Review on machine learning methods for convective cell identification and tracking using weather radar

V.A. Shapovalov, A.A. Adzhieva, M.M. Akhmatov, A.Zh. Khitieva

---

Abstract: Automatic detection and tracking of convective cells using radar data is a crucial task for the nowcasting of severe weather events. Traditional algorithms, such as threshold-based and object-oriented methods, are widely employed but suffer from limitations in accuracy.
Aim. To investigate and compare the performance of various machine learning models in detecting and tracking convective cells in radar imagery.
Results. A theoretical review of state-of-the-art approaches was conducted, covering classical algorithms (TITAN, SCIT), computer vision methods (threshold segmentation, clustering), and machine learning techniques, including fuzzy logic, decision trees, and neural networks (specifically deep convolutional networks). The performance characteristics of established machine learning models were evaluated based on quality metrics. The results demonstrate that such models can increase the probability of detection and reduce false alarms compared to threshold-based methods.
Conclusions. AI-based algorithms outperform traditional approaches across several metrics, enabling more reliable identification of dangerous convective cells and forecasting of their evolution. The practical application of these methods will improve the accuracy of thunderstorm and hail nowcasting; however, their implementation requires large, properly prepared training datasets that account for specific local conditions.

Keywords: weather radar, convective cells, detection, segmentation, tracking, optical flow, machine learning, deep learning, neural networks, nowcasting, severe weather events

For citation. Shapovalov V.A., Adzhieva A.A., Akhmatov M.M., Khitieva A.Zh. Review on machine learning methods for convective cell identification and tracking using weather radar. News of the Kabardino-Balkarian Scientific Center of RAS. Vol. 28. No. 1. Pp. 90–101. DOI: 10.35330/1991-6639-2026-28-1-90-101

