
38. Climate model configurations of the ECMWF Integrated Forecasting System (ECMWF-IFS cycle 43r1) for HighResMIP10.5194/gmd-11-3681-2018 ↗
39. Lang S, Alexe M, Chantry M, Dramsch J, Pinault F, Raoult B, et al. AIFS-ECMWF’s data-driven forecasting system. arXiv:2406.01465. 2024.
40. Coupling a new version of the common land model (CoLM) to the global/regional assimilation and prediction system (GRAPES): implementation, experiment, and preliminary evaluation10.3390/land11060770 ↗
41. Daytime along-valley winds in the Himalayas as simulated by the Weather Research and Forecasting (WRF) model10.5194/acp-23-821-2023 ↗
42. Representing low-intensity fire sensible heat output in a mesoscale atmospheric model with a canopy submodel: a case study with ARPS-CANOPY (version 5.2.12)10.5194/gmd-15-1713-2022 ↗
43. European centre for medium-range weather forecasts (ECMWF), “Confluence,” ECMWF [Internet]. [cited 2025 Apr 28]. Available from: https://confluence.ecmwf.int.
44. National Centers for Environmental Information (NCEI). International Best Track Archive, NCEI [Internet]. [cited 2025 Apr 28]. Available from: https://www.ncei.noaa.gov/products/international-best-track-archive.
45. Copernicus Climate Data Store (CDS). Copernicus climate data store, copernicus [Internet]. [cited 2025 Apr 28]. Available from: https://cds.climate.copernicus.eu/.
46. NOAA Physical Sciences Laboratory. Gridded Data, NOAA physical sciences laboratory [Internet]. [cited 2025 Apr 28]. Available from: https://psl.noaa.gov/data/gridded/index.html.
47. The Japanese 55-year reanalysis “JRA-55”: an interim report10.2151/sola.2011-038 ↗
48. Global Modeling and Assimilation Office (GMAO). MERRA-2 Data Access, NASA GMAO [Internet]. [cited 2025 Apr 28]. Available from: https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/data.
49. Keisler R. Forecasting global weather with graph neural networks. arXiv:2202.07575. 2022.
50. A structured graph neural network for improving the numerical weather prediction of rainfall10.1029/2023JD039011 ↗
51. Increasing trends in regional heatwaves10.1038/s41467-020-16970-7 ↗
52. Jeon HJ, Kang JH, Kwon IH, Lee OJ. CloudNine: analyzing meteorological observation impact on weather prediction using explainable graph neural networks. arXiv:2402.14861. 2024.
53. Feik M, Lerch S, Stühmer J. Graph neural networks and spatial information learning for post-processing ensemble weather forecasts. arXiv:2407.11050. 2024.
54. The EUPPBench postprocessing benchmark dataset v1.010.5194/essd-15-2635-2023 ↗
55. Forecasting PM2. 5 concentration in India using a cluster based hybrid graph neural network approach10.1007/s13143-022-00291-4 ↗
56. Observation impact explanation in atmospheric state estimation using hierarchical message-passing graph neural networks10.1088/2632-2153/ad8981 ↗
57. Alexe M, Boucher E, Lean P, Pinnington E, Laloyaux P, McNally A, et al. GraphDOP: towards skilful data-driven medium-range weather forecasts learnt and initialised directly from observations. arXiv:2412.15687. 2024.
58. National Centers for Environmental Information (NCEI). Global historical climatology network daily, NCEI [Internet]. [cited 2025 Apr 28]. Available from: https://www.ncei.noaa.gov/products/land-based-station/global-historical-climatology-network-daily.