©   Shapovalov V.A., Adzhieva A.A., Akhmatov M.M., Khitieva A.Zh., 2026

References

1. WMO-No. 1198. Guidelines for Nowcasting Techniques. Geneva, 2017. 184 p.
2. Abshaev M.T., Abshaev A.M., Malkarova A.M., Zharashuev M.V. Automated radar identification, measurement of parameters and classification of convective cells for hail protection and storm warning purposes. Meteorologiya i gidrologiya [Meteorology and Hydrology]. 2010. No. 3. Pp. 36–45. (In Russian)
3. Saraykin A.A., Chaykovskiy V.M. Application of a neural network in meteorological radar. Izvestiya vysshikh uchebnykh zavedeniy. Povolzhskiy region. Tekhnicheskiye nauki [University News. Volga Region. Technical Sciences]. 2024. No. 1(69). Pp. 83–90. (In Russian)
4. Shapovalov V.A., Tumgoyeva Kh.A. Recognition and tracking of cloud convective cells for nowcasting of hazardous weather phenomena. Izvestiya YuFU. Tekhnicheskiye nauki [Izvestiya SFedU. Engineering Sciences]. 2018. No. 3. Pp. 58–67. DOI: 10.23683/2311-3103-2018-3-58-67. (In Russian)
5. Bournas A., Baltas E. Development of a Storm-Tracking Algorithm for the Analysis of Radar Rainfall Patterns in Athens, Greece. Water. 2024. No. 16(20). P. 2905. DOI: 10.3390/w16202905
6. Lagerquist R., McGovern A., Gagne D. J. Deep Learning for Spatially Explicit Prediction of Synoptic-Scale Fronts. Weather and Forecasting. 2019. Vol. 34. No. 4. Pp. 1137–1160. DOI: 10.1175/WAF-D-18-0183.1
7. Dixon M., Wiener G. TITAN: Thunderstorm identification, tracking, analysis, and nowcasting – Radar-based methodology. Journal of Atmospheric and Oceanic Technology. 1993. Vol. 10. No. 6. Pp. 785–797. DOI: 10.1175/1520-0426(1993)010<0785:TTITAA>2.0.CO;2
8. Johnson J.T., MacKeen P.L., Witt A. et al. The storm cell identification and tracking algorithm: an enhanced WSR-88D Algorithm. Weather and Forecasting. 1998. Vol. 13. No. 2. Pp. 263–276. DOI: 10.1175/1520-0434(1998)013<0263: TSCIAT>2.0.CO;2
9. Pulkkinen S., Nerini D., Pérez Hortal A.A. et al. Pysteps: an open-source Python library for probabilistic precipitation nowcasting, including an ensemble of extrapolation, autoregressive and more advanced methods. Geoscientific Model Development. 2019. Vol. 12. Pp. 4185–4219. DOI: 10.5194/gmd-12-4185-2019
10. Jung S.-H., Lee G. Radar-based cell tracking with fuzzy logic approach. Meteorological Applications. 2015. Vol. 22. No. 4. Pp. 716–730. DOI: 10.1002/met.1509
11. Jergensen G.E., McGovern A., Lagerquist R., Smith T. Classifying convective storms using machine learning. Weather and Forecasting. 2020. Vol. 35, No. 2. Pp. 537–556.
12. Veillette M.S., Kurdzo J.M., Stepanian P.M. et al. A benchmark dataset for tornado detection and prediction using full-resolution polarimetric weather radar data (TorNet). Artificial Intelligence for the Earth Systems. 2024. Vol. 4. No. 1. DOI: 10.1175/AIES-D-24-0006.1
13. Nan G., Chen M., Qin R. et al. Identification, tracking and classification method of mesoscale convective system based on radar composite reflectivity mosaic and deep learning. Acta Meteorologica Sinica. 2021. Vol. 79(6). Pp. 1002–1021. DOI: 10.11676/qxxb2021.062
14. Ayzel G., Scheffer T., Heistermann M. RainNet v1.0: a convolutional neural network for radar-based precipitation nowcasting. Geoscientific Model Development. 2020. Vol. 13. No. 6. Pp. 2631–2644. DOI: 10.5194/gmd-13-2631-2020
15. Han L., Sun J., Zhang W. convolutional neural network for convective storm nowcasting using 3-d doppler weather radar data. IEEE Transactions on Geoscience and Remote Sensing. Vol. 58. No. 2. Pp. 1487–1495. DOI: 10.1109/TGRS.2019.2948070
16. Albu A.-I., Czibula G., Mihai A. et al. NeXtNow: A convolutional deep learning model for the prediction of weather radar data for nowcasting purposes. Remote Sens. 2022. No. 14. P. 3890. DOI: 10.3390/rs14163890
17. Ritvanen J., Pulkkinen S., Moisseev D., Nerini D. Cell-tracking-based framework for assessing nowcasting model skill in reproducing growth and decay of convective rainfall. Geoscientific Model Development. 2025. Vol. 18. Pp. 1851–1878. DOI: 10.5194/gmd-18-1851-2025
18. Ranganayakulu S.V., Subrahmanyam K.V., Niranjan A. A novel algorithm for convective cell identification and tracking based on optical character recognition neural network. Journal of Electromagnetic Waves and Applications. 2021. Vol. 35. No. 16. Pp. 2239–2255. DOI: 10.1080/09205071.2021.1941299
19. Liu J., Zhang Q. A novel algorithm for detecting convective cells based on h-maxima transformation using satellite images. Atmosphere. 2025. Vol. 16. No. 11. P. 1232. DOI: 10.3390/atmos16111232
20. Wang X., Liao R., Li J. et al. Thunderstorm identification algorithm research based on simulated airborne weather radar reflectivity data. EURASIP Journal on Wireless Communications and Networking. 2020. ID: 37. DOI: 10.1186/s13638-020-1651-6
21. Wang Y., Long M., Wang J. et al. PredRNN: Recurrent neural networks for predictive learning using spatiotemporal LSTMs. Advances in Neural Information Processing Systems. Vol. 30. Pp. 1–10.
22. Shi X., Chen Z., Wang H. et al. Convolutional LSTM Network: A machine learning approach for precipitation nowcasting. Advances in Neural Information Processing Systems (NIPS). 2015. Vol. 28. Pp. 802–810.
23. Shi X., Gao Z., Lausen L. et al. Deep Learning for precipitation nowcasting: a benchmark and a new model. Advances in Neural Information Processing Systems (NIPS). 2017. Vol. 30. Pp. 5617–5627.
24. Gao Z., Shi X., Wang H., et al. Earthformer: exploring space-time transformers for earth system forecasting. Advances in Neural Information Processing Systems (NeurIPS). 2022. Vol. 35. Pp. 25390–25403. arXiv:2207.05833
25. Hu J., Rosenfeld D., Zrnić D. et al. Tracking and characterization of convective cells through their maturation into stratiform storm elements using polarimetric radar and lightning detection. Atmospheric Research. 2019. Vol. 226. Pp. 192–207.
26. Xiao H., Wang Y., Zheng Y. et al. Convective-gust nowcasting by a deep learning algorithm. Geoscientific Model Development. 2023. Vol. 16. Pp. 3611–3628. DOI: 10.5194/gmd-16-3611-2023
27. Ravuri S., Lenc K., Willson M. et al. Skilful precipitation nowcasting using deep generative models of radar. Nature. 2021. Vol. 597. Pp. 672–677. DOI: 10.1038/s41586-021-03854-z

Information about the authors

Vitaliy A. Shapovalov, Doctor of Physics and Mathematics, Senior Researcher, High-Mountain Geophysical Institute;
2, Lenin avenue, Nalchik, 360004, Russia;
vet555_83@ mail.ru, ORCID: https://orcid.org/0000-0002-9701-6820, SPIN-code: 6938-9800
Aida A. Adzhieva, Doctor of Physics and Mathematics, Professor, Department of Higher Mathematics and Informatics, Kabardino-Balkarian State Agricultural University named after V.M. Kokov;
1v, Lenin avenue, Nalchik, 360030, Russia;
aida-adzhieva@mail.ru, ORCID: https://orcid.org/0000-0002-1047-8417, SPIN-code: 4128-9520
Mukhadin M. Akhmatov, Candidate of Physical and Mathematical Sciences, Associate Professor, Associate Professor of the Department of Higher Mathematics, Kabardino-Balkarian State Agricultural University named after V.M. Kokov;
1v, Lenin avenue, Nalchik, 360030, Russia;
m_ahmatov@mail.ru, ORCID: https://orcid.org/0009-0003-2941-7459, SPIN-code: 3620-2852
Aminat Zh. Khitieva, Candidate of Economic Sciences, Associate Professor, Associate Professor of the Department of Higher Mathematics and Informatics, Kabardino-Balkarian State Agricultural University named after V.M. Kokov;
1v, Lenin avenue, Nalchik, 360030, Russia;
aminkahitieva@mail.ru, ORCID: https://orcid.org/0009-0004-6847-6328, SPIN-code: 8085-7236

Funding

The study was performed without external funding.