59. An overview of nowcasting development, applications, and services in the Hong Kong Observatory10.1007/s13351-014-4048-9 ↗
60. Deep neural network based feature representation for weather forecasting
61. Eastern wind integration and transmission study
62. A graph neural network with spatio-temporal attention for multi-sources time series data: an application to frost forecast10.3390/s22041486 ↗
63. Regional heatwave prediction using graph neural network and weather station data10.1029/2023GL103405 ↗
64. ST-GNNs for weather prediction in South Africa
65. PM2.5-GNN: a domain knowledge enhanced graph neural network for PM2.5 forecasting
66. Spatiotemporal adaptive attention graph convolution network for city-level air quality prediction10.1038/s41598-023-39286-0 ↗
67. Promoting wind energy for sustainable development by precise wind speed prediction based on graph neural networks10.1016/j.renene.2022.09.036 ↗
68. Multistream graph attention networks for wind speed forecasting
69. Spatio-temporal graph deep neural network for short-term wind speed forecasting10.1109/TSTE.2018.2844102 ↗
70. HiSTGNN: hierarchical spatio-temporal graph neural network for weather forecasting10.1016/j.ins.2023.119580 ↗
71. A low rank weighted graph convolutional approach to weather prediction
72. National Centers for Environmental Information (NCEI). Integrated global radiosonde archive, NCEI [Internet]. [cited 2025 Apr 28]. Available from: https://www.ncei.noaa.gov/products/weather-balloon/integrated-global-radiosonde-archive.
73. National Oceanic and Atmospheric Administration (NOAA). Global surface summary of the day (GSOD), NOAA [Internet]. [cited 2025 Apr 28]. Available from: https://www.ncei.noaa.gov/access/metadata/landing-page/bin/iso?id=gov.noaa.ncdc:C00516.
74. A novel physics-aware graph network using high-order numerical methods in weather forecasting model10.1016/j.knosys.2024.112158 ↗
75. DGFormer: a physics-guided station level weather forecasting model with dynamic spatial-temporal graph neural network10.1007/s10707-024-00511-1 ↗
76. WeatherBench: a benchmark data set for data-driven weather forecasting10.1029/2020MS002203 ↗
77. GE-STDGN: a novel spatio-temporal weather prediction model based on graph evolution10.1007/s10489-021-02824-2 ↗
78. An adaptive adjacency matrix-based graph convolutional recurrent network for air quality prediction10.1038/s41598-024-55060-2 ↗
79. Innovating graph representation for dynamic weather forecasting10.22457/apam.v29n2a04938 ↗
80. Multi-information aggregation and estrangement HyperGraph convolutional networks for spatio-temporal weather forecasting
81. Group-aware graph neural network for nationwide city air quality forecasting
82. Joint air quality and weather prediction based on multi-adversarial spatiotemporal networks
83. Bién Data. KDD 2018 competition data, Bién data [Internet]. [cited 2025 Apr 28]. Available from: https://www.biendata.xyz/competition/kdd_2018/.
84. China National Environmental Monitoring Center (CNEMC). CNEMC-China National Environmental Monitoring Center, CNEMC [Internet]. [cited 2025 Apr 28]. Available from: http://www.cnemc.cn/en/.
85. A deep learning-based methodology for precipitation nowcasting with radar10.1029/2019EA000812 ↗
86. CNGAT: a graph neural network model for radar quantitative precipitation estimation
87. Spatio-temporal modeling of climate change impacts on drought forecast using Generative Adversarial Network: a case study in Africa10.1016/j.eswa.2023.122211 ↗
88. Probabilistic weather forecasting with machine learning10.1038/s41586-024-08252-9 ↗
89. Accurate medium-range global weather forecasting with 3D neural networks10.1038/s41586-023-06185-3 ↗
90. Physics-guided deep learning for skillful wind-wave modeling10.1126/sciadv.adr3559 ↗
91. Neural general circulation models for weather and climate10.1038/s41586-024-07744-y ↗
92. Multi-scale temporal features extraction based graph convolutional network with attention for multivariate time series prediction10.1016/j.eswa.2022.117011 ↗
93. Graphlime: local interpretable model explanations for graph neural networks10.1109/TKDE.2022.3187455 ↗
94. Multi-modal spatio-temporal meteorological forecasting with deep neural network10.1016/j.isprsjprs.2022.03.007 ↗
95. Multimodal learning with graphs10.1038/s42256-023-00624-6 ↗
96. A survey on heterogeneous graph embedding: methods, techniques, applications and sources
97. Knowledge graph embedding: a survey of approaches and applications10.1109/TKDE.2017.2754499 ↗
98. A comprehensive survey on graph neural networks10.1109/TNNLS.2020.2978386 ↗
99. Knowledge graphs: opportunities and challenges10.1007/s10462-023-10465-9 ↗
100. Know, know where, KnowWhereGraph: a densely connected, cross-domain knowledge graph and geo-enrichment service stack for applications in environmental intelligence